1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes in sbcl-1.0.27 relative to 1.0.26:
3 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
4 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
5 * improvement: the system either recovers from stack exhaustion or dies
6 properly as opposed to leaving the user uncertain of whether the handler
7 trampled on some random memory next to the stack or having to rely on
8 --lose-on-corruption (which is still a good idea to use in production
9 because stack exhaustion can happen in signal handlers which will likely
11 * bug fix: a type error is signaled for attempts to use the LOOP
12 keyword ACROSS for a NIL value. (thanks to Daniel Lowe)
13 * bug fix: fix gc related interrupt handling bug on ppc (regression from
14 1.0.25.37, reported by Harald Hanche-Olsen)
15 * bug fix: work around signal delivery bug in darwin (regression from
16 1.0.25.44, reported by Sidney Markowitz)
18 changes in sbcl-1.0.26 relative to 1.0.25:
19 * incompatible change: an interruption (be it a function passed to
20 INTERRUPT-THREAD or a timer function) runs in an environment where
21 interrupts can be enabled. The interruption can use
22 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
23 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
24 potentially running out of stack. Keep in mind that in the absence
25 of WITHOUT-INTERRUPTS some potentially blocking operation such as
26 acquiring a lock can enable interrupts.
27 * incompatible change: GC-OFF and GC-ON are removed, as they were
28 always unsafe. Use WITHOUT-GCING instead.
29 * new feature: runtime option --disable-ldb
30 * new feature: runtime option --lose-on-corruption to die at the
31 slightest hint of possibly non-recoverable errors: running out of
32 memory, stack, alien stack, binding stack, encountering a memory
33 fault, etc. In the absence of --lose-on-corruption a warning is
35 * enhancement: detect binding stack exhaustion
36 * enhancement: detect alien stack exhaustion on x86/x86-64
37 * improvement: generally more stable and reliable interrupt handling
38 * improvement: there is a per thread interruption queue,
39 interruptions are executed in order of arrival
40 * improvement: a repeating timer reschedules itself when the it has
41 finished, but expiration times are spaced equally. If an
42 expiration time is in the past it will trigger after a short grace
43 period that may give a chance to other things to run.
44 * optimization: slightly faster gc on multithreaded builds
45 * optimization: faster WITHOUT-GCING
46 * bug fix: when JOIN-THREAD signals an error, do it when not holding
47 important locks so that the debugger/handler doesn't produce
48 recursive errors or deadlock.
49 * bug fix: real-time signals are not used anymore, so no more
50 hanging when the system wide real-time signal queue gets full.
51 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
53 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
54 * bug fix: fix random memory faults related to interrupts on alpha
55 * bug fix: fix random memory faults related to interrupts on sparc
56 * bug fix: fix deadlocks related to starting threads
57 * bug fix: fix deadlines on locks on futex platforms
58 * bug fix: restore errno in signal handlers
59 * bug fix: fix deadlocks related to hash tables
60 * bug fix: fix deadlocks in pcl
62 changes in sbcl-1.0.25 relative to 1.0.24:
63 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
64 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
65 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
66 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
67 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
68 better name and does not return values so stale on multiprocessor systems.
69 Also, HOLDING-MUTEX-P was added for about the only sane usage of
71 * improvement: unithread builds keep track of MUTEX-VALUE.
72 * improvement: reading from a TWO-WAY-STREAM does not touch the output
73 stream anymore making it thread safe to have a concurrent reader and
74 a writer, for instance, in a pipe.
75 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
76 arguments, making compiler-macros for SETF-functions able to inspect
77 their constant arguments.
78 * improvement: COMPILE-FILE reports times with millisecond accuracy
79 (thanks to Luis Oliveira)
80 * optimization: CHAR-CODE type derivation has been improved, making
81 TYPEP elimination on subtypes of CHARACTER work better. (reported
82 by Tobias Rittweiler, patch by Paul Khuong)
83 * bug fix: setting alien structure fields of type struct by value now
84 computes the right offset for the memory copy.
85 * bug fix: compilation problem involving inlined calls to aliens with
86 result type VOID. (reported by Ken Olum)
87 * bug fix: #235a; sequential inline expasion in different policies no
88 longer reuses the functional from the previous expansion site.
89 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
90 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
91 * bug fix: named ENUMs in multiply-referenced alien record types no longer
92 cause an implied type redefinition. Regression from 1.0.21.29.
93 * bug fix: sign-extension of small signed return values now works with C
94 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
95 * improvements to the Windows port:
96 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
97 Westfold, thanks to Rudi Schlatte)
99 changes in sbcl-1.0.24 relative to 1.0.23:
100 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
101 vector of a multidimensional SIMPLE-ARRAY.
102 * new feature: the system now signals a continuable error if standard
103 readtable modification is attempted. (thanks to Tobias Rittweiler)
104 * new feature: DIRECTORY has been extended with a non-standard keyword
105 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
106 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
107 has been robustified and documented better. (thanks to Robert Goldman)
108 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
109 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
110 * tradeoff: constant FORMAT control strings are now compiled unless
111 SPACE > SPEED (previously only when SPEED > SPACE.)
112 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
114 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
116 * bug fix: lexical type declarations are now correctly reported by
117 SB-CLTL2. (reported by Larry D'Anna)
118 * bug fix: STRING-TO-OCTETS did not handle :START properly when
119 using UTF-8 as external format. (thanks to Luis Oliveira)
120 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
121 are signalled correctly. (thanks to Stas Boukarev)
122 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
123 designator. (thanks to Tobias Rittweiler)
124 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
125 designator, and returns T instead of the function. (thanks to
127 * bug fix: direct superclasses of STANDARD-CLASS and
128 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
129 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
130 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
131 when no class with the named slot yet exists no longer causes a
132 compile-time style-warning.
133 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
134 in safe code. (reported by Didier Verna)
135 * bug fix: #430; nested structure constructors can stack allocate.
136 * bug fix: on some 64-bit platforms dynamic space size was truncated
137 to #xffff0000 bytes. (reported by Benjamin Lambert)
138 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
139 REPL unusable. (reported by Daniel Herring)
140 * bug fix: return values of READ-SEQUENCE did not take :START into
141 account on file streams, regressions since 1.0.12.22. (reported by
142 Thomas Russ, patch by Paul Khuong)
143 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
144 method specializer used to confuse permuation vector optimization.
145 * bug fix: system inserted bogus implicit type declarations for local
146 special variables in DEFMETHOD bodies.
147 * bug fix: #354; duplicated frames in backtraces due to
148 non-tail-call-optimized XEPs to functions with return type NIL
150 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
151 initializes structure object slots according to DEFSTRUCT initforms,
152 and DEFSTRUCT forms :INCLUDEind structure classes defined using
153 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
154 (reported by Bruno Haible and Stephen Wilson)
155 * bug fix: #395; fill-pointer output streams used now support
156 element-type BASE-CHAR as well.
157 * bug fix: compiler error when attempting to derive return value of
158 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
160 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
162 * bug fix: Mac OS X binaries should now be portable between Leopard
165 changes in sbcl-1.0.23 relative to 1.0.22:
166 * enhancement: when disassembling method functions, disassembly
167 for the associated fast function is also produced.
168 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
170 * optimization: printing with *PRINT-PRETTY* true is now more
171 efficient as long as the object being printed doesn't require
172 special handling by the pretty printer.
173 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
174 now interact correctly with type declarations.
175 * partial bug fix: PCL detects infinite recursion during wrapper
176 validation. (thanks to Attila Lendvai)
177 * bug fix: #426; nested function calls are inlined properly.
178 Previously if FOO was an inline function, in calls of the form
179 (FOO (FOO ...)) the outer call was not inlined.
180 * bug fix: long long arguments passed in registers to alien
181 functions on PowerPC Linux and NetBSD are now handled correctly
182 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
184 changes in sbcl-1.0.22 relative to 1.0.21:
185 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
186 for the shared object in the current directory, but passes the native
187 namestring of the designated pathname to the operation system's shared
188 object loading function as-it.
189 * minor incompatible change: --disable-debugger toplevel option now takes
190 effect before processing of initialization files and --eval or --load
192 * new feature: new commandline argument: --script, which supports
193 shebang lines. See documentation for details. (based on work by
195 * new feature: SAVE-LISP-AND-DIE can save current values of
196 --dynamic-space-size and --control-stack-size in the executable core,
197 causing it to skip normal runtime option processing. See documentation
198 for details. (thanks to Zach Beane)
199 * enhancement: inoccous calls to EVAL or generic functions dispatching
200 on subclasses of eg. STREAM no longer cause compiler notes to appear.
201 * enhancement: the system no longer resignals errors from --load and
202 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
203 associated with the original error to be lost. (thanks to Ariel
205 * enhancement: :DONT-SAVE keyword argument has been added to
206 LOAD-SHARED-OBJECT for controlling interaction with
207 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
208 the effects of an earlier LOAD-SHARED-OBJECT call.
209 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
210 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
211 provided. (reported by Cedric St-Jean)
212 * bug fix: circularity handling in the reader did not treat raw
213 structure slots correctly. (reported by Cedric St-Jean)
214 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
215 file descriptors when there were none.
216 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
217 was broken. (reporteed by Willem Broekema)
218 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
219 pathnames without a directory.
220 * bug fix: COMPILE-FILE did not warn about undefined variable
221 references at toplevel, and LOAD of the resulting fasl did
223 * bug fix: functions with non-required arguments used to end up with
224 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
225 * bug fix: redefining a function with non-required arguments didn't
226 update the system's knowledge about its call signature properly.
227 * bug fix: fixed #431; incompatible alien record type redefinitions
228 are detected and handled. (reported by Neil Haven)
229 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
230 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
232 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
233 after alien stack frames.
234 * bug fix: strings passed to mk{s,d}temp lacked null termination
236 changes in sbcl-1.0.21 relative to 1.0.20:
237 * new feature: the compiler is able to track the effective type of a
238 generic function across method addition and removal.
239 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
240 appropriately typed structure slots without locking.
241 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
242 information like those gathered by TIME using a programming-friendly
244 * new feature: TIME reports time taken even if the form performs a
245 non-local transfer of control.
246 * enhancement: reduced conservativism on GENCGC platforms: on
247 average 45% less pages pinned (measured from SBCL self build).
248 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
249 mutate constant symbols or violate declaimed type of the symbol.
250 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
251 sleeping on the semaphore in heavily contested semaphores.
252 * bug fix: semaphores and condition variables were not interrupt
254 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
255 owned by other threads anymore.
256 * bug fix: FIND on lists called KEY outside the specified
257 subsequence. (reported by budden)
258 * bug fix: LOG doesn't use single-float intermediate results when
259 given mixed integer and double-float arguments, leading to better
260 precision. (reported by Bob Felts)
261 * bug fix: LOG with base zero returned values of inconsistent type.
263 changes in sbcl-1.0.20 relative to 1.0.19:
264 * minor incompatible change: OPTIMIZE qualities
265 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
266 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
267 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
268 * documentation: some slot access efficiency guidelines have been
269 added to the user manual.
270 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
271 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
272 as ASSOC and MEMEBER.
273 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
274 to more efficient EQ-comparison versions more often.
275 * optimization: enhanced derivation of DOLIST iteration variable type
277 * optimization: constant folding of simple (LIST ...) forms as DOLIST
279 * optimization: runtime lookup of function definitions can be
280 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
281 * optimization: compiler is able to derive the return type of
282 (AREF (THE STRING X) Y) as being CHARACTER.
283 * optimization: CLRHASH on empty hash-tables no longer does pointless
284 work. (thanks to Alec Berryman)
285 * optimization: REPLACE deftransforms don't punt when :START1 or
287 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
288 no longer emits a code deletion note. (reported by Andrew
290 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
291 functions with an odd number of &OPTIONAL arguments, a &REST
292 argument, and one or more &KEY arguments at the call site.
293 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
294 arguments appeared at call sites as well.
295 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
296 READ-LINE &co can no longer read from them afterwards. (reported
298 * bug fix: fixed #427: unused local aliens no longer cause compiler
299 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
301 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
302 alien-stack leakage. (reported by Andy Hefner)
303 * bug fix: PROGV signals an error when an attempt to violate declared
304 type of a variable or bind a constant is made.
305 * bug fix: SET signals an error when an attempt to violate declared
306 type of a variable is made.
307 * bug fix: restart computation during the execution of a restart
308 test function no longer causes infinite recursion. (reported by
310 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
311 signals an error. (thanks to Pierre Mai)
312 * bug fix: EXPORT left symbol unexported in conflict situations.
313 (thanks to Michael Weber)
314 * bug fix: correctly handle name conflicts involving CL:NIL.
315 (thanks to Michael Weber)
316 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
317 conflicts arising from USEing package with conflicting symbols
318 correctly. (thanks to Michael Weber)
320 changes in sbcl-1.0.19 relative to 1.0.18:
321 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
322 warnings that go otherwise unhandled will be muffled if they are
323 of the type that's the value of this variable.
324 * optimization: stack allocation is slightly more efficient on x86
326 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
327 where a raw slot always is initialized using the initform whose
328 type is not know sufficiently well a compile-time are now compiled
329 correctly. (reported by John Morrison)
330 * bug fix: compiler no longer makes erronous assumptions in the
331 presence of non-foldable SATISFIES types.
332 * bug fix: stack analysis missed cleanups of dynamic-extent
333 arguments in non-let-converted calls to local functions.
334 * improvements to the Windows port:
335 ** adjusted address spaces for building on both Win32 and
336 Win64. (thanks for John Connors)
337 * fixed some bugs revealed by Paul Dietz' test suite:
338 ** interval arithmetic during type derivation used inexact integer
339 to single-float coercions.
340 ** arithmetic operations involving large integers and single
341 floats give the same results in compiled and interpreted code.
342 ** deriving the result type of COERCE no longer signals an error
343 if the derived type of the second argument is a MEMBER type
344 containing invalid type specifiers.
345 ** ADJOIN with constant NIL as second argument works correctly.
346 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
348 changes in sbcl-1.0.18 relative to 1.0.17:
349 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
350 profiles only the current thread.
351 * minor incompatible change: changes to SYMBOL-VALUE of constants
352 defined with DEFCONSTANT now signal an error.
353 * enhancement: SB-SPROF now has support for wallclock profiling,
354 and is also able to profile specific threads. REPORT output
355 has also additional sorting options.
356 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
358 * optimization: structure allocation has been improved
359 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
360 ** out of line constructors are ~10% faster.
361 ** inline constructors are ~15% faster.
362 ** inline constructors are capable of dynamic extent allocation
363 (generally on x86 and x86-64, in some cases on other platforms
365 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
367 * optimization: file compiler is now able to coalesce non-circular
368 lists, non-base strings, and bit-vectors. Additionally, constants
369 are never referenced through SYMBOL-VALUE at runtime.
370 * optimization: code defining methods on PRINT-OBJECT (and other
371 generic functions in the COMMON-LISP package) now loads faster.
372 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
374 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
375 fasl is now deleted. (reported by Attila Lendvai)
376 * bug fix: READ-LINE always returned NIL for the last line in files.
377 (reported by Yoshinori Tahara)
378 * bug fix: more accurate disassembly annotations of foreign function
379 calls. (thanks to Andy Hefner)
380 * bug fix: trimming non-simple strings and non-string string
381 designators when the there is nothing to trim works properly.
382 (thanks to James Knight)
383 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
384 (thanks to Travis Cross)
385 * fixed some bugs revealed by Paul Dietz' test suite:
386 ** NIL is a valid function name (regression at 1.0.13.38)
387 ** FILL on lists was missing its return value (regression at 1.0.12.27)
388 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
389 fill pointers properly (regression at 1.0.12.23)
390 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
391 displacement indices properly (regression at 1.0.12.23)
393 changes in sbcl-1.0.17 relative to 1.0.16:
394 * temporary regression: user code can no longer allocate closure
395 variable storage on stack, due to bug 419 without explicitly
396 requesting it. Please consult sbcl-devel for advice if you need to
397 use this feature in the meanwhile.
398 * new feature: runtime argument --control-stack-size can be used to
399 adjust thread default control stack size.
400 * enhancement: improved TIME output
401 ** all times are reported using the measured accuracy (milliseconds
402 for real and GC times, microseconds for everything else.)
403 ** processor cycle counts on x86 and x86-64.
404 ** interpreted forms are counted for both evaluator modes.
405 ** number of lambdas converted by the compiler is reported.
406 ** CPU percentage report (computed from real and total run time.)
407 ** more comprehensive run time reporting, using a condenced format
408 ** interperted form, lambda, and page fault counts are omitted
410 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
412 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
413 * optimization: two argument forms of LAST are upto ~10% faster
414 in normal SPEED policies.
415 * optimization: NCONC no longer needs to heap cons its &REST list
416 in normal SPEED policies.
417 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
419 * bug fix: UNION and NUNION work with :TEST-NOT once more,
420 regression since 1.0.9.1. (thanks to Eric Marsden)
421 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
422 since 1.0.15.36. (thanks to Paul Khuong)
423 * bug fix: LAST when always returned the whole list when given a bignum
424 as the second argument.
425 * bug fix: dynamic extent allocation of nested lists and vectors
426 could leak to otherwise accessible parts.
427 * bug fix: invalid optimization of heap-allocated alien variable
429 * bug fix: fasl header checking is less vulnerable to different
430 platform word lengths.
431 * bug fix: more correct assembler syntax for GNU binutils
432 2.18.50.0.4 support. (thanks to Marijn Schouten)
433 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
435 * bug fix: the fopcompiler can handle LOCALLY forms (with no
436 declarations) successfully. (reported by Attila Lendvai)
438 changes in sbcl-1.0.16 relative to 1.0.15:
439 * minor incompatible change: revert the changes to sb-posix's error
440 signaling added in 1.0.14.
441 * minor incompatible change: change PROBE-FILE back to returning
442 NIL whenever we can't get a truename, as was the case before 1.0.14.
443 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
444 inherits from ERROR instead of just CONDITION.
445 * new feature: SB-INTROSPECT can provide source locations for instances
446 as well. (thanks to Tobias Rittweiler)
447 * optimization: binding special variables now generates smaller code
448 on threaded platforms.
449 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
450 and cases where no :TEST is given but the compiler can infer that
451 the element to search is of type (OR FIXNUM (NOT NUMBER)).
452 * optimization: better LOGNOT on fixnums.
453 * optimization: modular arithmetic for a particular requested width
454 is implemented using a tagged representation unless a better
455 representation is available.
456 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
457 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
458 for wellformedness and returns a specialized vector. (reported by
459 Francois-Rene Rideau)
460 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
461 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
462 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
463 unscheduling timer at the same time as another timer fires.
464 * bug fix: don't reschedule timers for dead threads.
465 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
466 * bug fix: copying output from RUN-PROGRAM to a stream signalled
467 bogus errors if select() was interrupted.
468 * enhancement: add support for fcntl's struct flock to SB-POSIX.
470 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
471 * enhancement: cleaner backtraces for interactive interrupts, as
472 well as other cases where the interesting frames used to be
473 obscured by interrupt handling frames.
474 * enhancement: untracing a whole package using (UNTRACE "FOO") is
475 now supported, and tracing a whole package using (TRACE "FOO") now
476 traces SETF-functions as well.
477 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
478 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
479 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
480 * unreadably printed representation of hash-tables now includes
482 * bug fix: partially fixed #188: type propagation from assignments
483 is now more efficient.
484 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
485 'DOUBLE-FLOAT) are not flushable.
486 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
487 full address of the object, and none of the tag bits.
488 * bug fix: readably printing hash-tables now respects other printer
489 control variables. (reported by Cedric St-Jean)
490 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
491 keyword with MAKE-HASH-TABLE.
492 * bug fix: export SB-POSIX:MKSTEMP.
493 * bug fix: SORT was not interrupt safe.
494 * bug fix: XREF accounts for the last node of each basic-block as
496 * bug fix: MAKE-INSTANCE optimizations interacted badly with
497 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
498 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
499 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
500 create obsolete instances in the case of redefinition or
501 obsoletion of a superclass. (thanks to Andy Hefner)
502 * bug fix: Support for the Alpha architecture has been revived; it had
503 suffered somewhat from lack of maintenance since sbcl-1.0.
504 * improvements to the Windows port:
505 ** The system detects the codepage to use at startup, instead of
506 using the value from the saved core. (thanks to Kei Suzuki)
508 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
509 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
510 (see documentation for details.)
511 * revived support for OpenBSD (contributed by Josh Elsasser)
512 * partially fixed bug #108: ROOM no longer suffers from occasional
513 (AVER (SAP= CURRENT END)) failures.
514 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
515 bodies no longer cause a WARNING to be signalled. (reported by
517 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
518 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
519 the bug in TRUNCATE.) (reported by Michael Weber)
520 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
521 no samples. (reported by Andy Hefner)
522 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
523 no longer appear as (NIL ...) frames in backtraces.
524 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
525 is now actually exported from SB-EXT as documented. (reported by
527 * bug fix: sb-aclrepl now correctly understands how to inspect
528 single-floats on 64-bit platforms where single-floats are not boxed.
529 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
530 yet been finalized. (reported by Levente Meszaros)
531 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
532 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
533 SBCL-specific optimize qualities.
535 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
536 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
537 an executable in the search path, and does so in the child
538 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
539 been removed; it can be found in the CVS history, for somebody who
540 needs that search behavior (see the manual).
541 * minor incompatible change: compiler policy re. weakening type
542 checks has changed: now type checks are weakened only if SAFETY < 2
544 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
545 unparsing of directory pathnames as files. Analogously,
546 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
547 filename to parse into a directory pathname.
548 * enhancement: implicit generic function creation now signals a
549 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
550 users can bind handlers for and muffle around calls to LOAD.
551 * enhancement: RUN-PROGRAM allows unicode arguments and environments
552 to be used (using the default stream external format), and allows
553 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
554 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
555 strings and vectors whose element-type or simplicity is not fully
556 known at compile-time.
557 * optimization: STRING-TRIM and related functions no longer allocate
558 a new string when no trimming needs to be performed. These functions
559 are also faster than before when the input string has been declared
561 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
562 * optimization: READ-LINE is significantly faster for files containing
564 * optimization: non-open coded uses of character comparison operators
565 (e.g. char=) no longer cons when called with more than one parameter
566 on platforms supporting dynamic-extent allocation.
567 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
568 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
569 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
571 * bug fix: some sequence functions elided bounds checking when
573 * bug fix: too liberal weakening of union-type checks when SPEED >
575 * bug fix: more bogus fixnum declarations in ROOM implementation
578 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
579 * new feature: MAKE-HASH-TABLE now experimentally accepts a
580 :SYNCHRONIZED argument, which makes the hash-table safe for
581 concurrent accesses (but not iteration.) See also:
582 SB-EXT:WITH-LOCKED-HASH-TABLE, and
583 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
584 * optimization: CONCATENATE on strings is an order of magnitude faster
585 in code compiled with (> SPEED SPACE).
586 * optimization: SUBSEQ is ~50% faster on lists.
587 * bug fix: bug 417 fixed -- source location reporting is now more robust.
588 * bug fix: SUBSEQ on a list will now correctly signal an error if if
589 END is smaller then START.
590 * bug fix: SB-PROFILE will no longer report extra consing for nested
591 calls to profiled functions.
592 * bug fix: ROOM implementation had bogus fixnum declarations which
593 could cause type-errors when calling ROOM on large images.
594 * bug fix: if file compilation is aborted, the partial fasl is now
595 deleted, and COMPILE-FILE returns NIL as the primary value.
596 * bug fix: number of thread safety issues relating to SBCL's internal
597 hash-table usage have been fixed.
598 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
599 be returned from its body when the values were being returned
600 using unknown-values return convection and the W-P-O was wrapped
601 inside an UNWIND-PROTECT.
602 * bug fix: sb-posix should now compile again under Windows, enabling
605 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
606 * incompatible change: hash-table accessor functions are no longer
607 automatically protected by locks. Concurrent accesses on the same hash-table
608 from multiple threads can give inconsistent results or even corrupt the
609 hash-table completely. Multi-threaded applications should do their own
610 locking at the correct granularity. In the current implementation it is
611 still safe to have multiple readers access the same table, but it's not
612 guaranteed that this property will be maintained in future releases.
613 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
614 and will signal an error at runtime.
615 * enhancement: SB-THREAD package now exports a semaphore interface.
616 * enhancement: CONS can now stack-allocate on x86 and
617 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
619 * enhancement: nested lists can now be stack allocated on
620 platforms providing stack allocation support.
621 * enhancement: dynamic-extent support has been extended to support
622 cases where there are multiple possible sources for the stack
624 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
625 if the mutex is uncontested on Linux.
626 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
627 as specified by the CLHS. (thanks to Tobias Rittweiler)
628 * bug fix: NaN comparison now works on x86-64.
629 * bug fix: CLOSE :ABORT T on a stream with pending output now
631 * bug fix: instances of non-standard metaclasses using standard
632 instance structure protocol sometimes missed the slot type checks
634 * bug fix: known functions can be inlined in high-debug code.
636 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
637 * minor incompatible change: the MSI installer on Windows no longer
638 associates .lisp and .fasl files with the installed SBCL.
639 * minor incompatible change: :UNIX is no longer present in *FEATURES*
640 on Windows. (thanks to Luis Oliviera)
641 * new feature: SB-CLTL2 contrib module now implements
642 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
643 * optimization: scavenging weak pointers is now more efficient,
644 requiring O(1) instead of O(N) per weak pointer to identify
645 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
646 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
647 method bodies using constant slot-names when the first argument is
648 a specializer parameter for the method.
649 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
650 names, when the class of the instance is a direct instance
651 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
652 as slow as the constant slot-name case.)
653 * optimization: member type construction is now O(N) instead
655 * optimization: UNION and NUNION are now O(N+M) for large
656 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
658 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
659 now more readable in environments like Slime which display it.
660 (thanks to Tobias C. Rittweiler)
661 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
662 was able to infer the correct argument type for the object on which
663 the CAS operation was being performed.
664 * bug fix: copy propagation interfered with parallel assignment
665 semantics in local calls. (reported by Paul Khuong)
666 * bug fix: the signed modular fixnum shift compiled to wrong code on
667 x86 and x86-64. (spotted by a slight modification to some of
670 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
671 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
672 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
673 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
675 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
676 and generic function now print the object identity for anonymous
677 instances. (thanks to Attila Lendvai)
678 * enhancement: as an extension to MOP, SBCL now supports SETF of
679 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
680 (thanks to Attila Lendvai)
681 * enhancement: sb-cover has an alternate annotation mode which puts
682 the form annotations at the car of the form, instead of annotating
683 it completely. The benefit of this mode is that it shows better
684 which forms were instrumented by the compiler.
685 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
686 by printer control variables. (reported by Dan Corkill)
687 * bug fix: system leaked memory when delayed output was performed by
688 the OS in smaller chunks then expected. (thanks to David Smith)
689 * bug fix: system leaked memory when file streams were not closed
691 * bug fix: large objects written to slow streams that were modified
692 after the write could end up with the modified state written to
693 the underlying file descriptor.
694 * bug fix: multiple threads operating in parallel on the same stream
695 could cause buffer-overflows.
696 * bug fix: source location information is stored correctly
697 (broken since 1.0.6). This bug would generally show up as the
698 Slime debugger highlighting the wrong form when the sldb-show-source
700 * bug fix: Branch forms are again annotated as branches in the sb-cover
702 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
703 on x86-64. (reported by Gregory Vanuxem)
704 * bug fix: EQUALP could return wrong results for structures with raw
705 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
706 word). (reported by Vjacheslav Fyodorov)
707 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
710 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
711 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
712 atomic compare-and-swap operations on threaded platforms.
713 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
714 allows assigning a global minimum value to optimization qualities
715 (overriding proclamations and declarations).
716 * enhancement: closed over variables can be stack-allocated on x86
718 * performance bug fix: GETHASH and (SETF GETHASH) are once again
720 * optimization: slot definition lookup is now O(1). This speeds up
721 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
722 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
723 * optimization: ASSOC and MEMBER can now be open-coded for all
724 combinations of keyword arguments when second argument is constant
725 and SPEED >= SPACE. In other cases a specialized version is
727 * bug fix: using obsoleted structure instances with TYPEP and
728 generic functions now signals a sensible error.
729 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
730 (reported by Kristoffer Kvello)
731 * bug fix: backtrace construction is now more careful when making
732 lisp-objects from pointers on the stack, to avoid creating bogus
733 objects that can be seen by the GC.
734 * bug fix: defaulting of values in contexts expecting more than 7
735 variables now works on x86-64. (reported by Christopher Laux)
736 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
738 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
739 as the property-list of a symbol.
740 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
741 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
742 situations requested, are once again file-compileable. (reported
745 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
746 * MOP improvement: support for user-defined subclasses of
747 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
748 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
749 the expansion of DEFMETHOD, is responsible for generating a form
750 which creates a list of specializers when evaluated. Additional
751 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
752 debugging and introspective support.
753 * minor incompatible change: the (unsupported) spinlock interface
754 has changed: free spinlock now has the value NIL, and a held spinlock
755 has the owning thread as its value.
756 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
757 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
759 * enhancement: name of a socket-stream is now "a socket" instead of
761 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
762 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
763 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
764 for generic functions. (Thanks to Tobias C. Rittweiler)
765 * optimization: bignum printing speed has been improved by 20-40%
766 (depending on the bignum size.)
767 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
769 * bug fix: the cache used by the CLOS to store precomputed effective
770 methods, slot offsets, and constant return values is now thread and
772 * bug fix: generic function dispatch function updating is now thread
773 and interrupt safe (in the sense that the known issues have been
775 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
776 * bug fix: interning EQL-specializers is now thread and interrupt safe.
777 * bug fix: asdf systems with dependencies to the SB-POSIX or
778 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
779 * bug fix: interrupt safety of applicable method computation has been
782 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
783 * new contrib: sb-cover, an experimental code coverage tool, is included
785 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
786 significantly faster.
787 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
788 produce more efficient code on x86-64 if the compiler can determine
789 that (+ INDEX <constant>) does not require a bounds check and FOO
790 has an element type at least 8 bits wide.
791 * enhancement: a new, experimental synchronous timeout facility is
792 provided. Refer to SB-SYS:WITH-DEADLINE for details.
793 * enhancement: when a symbol name conflict error arises, the
794 conflicting symbols are always printed with a package prefix.
795 (thanks to Kevin Reid)
796 * enhancement: stepping is now once again supported on the SPARC and
797 MIPS platforms. (It is also now more likely to work on CheneyGC
799 * enhancement: sb-sprof can now also track and report accurate call
801 * bug fixes: the treatment of non-standard subclasses of
802 SB-MOP:SPECIALIZER is more correct.
803 * incompatible change: PURIFY no longer copies the data from the
804 dynamic space into the static and read-only spaces on platforms
805 that use the generational garbage collector
806 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
808 * bug fix: GC race condition occasionally resulting in crashes with
809 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
811 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
813 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
814 a STYLE-WARNING to be signalled (regression from 1.0.4.)
815 * bug fix: an asynchronous interrupt could previously leave the
816 system running with GC inhibited.
817 * bug fix: a DECLARE form evaluated at top-level now causes an error
818 rather than silently (or verbosely) returning NIL.
819 * bug fix: trying to dissassemble functions compiled at high DEBUG could
820 result in a error being signalled due to source form lookup errors.
821 (reported by Peter Graves)
823 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
824 * incompatible change: removed writer methods for host-ent-name,
825 host-ent-addresses -- changing the values did not update the DNS
827 * minor incompatible change: changed experimental JOIN-THREAD interface
828 * documentation: the manual now lists reader and writer methods
829 in class slot documentation sections. (thanks to Richard M Kreuter)
830 * documentation: unwinding from asyncronous events has been
831 documented as unsafe.
832 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
833 in multithreaded application code.
834 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
835 platforms. (thanks to James Anderson for the optimization hint)
836 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
838 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
839 variants no longer cons.
840 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
841 their NOT- variants no longer cons.
842 * optimization: EQUAL hash tables no longer use SXHASH for objects
843 of all data types, but instead use an EQL hash for types for which
844 EQUAL is the same as EQL.
845 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
846 are significantly faster.
847 * optimization: new STRING-OUTPUT-STREAM implementation is much
848 faster for large outputs and conses 30% less on average.
849 * enhancement: XREF information is now collected for references made
850 to global variables using SYMBOL-VALUE and a constant argument.
851 * enhancement: SIGINT now causes a specific condition
852 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
853 * bug fix: FILE-POSITION used to signal an error in some cases where
854 ANSI requires it to return NIL.
855 * bug fix: ADJUST-ARRAY is now interrupt-safe.
856 * bug fix: adding and removing fd-handlers is now interrupt-safe.
857 * bug fix: inlined calls to C now ensure 16byte stack alignment on
859 * bug fix: bad type declaration in the CLOS implementation has
860 been fixed. (reported by James Anderson)
861 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
862 (thanks to Lutz Euler)
863 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
864 required by the CLHS. (reported by Samium Gromoff)
865 * bug fix: dead unbound variable references now signal an error.
866 (reported by Marco Monteiro)
867 * bug fix: / with an unused value was being deleted in safe code.
868 (thanks to Marco Monteiro and Kevin Reid)
869 * bug fix: number of characters that can be written onto a single
870 line in a file is unlimited.
871 * bug fix: some GC deadlocks caused by asynchronous interrupts have
872 been fixed by inhibiting interrupts for when GC is disabled.
873 * bug fix: some interrupt safety issues with GETHASH, (SETF
874 GETHASH), CLRHASH and REMHASH have been fixed.
875 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
876 specifier no longer causes infinite recursion.
877 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
878 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
879 * bug fix: an &environment argument with the correct variable information
880 is passed to macros that are expanded during byte compilation.
881 (reported by Samium Gromoff)
882 * bug fix: a logic error was causing the hash function for bignums to
883 have excessive amounts of collisions. (reported by Faré Rideau)
884 * bug fix: modifying the contents of an array could change the return
885 value of SXHASH on that array, which is only allowed for strings
886 and bit vectors. (bug introduced in 0.9.16)
887 * improvement: the x86-64/darwin port now passes all tests (except
888 for the debugger tests) but should still be considered
889 experimental until this is fixed.
890 * improvement: a style-warning is signaled for CASE (etc) clauses with
891 duplicate keys. (thanks to Kevin Reid)
892 * improvement: macos/x86 and macos/x86-64 now use mach exception
893 handlers for dealing with illegal instructions (for trapping and
894 error handling) and memory protection violations (for GC).
896 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
897 * new platform: experimental support for x86-64/darwin (MacOS).
898 * incompatible change: the thread-safe (on most platforms) getaddrinfo
899 and getnameinfo sockets functions are used instead of gethostbyaddr
900 and gethostbyname, on platforms where the newer functions are available.
901 As a result, the ALIASES field of HOST-ENT will always be NIL on these
903 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
904 don't print the startup banner, but behave as if --noinform was passed
905 as a command line argument. (thanks to Kevin Reid)
906 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
907 * optimization: code using alien values with undeclared types is much faster.
908 * optimization: the compiler is now able to open code SEARCH in more cases.
909 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
910 * bug fix: using standardized COMMON-LISP special variables as loop
911 variables no longer signals bogus package lock violations. (reported
913 * bug fix: declaring local loop variables to be of a range-limited type
914 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
915 (reported by Andras Simon)
916 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
917 bugs remain on x86-64.)
918 * bug fix: the #= and ## reader macros now interact reasonably with
919 funcallable instances.
920 * bug fix: type-checks for function arguments were compiled using the
921 compiler policy settings of the wrong lexical environment (in 1.0.2
923 * bug fix: SHADOW accepts characters as string designators, as required
924 by the spec (thanks to Eric Marsden)
925 * bug fix: fixed GC safety issues when foreign functions are called with
926 non-base strings as arguments
927 * bug fix: more consistent error output for fatal-compiler-warnings, like
929 * bug fix: functions with proper names were showing up as NIL in
930 backtraces (reported by Edi Weitz, regression in 1.0.2)
932 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
933 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
934 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
935 produce more efficient code on the x86 if the compiler can determine
936 that (+ INDEX <constant>) does not require a bounds check and FOO
937 has an element type at least 8 bits wide.
938 * bug fix: references to undefined variables are handled the same way
939 in toplevel forms as in the normal compiler
940 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
942 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
943 (thanks to Jon Buller)
944 * improvement: faster compilation times for complex functions
945 * improvement: added readlink support to SB-POSIX (thanks to Richard
948 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
949 * improvement: experimental support for mach exception handling on
950 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
952 * improvement: support for GBK external format.
953 (thanks to Chun Tian (binghe))
954 * improvement: the debugger now displays variables that have been closed
955 over, in code compiled with (DEBUG 2) or higher
956 * improvement: support for executable cores on NetBSD (thanks to
958 * new feature: added a RESTART-FRAME debugger command
959 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
960 be used to provide an implementation for FILE-POSITION on Gray streams
961 (thanks to Eric Marsden)
962 * new feature: add syslog, openlog and closelog support to SB-POSIX
963 (thanks to Richard Kreuter)
964 * optimization: the function call overhead in code compiled with
965 a high DEBUG optimization setting is significantly
966 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
967 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
968 (thanks to Tony Martinez)
969 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
970 evaluated once (reported by Kevin Reid)
971 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
972 works again (thanks to Kilian Sprotte)
973 * bug fix: an error is signaled for tagbodies with duplicate tags
974 (thanks to Stephen Wilson)
975 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
976 * bug fix: Win32 port can now handle foreign code unwinding Lisp
977 stack frames from alien callbacks.
978 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
979 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
980 65k pages (thanks to David Woodhouse)
981 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
983 changes in sbcl-1.0.1 relative to sbcl-1.0:
984 * new platform: FreeBSD/x86-64, including support for threading.
985 * new feature: the compiler stores cross-referencing information
986 about function calls (who-calls), macroexpansion (who-macroexpands)
987 and special variables (who-binds, who-sets, who-references) for code
988 compiled with (< SPACE 3). This information is available through the
989 sb-introspect contrib.
990 * new feature: users may subclass SEQUENCE, and have instances of
991 these classes interoperate with standard Common Lisp functions if
992 a number of methods are defined. (This feature is experimental
993 and the interface subject to change based on feedback from SBCL
994 users and the general community)
995 * improvement: sb-sprof traces call stacks to an arbitrary depth on
996 x86/x86-64, rather than the previous fixed depth of 8
997 * improvement: another pthread back-end of mutex "pthread-futex".
998 * bug fix: non-ascii command-line arguments are processed correctly
999 (thanks to Yaroslav Kavenchuk)
1000 * bug fix: non-required arguments were not passed correctly when a method
1001 defined using DEFMETHOD was called from a mop-generated method using
1002 CALL-NEXT-METHOD (reported by Pascal Costanza)
1003 * bug fix: recursion is now permitted in accessors through
1004 SLOT-UNBOUND. (reported by Pascal Costanza)
1005 * bug fix: an error was signaled at startup if the HOME environment
1006 variable was defined, but had an empty value (reported by Peter Van Eynde)
1007 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1008 signal a PROGRAM-ERROR, not a BUG.
1009 * bug fix: SB-POSIX:READDIR works when built with large file support.
1010 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1012 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1013 for better automatic indentation support. (thanks to Matt Pillsbury)
1014 * optimization: loading generic functions no longer takes O(n^2) time,
1015 proportional to the amount of methods in the generic function
1016 (reported by Todd Sabin and Jeremy Brown)
1017 * optimization: the FIND and POSITION family of sequence functions
1018 are significantly faster on arrays whose element types have been
1020 * improvements to the Windows port:
1021 ** Intermittent heap corruption problems have been fixed. (thanks
1022 to Alastair Bridgewater)
1023 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1025 ** Lisp is able to unwind foreign exception frames from alien
1026 callbacks. (thanks to Alastair Bridgewater)
1028 changes in sbcl-1.0 relative to sbcl-0.9.18:
1029 * improvement: experimental support for threading on FreeBSD/x86.
1030 (thanks to NIIMI Satoshi)
1031 * improvement: runtime option --dynamic-space-size can be used
1032 to set the size of the dynamic space reserved on startup.
1033 * improvement: floating point modes in effect are now saved in
1034 core, and restored on startup.
1035 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1036 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1037 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1038 threads are not running after *SAVE-HOOKS* have run.
1039 * improvement: writes to CLOS instance slots are type-checked in code
1040 compiled with (SAFETY 3)
1041 * improvement: floating-point exception handling on FreeBSD (thanks to
1043 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1044 (thanks to Zach Beane)
1045 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1047 * improvement: added support for the Shift-JIS external format.
1048 (contributed by NIIMI Satoshi)
1049 * improvement: callbacks are supported on Linux/PPC. (thanks to
1051 * bug fix: compiler bug triggered by a (non-standard) VALUES
1052 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1053 * bug fix: file compiler no longer confuses validated and already
1054 dumped structurres. (reported by Kaersten Poeck)
1055 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1056 fill-pointer signals a type-error as required. (thanks to
1058 * bug fix: disassemly of funcallable instances works.
1059 * bug fix: single stepping on PPC.
1060 * bug fix: fix thread-safety problems in the type system (generally
1061 manifesting as nonsensical errors like "STRING is a bad type specifier
1062 for sequences" or "The value 1 is not of type FIXNUM")
1063 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1064 (reported by Josip Gracin)
1065 * bug fix: an error is signaled for attempts to displace arrays with
1066 incompatible element types (thanks to Mario Mommer)
1067 * bug fix: more correct handling of wide characters in debug info
1068 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1069 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1070 and don't cause extra consing
1071 * optimization: MAP and MAP-INTO are significantly faster on vectors
1072 whose elements types have been declared.
1073 * Improvements to SB-SPROF:
1074 ** Support for allocation profiling
1075 ** Reduced profiling overhead, especially for long profiling runs
1076 * Improvements to the Windows port:
1077 ** floating point exceptions are now reported correctly.
1078 ** stack exhaustion detection works partially.
1079 ** more accurate GET-INTERNAL-REAL-TIME.
1080 ** better RUN-PROGRAM behaviour in case of errors in the child
1082 ** PROBE-FILE now simplifies pathnames correctly.
1083 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1084 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1085 1st 1970 (although time timezone data will be unreliable).
1086 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1088 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1089 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1090 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1092 * bug fix: two potential GC deadlocks affecting threaded builds.
1093 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1094 returns T (reported by Anton Kazennikov)
1095 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1096 with non-variable places
1097 * bug fix: remove a race condition in the setting of
1098 funcallable-instance functions, this should make threaded CLOS
1099 code more stable against memory faults.
1100 * bug fix: corruption of specials when unbinding is interrupted by an
1101 asynchronous unwind (reported by Hannu Koivisto)
1102 * improvement: the debugger will now also display local variables that
1103 are only used once, for code compiled with a DEBUG optimization quality
1106 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1107 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1108 * incompatible change: External-format support for FFI calls. The
1109 SB-ALIEN:C-STRING no longer implies an ASCII
1110 external-format. Instead the string is subject to external-format
1111 conversion. Additionally return values of type SB-ALIEN:C-STRING
1112 are by default of type (SIMPLE-ARRAY CHARACTER), not
1113 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1114 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1115 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1116 * incompatible change: SB-EXT package no longer contains the
1117 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1118 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1120 * incompatible change: the single-stepper is no longer available
1121 on Alpha, Mips and Sparc platforms.
1122 * minor incompatible change: the direct superclasses of
1123 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1124 not (STANDARD-OBJECT FUNCTION). This makes the
1125 class-precedence-lists of GENERIC-FUNCTION and
1126 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1128 * minor incompatible change: the default stream external format on
1129 non-unicode SBCL is no longer detected from the locale, but is always
1131 * new feature: Add a version of evaluator that uses an interpreter instead
1132 of the compiler. EVAL still uses the compiler by default, to switch it
1133 to use the interpreter, set the value of the variable
1134 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1135 * minor incompatible change: the single-stepper REPL has been merged
1136 with the normal debugger (see the "Stepping" heading of the debugger help
1137 for more details). The debugger command STEP will no longer switch
1138 to the single-stepper REPL.
1139 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1140 for a type now works.
1141 * bug fix: loading of default sysinit file works. (thanks to Leonid
1143 * bug fix: better detection of circularities in the file-compiler.
1144 (reported by Marco Monteiro)
1145 * bug fix: the CL pathname functions now work with files that have
1146 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1147 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1148 systems with Unix98 pty semantics.
1149 * bug fix: ASDF-INSTALL will now work with bsd tar.
1150 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1152 * bug fix: timers expiring in dead threads no longer cause a
1153 type-error (reported by Paul "Nonny Mouse").
1154 * bug fix: thanks to more lightweight single-stepper instrumentation,
1155 code compiled with (DEBUG 3) will compile and execute significantly faster,
1156 and will have more accurate type-inferencing than before
1157 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1158 whose bindings are modified
1159 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1160 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1161 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1162 CL:SOFTWARE-VERSION with proper version
1164 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1165 * feature: implemented the READER-METHOD-CLASS and
1166 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1167 as specified by AMOP.
1168 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1170 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1171 (thanks to Lutz Euler)
1172 * optimization: hashing of general arrays and vectors has been
1173 improved. (reported by Any Fingerhut)
1174 * enhancement: SB-INTROSPECT is now able to find definitions of
1175 profiled functions. (thanks to Troels Henriksen)
1176 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1177 as well, allowing compiler-macros for SETF-functions to expand.
1178 * enhancement: step-instrumentation no longer wraps known
1179 single-value functions with multiple-value context, allowing
1180 better type inference.
1181 * fixed bug #337: use of MAKE-METHOD in method combination now works
1182 even in the presence of user-defined method classes. (reported by
1183 Bruno Haible and Pascal Costanza)
1184 * fixed bug #339(c): if there are applicable methods not part of any
1185 long-form method-combination group, call INVALID-METHOD-ERROR.
1186 (reported by Bruno Haible)
1187 * fixed bug #361: the :FUNCTION initarg in the protocol for
1188 initialization of methods can now be used to override
1189 internally-produced optimized functions. (reported by Bruno
1191 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1192 system-provided lambda expression no longer cause warnings about
1193 unbound #:|pv-table| symbols.
1194 * bug fix: improved the handling of type declarations and the
1195 detection of violations for keyword arguments with non-constant
1197 * bug fix: potentially erronous calls to PATHNAME and
1198 MERGE-PATHNAMES were being flushed in some cases.
1199 (reported by Richard Kreuter)
1200 * bug fix: compiled calls to TYPEP were mishandling obsolete
1201 instances. (reported by James Bielman and Attila Lendvai)
1202 * bug fix: format strings with ~> without matching ~< no longer
1203 trigger an AVER, but signal an understandable error instead.
1204 (reported by Antonio Martinez)
1205 * bug fix: specifying an output-file without a pathname-type for
1206 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1207 of input-file instead of "fasl". (reported by Robert Dodier)
1208 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1209 (reported by James Y Knight).
1210 * bug fix: compiler-macros-function did not consider the environment
1211 argument for shadowing by local functions.
1212 * bug fix: compiler-macros expansion was inhibited by local INLINE
1214 * bug fix: inline expansions of known functions were subject to
1215 step-instrumentation in high debug policies, leading to problems
1216 with type-inference.
1217 * bug fix: compiler failed to differentiate between different CONS
1218 types in some cases.
1219 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1220 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1221 * bug fix: #368: incorrect use of expressed vs. upgraded array
1223 * bug fix: #306a: more precise unions of array types.
1224 * thread-safety improvements:
1225 ** CONDITION-WAIT could return early on Linux, if the thread was
1226 interrupted and subsequently continued with SIGCONT.
1227 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1229 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1230 * added support for the ucs-2 external format. (contributed by Ivan
1232 * minor incompatible change: pretty printing of objects of type
1233 (cons symbol) is, in the default pprint-dispatch-table, now
1234 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1236 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1237 called later in a class's lifetime, possibly as late as when the
1238 first instance of the class is created. Previously,
1239 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1240 class became finalizeable.
1241 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1242 for multibyte external-format streams. (thanks to "vbzoli")
1243 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1244 with :ALLOCATION :CLASS inherited from superclasses of the
1246 * fixed bug: anonymous classes can now be created using the :NAME
1247 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1248 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1249 * fixed bug: core-files saved with :EXECUTABLE T can again be
1250 executed when SBCL_HOME isn't set. (reported by James Knight)
1251 * fixed bug: toplevel LOCALLY forms with declarations could
1252 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1253 * fixed bug: printing from several different threads using different
1254 values of *print-case* could cause invalid output, due to
1255 some internal special variables of the printer not being bound
1256 thread-locally (reported by Max Mikhanosha)
1257 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1258 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1259 (reported by Pascal Costanza)
1260 * fixed bug: FINALIZE-INHERITANCE is called from
1261 REINITIALIZE-INSTANCE on classes when the class has previously
1262 been finalized, as required by AMOP.
1263 * minor code generation optimizations:
1264 ** better register allocation in CLOS dispatching functions
1265 ** overflow detection when coercing signed bytes to fixnums on x86-64
1266 ** is now implemented with one IMUL instruction instead of three shifts
1267 ** more efficient bit-vector access on x86 and x86-64
1268 ** more efficient access to raw structure slots on x86 and x86-64
1269 * fixed some bugs revealed by Paul Dietz' test suite:
1270 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1271 return its argument.
1273 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1274 * feature: thread support on Solaris/x86, and experimental thread support
1276 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1278 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1279 default method for SLOT-UNBOUND.
1280 * minor incompatible change: prevent the user from specializing the
1281 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1282 somewhat counter to the intent of the protocol, I (CSR) think, and
1283 additionally it just doesn't work in SBCL as currently
1284 implemented, thanks to optimizations (that are always valid for
1285 the other three SLOT-VALUEish functions, but not for the setter).
1286 * minor incompatible change: the :SB-LDB feature is now enabled by
1287 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1288 the low-level debugger.
1289 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1290 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1291 down to an SB-SYS:FD-STREAM, allowing direct communication in
1292 more cases. Temporary files and pipes are still used as a fallback
1294 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1295 on the non-reentrant C functions localtime(3) and gmtime(3).
1296 * bug fix: global optimization policy was not visible in LOCALLY and
1298 * bug fix: class objects can be used as specializers in methods.
1299 (reported by Pascal Costanza)
1300 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1301 with a type error. (reported by blitz_ on #lisp)
1302 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1303 failed with a type error. (reported by Pascal Bourguignon)
1304 * bug fix: merging pathnames against defaults with :DIRECTORY
1305 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1306 (reported by James Y Knight)
1307 * bug fix: saving large (>2GB) cores on x86-64 now works
1308 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1309 constant index and a (simple-array (signed-byte 32)) array
1310 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1311 error (patch by Robert J. Macomber)
1312 * feature: TIME now displays the amount of run-time spent in GC
1313 * bug fix: The debugger now does a better job of respecting
1314 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1315 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1316 workaround for bug 403.)
1317 * bug fix: floating point exception handling now works on FreeBSD
1318 (thanks to NIIMI Satoshi)
1319 * fixed some bugs revealed by Paul Dietz' test suite:
1320 ** MISC.641: LET-conversion were not supposed to work in late
1322 * improvements to the Win32/x86 port:
1323 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1324 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1326 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1327 * new feature: source path information is generated for macro-expansion
1328 errors for use in IDE's like Slime (thanks to Helmut Eller)
1329 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1330 * bug fix: compiler does not loop forever on an invalid type in
1332 * improvement: compilation of most CLOS applications is significantly
1334 * optimization: added a limited bytecode compiler for simple toplevel
1335 forms, speeding up compilation and FASL loading
1336 * bug fix: the statistical profiler now properly distinguishes anonymous
1339 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1340 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1341 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1342 into subclasses of the finalized class. Now user calls to
1343 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1344 system return before any subclasses are finalized.
1345 * minor incompatible change: The reader no longer ignores errors
1346 regarding non-existent packages in #+ and #- feature tests.
1347 * new feature: command line options --no-sysinit, --no-userinit to
1348 inhibit loading the corresponding init files
1349 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1350 for sending data through UDP sockets (thanks to François-René Rideau)
1351 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1352 error is signalled instead (thanks to François-René Rideau)
1353 * minor incompatible change: Error signalling behaviour of lexical
1354 operations violating package locks has changed slightly. Refer to
1355 documentation on package locks for details.
1356 * enhancement: EVAL can process IF-expressions without resorting to the
1358 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1359 constant-folded properly due to missing out-of-line functions.
1360 (reported by Mika Pihlajamäki)
1361 * bug fix: LISTEN sometimes returned T even in cases where no data was
1362 immediately available from the stream
1363 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1364 were reversed. (reported by Levente Mészáros)
1365 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1366 when given a vector displaced to an adjustable array. (reported
1368 * bug fix: derivation of float boundaries from numbers outside the
1369 appropriate float range (reported by John Wiseman)
1370 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1372 * bug fix: precomputing cache entries for generic functions with
1373 some subclasses of specializers as yet invalid does not attempt to
1374 fill a cache line with a negative offset. (reported by Levente
1376 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1377 allow condition class objects as arguments to DOCUMENTATION and
1378 (SETF DOCUMENTATION); only find and set documentation for
1379 structure names for the STRUCTURE doc-type. (suggested by Gary
1381 * improvements to the Win32/x86 port:
1382 ** (user-homedir-pathname) and default initialization file
1383 locations now know about the user's "Documents and Settings"
1384 directory (thanks to Yaroslav Kavenchuk)
1385 ** run-program is implemented (thanks to Mike Thomas)
1386 ** sockets support (thanks to Timothy Ritchey)
1387 ** better backtrace support (thanks to Alastair Bridgewater)
1388 ** sb-grovel supported
1389 ** asdf-install and sb-posix work somewhat
1390 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1391 * improvements to the Solaris/x86 port:
1392 ** works on Solaris 11/Solaris Express
1393 ** floating-point exception handling support
1394 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1395 * fixed some bugs revealed by Paul Dietz' test suite:
1396 ** REMOVE-METHOD returns its generic function argument even when
1397 the method is not one of the generic functions' methods.
1398 ** objects declared as MEMBER types can be admissible arguments to
1399 structure accessors.
1400 ** printing characters should simply be printed by the FORMAT ~:C
1402 ** compiler failure when compiling functions with hairy constant
1403 defaults for optional parameters.
1404 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1405 function, which is already optimized.
1407 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1408 * new platform: experimental support for SBCL x86/Darwin, including
1409 MacOS X 10.4.5 on Intel.
1410 * new feature: Unicode character names are now known to the system
1411 (through CHAR-NAME and NAME-CHAR).
1412 * new feature: the filesystem location of SBCL's core file is
1413 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1414 * minor incompatible change: the contrib modules SB-POSIX and
1415 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1416 this change is to make it easier to distribute
1417 location-independent binaries.
1418 * bug fix: as implied by AMOP, standardized classes no longer have
1419 slots named by external symbols of public packages. (reported by
1421 * optimization: calling functions via constant symbols -- as in
1422 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1423 via the function object as in (FUNCALL #'FOO).
1424 * optimization: on x86 and x86-64, the calling convention for the
1425 case of dealing with an unknown number of values has been altered
1426 to be friendlier to the prediction heuristics implemented,
1427 particularly with reference to CALL and RET pairing. (thanks to
1428 Alastair Bridgewater)
1429 * optimization: on x86, the code for access to thread-local symbol
1430 values and binding/unbinding of thread-local symbols is smaller.
1431 (thanks to Alastair Bridgewater)
1432 * enhancement: CONSTANTP is now able to determine constantness of
1433 more complex forms, including calls to constant-foldable standardized
1434 functions and some special forms beyond QUOTE.
1435 * fixed bug: occasional GC crashes on Solaris/x86
1436 * optimization: x86-64 supports stack allocation of results of simple
1437 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1438 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1439 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1440 (thanks to James Knight)
1441 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1442 by the standard (thanks to Stephen Compall)
1444 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1445 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1446 be used for bundling the runtime and the core file into one
1447 executable binary. This feature is not currently supported on all SBCL
1448 platforms. (thanks to James Bielman and NIIMI Satoshi)
1449 * new feature: a generational or ephemeral garbage collector is now
1450 the default on the PowerPC platform (both Linux and Darwin). The
1451 old Cheney (stop and copy) collector is a build-time option.
1452 (thanks to Cyrus Harmon, after Raymond Toy)
1453 * minor incompatible change: the method by which SBCL finds its
1454 contributed modules has changed; it no longer relies on symbolic
1455 links from an $SBCL_HOME/systems directory, but searches directly
1456 in the subdirectories of $SBCL_HOME.
1457 * enhancement: the dynamic heap size on the Linux/PPC platform is
1458 markedly larger, even using the older Cheney garbage collector.
1459 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1460 certain complicated string types. (reported by Gary King)
1461 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1462 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1463 * fixed bug: DOCUMENTATION on structures no longer signals an
1464 error if no structure type of the right name exists. (reported by
1466 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1467 applicable non-standard primary or :AROUND method on
1468 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1469 no longer cause unbound variable errors. (reported by Kevin Reid)
1470 * fixed bug: metacircle resolution in cases where methods have slots
1471 added before the slots from STANDARD-METHOD. (reported by Jean
1473 * fixed bug: the Power PC assembly code for calling into lisp
1474 sometimes computed its return address wrongly (depending on the
1475 15th bit of the address, fixed at link time). This used to
1476 manifest itself in a segmentation violation while building PCL.
1477 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1478 many others over the years)
1479 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1480 the stack pointer (treated as a signed integer) is negative.
1481 (thanks to Peter van Eynde)
1483 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1484 * new platform: experimental support for the Windows operating
1485 system has been added. (thanks to Alastair Bridgewater)
1486 * fixed several bugs in and robustified the PPC FFI (including
1487 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1488 * bug fix: allow non-simple string symbol names (reported by Paul
1490 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1492 * bug fix: saving a core corrupted callbacks on x86/x86-64
1493 * bug fix: closed a loophole in metacircularity detection and
1494 grounding in the PCL implementation of CLOS.
1495 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1496 (reported by tomppa on #lisp)
1497 * bug fix: clear the direction flag on Lisp -> C transitions, as
1498 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1499 (reported by Andrej Grozin and Hendrik Maryns)
1500 * optimization: major improvements to GC efficiency on GENCGC platforms
1501 * optimization: faster implementation of EQUAL
1502 * optimization: emit more efficient opcodes for some common
1503 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1504 * optimization: several other minor code-generation improvements on x86-64
1505 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1507 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1508 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1509 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1510 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1511 by ANSI as a generic function, and for consistency (SETF
1512 GENERIC-FUNCTION-NAME) is treated likewise.
1513 * fixed bug #233.b: make constraint propagation notice when a variable
1514 value is changed after it is referenced but before it is used
1515 * fixed bug #296: no more arbitrary behaviour from filenames with
1516 odd characters as --load arguments. (reported by Adam Warner)
1517 * fixed bug #390: :CHARACTER-SET pathname components now work as
1518 expected. (reported by Tim Daly Jr)
1519 * fixed bug #391: complicated :TYPE intersections in slot
1520 definitions no longer cause an error in PCL internals.
1521 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1522 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1523 returning the number of octets which would be written to the
1524 file-stream. (thanks to Robert J. Macomber)
1525 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1526 arguments to the internal stream functions. (thanks to David
1528 * bug fix: stack exhaustion is now handled on the x86/SunOS
1529 platform. (thanks to Daisuke Homma)
1530 * bug fix: more accurate ROOM results on GENCGC platforms
1531 * optimization: improved type inference for arithmetic-for
1532 index variables in LOOP
1533 * optimization: faster floating-point SQRT on x86-64
1534 * fixed some bugs revealed by Paul Dietz' test suite:
1535 ** DOCUMENTATION returns NIL instead of "" for method combinations
1536 that don't have a docstring
1538 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1539 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1540 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1541 therefore are not customizeable by user code (as seems to be at
1542 least permitted and maybe required by AMOP). As a consolation,
1543 however, the SBCL implementation of these functions now calls
1544 REINITIALIZE-INSTANCE as specified by AMOP.
1545 * bug fix: slot-definition documentation is propagated as per ANSI
1546 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1547 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1548 argument, as do eslotd initialization methods. (from Pascal
1549 Costanza's "Closer" project)
1550 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1551 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1553 * bug fix: it is now possible to have more than one subclass of
1554 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1555 (reported by Bruno Haible, Pascal Costanza and others)
1556 * bug fix: the dependent update protocol now works for generic
1557 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1558 and Pascal Costanza)
1559 * bug fix: condition-class instances corresponding to
1560 DEFINE-CONDITION forms are now created eagerly. (reported by
1561 Kalle Olavi Niemitalo on comp.lang.lisp)
1562 * bug fix: floating point printing is more accurate in some
1563 circumstances. (thanks to Simon Alexander)
1564 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1565 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1566 * bug fix: callbacks on OS X now preserve stack-alignment.
1567 (thanks to Cyrus Harmon)
1568 * enhancement: the x86-64 disassembler is much better at
1569 disassembling SSE instructions. (thanks to Lutz Euler)
1570 * enhancement: CHECK-TYPE on a variable now allows type-inference
1571 (thanks to Vincent Arkesteijn)
1572 * optimization: improved performance of EUC-JP external format.
1573 (thanks to NIIMI Satoshi)
1574 * optimization: performance improvements to IO on file streams of
1575 :ELEMENT-TYPE CHARACTER
1576 * optimization: much faster memory allocation on x86-64
1577 * optimization: faster garbage collections (latency and throughput)
1579 * optimization: faster bignum subtraction and division on x86 and x86-64
1580 (thanks to Lutz Euler)
1581 * optimization: more accurate type-inference for loops that use a
1582 floating point index variable or a negative step.
1584 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1585 * bug fix: add a workaround to SBCL looping infinitely at startup on
1586 Linux kernels with apparently buggy implementations of personality().
1587 (thanks to Svein Ove Aas)
1588 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1589 (thanks to David Lichteblau)
1590 * bug fix: local bindings shadow global symbol macros for
1591 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1592 * optimization: non-open coded uses of numeric comparison operators
1593 (e.g. >) no longer cons when called with more than one parameter
1594 on platforms supporting dynamic-extent allocation.
1595 * enhancement: saving cores with foreign code loaded is now supported
1596 on MIPS/Linux in addition to the previously supported platforms.
1597 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1598 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1599 * bug fix: division by zero in sb-sprof when no samples were collected
1600 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1601 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1602 (thanks to Alexander Kjeldaas)
1603 * new feature: ignoring the return values of destructive functions
1604 like NREVERSE signals a compile-time style-warning.
1605 (thanks to Kevin Reid)
1606 * new feature: the alignment of alien structure fields can be
1607 explicitly specified. (thanks to Cyrus Harmon)
1608 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1609 and dump core on SIGQUIT
1611 ** incompatible change: threads do not inherit values of specials
1612 from their parents (see manual)
1613 ** bug fix: threads stacks belonging to dead threads are freed by the
1614 next exiting thread, no need to gc to collect thread stacks anymore
1615 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1616 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1617 ** bug fix: dynamic variable and thread start related gc lossage
1618 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1620 * fixed some bugs revealed by Paul Dietz' test suite:
1621 ** SUBTYPEP is slightly more accurate on heinously complicated
1622 CONS types where some of the members have uncertain (in the
1623 NIL, NIL sense) type relationships to each other.
1625 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1626 no longer purified unless :PURIFY T is explicitly specified.
1627 ** Non-purified cores are significantly smaller than before
1629 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1630 * new feature: timers based on Zach Beane's excellent timer package
1631 * added support for the following external formats: koi8-u,
1632 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1633 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1634 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1635 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1636 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1637 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1638 * incompatible change: a threaded SBCL will no longer revert to
1639 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1640 * bug fix: interrupts are disabled until startup is complete; no
1641 more sigsegvs when receiving a signal to soon
1642 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1644 * bug fix: add a workaround for the memory randomization features in
1645 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1646 workaround will only be in effect on systems with the proc filesystem
1648 * bug fix: printing objects of type HASH-TABLE signals a
1649 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1651 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1652 have been running for over 50 days. (reported by Gilbert Baumann)
1653 * bug fix: the logic for getting names of functions gets less
1654 confused when confronded with alternate-metaclass
1655 funcallable-instances. (reported by Cyrus Harmon)
1656 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1657 consistent, even on internal alternate-metaclass objects.
1658 * bug fix: flush closure information collected by physical
1659 environment analysis prepass before the main pass. (bug reported
1661 * bug fix: compiler pack phase does not modify a hashtable, which is
1662 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1663 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1664 may be expanded. (reported by Alexander <malishev> on c.l.l)
1665 * bug fix: no more style-warnings when compiling code using LAST
1666 under high-SPEED low-SPACE settings. (reported by David Wallin
1668 * fixed some incompatibilities between SBCL's MOP and the MOP
1670 ** the METAOBJECT class is now implemented;
1671 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1672 STANDARD-OBJECT, as required;
1673 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1674 now compatible (as required by VALIDATE-SUPERCLASS); there
1675 remains a consistency requirement over the presence or absence
1676 of the FUNCTION class in the superclass list of finalized
1677 classes; see the manual for more details;
1678 ** the :AROUND method for COMPUTE-SLOTS on
1679 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1680 requested slot ordering.
1682 ** bug fix: parent thread now can be gc'ed even with a live
1684 ** bug fix: binding a special with PROGV to no value is not
1685 the same as that symbol not having been bound (thanks to
1687 ** bug fix: binding specials is thread safe (thanks to
1689 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1690 and SB-SPROF do not die with 'no handler for signal XX in
1691 interrupt_handle_now(..)' anymore
1692 ** bug fix: WITH-TIMEOUT works with multiple running threads
1693 * fixed some bugs revealed by Paul Dietz' test suite:
1694 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1695 the :method-class keyword argument.
1697 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1698 * new port: the Solaris operating system on x86 processors is now
1699 mostly supported, though some rough edges in the environment
1700 remain. (thanks to Daisuke Homma)
1701 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1702 space for its heap. (thanks to Thiemo Seufer)
1703 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1704 closures. (thanks to Thiemo Seufer)
1705 * minor incompatible change: eof selects abort in the debugger.
1706 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1707 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1709 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1710 deprecated, and will go away in a future revision of SBCL.
1711 * minor incompatible change: GC-ON and GC-OFF are no longer
1712 implemented with a counter, it does not matter how many times gc
1713 is switched on or off
1714 * bug fix: discriminating functions for generic function classes
1715 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1716 make invalid assumptions about method precedence order. (reported
1718 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1719 platforms. (thanks to Thiemo Seufer)
1720 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1721 * bug fix: correct pathname printing with printer escaping is on.
1722 (thanks to Kevin Reid)
1723 * bug fix: complex VOP definitions in "user-space" no longer trigger
1724 package locks. (reported by Zach Beane)
1725 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1726 for funcallable-instance functions, and is no different from
1728 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1729 characters in the current readtable. (reported by Nicholas Neuss)
1730 * bug fix: SUBTYPEP on various CONS types returns more a more
1731 accurate acknowledgment of its certainty.
1732 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1733 lists in some cases. This partially fixes bug 384.
1734 * flush all standard streams before prompting in the REPL and the
1736 * bug fix: signal handling and triggering gc do not conflict
1737 directly anymore, in particular a high frequency sb-sprof does
1738 not prevent gc from running
1739 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1740 approximation for timezone and DST information between the
1741 universal time epoch and the smallest negative 32-bit time_t.
1742 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1743 year 1899 so long as the encoded time is non-negative.
1744 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1745 the correct answer for non-constant rotate amounts when performing
1746 an inline 32-bit rotation.
1748 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1749 there is only one thread in the session
1750 ** bug fix: memory leak for streams created in one thread and
1751 written to in another
1752 ** bug fix: lockup when compiled with gcc4
1753 ** bug fix: race that allows the gc to be triggered when gc is
1755 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1756 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1758 ** bug fix: finished threads are gc'ed properly
1759 * fixed some bugs revealed by Paul Dietz' test suite:
1760 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1761 has arguments with a different set of applicable methods from
1762 the orignal arguments.
1763 ** The type error thrown by MAP now has a correct expected-type
1765 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1766 name a compiled function.
1767 ** Three MISC tests where a large bignum was improperly coerced to
1768 a float that couldn't represent that bignum during type
1769 derivation were fixed.
1770 ** SUBTYPEP can now handle types involving the negation of a
1771 list-form FUNCTION type.
1772 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1773 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1774 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1776 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1777 * New feature: Experimental support for bivalent streams: streams
1778 opened with :element-type :default now allow character and binary
1779 (unsigned-byte 8) I/O
1780 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1781 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1783 * bug fix: no more highly sporadic "couldn't check whether ~S is
1784 readable" when reading a stream and an interrupt hits in the middle
1785 of a select system call
1786 * compiler better recognizes complex arrays (reported by Eduardo
1788 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1789 for variables on SBCL built with linkage-tables. (reported by Luis
1791 * various error reporting improvements.
1792 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1793 (Thanks to Thiemo Seufer)
1794 * bug fix: nested reader invokations work correctly
1795 * bug fix: it is possible to have simultaneous references to foreign
1796 code and foreign data with the same name.
1798 ** added x86-64 support
1799 ** incompatible change: the threading api now works with thread
1800 objects instead of thread ids
1801 ** bug fix: threads are protected from signals and interruption when
1802 starting up or going down
1803 ** bug fix: a race where an exiting thread could lose its stack to gc
1804 ** bug fix: don't halt on infinite error in threads if possible
1805 ** fixed numerous gc deadlocks introduced in the pthread merge
1806 ** bug fix: fixed thread safety issues in read and print
1807 ** bug fix: debugger doesn't hang on session lock if interrupted at
1808 an inappropriate moment
1809 ** bug fix: run-program is now thread safe(r)
1810 ** bug fix: inner with-recursive-lock no longer releases the mutex
1811 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1812 * fixed some bugs revealed by Paul Dietz' test suite:
1813 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1814 EXPECTED-TYPE slots filled.
1815 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1817 ** MAKE-CONDITION accepts classes as type-designators.
1818 ** COMPILE may never return NIL.
1819 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1820 range before calling Unix time functions
1822 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1823 * numerous signal handling fixes to increase stability
1824 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1825 * minor incompatible change: we now correctly canonize default
1826 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1827 as per the MOP, rather than the historical (INITARG INITFUNCTION
1828 INITFORM). (reported by Bruno Haible)
1829 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1830 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1831 for more information.
1832 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1833 pathname is a directory pathname.
1834 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1835 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1837 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1838 :METACLASS argument in addition to a class name. (reported by
1839 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1840 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1841 (reported by Bruno Haible; patch from Gerd Moellmann)
1842 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1844 * bug fix: more cleanups to the floating point exception handling on
1845 x86-64 (thanks to James Knight)
1846 * bug fix: the compiler does not try to constant fold calls of
1847 COERCE and BIT-* functions when they return freshly constructed
1848 objects. (reported by Nikodemus Siivola and Paul Dietz)
1849 * optimization: improved the MIPS versions of generic (in the
1850 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1851 * optimization: direct conversion of (unsigned-byte 32) to floats on
1852 the PowerPC platform.
1853 * optimization: structure instances with raw slots now use less
1854 memory, and probably show better memory locality. (thanks to
1856 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1857 are treated as requests for stack allocation on the x86-64,
1858 Alpha-32, PPC and SPARC platforms.
1859 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1860 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1861 (thanks to Tony Martinez)
1863 ** gcing a dead thread can no longer lead to lockups
1864 ** threads block signals until they are set up properly
1865 ** errno is no longer shared by threads
1866 ** interrupt-thread restores the eflags register on x86
1867 ** fixed some lockups due to gc/thread interaction
1868 * fixed some bugs revealed by Paul Dietz' test suite:
1869 ** invalid dotted lists no longer raise a read error when
1870 *READ-SUPPRESS* is T
1871 ** signal an error if a symbol that names a declaration is used
1872 as the name of a type, or vice versa
1873 ** allow using the (declare (typespec var*)) abbreviation for
1874 (declare (type typespec var*)) with all type specifiers
1875 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1876 ** always return NIL from PPRINT-POP when OBJECT is NIL
1877 ** don't signal errors when pretty-printing malformed LABELS,
1878 FLET or MACROLET forms
1879 ** declarations in a DOLIST body are also in scope for the
1881 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1883 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1886 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1887 * fixed cross-compiler leakages that prevented building a 32-bit
1888 target with a 64-bit host compiler.
1889 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1890 opened with :IF-EXISTS :OVERWRITE.
1891 * fixed bug 281: error for an invalid qualifier in a short-form method
1892 combination method is not signalled until the faulty method is called.
1893 * bug fix: iteration variable type inferrer failed to deal with open
1894 intervals. (reported by Alan Shields)
1895 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1896 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1897 or more runtime options were provided to the sbcl binary.
1898 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1899 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1900 * optimization: there's now a fast-path for fixnum arguments in the
1901 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1902 * optimization: the code generated on x86-64 is more compact thanks
1903 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1904 * bug fix: floating-point exception handling now partly works on
1905 x86-64. (Thanks to James Knight)
1906 * improvement to the MIPS backend from Thiemo Seufer: C-style
1907 64-bit long long arguments and return values to alien functions
1909 * fixed some bugs revealed by Paul Dietz' test suite:
1910 ** the type-error signalled from WARN has a filled-in DATUM slot.
1911 ** the type-error required when a stream is not associated with
1912 a file has the stream as its datum.
1913 ** type-errors on single-floats on x86-64 no longer have
1914 :INVALID-OBJECT as the datum
1915 ** the type-errors signalled for invalid function names now have
1916 a correct expected type
1917 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1918 for structures defined with a :TYPE. Documentation strings for
1919 typed structures are no longer immediately discarded
1920 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1921 broadcast streams and synonym streams. FILE-LENGTH now also works
1922 on broadcast streams.
1924 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1925 * incompatible change: the --noprogrammer option, deprecated since
1926 version 0.7.5, has been removed. Please use the equivalent
1927 --disable-debugger option instead.
1928 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1930 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1931 inoperational for a while now) has been completely removed.
1932 * null lexical environments are now printed as #<NULL-LEXENV>,
1933 significantly reducing the amount of clutter in typical backtraces.
1934 * documentation on weak pointers, finalization, and after GC hooks
1935 has been added to the manual.
1936 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1937 as other specialized array types, is much faster. SUBSEQ and
1938 COPY-SEQ on such arrays have also been sped up.
1939 * optimization: EQL is now more efficient when at least other argument
1940 is known to be of type (OR FIXNUM (NOT NUMBER)).
1941 * fixed bug: the runtime is now less vulnerable to changes in the
1942 size of the SBCL object on OS X, and virtual memory is reserved for
1943 all spaces that need to be at a fixed address.
1944 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1945 * fixed bug: finalizers and after GC hooks that cause consing are now
1947 * fixed bug: compiler error messages and summaries are now printed to
1948 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1949 * fixed inference of the upper bound of an iteration variable.
1950 (reported by Rajat Datta).
1951 * fixed bug 211e: calling local functions with duplicated constant
1952 keyword argument no longer causes a bogus style warning about an
1954 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1955 ftype declaration to be disregarded. (reported by Dave Roberts)
1956 * fixed bug 373: caused by erronous compilation of references to alien
1957 variables in the runtime on ppc/darwin.
1958 * fixed bug 376: CONJUGATE type deriver.
1959 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1960 (reported by Baughn on #lisp)
1961 * a cross-compiler bug on non-x86 platforms has been identified and
1962 fixed. (thanks to Bruno Haible)
1963 * improvements to the MIPS runtime code for increased stability.
1964 (thanks to Thiemo Seufer)
1965 * increased the maximimum compact environment size to allow
1966 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1967 * partial workaround for bug 135: don't name the function we're
1968 calling for hairy cases of EVAL, so as not to accumulate one environment
1969 entry per call to EVAL. (reported by Kevin Reid)
1970 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1971 * optimization: type testing for non-vector arrays should be faster.
1972 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1974 * bug fix: setting 31st element of a bit vector to zero did not work
1976 * bug fix: redefining a class definition which failed due to a
1977 previous accessor / function clash now works (but see BUGS entry
1978 #380 for more problems in this area). (thanks to Zach Beane)
1979 * the long form of DEFINE-METHOD-COMBINATION disables method group
1980 checking when given a single method group with pattern *. (thanks
1982 * on x86 compiler supports stack allocation of results of simple
1983 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1984 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1985 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1986 * fixed some bugs related to Unicode integration:
1987 ** the restarts for recovering from input and output encoding
1988 errors only appear when there is in fact such an error to
1990 * fixed some bugs revealed by Paul Dietz' test suite:
1991 ** MISC.549 and similar: late transformation of unsafe type
1992 assertions into derived types caused unexpected code
1994 ** SCALE-FLOAT type deriver is less wrong.
1995 ** type derivers for EXP, LOG and similar functions compute result
1996 types for complex arguments better.
1997 ** (MISC.563) CONJUGATE type deriver works for very restricted
1999 ** out-of-line type testers for character strings are available.
2000 ** EQUAL compiler transform understands specialness of objects
2002 ** accessing double-floats stored on the stack now works on x86-64.
2003 ** debugger internals could sometimes create invalid lispobjs,
2004 resulting in GC crashes.
2005 ** MISC.548: type check weakening can convert required type into
2007 ** initialization forms for bindings are not in scope of free special
2010 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2011 * incompatible change: thread support for non-NPTL systems has
2012 been removed - locking is buggy and unreliable. A threaded
2013 SBCL build will now warn at startup and refuse to create
2014 new threads, unless futex support is detected in the kernel
2015 * incompatible change: the top level REPL now has only an ABORT
2016 restart associated with it, not TOPLEVEL and ABORT as it used to.
2017 TOP and TOPLEVEL are now available as debugger commands for
2018 returning to the top level.
2019 * incompatible change: forms evaluated in the REPL now use the
2020 global optimization policy.
2021 * incompatible change: user- and system-initialization files are
2022 no longer processed with LOAD, but by READ and EVAL; hence the
2023 global optimization policy, startup package, readtable, etc,
2025 * The .fasl file format number has been incremented because of
2026 various incompatible changes.
2027 * internal entry point details and argument counts no longer appear
2028 in backtraces unless explicitly requested by setting
2029 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2030 * built-in and standard functions no longer have names like "top
2031 level local call to FOO".
2032 * fixed bug 32: functions defined in non-null lexical environments
2033 now have more legible printed representation
2034 * fixed bug 33: functions defined in non-null lexical environemnts
2035 are now more amenable to inspection by INSPECT.
2036 * workaround for bug 354: XEPs no longer appear in backtraces unless
2037 explicitly requested.
2038 * fixed bug: receiving the signal which results from attempting to
2039 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2040 SIGBUS on other *BSDs) on architectures where the C stack is also
2041 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2042 to debugger confusion or wild instability. (reported by Cheuksan
2044 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2045 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2046 (reported by Lutz Euler)
2047 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2048 compiled code. (reported by Paul Dietz)
2049 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2050 specialized parameter to a method to something that is not TYPEP
2051 the specializer is now possible.
2052 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2053 face of package deletion.
2054 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2055 pathname. (reported by David Tolpin on c.l.l)
2056 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2057 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2058 STRING-STREAM and FILE-STREAM, respectively.
2059 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2060 than previously; in addition, its test suite is now run on build.
2061 (thanks to Luke Gorrie)
2062 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2063 * optimization: added a immediate representation for single-floats
2065 * fixed some bugs revealed by Paul Dietz' test suite:
2066 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2067 ** attempting to create a package with a colliding nickname causes
2068 correctable errors to be signalled.
2069 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2070 ** several x86-64 backend bugs related to sign-extension of immediate
2073 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2074 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2075 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2077 * fixed loading of multiply forward-referenced layouts.
2078 (thanks to Cheuksan Wang)
2079 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2080 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2081 (see COMPILE-FILE documentation for details of :PRINT option)
2082 * fixed bugs 19 and 317: fixed-format floating point printing is
2083 more accurate. This also fixes a bug reported by Adam Warner
2084 related to the ~@F format directive.
2085 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2087 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2088 dispatch table if the from-char is a dispatch macro character.
2089 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2090 by the wordsize no longer ignore the last word. (reported by Lutz
2092 * fixed bug in type checking of dynamic-extent variables. (reported
2094 * optimization: sequence traversal functions use their freedom to
2095 coerce function designators to functions.
2096 * optimization: code with many calls to CLOS methods specialized on
2097 CLOS classes has had redundant type checks removed.
2098 * optimization: added declarations to speed up operations that access
2099 the internal character database (for example STRING-UPCASE)
2100 * optimization: comparison operations between floats and sufficiently small
2101 fixnums no longer create extra rationals
2102 * fixed some bugs related to Unicode integration:
2103 ** portions of multibyte characters at the end of buffers for
2104 character-based file input are correctly transferred to the
2105 start of the buffer at the next read.
2106 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2107 passing it through to OPEN.
2108 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2109 argument given to internal calls to OPEN.
2110 ** the built-in comment readers (introduced by character sequences
2111 ";" and "#|") are more forgiving to encoding errors; they will
2112 STYLE-WARN and then attempt to resync the stream at a character
2113 boundary. (thanks to Teemu Kalvas)
2114 * fixed some bugs in the x86-64 port:
2115 ** Negative short int return values from c-calls are sign-extended
2117 ** The stack is aligned to 16-bytes for c-calls, as required by
2118 the ABI. (thanks to Cheuksan Wang)
2119 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2120 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2121 for enums. (thanks to Vincent Arkesteijn)
2122 ** Multiple small optimizations and bugfixes for floating point
2124 * fixed some bugs revealed by Paul Dietz' test suite:
2125 ** Space, Tab, Linefeed, Return and Page have the invalid
2126 secondary constituent character trait.
2127 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2129 ** WITH-INPUT-FROM-STRING should only update the index place on
2131 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2132 works more reliably.
2133 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2134 with one-dimensional arrays with fill pointers.
2135 ** TYPE-OF failed on a complex with an integer realpart and a
2137 ** compiler failure during type inference for the code of form
2138 (IF (EQL X (THE ...)) ...) (MISC.535).
2140 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2141 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2142 platform. The port supports 61-bit fixnums, large memory spaces
2143 and reloading shared object files.
2144 * enhancement: saving cores with foreign code loaded is now
2145 supported on ppc/Darwin in addition to the previously supported
2147 * enhancement: the statistical profiler now walks deeper into the
2148 call stack for more meaningful call-graphs and accrued time
2149 reports (x86/x86-64 only). It also now reports time spent in
2151 * enhancement: it is now possible to trace most individual methods
2152 of a generic function in addition to tracing the generic function
2154 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2156 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2157 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2158 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2159 produces an error. (thanks to Vincent Arkesteijn)
2160 * bug fix: NAMESTRING on pathnames with :WILD components in their
2161 directories works correctly. (thanks to Artem V. Andreev)
2162 * fixed bug 125: compiler preserves identity of closures. (reported
2164 * bug fixed: functions with &REST arguments sometimes failed with
2165 "Undefined function" when compiled with (DEBUG 3). (reported by
2167 * bug fix: overflow during compiling of setting element of a bit
2168 vector with constant index and value. (reported by Timmy Douglas)
2169 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2170 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2171 * on x86 compiler partially supports stack allocation of dynamic-extent
2173 * GO and RETURN-FROM do not check the extent of their exit points
2174 when compiled with SAFETY 0.
2175 * fixed some bugs related to Unicode integration:
2176 ** encoding and decoding errors are now much more robustly
2177 handled; it should now be possible to recover even from invalid
2178 input or output to the terminal. (thanks to Teemu Kalvas)
2179 ** provided a first cut at implementing STRING-TO-OCTETS and
2180 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2181 ** altered the SB-MD5 contributed module slightly, changing the
2182 interface just enough for it to be supportable for builds where
2183 lisp characters are not eight bits.
2184 * fixed some bugs revealed by Paul Dietz' test suite:
2185 ** the FORMATTER-generated functions for ~V[ conditionals require
2186 the correct number of arguments.
2187 ** READ-FROM-STRING returns the mandated second value when applied
2188 to displaced strings.
2189 ** the #\Rubout and #\Backspace characters are treated as invalid
2190 constituent characters by the tokenizer.
2192 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2193 * new feature: reloading changed shared object files with
2194 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2195 * new feature: references to foreign variables and functions
2196 can now be compiled and loaded before the corresponding shared
2197 object file is loaded, as long as the foreign definitions are
2198 available at runtime.
2199 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2200 supported. (thanks to Dan Debertin)
2201 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2202 just keywords. (thanks to Vincent Arkesteijn)
2203 * fixed bug #331: structure-class instances corresponding to
2204 DEFSTRUCT forms are now created eagerly.
2205 * fixed bug #345: backtraces from calls to undefined functions work
2206 on x86 as well. Related bug #61 is now also partially fixed on x86
2207 (backtraces from throws to unknown catch tags.)
2208 * bug fix: lambda-list parsing is now stricter vrt. order and number
2209 of lambda-list keywords.
2210 * bug fix: as specified by AMOP, an error is signalled if a
2211 class-option appears multiple times in a DEFCLASS form. (reported
2213 * bug fix: RANDOM can be compiled when the compiler derives the type
2214 of its numeric argument as a disjoint set of small integers.
2215 (reported by Paul Dietz)
2216 * bug fix: starting a core saved with shared objects loaded when
2217 those objects are not available no longer causes threaded SBCL to
2218 hang. (reported by Sean Ross)
2219 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2221 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2222 argument, which must always be NIL. (reported by Kalle Niemitalo)
2223 * bug fix: printing 1.0d+23 no longer results in an error.
2224 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2225 * bug fix: structure slot setters preserve evaluation order. (thanks
2227 * fixed some bugs related to Unicode integration:
2228 ** RUN-PROGRAM can allow its child to take input from a Lisp
2229 stream. (reported by Stefan Scholl)
2230 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2231 locales. (reported by Ken Causey)
2232 * fixed some bugs revealed by Paul Dietz' test suite:
2233 ** INCF, DECF and REMF evaluate their place form as specified in
2235 ** FORMATTER expands ~{ ~} iteration directives with V or #
2236 parameters correctly.
2237 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2238 consequent uses no arguments correctly.
2239 ** the system has a partial understanding of the (COMPLEX RATIO)
2241 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2244 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2245 * new feature: a build-time option (controlled by the :SB-UNICODE
2246 keyword feature, enabled by default) for building the system with
2247 support for the entire 21-bit character space defined by the
2249 * new feature: the system now has rudimentary external-format
2250 support; the primary user-visible change at this time is that
2251 characters with the high bit set (such as lower-case-e-acute) will
2252 print correctly to a terminal in a UTF-8 environment.
2253 * minor incompatible change: BASE-CHAR no longer names a class;
2254 however, CHARACTER continues to do so, as required by ANSI.
2255 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2256 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2257 has been moved to the SB-EXT package (temporarily re-exported from
2259 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2260 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2262 * fixed bug #351: better error-handlind and reporting for malformed
2264 * fixed bug #350: bignum-printing is now more memory-efficient,
2265 allowing printing of very large bignums, eg. (expt 2 10000000).
2266 (reported by Bruno Haible)
2267 * fixed bug #302: better primitive-type selection for intersection
2269 * fixed bug #308: non-graphic characters now all have names, as
2270 required. (reported by Bruno Haible)
2271 * bug fix: redefining a class with different superclasses now correctly
2272 removes it from the direct-subclasses of its previous superclasses.
2273 (reported by David Morse)
2274 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2275 new value now works. (reported by Bruno Haible)
2276 * bug fix: correct canonicalization of multiple non-standard slot
2277 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2278 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2279 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2280 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2281 now exists, an signals an error.
2282 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2283 during macroexpansion for non-iteration variables as well. (reported
2284 by Bruno Haible for CMUCL)
2285 * bug fix: Cyclic structures and unprintable objects in compiler
2286 messages no longer cause errors. (reported by Bruno Haible)
2287 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2288 and READ-FROM-STRING all now return a primary value of NIL if
2289 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2290 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2291 (reported by Bruno Haible for CMUCL)
2292 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2293 do not designate a condition. (reported by Bruno Haible for
2295 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2296 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2297 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2298 permitting resolution in favour of any of the conflicting symbols.
2299 (reported by Bruno Haible for CMUCL)
2300 * bug fix: EQUAL compiler optimizations is less aggressive on
2301 strings which can potentially compare true despite having distinct
2302 specialized array element types.
2303 * bug fix: unit enumerations can be defined without dividing by
2304 zero. (thanks to Vincent Arkesteijn)
2305 * FORMAT compile-time argument count checking has been enhanced.
2306 (report from Bruno Haible for CMUCL)
2307 * a partial workaround for the bug 262: the compiler does not try to
2308 inline-expand a local function doing RETURN-FROM from a deleted
2309 BLOCK. (thanks to Peter Denno for the bug report and to David
2310 Wragg for the simple test case)
2311 * fixed some bugs revealed by Paul Dietz' test suite:
2312 ** RENAME-PACKAGE allows all package designators as new package
2314 ** constraint propagation and lambda variable substitution are
2315 more cautious in dealing with partially deleted code.
2316 ** compiler sometimes lost reoptimization passes.
2317 ** CERROR, given a condition as condition designator, passes its
2318 remaining arguments to the continue format control without
2320 ** Case-altering FORMAT directives work correctly on non-ASCII
2322 ** The REAL type specifier handles bounds outside the floating
2323 point ranges without signalling FLOATING-POINT-OVERFLOW.
2324 ** Functions with IR1-transformations can create intercomponent
2325 references to global functions.
2326 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2328 ** FORMAT ~R treats a nil value for its first parameter correctly.
2330 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2331 * enhancement: saving cores with foreign code loaded is now
2332 supported on x86/NetBSD and sparc/Linux in addition to the previously
2333 supported platforms.
2334 * bug fix: on some platforms repeated installations caused multiple
2335 copies of HTML documentation to be installed -- should not happen
2336 any more. (reported by Stefan Scholl)
2337 * bug fix: parsing self-recursive alien record types multiple times
2338 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2339 original patch by Helmut Eller for CMUCL)
2340 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2341 (thanks to Richard Kreuter)
2342 * bug fix: defining classes whose accessors are methods on existing
2343 generic functions in other (locked) packages no longer signals
2344 bogus package lock violations. (reported by François-René Rideau)
2345 * bug fix: special variables as DEFMETHOD parameters no longer have
2346 associated bogus type declarations. (reported by David Wragg and
2348 * bug fix: read-write consistency on streams of element-type
2349 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2350 * bug fix: redefiniton of the only method of a generic function with
2351 no DEFGENERIC no longer emits a full WARNING. In addition,
2352 redefinition of generic functions with no DEFGENERIC to an
2353 incompatible lambda list now signals an error. (thanks to Zach
2355 * bug fix: DEFGENERIC now works even when there's a function of the
2356 same name in an enclosing lexical environment. (thanks to Zach
2358 * fixed compiler failure, caused by instrumenting code during
2359 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2360 * optimization: added loop analysis and improved register allocation
2361 * fixed some bugs revealed by Paul Dietz' test suite:
2362 ** POSITION on displaced vectors with non-zero displacement
2363 returns the right answer.
2364 ** (SIMPLE-STRING) is a valid type specifier for sequence
2366 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2368 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2369 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2371 ** The justification version of the FORMAT ~< directive treats
2372 non-zero minpad parameter correctly.
2374 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2375 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2376 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2377 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2378 the supported interface.
2379 * new feature: Single-stepping of code compiled with DEBUG 2 or
2380 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2381 * new feature: saving cores with foreign code loaded is now
2382 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2383 Timothy Moore's work for CMUCL)
2384 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2385 parameters to * instead of NIL if no initform is supplied.
2386 (reported by Johan Bockgaard)
2387 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2388 correctly when FUNCALL appears as the car of the form. Note:
2389 despite this FUNCALL forms are not currently subject to
2390 compiler-macro expansion. (port of Raymond Toy's fix for the
2391 same from CMUCL, reported by Johan Bockgaard)
2392 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2393 (thanks for Teemu Kalvas)
2394 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2395 case instead of terminating the iteration (thanks for Julian
2396 Squires, Sean Champ and Raymond Toy)
2397 * bug fix: incorrect expansion of defgeneric that caused a style
2398 warning. (thanks for Zach Beane)
2399 * bug fix: slot accessor effective method computation works properly
2400 for classes with multiple non-standard applicable methods on
2401 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2402 * on x86 compiler supports stack allocation of results of LIST and
2403 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2404 CMUCL implementation by Gerd Moellmann)
2405 * fixed some bugs revealed by Paul Dietz' test suite:
2406 ** FORMAT strings with both the ~<~:;~> form of the justification
2407 directive and pretty-printing directives cause an error.
2409 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2410 * incompatible change: the internal functions
2411 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2412 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2413 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2414 instead of the old functions.
2415 * new feature: on platforms where "dladdr" is available foreign
2416 function names now appear in backtraces. (based on Helmut Eller's
2418 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2419 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2421 * documentation: ASDF manual is now distributed with SBCL.
2422 * bug fix: SBCL can now load its contributed modules with REQUIRE
2423 even if the system-provided entries have been removed from
2424 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2426 * bug fix: backtraces involving undefined functions or assembly
2427 routines are more informative. (thanks to Brian Downing)
2428 * bug fix: mutually referent alien structures now work correctly.
2429 (reported by Rick Taube)
2430 * bug fix: structures defined by WITH-ALIEN can be referred to
2431 within other definitions in the same WITH-ALIEN.
2432 * bug fix: division operators (MOD, TRUNCATE and the like) with
2433 constant zero divisors and integer dividends no longer generate
2435 * bug fix: provide default methods for INPUT-STREAM-P and
2436 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2437 * bug fix: improve the detection and resolution of MOP metacycles
2438 (where effective method computation depends on the generic function
2439 having its effective method computed). (reported by Bruno Haible)
2440 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2441 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2442 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2443 * bug fix: pathnames with relative directory components can be
2444 represented relative to default pathnames.
2445 * optimization: in taking the GCD of bignums, reduce the two bignums
2446 to approximately the same size (using Euclid's algorithm) before
2447 applying the more sophisticated binary GCD. (thanks to Juho
2449 * optimization: COUNT on bitvectors now operates word-at-a-time.
2450 * optimization: ASH with a positive, but not necessarily constant,
2451 (leftwards) shift, when in a modular context, is compiled to a
2453 * fixed some bugs revealed by Paul Dietz' test suite:
2454 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2455 if the corresponding argument is NIL.
2456 ** FORMAT directives accepting a minpad parameter treat negative
2458 ** literal commas as character parameters to FORMAT directives are
2460 ** literal spaces directly after ~<Newline> directives within a
2461 format-logical-block (~:< ~@:>) do not induce :FILL-style
2462 conditional newlines.
2463 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2464 as specified: it no longer includes conditional newlines).
2465 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2467 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2468 :PER-LINE-PREFIX argument does not evaluate to a string.
2470 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2471 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2472 the manual for details; for now, package locks can be disabled by
2473 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2474 if no major problems are found then it is likely that they will be
2475 compiled in unconditionally.
2476 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2477 unsupported operators on all platforms. To load a shared library
2478 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2479 object file, link it into a shared library outside of SBCL and
2480 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2482 * bug workaround: ROOM T and the GC were not working together
2483 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2484 reporting facility aren't preserved. That reporting has been
2485 disabled, so now until and unless someone figures out how to make it
2486 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2487 * minor incompatible change: as threatened around sbcl-0.8.0, the
2488 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2489 The symbols are also exported from SB-PCL for backwards
2490 compatibility, but more so than before SB-PCL should be treated as
2491 an implementation-internal package.
2492 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2494 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2495 It is known as of this release not to work on the Alpha, however.
2496 * fixed bug #167: errors signalled due to illegal syntax in method
2497 bodies are now more legible.
2498 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2499 designators and can hence be used with TYPEP.
2500 * fixed bug #333: CHECK-TYPE now ensures that the type error
2501 signalled, if any, has the right object to be accessed by
2502 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2503 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2504 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2506 * fixed bug #334: programmatic addition of slots using specialized
2507 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2508 and :ALLOCATION :CLASS slots.
2509 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2510 float exponents. (rereported by Peter Seibel)
2511 * fixed a bug: #\Space (and other whitespace characters) are no
2512 longer considered to be macro characters in standard syntax by
2513 GET-MACRO-CHARACTER.
2514 * fixed bug: initialization of condition class metaobjects no longer
2515 causes an instance of the condition to be created. (reported by
2517 * fixed bug: it is now possible to have slots such that
2518 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2519 description is neither :INSTANCE nor :CLASS.
2520 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2521 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2522 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2523 non-standard SLOT-DEFINITION-ALLOCATION values.
2524 * fixed bugs: various race conditions handling exiting threads.
2525 CL-PPCRE's thread torture test now passes.
2526 * fixed arguably-a-bug: GC time proportional to number of threads in
2527 system even when most of them are idle
2528 * optimization: improved performance of BIT and SBIT on bit-vectors.
2529 * .fasl file incompatibility: The fasl file version number has
2530 been incremented because of changes associated with package locks.
2532 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2533 * minor incompatible change: the system no longer provides
2534 optimization hints (conditions which are TYPEP
2535 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2536 mode; these hints are still emitted when higher SPEED optimization
2538 * new contrib module: a sampling profiler (profiling by statistical
2539 sampling, rather than by instrumenting functions) is available as
2540 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2541 * the behaviour of the standard function ED is now customizeable by
2542 third parties through a hook variable: see ED's documentation
2543 string for information on the protocol.
2544 * the compiler no longer emits efficiency notes for (FUNCALL X)
2545 when the type of X is uncertain under default optimization
2547 * fixed bug 276: mutating a binding of a specialized parameter to a
2548 method to something that is not TYPEP the specializer is now
2550 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2551 exhibit the required behaviour on the x86 platform. (thanks to
2552 Peter van Eynde, Eric Marsden and Bruno Haible)
2553 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2554 even for difficult arguments. (reported by Peter Graves)
2555 * fixed bug 141a: the backquote printer now descends quoted
2557 * fixed another bug in backquote printing: no more destructive
2558 modification of the form's list structure. (reported by Brian
2560 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2561 move between its address being taken and the call to
2562 interrupt_thread, fixing a crashing race condition.
2563 * the SB-POSIX contrib implementation has been adjusted so that it
2564 no longer exhibits ridiculously poor performance when constructing
2565 instances corresponding to C structs.
2567 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2568 * minor incompatible change: the sb-grovel contrib now treats C
2569 structures as alien (in the sense of SB-ALIEN) objects rather than
2570 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2571 has implications for memory management of client code
2572 (sb-grovel-returned objects must now be manually managed) and for
2573 type safety (alien objects now have full types).
2574 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2575 used to control emission of compiler diagnostics, rather than the
2576 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2577 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2578 quality should be considered deprecated.
2579 * install.sh now installs the user manual as well
2580 * (not quite a new documentable feature, but worth considering in
2581 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2582 beginnings of a semantically meaningful condition hierarchy is
2583 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2585 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2586 designator as the defaults argument.
2587 * fixed bug: Displaced arrays whose displaced-to array has become
2588 too small now cause ARRAY-DIMENSION to signal an error, providing
2589 for safer bounds-checking. (reported by Bruno Haible)
2590 * fixed bug: DEFCLASS slot definitions with identical :READER and
2591 :WRITER names now signal a reasonable error. (reported by Thomas
2593 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2595 * fixed bug: Invalid slot specification errors now print correctly.
2596 (thanks to Zach Beane)
2597 * fixed bug 320: Shared to local slot value transfers in class
2598 redefinitions now happen corrently from superclasses as
2599 well. (reported by Bruno Haible)
2600 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2602 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2603 improper lists correctly. (reported by Bruno Haible)
2604 * fixed bug 313: source-transform for <fun-name> was erroneously
2605 applied to a call of a value of a variable with name <fun-name>.
2606 (reported by Antonio Menezes Leitao)
2607 * fixed bug 307: The obsolete instance protocol ensures that
2608 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2609 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2610 remove dead unknown-values globs from the middle of the stack.
2611 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2612 easier to resume long computations after using *BREAK-ON-SIGNALS*
2613 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2614 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2615 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2616 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2617 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2618 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2620 * fixed a (fairly theoretical) bug in string printing: if
2621 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2622 does not have array-element-type equal to the most general string
2624 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2625 function name. (thanks to Nikodemus Siivola)
2626 * fixed bug: compile-time format string checker failed on
2627 non-closed ~{. (reported by Thomas F Burdick)
2628 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2629 don't warn on legal loop code involving more than one aggregate
2630 boolean. (thanks to Nikodemus Siivola)
2631 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2632 in keyword position to MAKE-PACKAGE was not being reported as
2634 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2635 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2637 * fixed bugs 280 and 312: the checking for multiple definitions in a
2638 file is less likely to become confused by uses of inline
2640 * fixed bug: the #S reader macro performs the keyword coercion
2641 specified for slot names. (reported by Kalle Niemitalo)
2642 * fixed bug: lambda lists may contain symbols whose names start with
2643 & but are not lambda-list-keywords; their occurrence triggers a
2645 * fixed bug 321: define-method-combination argument lambda lists do
2646 not cause invalid code to be generated when &optional or &aux
2647 variables have default values. (reported by Bruno Haible)
2648 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2649 have CLOS classes; this appears also to have fixed reported
2650 instability in stack exhaustion detection.
2651 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2652 MUFFLE-WARNING when no associated restart is present is now
2654 * optimization: rearranged the expansion of various defining macros
2655 so that each expands into only one top-level form in a
2656 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2658 * optimization: used a previously little-used slot in symbols to
2659 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2661 * fixed some bugs revealed by Paul Dietz' test suite:
2662 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2664 ** Fixed an optimization bug related to inheritance of initforms
2665 from local to shared slots.
2666 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2667 ** CAST optimizer forgot to flush argument derived type.
2668 ** print/read consistency on floats is now orders of magnitude
2669 more likely. (thanks also to Bruno Haible for a similar report
2671 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2672 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2673 for objects of type REAL. Make it so.
2674 ** SXHASH is specified (infelicitously) to respect similarity,
2675 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2676 it so. (thanks to Markus Ziegler)
2677 ** on the Alpha, the compiler succeeds in compiling functions
2678 returning a known number of arguments greater than 63.
2679 ** fixed handling of invalid NIL arguments in keyword position
2681 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2683 ** provided a readably-printable representation for RANDOM-STATE
2685 ** ensured that pathnames, where they have a namestring, always
2686 print using #P"..." syntax.
2688 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2689 * Support for the forthcoming 2.0 version of the NetBSD kernel
2690 running on x86 hardware has been added. (thanks to Perry
2691 E. Metzger most immediately, and others for their past work)
2692 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2693 current release version; SBCL's OpenBSD support had been broken
2694 since about the time of OpenBSD's switch to ELF binary format.)
2695 * [placeholder for DX summary]
2696 ** user code with &REST lists declared dynamic-extent, under high
2697 speed or space and low safety and debug optimization policy.
2698 * The manual has been converted to Texinfo format and the debugger
2699 chapter from the cmucl manual has been added.
2700 * A facility has been added to extract documentation strings from
2701 sbcl and store them as Texinfo-formatted snippets for inclusion in
2702 the manual (via Texinfo's @include directive)
2703 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2704 the test case to Dave Roberts)
2705 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2706 pointers. (reported by Sean Ross)
2707 * bug fix: PROFILE output is printed nicely even for large numerical
2708 values. (thanks to Zach Beane)
2709 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2710 greater than 32 handle EOF correctly.
2711 * bug fix: on X86 an immediate argument of the IMUL instruction is
2712 correctly printed in disassembly. (thanks to Lutz Euler)
2713 * bug fix: class slots in redefined classes preserve their old
2714 values. (thanks to Bruno Haible and Nikodemus Siivola)
2715 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2716 succeeds. (reported by Marco Baringer)
2717 * fixed some bugs revealed by Paul Dietz' test suite:
2718 ** READ-SEQUENCE now works on ECHO-STREAMs.
2719 ** RATIONALIZE works more according to its specification. (thanks
2722 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2723 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2724 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2725 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2726 should matter to you only if you rebind the printer control
2727 variables and then find you want different bindings in the
2728 debugger than in the ordinary execution of your program.)
2729 * The runtime build system has been tweaked to support building
2730 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2731 assembler and linker. (thanks to Nikodemus Siivola)
2732 * Unbound, undefined, undeclared variables now trigger full
2733 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2734 more often programmer error than deliberate exploitation of undefined
2736 * optimization: the hash algorithm for strings has changed to one
2737 that is less vulnerable to spurious collisions. (thanks to Juho
2739 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2740 needless bounds checking. (thanks to Juho Snellman)
2741 * optimization: implemented multiplication as a modular
2742 (UNSIGNED-BYTE 32) operation on the PPC backend.
2743 * fixed some bugs revealed by Paul Dietz' test suite:
2744 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2746 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2747 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2748 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2750 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2752 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2754 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2755 their output stream on EOF from read.
2756 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2757 have been read to end-of-file.
2758 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2760 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2761 description of determination of which consecutive characters
2763 ** Printing the "Space" character with escaping on now yields "#\\ ",
2764 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2765 ** Reading floating-point numbers with *READ-BASE* set to a number
2766 less than 10 works correctly.
2767 ** Reading floating-point numbers with *READ-BASE* set to a number
2768 more than 10 works correctly.
2769 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2770 the readtable currently in effect.
2772 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2773 * minor incompatible change: parsing of namestrings on a physical
2774 (Unix) host has changed; numbers after the final #\. in a
2775 namestring are no longer interpreted as a version field. This is
2776 intented to be largely invisible to the user, except that the
2777 meaning of the namestring "*.*.*" has changed: it now refers to a
2778 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2779 should usually be replaced by
2780 (make-pathname :name :wild :type :wild :version :wild)
2781 with the added benefit that this is more likely to be portable.
2782 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2783 option to OPEN now signals an error if the file being opened
2784 exists; this may have an impact on existing code.
2785 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2786 signals received, on the PowerPC platforms (both Linux and
2787 Darwin). (thanks to Pierre Mai for pointing out the location of
2789 * several fixes on OS X: The system now builds and runs cleanly on
2790 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2791 (thanks to Brian Mastenbrook)
2792 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2793 second-resolution: integer multiples of 1/3600 between -24 and 24.
2794 (thanks to Vincent Arkesteijn)
2795 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2796 type when called with 1 argument; PEEK-CHAR checked type of
2797 PEEK-TYPE only after having read first character from a
2798 stream. (reported by Peter Graves)
2799 * bug fix: the garbage collector now has much better locality
2800 behaviour, and in particular no longer treats all memory as being
2801 exhausted when many small objects point to each other in a deeply
2803 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2804 recognized as being TYPEP their class.
2805 * bug fix: the PUSHNEW documentation string has been corrected.
2806 (thanks to Vincent Arkesteijn)
2807 * bug fix: defaulting of the value for the last of an atypically
2808 large number of multiple values being bound was not being
2809 performed correctly on the Alpha or PPC platforms
2810 * optimization: implemented multiplication as a modular
2811 (UNSIGNED-BYTE 32) operation on the x86 backend.
2812 * optimization: SEARCH on simple-base-strings can now be open-coded.
2813 (see also contrib/compiler-extras.lisp for inspiration for
2814 teaching the compiler about the Boyer-Moore algorithm).
2815 * value, returned by MAX (and MIN) called with several EQUALP, but
2816 not EQL, arguments now does not depend on compiler settings.
2817 * fixed some bugs revealed by Paul Dietz' test suite:
2818 ** in stack analysis liveness information is propagated from
2819 non-local entry points.
2820 ** pathwise CAST removing failed when the CAST node did not start
2822 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2823 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2825 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2826 host is already defined.
2827 ** RENAME-FILE works on streams instead of signalling an internal
2829 ** PEEK-CHAR uses the current readtable when determining whether
2830 or not a character is whitespace.
2831 ** MERGE-PATHNAMES handles the case when the pathname does not
2832 specify a name while the default-pathname specifies a version.
2833 ** Pathnames now stand a better chance of respecting print/read
2835 ** Attempting to use standardized file system operators with a
2836 pathname with invalid :DIRECTORY components signals a
2838 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2840 ** DIRECTORY on logical pathnames is more correct.
2841 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2842 signal a TYPE-ERROR if their argument is not a stream.
2843 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2844 designator argument does not designate a stream.
2845 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2846 examining the synonym.
2847 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2849 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2850 element-type larger than ([UN]SIGNED-BYTE 32).
2852 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2853 * When built with the :SB-FUTEX feature, threaded builds now take
2854 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2855 for faster/more reliable mutex and condition variable support.
2856 * Incompatible change (but one you probably shouldn't have been using
2857 anyway): the interface and code for arbitrating between multiple
2858 threads in the same user session has been redesigned.
2859 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2860 error when called without an explicit environment argument.
2861 (thanks to Nikodemus Siivola)
2862 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2863 confused about their position. (thanks to Adam Warner and Gerd
2865 * bug fix: Pretty printing backquoted forms with unquotations in the
2866 argument list position of various code constructs such as LAMBDA
2867 now works correctly. (reported by Paul Dietz)
2868 * bug fix: Pretty printing unquotations no longer loses all
2869 stream position information.
2870 * optimization: performance of string output streams is now less
2871 poor for multiple small sequence writes.
2872 * optimization: performance of CSUBTYPEP in the presence of complex
2873 expressions involving CONS and NOT many times has been improved.
2874 (reported by Paul Dietz)
2875 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2877 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2878 streams for the REPL, for the debugger, and so on. (thanks to
2880 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2881 * fixed some bugs revealed by Paul Dietz' test suite:
2882 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2883 is ignored if it is too small, rather than propagating through
2884 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2885 ** extremely complex negations of CONS types were not being
2886 sufficiently canonicalized, leading to inconsistencies in
2888 ** VALUES tranformer lost derived type.
2890 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2891 * fixed a bootstrapping bug: the build process no longer assumes
2892 that the various BOOLE-related constants have the same value in
2893 host and target lisps. (noted by Paul Dietz' test suite on an
2894 SBCL binary built from CLISP)
2895 * The system can now be dynamically linked on the MIPS platform,
2896 which enables dynamic loading of foreign code from Lisp. (thanks
2897 to Ralf Baechle for discussions on the MIPS ABI)
2898 * The system now records debugging information for its own source
2899 files in a filesystem-position-independent manner, relative to
2900 the "SYS" logical host.
2901 * fixed a compiler bug: MV-LET convertion did not check references
2902 to the "max args" entry point. (reported by Brian Downing)
2903 * tweaked disassembly notes to be less confident about proclaiming
2904 some instruction as an LRA. (thanks to Brian Downing)
2905 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2906 now each have their own history, command character, and other
2907 characteristics. (thanks to David Lichteblau)
2908 * fixed some bugs revealed by Paul Dietz' test suite:
2909 ** compiler failure in compiling LOGAND expressions including a
2911 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2912 shift greater than 32.
2913 ** FLUSH-DEST did not mark blocks for type check regeneration.
2914 ** HANDLER-CASE failed to accept declarations in handler clauses
2915 in some circumstances.
2917 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2918 * New code in contrib/sb-introspect (still probably not entirely
2919 stable yet) provides some support for smart Lisp development
2920 environments like SLIME.
2921 * The conditions signalled for errors occurring when loading .fasl
2922 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2923 in a way which should help ASDF recover gracefully.
2924 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2925 stuff has been cleaned up. If you code contrib/ stuff, this might
2926 affect you, and you can look at contrib/README, contrib/STANDARDS,
2927 and/or the 0.8.4.27 diff to check.
2928 * In full calls the compiler now does not generate checks for declared
2929 argument types for all arguments.
2930 * various threading fixes
2931 ** and some experimental patches which didn't make it into
2932 the main tree for this release, but which are shipped in
2933 contrib/experimental-thread.patch as a possible fix for some
2934 failures (deadlock, spinning...) in GC-intensive multithreaded
2936 * fixed PPC build problem (source code incompatibility of different
2937 library versions): added offsetof() hackery which attempts to divine
2938 where glibc maintainers put uc_mcontext today
2939 * fixed bug 282: compiler does not trust type assertions while passing
2940 arguments to a full call.
2941 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2942 &OPTIONAL VALUES type parameter.
2943 * fix bug 214: algorithm for noting rejected templates is now more
2944 similar to that of template seletion. (also reported by rydis on
2946 * fixed bug 141b: printing backquoted information readably and prettily
2947 inserts a space where necessary.
2948 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2949 (CHARACTER 10) are now reported as errors, rather than propagated
2950 as unknown types. (reported by piso on #lisp)
2951 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2952 with values NIL and :ERROR. (thanks to Milan Zamazal)
2953 * fixed bug 191c: CLOS now does proper keyword argument checking as
2954 described in CLHS 7.6.5 and 7.6.5.1.
2955 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2956 longer raise an error; further, using a list as a for-as-arithmetic
2957 counter now raises a meaningful error.
2958 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2959 now understood by sequence creation functions such as MAKE-SEQUENCE
2961 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2962 a STRING-INPUT-STREAM.
2963 * compiler enhancement: SIGNUM is now better able to derive the type
2965 * type declarations inside WITH-SLOTS are checked. (reported by
2967 * fixed some bugs revealed by Paul Dietz' test suite:
2968 ** incorrect optimization of TRUNCATE for a positive first
2969 argument and negative second.
2970 ** compiler failure in let-convertion during flushing dead code.
2971 ** compiler failure while deriving type of TRUNCATE on an
2972 interval, containing 0.
2973 ** ASH of a negative bignum by a negative bignum count now returns
2975 ** intersection of CONS types now canonicalizes properly, fixing
2976 inconsistencies in SUBTYPEP.
2978 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2979 * incompatible change: The --disable-debugger command line
2980 option now clobbers the debugger at a more fundamental
2981 level, by redefining #'INVOKE-DEBUGGER instead of by
2982 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2983 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2984 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2985 enter the debugger REPL and then suffer recursive errors
2986 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2987 while under the new system, BREAK in --disable-debugger mode
2988 terminates the system just as an unhandled error would.
2989 * fixed compiler performance when processing loops with a step >1;
2990 * bug fix: DOCUMENTATION now retrieves generic function
2991 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2992 support has been systematized, and now supports the methods
2993 specified by ANSI, along with a default method and a method for
2994 slot documentation. (reported by Nathan Froyd)
2995 * bug fix: effective methods associated with a generic function are
2996 no longer cached over a change of that generic function's method
2997 combination. (reported by Andreas Fuchs)
2998 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2999 names a non-existent directory. (thanks to Andreas Fuchs)
3000 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3001 the PPC platform, lead to stack corruption; this has been fixed.
3002 (reported by Rainer Joswig)
3003 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3004 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3005 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3006 arithmetic in general, is now much more reliable on the PPC
3008 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3009 platform now returns the right answer.
3010 * optimization: restored some effective method precomputation in
3011 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3012 precomputation is now tunable.
3013 * optimization: compiler-internal data structure use has been
3014 reviewed, and changes have been made that should improve the
3015 performance of the compiler by about 20%.
3016 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3017 simple-base-strings and simple-bit-vectors is improved.
3018 * optimization: the optimization of 32-bit logical and arithmetic
3019 functions introduced in version 0.8.3 on the x86 has been
3020 implemented on the mips, ppc and sparc platforms; an
3021 implementation of the same facility, but for 64-bit arithmetic,
3022 has been added for the alpha.
3023 * microoptimization: the compiler is better able to make use of the
3024 x86 LEA instruction for multiplication by constants.
3025 * bug fix: in some situations compiler did not report usage of
3026 generic arithmetic in (SPEED 3) policy.
3027 * bug 145b fix: compiler used wrong type specifier while converting
3028 MEMBER-types to numeric.
3029 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3031 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3032 index argument now works properly on the Alpha platform.
3033 * bug fix: floating point exception treatment on the Alpha platform
3035 * bug fix: FILE-POSITION works much better on string input and
3036 output streams. (thanks to Nikodemus Siivola)
3037 * bug fix: many threading/garbage collection symptoms sorted.
3038 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3039 might be pseudo-atomic.
3040 * internal change: Stopping for GC is now done with signals not
3041 ptrace. GC is now done in whichever thread wanted it, instead of
3043 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3045 * bug fix: result form in DO is not contained in the implicit
3047 * incompatible change: ICR structure is changed; the value part of
3048 CONTINUATION is now called LVAR; corresponding functions are
3049 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3050 * added type deriver for ISQRT (thanks to Robert E. Brown).
3051 * bug fix: better support for loading from the command line when an
3052 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3054 * fixed some bugs revealed by Paul Dietz' test suite:
3055 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3056 ** ROUND and FROUND now give the right answer when given very
3057 small float arguments.
3058 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3060 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3061 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3062 ** LCM should return a non-negative integer.
3063 ** PARSE-INTEGER returned the index of a terminator instead of the
3064 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3065 ** PARSE-INTEGER returned an incorrect index being applied to a
3067 ** LCM with two arguments of 0 returns 0 rather than signalling
3069 ** unsigned addition of a 32-bit constant with the high bit set no
3070 longer causes an internal compiler error on the x86.
3071 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3072 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3073 with negative last argument.
3074 ** byte specifiers with zero size and position no longer cause
3075 an error during type derivation.
3076 ** bignum multiplication on the Alpha platform now returns the
3078 * porting: The system now builds on SuSE AMD64, although it still
3079 generates a 32-bit binary.
3080 * .fasl file incompatibility: The fasl file version number has
3081 been incremented (because of the changes to internal compiler
3082 data structures referred to above).
3084 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3085 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3086 more accurately, on the Darwin kernel running on PowerPC hardware.
3087 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3088 * Compiler code deletion notes now signal a condition of type
3089 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3090 an associated MUFFLE-WARNING restart.
3091 * The compiler now performs limited argument count validation of
3092 constant format strings in FORMAT, and where appropriate in ERROR,
3093 CERROR and WARN. (thanks to Gerd Moellmann)
3094 * New ASDF-INSTALL contrib can be used for automatic download and
3095 installation of third-party Lisp code from CCLAN or other sites
3097 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3098 which forces another thread to execute a function supplied by the
3100 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3101 now accept and act upon their :ELEMENT-TYPE keyword argument.
3102 (reported by Martin Atzmueller, Edi Weitz)
3103 * bug fix: FILE-POSITION now accepts position designators up to
3104 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3105 is the greater. (thanks to Patrik Nordebo)
3106 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3107 Kalle Olavi Niemitalo)
3108 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3109 printable. (reported by Eric Marsden)
3110 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3111 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3112 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3113 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3114 (thanks to Stig Sandoe)
3115 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3116 UNREAD-CHAR now correctly includes the unread character in the
3117 target sequence. (thanks to Gerd Moellmann)
3118 * bug fix in threaded builds: the system can now be suspended and
3119 resumed by shell job control with minimal disruption.
3120 * bug fixes in times and timezones >2038AD
3121 * better handling of "where is GNU make?" problem in build scripts
3122 (thanks to Nikodemus Siivola)
3123 * new optimization: inside a named function any reference to a
3124 function with the same name is considered to be a self-reference;
3125 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3126 optimization quality.
3127 * new optimization on x86: logical functions and + now have
3128 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3129 used when the result is truncated to 32 bits.
3130 * VALUES declaration is partially enabled.
3131 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3132 * fixed some bugs revealed by Paul Dietz' test suite:
3133 ** The system now obeys the constraint imposed by
3134 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3135 types form a lattice under type intersection.
3136 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3137 ** ASSOC now ignores NIL elements in an alist.
3138 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3139 and (1+ MOST-POSITIVE-FIXNUM) answers.
3140 ** The addition of a method with invalid qualifiers to a generic
3141 function does not cause an error to be signalled immediately;
3142 a warning is signalled, and the error is generated only on
3143 calling the generic function.
3144 * changed .fasl file version number, in part to add type codes for
3145 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3146 obscure ANSI requirements
3148 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3149 * fixed bug 148: failure to inline-expand a local function left
3150 garbage, confusing the compiler.
3151 * fixed bugs 3cd: structure slot readers perform type check if the
3152 slot can have an invalid value (i.e. it is either not initialized
3153 or can be written with a less specific slot writer).
3154 * bug fix: the compiler now traps array references to elements off
3155 the end of an array; previously, the bounds checking in some
3156 circumstances could go off-by-one.
3157 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3159 * type declarations for array element types now obey the description
3160 on the CLHS page "Declaration TYPE", as per discussions on
3161 sbcl-help around 2003-05-08. This means that a declaration
3162 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3163 declaration, all references to BAR will be asserted or assumed
3164 (with THE, so dependent on compiler policy) to involve objects of
3165 type FOO. Note that no such declaration is implied in
3166 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3167 * declared types of functions from the "Conditions"
3168 chapter. (reported by Paul Dietz)
3169 * bug fix: CERROR accepts a function as its first argument.
3170 * bug fix: NTH an NTHCDR accept a bignum as index
3171 arguments. (reported by Adam Warner)
3172 * optimization: character compare routines now optimize comparing
3173 against a constant character. (reported by Gilbert Baumann)
3174 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3175 argument now works properly on the MIPS platform.
3176 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3177 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3178 declared to have a functional type, does not cause an error
3180 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3181 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3182 * fixed bug: if last continuation of a deleted block has a
3183 destination, this destination should be deleted too. (reported by
3185 * fixed a bug in the bootstrap process: the host compiler's values
3186 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3187 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3188 test case from Patrik Nordebo)
3189 * improved the ability of the disassembler on the PPC platform to
3190 provide helpful disassembly notes.
3191 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3192 the class in more cases than previously.
3193 * bug fix: FILE-POSITION now understands :START and :END for
3194 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3195 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3196 8) data. (thanks to David Lichteblau)
3197 * bug fix: it is possible to add a method to a generic function
3198 without lambda list.
3199 * bug fix: reader failed to signal END-OF-FILE inside an
3200 object representation. (reported by Nikodemus Siivola)
3201 * fixed some bugs revealed by Paul Dietz' test suite:
3202 ** LAST and [N]BUTLAST should accept a bignum.
3203 ** condition slot accessors are methods.
3204 ** (VECTOR NIL) is a subtype of STRING.
3206 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3207 * minor incompatible change: some nonsensical specialized lambda
3208 lists (used in DEFMETHOD) which were previously ignored now signal
3210 * minor incompatible change: the system is now aware of the types of
3211 variables in the COMMON-LISP package, and will signal errors for
3212 most violations of these type constraints (where previously they
3213 were silently accepted).
3214 * minor incompatible change: COMPILE-FILE now uses the freedom
3215 afforded (ANSI 3.2.2.3) to use derived function types for
3216 functions defined in the same file. This also permits the system
3217 to warn on static type mismatches and function
3218 redefinition. (Currently it does not work with high DEBUG level.)
3219 * minor incompatible change: VALUES declaration is disabled.
3220 * When issuing notes, the compiler now signals a condition of type
3221 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3222 restart for use in user handlers. It is expected that the
3223 COMPILER-NOTE condition will eventually become a condition
3224 supertype to a hierarchy of note types, which will then be
3225 handleable in a similar fashion. However, at the moment, no such
3226 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3227 but it's an implementation detail, not a classification for the
3229 * Changes in type checking closed the following bugs:
3230 ** type checking of unused values (192b, 194d, 203);
3231 ** template selection based on unsafe type assertions (192c, 236);
3232 ** type checking in branches (194bc).
3233 * A short form of VALUES type specifier has ANSI meaning (it has
3234 increased the number of situations when SBCL cannot perform type
3236 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3237 keywords or constants is permissible.
3238 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3239 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3240 classes. (thanks to Antonio Martinez)
3241 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3242 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3243 argument to a pretty-print stream. Instead, it leaves any such
3244 operations to DESCRIBE-OBJECT methods.
3245 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3247 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3248 lambda lists. (thanks to David Lichteblau)
3249 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3250 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3251 (thanks to Gerd Moellmann)
3252 * fixed bug 246: increased compilation speed of long
3253 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3255 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3257 * DEFCONSTANT now throws a condition of type
3258 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3259 constant to a non-EQL value; CONTINUE and ABORT restarts
3260 respectively change and preserve the value.
3261 * fixed bug 63: The code walker, part of the implementation of CLOS,
3262 is now better at handling symbol macros.
3263 * bug fix: There is no longer an internal implementation type named
3264 CL:LENGTH. (reported by Raymond Toy)
3265 * bug fix: In macro-like defining macros/special operators the
3266 implicit block does not enclose the lambda list.
3267 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3268 enclosed in parentheses) are not suitable as type specifiers, and
3269 their use properly signals an error now.
3270 * bug fix: An argument count mismatch for a type specifier in code
3271 being compiled no longer causes an unhandled error at compile
3272 time, but signals a compile-time warning.
3273 * fixed simple vector readable printing
3274 * bug fix: DESCRIBE takes more care over whether the class
3275 precedence list slot of a class is bound before accessing it.
3276 (reported by Markus Krummenacker)
3277 * bug fix: FORMATTER can successfully compile pretty-printer format
3278 strings which use variants of the ~* directive inside.
3279 * bug fix: SEARCH now applies its TEST predicate to the elements of
3280 the arguments in the correct order. (thanks to Wolfhard Buss)
3281 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3282 in type inference. (thanks to Robert E. Brown)
3283 * bug fix: Reading in symbols with an explicit package name of ""
3284 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3285 * fixed some bugs revealed by Paul Dietz' test suite:
3286 ** NIL is now allowed as a structure slot name.
3287 ** Arbitrary numbers, not just REALs, are allowed in certain
3288 circumstances in LOOP for-as-arithmetic clauses.
3289 ** Multiple class redefinitions before slot access no longer
3290 causes a type error.
3291 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3292 association between the name and a class.
3293 ** Generic functions with non-standard method-combination and over
3294 five methods all of which return constants no longer return NIL
3295 after the first few invocations. (thanks to Gerd Moellmann)
3296 ** CALL-NEXT-METHOD with no arguments now passes the original
3297 values of the arguments, even in the presence of assignment.
3298 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3300 ** DISASSEMBLE works with closures and funcallable instances.
3301 ** ADD-METHOD now returns the generic function, not the new method.
3302 ** FIND-METHOD signals an error if the lengths of the specializers
3303 is incompatible with the generic function, even if the ERRORP
3305 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3306 which its argument is a member.
3307 ** DEFCLASS only redefines the class named by its class-name
3308 argument if that name is the proper name of the class;
3309 otherwise, it creates a new class.
3310 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3311 of the UNBOUND-SLOT condition to the name of the slot.
3312 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3313 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3314 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3315 SLOT-MAKUNBOUND in the specified fashion.
3317 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3318 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3319 cross-compilation host. As a consequence, we can now bootstrap our
3320 way up to SBCL starting with a bare gcc toolchain and human-readable
3321 source code (first the source to CLISP, then the source to SBCL).
3322 * A contributed module containing a partial implementation of the
3323 simple-streams interface has been included. (thanks to Rudi
3325 * A contributed module implementing the RFC1321 Message Digest
3326 Algorithm, known as MD5, has been included.
3327 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3328 no longer has any effect, as the code controlled by this feature
3329 has been deleted. (As far as we know, no-one has ever built using
3330 this feature, and its semantics were confused in any case).
3331 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3332 work on conditions (as required by the ANSI specification),
3333 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3334 expected behaviour on conditions. Users should note, however,
3335 that such behaviour is not required by the ANSI specification,
3336 and so use of this behaviour may render their code unportable.
3337 * fixed some bugs revealed by Paul Dietz' test suite:
3338 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3340 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3341 ** SXHASH on condition objects no longer returns NIL.
3342 ** :ALLOCATION :CLASS slots are better treated; their values are
3343 updated on class redefinition, and initforms inherited from
3344 superclasses are applied.
3345 ** REMOVE-METHOD returns its generic function argument even when
3346 no method was removed.
3347 ** SHARED-INITIALIZE now initializes the values of the requested
3348 slots, including those with :ALLOCATION :CLASS.
3349 ** ALLOCATE-INSTANCE now works on structure classes defined via
3350 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3352 ** SLOT-EXISTS-P now works on conditions, as well as structures
3354 ** MAKE-LOAD-FORM now has the required methods on
3355 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3356 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3357 keyword, and now implements the SLOT-NAMES argument.
3358 ** methods with &OPTIONAL arguments no longer allow too many
3359 arguments to be passed in the call without error.
3360 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3361 option is consistent with the required arguments of the generic
3362 function lambda list.
3363 * bug fix: REQUIRE accepts a string designator. (Thanks to
3365 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3366 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3367 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3368 * bug fix: The compiler now checks for duplicated variables in macro
3370 * bug fix: SETQ on globals returns the correct value.
3371 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3372 not a COMPILER-ERROR (followed by some other strange error on
3373 choosing the CONTINUE restart).
3374 * bug fix: make.sh and friends are now more consistent in the way that
3375 they look for GNU "make".
3377 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3378 * experimental native threads support (on x86 Linux >=2.4 only).
3379 This is not compiled in by default: you need to add :SB-THREAD to
3380 the target features. See the "Beyond ANSI" chapter of the manual
3382 * fix for longstanding nonANSIism: The old distinction between
3383 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3384 The return value from CL:FIND-CLASS is now a CLOS class, and
3385 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3386 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3387 * An interface to the MetaObject Protocol, as described in Kiczales,
3388 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3389 MIT Press, 1991, is available from the SB-MOP package.
3390 * incompatible change: the SB-PCL package should now be considered
3391 a private implementation detail, and no longer a semi-private MOP
3393 * minor incompatible change: due to rearrangement for threads, the
3394 control stack and binding stack are now allocated at arbitrary
3395 addresses instead of being hardcoded per-port. Users affected by
3396 this probably have to be doing advanced things with shared
3397 libraries, and will know who they are.
3398 * minor incompatible change: Previously, all --eval forms used were
3399 processed with READ before any of them were processed with EVAL.
3400 Now each --eval form is processed with both READ and EVAL before
3401 the next --eval form is processed. (Thus package operations like
3402 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3403 work as the user might reasonably expect.)
3404 * minor incompatible change: *STANDARD-INPUT* is now only an
3405 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3407 * minor incompatible change: Y-OR-N-P is now character-oriented,
3408 not line oriented. Also, YES-OR-NO-P now works without errors.
3409 (thanks to Antonio Martinez)
3410 * sb-aclrepl module improvements: an integrated inspector, added
3411 repl features, and a bug fix to :trace command.
3412 * Known functions, which cannot be open coded by the backend, are
3413 considered to be able to check types of their arguments. (fixing
3414 a bug report by Nathan J. Froyd)
3415 * fixed a bug in computing method discriminating functions: It is
3416 now possible to define methods specialized on classes which have
3417 forward-referenced superclasses. (thanks to Gerd Moellmann)
3418 * fixed evaluation order in optional entries (reported by Gilbert
3420 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3421 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3422 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3423 required-by-AMOP NAME argument, as well as CLASS and
3424 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3425 * fixed bug 20: DEFMETHOD can define methods using names that are
3426 not the proper names of classes to designate class specializers.
3427 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3428 Unix file descriptors, instead of blowing up. (thanks to Antonio
3430 * Garbage collection refactoring: user-visible change is that a
3431 call to the GC function during WITHOUT-GCING will not do garbage
3432 collection until the end of the WITHOUT-GCING. If you were doing
3433 this you were probably losing anyway.
3434 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3435 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3436 TYPEP the latter but not the former.
3437 * The compiler issues a full WARNING for calls to undefined functions
3438 with names from the CL package.
3439 * MAP-INTO for a vector destination is open coded. (reported by
3440 Brian Downing on c.l.l)
3441 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3442 documentation string.
3443 * fixed some bugs revealed by Paul Dietz' test suite:
3444 ** COPY-ALIST now signals an error if its argument is a dotted
3446 ** Condition slots are now accessed more correctly in the presence
3447 of multiple initargs for a given slot.
3448 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3449 exclude restarts of the same name associated with a different
3451 ** DEFCLASS of forward-referenced classes with another
3452 forward-referenced class in the superclasses list no longer
3454 ** Condition slots are now initialized once each, not multiple
3455 times. (thanks to Gerd Moellmann)
3456 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3458 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3459 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3460 arguments contain duplicated elements.
3461 ** RESTART-CASE understands local macros.
3462 ** RESTART-CASE associates exactly its own restarts with a condition.
3463 ** ENDP in safe mode checks its argument to be of type LIST.
3464 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3465 in question is unbound.
3466 ** Optimized MAKE-INSTANCE functions no longer cause internal
3467 assertion failures in the presence of duplicate initargs.
3468 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3469 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3470 propagated even in the ctor optimized implementation of
3472 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3474 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3475 * a better implementation of SXHASH on (simple) bit vectors,
3476 measured both in execution speed and in distribution of results
3477 over the positive fixnums, has been installed. Likewise, a better
3478 implementation of EQUAL for simple bit vectors is now available.
3479 * fixed CEILING optimization for a divisor of form 2^k.
3480 * fixed bug 240 (emitting extra style warnings "using the lexical
3481 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3482 by Antonio Martinez)
3483 * fixed SXHASH, giving different results for NIL depending on type
3484 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3485 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3486 variable. (found by Rolf Wester)
3487 * SBCL does not ignore type declarations for special
3488 variables. (reported by rif on c.l.l 2003-03-05)
3489 * some bug fixes in contrib/sb-aclrepl/
3490 * fixed some bugs revealed by Paul Dietz' test suite:
3491 ** a bug in the CONS type specifier, whereby the CAR and CDR
3492 types got intertwined, has been fixed;
3493 ** the type system is now able to reason about the interaction
3494 between INTEGER and RATIO types more completely;
3495 ** APPEND, [N]REVERSE and NRECONC check that those their
3496 arguments, which must be proper lists, are really so;
3497 ** An array specialized to be unable to hold elements has been
3498 implemented, as required -- yes, really -- by ANSI;
3499 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3500 on malformed property lists;
3502 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3503 * incompatible packaging change: in line with Unix convention,
3504 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3505 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3506 * REQUIRE and PROVIDE are now optionally capable of doing something
3507 useful. See the documentation string for REQUIRE.
3508 * infrastructure for a managed SBCL contrib system: contributed
3509 modules in this release include:
3510 ** the ASDF system definition facility;
3511 ** an interface to the BSD Sockets API;
3512 ** an ACL-like convenience interface to the repl;
3513 (thanks to Kevin Rosenberg)
3514 ** an implementation of ROTATE-BYTE, with efficient implementation
3516 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3517 gives a closer approximation to the right answer than previously.
3518 (thanks to Raymond Toy)
3519 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3520 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3521 optional environment arguments, as required by ANSI.
3522 * fixed bugs in other functions taking environment objects, allowing
3523 calls with an explicit NIL environment argument to be compiled
3525 * fixed bug 228: primary return values from
3526 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3527 COMPILE or FUNCTION.
3528 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3529 :TYPE will no longer signal errors on innocuous objects.
3530 * fixed bug 231b: SETQ is better at respecting type declarations in
3531 the lexical environment.
3532 * fixed a bug in DEFCLASS: classes named by symbols with no or
3533 unprintable packages can now be defined.
3534 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3535 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3536 * fixed compiler failure related to checking types of functions.
3537 (reported by Robert E. Brown)
3538 * the compiler is now much more consistent in its error-checking
3539 treatment of bounding index arguments to sequence functions: in
3540 (SAFETY 3) code, errors will be signalled in almost all cases if
3541 invalid sequence bounding indices are passed to functions defined
3542 by ANSI to operate on sequences.
3543 * fixed a bug in the build procedure: documentation of SBCL-specific
3544 packages is now preserved and available in the final Lisp image.
3545 * lifted FDEFINITION lookup out of loops in the implementation of
3546 many list operations. (thanks to Robert E. Brown)
3547 * fixed a bug in the reader: the #n# reader macro now works for
3548 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3549 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3550 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3551 E. Brown and Tony Martinez respectively)
3552 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3553 SLOT-BOUNDP: in particular, it is now able to optimize them much
3554 better, and is now not vulnerable to having packages renamed.
3555 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3556 always called when a slot is not present in an instance. (thanks
3558 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3559 longer signal an error on unreading a character following EOF on
3560 the previous constituent stream. (thanks to Tony Martinez)
3561 * fixed some bugs revealed by Paul Dietz' test suite:
3562 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3563 not just nonnegative fixnums;
3564 ** the logical bit-array operators such as BIT-AND now accept an
3565 explicit NIL for their "opt-arg" argument (to indicate a
3566 freshly-consed result bit-array);
3567 ** ELT now signals an error on an invalid sequence index in safe
3569 ** the type system is now cleverer about negations of numeric
3570 types, and consequently understands the BIGNUM and RATIO types
3572 ** the type system is now cleverer about the interaction between
3573 INTEGER and RATIO types: while bugs still remain, many more
3574 cases are accurately computed;
3575 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3576 if it is in the last clause;
3577 ** CONSTANTLY now correctly returns a side-effect-free function in
3579 ** DECLARE is no longer treated as a special-operator; in
3580 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3581 * incremented fasl file version number due to the change in the
3582 DEFSTRUCT-SLOT-DESCRIPTION structure.
3584 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3585 * minor incompatible change: code processed by the "interpreter" or
3586 EVAL now has a compilation optimization policy of (DEBUG 2)
3587 (changed from (DEBUG 1)) to improve debuggability of interactive
3588 development, and to allow the use of the debug RETURN command in
3590 * an experimental implementation of the RETURN command for the
3591 debugger has been included. (thanks to Frederik Kuivinen)
3592 * fixed bug 62: constraints were not propagated into a loop.
3593 * fixed bug in embedded calls of SORT (reported and investigated by
3595 * fixed some bugs revealed by Paul Dietz' test suite:
3596 ** printing and reading of arrays with some dimensions having
3597 length 0 (thanks to Gerd Moellmann);
3598 ** BOA constructor with &AUX argument without a default value does
3599 not cause a type error;
3600 ** CONSTANTP now returns true for all self-evaluating objects.
3602 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3603 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3604 accessors that are related by inheritance, as specified in the
3605 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3606 Valtteri Vuorikoski)
3607 * The compiler is now able to inline functions that were defined in
3608 a complex lexical environment (e.g. inside a MACROLET).
3609 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3611 * Support for the upcoming FreeBSD-5.0 release has been included.
3612 (thanks to Dag-Erling Smorgrav)
3613 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3614 effect when it is not in a toplevel context.
3615 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3616 stand a better chance of being correct. (thanks to Gerd
3618 * fixed bug in COERCE, which now signals an error on coercing a
3619 rational to a bounded real type which excludes the expected
3621 * The compiler is now able to derive types more accurately from the
3622 COERCE and COMPILE functions.
3623 * fixed bug 223: functional binding is considered to be constant
3624 only for symbols in the CL package.
3625 * fixed bug 231: SETQ did not check the type of a variable being set
3626 (reported by Robert E. Brown)
3627 * A new optimization for MAKE-INSTANCE has been included, fixing
3628 various bugs (including relating to :ALLOCATION :CLASS slots and
3629 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3631 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3632 ** As required by ANSI, LOOP now disallows anonymous collection
3633 clauses such as COLLECT I in conjunction with aggregate boolean
3634 clauses such as THEREIS (= I 1);
3635 ** LOOP now signals an error when any variable is reused in the
3636 same loop (including the potentially useful construct analogous
3637 to WITH A = 1 WITH A = (1+ A);
3638 ** IT is only a special loop symbol within the first clause of a
3639 conditional loop clause;
3640 ** LOOP with a typed iteration variable over a hashtable now
3641 signals a type error iff it should.
3642 * fixed some other bugs revealed by Paul Dietz' test suite:
3643 ** FILE-STREAM now names the class previously known as FD-STREAM;
3644 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3645 argument) no longer signals an error;
3646 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3647 of the slot symbol, rather than using the current package
3648 ((:CONC-NAME "") continues to intern the slot's name in the
3650 * incremented fasl file version number, because of the incompatible
3651 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3652 of the new implementation of DEFINE-COMPILER-MACRO.
3654 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3655 * Support for building SBCL for MIPS platforms running in
3656 little-endian mode has now been checked in, and basic
3657 functionality on said platforms verified.
3658 * minor incompatible change: PCL now records the pathname of a file
3659 in which methods and the like are defined, rather than its
3661 * minor incompatible change: TRUENAME now considers the truename of
3662 a file naming a directory to be the pathname with :DIRECTORY
3663 component indicating that directory.
3664 * minor incompatible change: a NAMED clause in the extended form of
3665 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3666 reason for the previous behaviour is unclear.
3667 * more systematization and improvement of CLOS and MOP conformance
3668 in PCL (thanks to Gerd Moellman and Pierre Mai):
3669 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3671 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3672 primary methods with no specializers;
3673 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3675 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3676 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3677 FORWARD-REFERENCED-CLASSes; error reporting on
3678 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3680 ** SXHASH on CLOS instances now uses a slot internal to the
3681 instance to return different numbers on distinct instances,
3682 while preserving the same return value through invocations of
3684 ** DEFMETHOD signals errors when methods with longer incongruent
3685 lambda lists are added to generic functions;
3686 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3687 CLASS, as specified in AMOP;
3688 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3689 on the order returned by the primary method for classes of
3690 class STANDARD-CLASS;
3691 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3692 * fixed some bugs shown by Paul Dietz' test suite:
3693 ** DOLIST puts its body in TAGBODY;
3694 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3696 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3697 value producing form;
3698 ** if more variables are given to PROGV than values, extra
3699 variables are bound and made to have no value;
3700 ** NSUBSTITUTE on list arguments gets the right answer with
3702 ** ELT signals an error of type TYPE-ERROR when the index argument
3703 is not a valid sequence index;
3704 ** LOOP signals (at macroexpansion time) an error of type
3705 PROGRAM-ERROR when duplicate variable names are found;
3706 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3707 ** FUNCALL of special-operators now cause an error of type
3709 ** PSETQ now works as required in the presence of side-effecting
3710 symbol-macro places;
3711 ** NCONC accepts any object as its last argument;
3712 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3714 ** loop-for-as-package does not require a package to be explicitely
3716 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3718 * fixed bug 166: compiler preserves "there is a way to go"
3719 invariant when deleting code.
3720 * fixed bug 172: macro lambda lists with required arguments after
3721 &REST arguments now cause an error to be signalled. (thanks to
3723 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3724 bug. (thanks to Matthew Danish)
3725 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3727 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3728 arguments are lexically rebound. (thanks to Gerd Moellmann and
3730 * fixed bug 194: error messages are now more informative when there
3731 is no primary method applicable in a call to a generic
3732 function. (thanks to Gerd Moellmann)
3733 * fixed bug in command line argument checking (thanks to Julian
3735 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3736 (thanks to Matthew Danish)
3737 * incremented fasl file version number, because of the
3738 SXHASH-related changes in the layout of CLOS data structures
3740 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3741 * minor incompatible change: The runtime (the Unix executable named
3742 "sbcl") is now much pickier about the .core files it will load.
3743 Essentially it now requires .core files to descend from the same
3744 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3745 as the runtime does. (The intent is to prevent the crashes which
3746 can occur, and which can even be reported as mysterious failures,
3747 when people patch the sources or change the build parameters
3748 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3749 sbcl and .core files.)
3750 * fixed bug: VALUES-LIST is no longer optimized away.
3751 * fixed bug 142: The FFI conversion of C string values to Lisp
3752 string values no longer conses excessively. (thanks to Nathan
3753 Froyd porting Raymond Toy's fix to CMU CL)
3754 * began to systematize and improve MOP conformance in PCL (thanks to
3755 Nathan Froyd, Gerd Moellman and Pierre Mai):
3756 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3758 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3759 ** FINALIZE-INHERITANCE is now called on class finalization;
3760 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3761 argument precedence order.
3762 * fixed bug 202: The compiler no longer fails on functions whose
3763 derived types contradict their declared type.
3764 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3765 so it can be non-toplevel.
3766 * The fasl file version number has changed (because of the new
3767 implementation of DEFMACRO).
3768 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3769 safe code, that any length requirement by their type-specifier
3770 argument is valid. The exceptions to this are described in bug
3772 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3773 functions now check the types of their inputs as required by ANSI.
3774 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3775 introduced symbol is DECLAREd to be SPECIAL.
3776 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3777 * fixed bug: :COUNT argument to sequence functions may be negative
3778 * fixed bug: body of DO-SYMBOLS may contain declarations
3779 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3780 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3781 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3782 symbol macro only once
3783 * fixed printing of call frame when argument list is unavailable
3784 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3785 * compiler no longer signals WARNING on unknown keyword
3788 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3789 * A beta-quality port to the mips architecture running Linux,
3790 based on the old CMUCL backend, has been made. It has been tested
3791 on a big-endian kernel, and works sufficiently well to be able to
3792 rebuild itself; it has not been tested in little-endian mode.
3793 * fixed an inconsistency between gencgc.c and purify.c which made
3794 dumping/loading .core files unreliable
3795 * fixed bug 120a: The compiler now deals correctly with IFs where
3796 the consequent is the same as the alternative, instead of
3797 misderiving the return type. (thanks to Alexey Dejneka)
3798 * fixed bug 113: Logical pathnames are now dumpable (the logical
3799 host is resolved at load-time, throwing an error if it is not
3801 * fixed bug 174: FORMAT's error message is slightly clearer when a
3802 non-printing character is used in a format directive.
3803 * fixed several bugs in compiler checking of type declarations, i.e.
3804 violations of the Python "declarations are assertions" principle
3805 (thanks to Alexey Dejneka)
3806 * fixed several bugs in PCL's error checking (thanks to Gerd
3808 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3810 * fixed bug in compilation of functions as first class values
3811 (thanks to Antonio Martinez-Shotton)
3812 * The compiler's handling TYPE-ERRORs which it can prove will
3813 inevitably happen at runtime has been cleaned up and corrected
3814 in several ways. (thanks to Alexey Dejneka)
3815 * improved argument type checking for various basic arithmetic
3816 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3817 lavished on them in the past that they can be compiled in many
3818 ways in different special cases
3819 * fixed bug 181: compiler checks validity of user supplied type
3821 * cleaned up code flushing in optimization: Function calls which
3822 should signal errors for safety purposes (e.g. which ANSI says
3823 should signal errors when their arguments are of incorrect type)
3824 are no longer optimized away.
3825 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3826 * incremented fasl file version number, because changes in the
3827 implementation of sequence functions like COERCE caused
3828 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3829 in old inline expansions) to become undefined. (Actually these
3830 changes were later undone, so we might very well be binary
3831 compatible with 0.7.7 after all, but leaving the version number
3832 incremented seemed like the simplest and most conservative
3835 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3836 * An alpha-quality port to the parisc architecture running Linux,
3837 based on the old CMUCL backend, has been made. This, even more so
3838 than the other backends, should be considered still a work in
3839 progress; known problems include that the Linux kernel in 64-bit
3840 mode does not propagate the correct sigcontext structure to
3841 userspace, and consequently SBCL on a parisc64 kernel will not
3843 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3844 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3845 suppressed.) Also now that the compiler is looking at declarations
3846 in the environment, it checks optimization declarations as well,
3847 and suppresses inlining when (> DEBUG SPEED).
3848 * More fixes have been made to treatment of floating point exception
3849 treatment and other Unix signals. In particular, floating point
3850 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3851 * The detection and handling of control stack exhaustion (infinite
3852 or very deeply nested recursion) has changed. Stack exhaustion
3853 detection is now done by write-protecting pages at the OS level
3854 and applies at all optimization settings; when found, a
3855 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3856 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3857 be caught using IGNORE-ERRORS.
3858 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3859 correctly when transforming recursive LABELS functions to LETs.
3860 (thanks to Alexey Dejneka)
3861 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3862 that are names of constants or global variables.
3863 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3864 alien routines with docstrings.
3865 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3866 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3868 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3869 (thanks to Eric Marsden)
3870 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3871 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3872 * Bug fix: The SPARC backend can now compile functions involving
3873 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3874 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3875 to a functions expecting a PATHNAME-DESIGNATOR.
3876 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3877 lambda lists. (thanks to Alexey Dejneka)
3878 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3879 functions so that EXPORTing the name of the class doesn't cause
3880 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3881 to Antonio Martinez for reporting this)
3882 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3883 constant-folded without failing an assertion. (thanks to Einar
3884 Floystad Dorum for reporting this)
3885 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3886 types are now dealt with more correctly by the compiler.
3887 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3888 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3889 behaviour analogous to RENAME-FILE. This puts its behaviour more
3890 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3892 * The fasl file version number has changed again. (because of the
3893 bug fix involving the names of PCL MAKE-INSTANCE functions)
3895 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3896 * bug fix: Floating point exceptions are treated much more
3897 consistently on the x86/Linux and PPC/Linux platforms.
3898 * Array initialization with :INITIAL-ELEMENT is now much faster for
3899 cases when the compiler cannot open code the array creation, but
3900 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3901 array accesses have also seen a speed increase.
3902 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3903 specified. (This was at the root of some bad interactions between
3904 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3906 * bug fix: Internal error arguments for undefined functions are now
3907 computed correctly on the PPC/Linux platform.
3908 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3909 Raymond Toy's patch for CMU CL)
3910 * Support for the Solaris 9 operating environment has been included
3911 (thanks to Daniel Merritt)
3912 * A very ugly but hopefully complete draft of the missing FFI chapter
3913 of the manual has been created by reformatting the corresponding
3914 CMU CL manual chapter into (currently very ugly and incoherent)
3915 DocBook and bringing it up to date for SBCL behavior. Thus, the
3916 manual is now essentially complete, at least by my extreme
3917 once-and-only-once standards, whereby it's acceptable to refer to
3918 the doc strings of SB-EXT functions as the primary documentation.
3919 * The fasl file version number has changed again, due to cleanup of
3920 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3921 is no longer a static symbol.)
3923 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3924 * SBCL now builds with OpenMCL (version 0.12) as the
3925 cross-compilation host; also, more progress has been made toward
3926 bootstrapping under CLISP.
3927 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3929 * bug 158 fixed: The compiler can now deal with integer loop
3930 increments different from 1; fixing this turned out also to fix
3932 * bug 169 fixed: no more bogus warnings about using lexical bindings
3933 despite the presence of perfectly good SPECIAL declarations (thanks
3934 to David Lichteblau)
3935 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3936 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3937 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3939 * bug 180 fixed: Method combination specifications no longer ignore
3940 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3941 * bug fix: Structure type predicate functions now check their argument
3942 count as they should.
3943 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3944 correctly. (thanks to Pierre Mai)
3945 * minor incompatible change: The --noprogrammer option is deprecated
3946 in favor of the new --disable-debugger option, which is very similar.
3947 (The major difference is that it takes effect at a slightly different
3948 time at startup, causing handling of errors in --sysinit and
3949 --userinit files will be affected differently.) The
3950 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3951 been added to allow this functionality to be controlled from ordinary
3952 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3953 maintainers, who might want to run non-interactive scripts to
3954 build SBCL cores which will later be used interactively.)
3955 * minor incompatible change: The LOAD function no longer, when given
3956 a wild pathname to load, loads all files matching that pathname.
3957 Instead, an error of type FILE-ERROR is signalled.
3959 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3960 * bug 147 fixed: The compiler preserves its block link/count
3961 invariants more correctly now so that it doesn't crash. (thanks
3963 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3965 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3966 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3967 no-value-for-this-character as NIL (as specified by ANSI).
3968 * HOST-NAMESTRING on physical pathnames now returns a string that is
3969 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3970 (thanks to Christophe Rhodes)
3971 * The Alpha port handles icache flushing more correctly. (thanks to
3973 * More progress has been made toward bootstrapping under CLISP. (thanks
3974 to Christophe Rhodes)
3975 * The fasl file format has changed again, because dynamic loading
3976 on OpenBSD (which has non-ELF object files) motivated some cleanups
3977 in the way that foreign symbols are transformed and passed around.
3978 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3979 is no longer treated as whitespace by the reader, but instead as
3980 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3981 |A
\7fB|, instead of A as it used to.
3983 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3984 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3985 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3986 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3987 this as of 0.7.1.45, but was left out of the previous news section
3988 (thanks to Dan Barlow)
3989 * SBCL now runs on the Solaris operating system on SPARC architectures
3990 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3991 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3992 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3993 * SPARC backend cleanups, allowing builds of cores optimized for V8
3994 and V9 SPARCS, and also emission of code targeted to a particular
3995 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3997 * SBCL is closer to bootstrapping under CLISP, thanks to various
3998 fixes by Christophe Rhodes.
3999 * The fasl file format has changed again, to allow the compiler's
4000 INFO database to support symbol macros.
4001 * The user manual (in doc/) is formatted into HTML more nicely.
4002 (thanks to coreythomas)
4003 * The system is smarter about SUBTYPEP relationships, especially
4004 those involving NOT types (including types such as ATOM which are
4005 represented internally using NOT types). Thus SUBTYPEP is less
4006 likely to return (VALUES NIL NIL) in general, and in particular
4007 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4009 * The fasl file format has changed again, because the internal
4010 representation of types now includes a new slot to support the new
4011 SUBTYPEP-of-NOT-types logic.
4012 * (not a change in the main branch of SBCL, but a related prototype
4013 which can hopefully be merged into the main branch of SBCL in the
4014 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4015 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4016 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4017 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4018 dispatch macro character combinations. (thanks to Alexey Dejneka)
4019 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4020 that superficially look like logical namestrings correctly.
4021 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4023 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4024 default supplied-p) for &optional and &key arguments. (thanks to
4027 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4028 * incompatible change: The compiler is now less aggressive about
4029 tail call optimization, doing it only when (> SPACE DEBUG) or
4030 (> SPEED DEBUG). (This is an incompatible change because there are
4031 programs which relied on the old CMU-CL-style behavior to optimize
4032 away their unbounded recursion which will now die of stack overflow.)
4033 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4034 for non-GENCGC systems has been increased to 20M (since that
4035 seems much closer to the likely performance optimum for modern
4036 systems than the old 4M value was)
4037 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4038 and *DEBUG-PRINT-LEVEL*
4039 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4040 of CMU CL's support for SPARC, and various endianness and other
4041 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4042 * new syntactic sugar for the Unix command line: --load foo.bar is now
4043 an alternate notation for --eval '(load "foo.bar")'.
4045 ** The system now detects stack overflow and handles it gracefully,
4046 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4047 optimization settings. (This is a good thing in general, and
4048 its introduction in this version should be particularly timely
4049 for anyone whose code fails because of suppression of tail
4051 ** The system now hunts for the C variable "environ" in a more
4052 devious way, to avoid segfaults when the C library version differs
4053 between compile time and run time. (thanks to Christophe Rhodes)
4054 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4055 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4056 ** The compiler no longer issues bogus style warnings for undefined
4057 classes in the same source file as the DEFCLASSes which defined
4058 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4060 ** fixes in CONDITION class precedence list for undefined function
4061 errors (thanks to Alexei Dejneka)
4062 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4063 correctly. (thanks to Dan Barlow)
4064 ** portability fixes aiming at bootstrapping under CLISP (thanks
4065 to Dave McDonald and Christophe Rhodes)
4066 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4067 ** fixes in type translation and and type inference (thanks to
4069 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4070 ** various fixes in the new ports (thanks to Dan Barlow)
4071 * several changes related to debugging:
4072 ** suppression of tail recursion, as noted above
4073 ** stack overflow detection, as noted above
4074 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4075 is now the default. (For some time encapsulation has been more
4076 reliable than the breakpoint-based :ENCAPSULATE NIL
4077 implementation, at least on X86 systems; and I just noticed that
4078 encapsulation also seems closer to the spirit of the ANSI
4081 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4083 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4084 up properly again. (There was a packaging bug in 0.7.0 which
4085 left their definitions in SB-SYS::LOAD-FOREIGN and
4086 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4087 vital for most things which interface to C-level interfaces,
4088 like extensions working with sockets or databases or
4089 Perl-compatible regexes or whatever, and the need to fix
4090 this bug is the main reason that 0.7.1 was released so
4092 ** DEFGENERIC is now choosier about the methods it redefines, so that
4093 reLOADing a previously-LOADed file containing DEFGENERICs does
4094 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4095 cycle now works as it should. (thanks to Alexey Dejneka)
4096 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4097 fixed by Christophe Rhodes. (He actually submitted this patch
4098 months ago, and I delayed until after 0.7.0.)
4099 ** Bug 111 (internal compiler confusion about runtime checks on
4100 FUNCTION types) was fixed by Alexey Dejneka.
4101 * Some internal cleanups (getting rid of variables which aren't
4102 needed now that the byte interpreter is gone) caused the fasl
4103 file format number to change again.
4105 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4106 * major incompatible change: The default fasl file extension, i.e. the
4107 default extension for files produced by COMPILE-FILE, has changed
4108 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4110 ** There are many changes in the implementation of the compiler.
4111 SBCL is now essentially a compiler-only implementation of ANSI
4112 Common Lisp. EVAL still "interprets" a few special cases, but
4113 almost all the interesting cases are handled by creating
4114 a LAMBDA expression, calling COMPILE on it, then calling
4115 FUNCALL on the result.
4116 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4117 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4118 Since the code is newer, there might still be some new bugs
4119 (though not as many as before Martin Atzmueller's fixes:-). But
4120 the new code is substantially simpler and clearer, and hopefully
4121 any remaining bugs will be simpler, less fundamental, and more
4122 fixable then the bugs in the old code.
4123 ** The revised compiler is still a little unsteady on its feet.
4125 *** The debugging information it produces (particularly the names
4126 of FUNCTION objects) is sometimes much less useful than what
4127 the old compiler produced.
4128 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4129 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4130 within a MACROLET) has been temporarily weakened.
4131 ** There are new compiler optimizations for various functions:
4132 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4133 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4134 *** the math functions TRUNCATE, FLOOR, and CEILING
4135 *** the function-of-all-trades COERCE
4136 Mostly these should be transparent, but there's one
4137 potentially-annoying problem (bug 117): when the compiler
4138 inline-expands a function and does type analysis on the result,
4139 it can create control paths which have type mismatches, and
4140 when it can't prove that those control paths aren't taken,
4141 it will issue WARNINGs about the type mismatches. This is
4142 a particular problem in practice for the new sequence functions.
4143 It's not clear how this should be fixed, and for now, a
4144 workaround is given in the entry for 117 in the BUGS file.
4145 ** (Because of the interaction between the two previous items --
4146 occasional inlining problems and new inline expansions -- some
4147 of the new sequence function optimizations won't really kick in
4148 completely until debugging information, and then inlining, are
4149 straightened out in some future version.)
4150 * minor incompatible changes:
4151 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4152 DIRECTORY no longer implicitly promotes NIL slots of its
4153 pathname argument to :WILD. In particular, when you ask for the
4154 contents of a directory (which you used to be able to do without
4155 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4156 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4157 ** changes in behavior that ANSI explicitly defines to be
4158 implementation dependent:
4159 *** The new compiler-only implementation still conforms with ANSI,
4160 but acts a little different than before. Besides the obvious
4161 changes in performance tradeoffs (that the cost per form passed
4162 to EVAL has gone up, and the cost per form executed by EVAL
4163 has gone down), the behavior of the system changes a little
4164 because there are no longer any interpreted function objects.
4165 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4166 e.g. doing COMPILE on the output of interactive DEFUN is
4168 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4170 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4171 has changed from (:CL) to NIL.
4172 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4173 e.g. #\Newline and #\Tab, don't have names specified in the
4174 ANSI Common Lisp standard, is now based on their ASCII symbolic
4175 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4176 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4177 are no longer used for output.
4178 ** changes in internal implementation constants:
4179 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4180 4 million. (If your application spends a lot of time GCing and
4181 you have a lot of RAM, you might want to experiment with
4182 increasing it even more.)
4183 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4184 However, almost all old code should still continue to work without
4185 immediate update, as SB-C-CALL is now a (deprecated) nickname
4187 ** Old operator names in the style DEF-FOO are now deprecated in
4188 favor of new corresponding names DEFINE-FOO, for consistency with
4189 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4190 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4191 internal symbols, but a few supported extensions like
4192 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4193 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4194 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4195 etc. as you get deeper into recursive calls to the debugger
4196 command loop, instead of the old "5]", "5]]", "5]]]"
4197 sequence. (I was motivated to do this when squabbles between
4198 ILISP and SBCL left me very deeply nested in the debugger. In the
4199 short term, this change will probably provoke more ILISP/SBCL
4200 squabbles, but hopefully it will be an improvement in the long run.)
4201 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4202 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4203 function, since declaiming return types involving aliens is
4204 (1) annoyingly messy to do by hand and (2) vital to efficient
4205 compilation of code which calls such functions.
4206 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4207 longer reexported by the SB-EXT package. They're solely useful
4208 for alien code, so it seems more logical that you should get
4209 them from the SB-ALIEN package, not in SB-EXT.
4210 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4211 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4212 features. Instead, the code that they used to control is always
4213 built into the system.
4214 * many other bug fixes
4215 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4216 advantage of the new EVAL-WHEN stuff and to clean them up in
4217 general, and they are now more ANSI-compliant in a number of
4218 ways. Martin Atzmueller is responsible for a lot of this.
4219 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4221 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4222 *** correct ERROR type for various file operations
4223 *** some fixes for Lisp streams
4224 *** DEFMETHOD syntax checking
4225 *** changing old weird representation of debug information as
4226 strings (which, among their other deficiencies, don't transform
4227 correctly when you rename packages, and don't change their
4228 print representation when you change things like *PACKAGE*
4229 and *PRINT-LENGTH*) to symbols and lists of symbols
4230 He also made several improvements and fixed several bugs in DESCRIBE.
4231 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4233 *** misbehavior of WRITE-STRING/WRITE-LINE
4234 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4235 and several other LOOP problems as well
4236 *** DIRECTORY when similar filenames are present
4237 *** DEFGENERIC with :METHOD options
4238 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4239 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4240 *** argument ordering in FIND with :TEST option
4241 *** mishandled package designator argument in APROPOS-LIST
4242 *** various problems in the backquote readmacro
4243 *** a bug in APROPOS
4244 *** probably some others that I'm not describing very well here,
4245 since the CVS log documents them by reference to sbcl-devel
4246 messages, and the SourceForge archives aren't working well.:-(
4247 ** Dan Barlow improved the Alpha port (and is making progress on the
4248 PPC port, for those of you who think different).
4249 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4250 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4251 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4252 continues to work on the SPARC port (for those of you in a position
4253 to look down upon our little PC-compatible boxes from a great height).
4254 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4255 and uses that copy, rather than the current dynamic value, when
4256 it's trying to decide whether to truncate output. Thus e.g.
4257 (let ((*print-lines* 50))
4258 (pprint-logical-block (stream nil)
4260 (let ((*print-lines* 8))
4261 (print (aref possiblybigthings i) stream)))))
4262 should now truncate the logical block only at 50 lines, instead of
4263 often truncating it at 8 lines, as it did before.
4264 * The doc/cmucl/ directory, containing old CMU CL documentation
4265 from the time of the fork, is no longer part of the base system.
4266 SourceForge has shut down its anonymous FTP service, and with it
4267 my original plan for distributing the old CMU CL documentation
4268 there. For now, if you need these files you can download an old
4269 SBCL source release and extract them from it.
4270 * The fasl file version number changed again, for dozens of reasons,
4271 some of which are apparent above.
4273 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4274 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4275 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4276 patch, so that DISASSEMBLE output is much nicer.
4277 * The code in the SB-PROFILE package now seems reasonably stable.
4278 I still haven't decided what the final interface should look like
4279 (I'd like PROFILE to interact cleanly with TRACE, since both
4280 facilities use function encapsulation) but if you have a need
4281 for profiling now, you can probably use it successfully with
4282 the current CMU-CL-style interface.
4283 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4284 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4285 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4286 link now returns the dangling link itself, and for similar
4287 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4288 link itself. (In these cases the old code signalled an error and
4289 looped endlessly, respectively.) Thus, DIRECTORY now works even
4290 in the presence of dangling and cyclic symbolic links.
4291 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4292 is now a supported extension again, since the consensus on
4293 sbcl-devel was that it can be useful for ordinary development
4294 work, not just for debugging SBCL itself.
4295 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4296 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4297 that currently-defined functions might be redefined later with
4298 different return types.
4299 * Hash tables can be printed readably, as inspired by CMU CL code
4300 of Eric Marsden and SBCL code of Martin Atzmueller.
4301 * better error handling in CLOS method combination, thanks to
4302 Martin Atzmueller porting Pierre Mai's CMU CL patches
4303 * more overflow fixes for >16Mbyte I/O buffers
4304 * A bug in READ has been fixed, so that now a single Ctrl-D
4305 character suffices to cause end-of-file on character streams.
4306 In particular, now you only need one Ctrl-D at the command
4307 line (not two) to exit SBCL.
4308 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4310 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4311 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4312 does the right thing.
4313 * The implementation of some type tests, especially for CONDITION
4314 types, is now tidier and maybe faster, due to CMU CL code
4315 originally by Douglas Crosher, ported by Martin Atzmueller.
4316 * Some math functions have been fixed, and there are new
4317 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4318 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4319 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4320 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4321 likely to be built into sbcl-0.7.0.)
4322 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4323 CL, ported to SBCL by Martin Atzmueller)
4324 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4325 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4326 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4327 * There's a new slam.sh hack to shorten the edit/compile/debug
4328 cycle for low-level changes to SBCL itself, and a new
4329 :SB-AFTER-XC-CORE target feature to control the generation of
4330 the after-xc.core file needed by slam.sh.
4331 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4332 COMPILE-FILE is no longer supported, so that now every function
4333 gets an entry point, so that block compilation looks a little
4334 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4335 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4336 returns the number of bytes consed since the system started,
4337 rather than the number consed since the first time the function
4338 was called. (The new definition parallels ANSI functions like
4339 CL:GET-INTERNAL-RUN-TIME.)
4340 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4341 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4342 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4343 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4344 * Fasl file version numbers are now independent of the target CPU,
4345 since historically most system changes which required version
4346 number changes have affected all CPUs equally. Similarly,
4347 the byte fasl file version is now equal to the ordinary
4350 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4351 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4352 is no longer recognized. I apologize for this, because it was
4353 listed in SB-EXT as a supported extension, but I found that
4354 its existing behavior was poorly specified, as well as incorrectly
4355 specified, and it looked like too much of a mess to straighten it
4356 out. I have enough on my hands trying to get ANSI stuff to work..
4357 * many patches ported from CMU CL by Martin Atzmueller, with
4358 half a dozen bug fixes in pretty-printing and the debugger, and
4359 half a dozen others elsewhere
4360 * fixed bug 13: Floating point infinities are now supported again.
4361 They might still be a little bit flaky, but thanks to bug reports
4362 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4363 as flaky as they were.
4364 * The --noprogrammer command line option is now supported. (Its
4365 behavior is slightly different in detail from what the old man
4366 page claimed it would do, but it's still appropriate under the
4367 same circumstances that the man page talks about.)
4368 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4369 are now supported, and enabled by default. Thus, the compiler can
4370 handle many floating point and complex operations much less
4371 inefficiently. (Thus e.g. you can implement a complex FFT
4373 * The compiler now detects type mismatches between DECLAIM FTYPE
4374 and DEFUN better, and implements CHECK-TYPE more correctly, and
4375 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4376 thanks to patches from Martin Atzmueller.
4377 * various fixes to make the cross-compiler more portable to
4378 ANSI-conforming-but-different cross-compilation hosts (notably
4379 Lispworks for Windows, following bug reports from Arthur Lemmens)
4380 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4381 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4382 fixed thanks to Pierre Mai's CMU CL patches.
4383 * fixes to keep the system from overflowing internal counters when
4384 it tries to use i/o buffers larger than 16M bytes
4385 * fixed bug 45a: Various internal functions required to support
4386 complex special functions have been merged from CMU CL sources.
4387 (When I was first setting up SBCL, I misunderstood a compile-time
4388 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4389 * improved support for type intersection and union, fixing bug 12
4390 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4391 more obscure bugs as well
4392 * some steps toward byte-compiling non-performance-critical
4393 parts of the system, courtesy of patches from Martin Atzmueller
4394 * Christophe Rhodes has made some debian packages of sbcl at
4395 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4396 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4397 stable, but are nonetheless usable. When he's ready, I'd be happy
4398 to add them to the SourceForge "File Releases" section. (And if
4399 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4400 * new fasl file format version number (because of changes in
4401 internal representation of (OR ..) types to accommodate the new
4402 support for (AND ..) types, among other things)
4404 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4405 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4407 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4409 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4410 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4411 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4412 * DESCRIBE now works on structure objects again.
4413 * Most function call argument type mismatches are now handled as
4414 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4415 know whether the function will be redefined before the call is
4416 executed. (The compiler could flag local calls with full WARNINGs,
4417 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4418 it doesn't keep track of enough information to know whether calls
4419 are local in this sense.)
4420 * Compiler output is now more verbose, with messages truncated
4421 later than before. (There should be some supported way for users
4422 to override the default verbosity, but I haven't decided how to
4423 provide it yet, so this behavior is still controlled by the internal
4424 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4425 src/compiler/ir1util.lisp.)
4426 * Fasl file format version numbers have increased again, because
4427 support for the Gray streams extension changes the layout of the
4428 system's STREAM objects.
4429 * The Gray subclassable streams extension now works, thanks to a
4430 patch from Martin Atzmueller.
4431 * The full LOAD-FOREIGN extension (not just the primitive
4432 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4433 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4434 but like most other programs, it defaults to copying the Unix
4435 environment from the original process instead of starting the
4436 new process in an empty environment.
4437 * Extensions which manipulate the Unix environment now support
4438 an :ENVIRONMENT keyword option which doesn't smash case or
4439 do other bad things. The CMU-CL-style :ENV option is retained
4440 for porting convenience.
4441 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4442 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4444 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4446 * A patch from Martin Atzmueller seems to have solved the SIGINT
4447 problem, and as far as we know, signal-handling now works cleanly.
4448 (If you find any new bugs, please report them!)
4449 * The system no longer defaults Lisp source file names to types
4450 ".l", ".cl", or ".lsp", but only to ".lisp".
4451 * The compiler no longer uses special default file extensions for
4452 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4453 seems to expect a single default extension for all compiled code,
4454 and there's no compelling reason to try to stretch the standard
4455 to allow two different extensions.) Instead, byte-compiled files
4456 default to the same extension as native-compiled files.
4457 * Fasl file format version numbers have increased again, because
4458 a rearrangement of internal implementation packages made some
4459 dumped symbols in old fasl files unreadable in new cores.
4460 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4461 many fewer weird special cases.
4462 * Bug #17 (differing COMPILE-FILE behavior between logical and
4463 physical pathnames) has been fixed, and some related misbehavior too,
4464 thanks to a patch from Martin Atzmueller.
4465 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4466 by Tim Moore, ported to SBCL by Martin Atzmueller.
4467 * Martin Atzmueller fixed several filesystem-related problems,
4468 including bug #36, in part by porting CMU CL patches, which were
4469 written in part by Paul Werkowski.
4470 * More compiler warnings in src/runtime/ are gone, thanks to
4471 more patches from Martin Atzmueller.
4472 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4475 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4477 * DESCRIBE now works on CONDITION objects.
4478 * The debugger now handles errors which arise when trying to print
4479 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4481 * The build system now uses an additional file, customize-target-features.lisp,
4482 to allow local modifications to the target *FEATURES* list. (The point of
4483 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4484 debugging features enabled, without having to worry about propagating it
4485 into everyone's system when I do a "cvs update".) When no
4486 customize-target-features.lisp file exists, the target *FEATURES* list
4487 should be constructed the same way as before.
4488 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4489 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4490 warn about behavior which is undefined under the ANSI spec. Note: This
4491 is specified by ANSI, but it's not very popular with programmers.
4492 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4493 macro in the SBCL sources for an example of a workaround which you
4494 might use to make portable ANSI-standard code which does what you want.
4495 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4496 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4497 when it's used as a non-toplevel form. (This is required in order
4498 to implement the DEFCONSTANT-EQX macro.)
4499 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4500 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4501 funny things, due to bugs in the implementation of EVAL-WHEN
4502 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4503 non-toplevel DEFCONSTANTs are uncommon.)
4504 * The core file version number and fasl file version number have been
4505 incremented, because the old noncompliant DEFCONSTANT behavior involved
4506 calling functions which no longer exist, and because I also took the
4507 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4508 * fixed bug 1 (error handling before read-eval-print loop starts), and
4509 redid debugger restarts and related debugger commands somewhat while
4511 ** The QUIT debugger command is gone, since it did something
4512 rather different than the SB-EXT:QUIT command, and since it never
4513 worked properly outside the main toplevel read/eval/print loop.
4514 Invoking the new TOPLEVEL restart provides the same functionality.
4515 ** The GO debugger command is also gone, since you can just invoke
4516 the CONTINUE restart directly instead.
4517 ** The TOP debugger command is also gone, since it's redundant with the
4518 FRAME 0 command, and since it interfered with abbreviations for the
4520 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4522 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4523 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4524 some porting work by Martin Atzmueller)
4525 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4526 with a boatload of patches which he ported to SBCL. Now that those
4528 ** The system tries to make sure that its low-priority messages
4529 are prefixed by semicolons, to help people who like to use
4530 syntax highlighting in their ILISP buffer. (This patch
4531 was originally due to Raymond Toy.)
4532 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4533 patches originally written by Raymond Toy.
4534 ** The compiler understands coercion between single-value and
4535 multiple-VALUES type expressions better, getting rid of some very
4536 weird behavior, thanks to patches originally by Robert MacLachlan
4537 and Douglas Crosher.
4538 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4539 lambda lists, thanks to a patch originally by Pierre Mai.
4540 ** The system no longer bogusly warns about "abbreviated type
4542 ** The compiler gets less confused by inlining and RETURN-FROM,
4543 thanks to some patches originally by Tim Moore.
4544 ** The system no longer hangs when dumping circular lists to fasl
4545 files, thanks to a patch originally from Douglas Crosher.
4546 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4547 undefined function error.
4548 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4549 in structure slot initforms) for now, documented workaround instead:-|
4550 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4551 * fixed bug 5: added stubs for various Gray stream functions called
4552 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4553 installed, at least appropriate type errors are generated
4554 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4555 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4556 consistently in DEFMETHOD forms.
4557 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4558 it doesn't seem to affect SBCL after all
4559 * The C runtime system now builds with better optimization and many
4560 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4562 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4564 * The system is now under CVS at SourceForge (instead of the
4565 CVS repository on my home machine).
4566 * The new signal handling code has been tweaked to treat register
4567 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4568 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4569 downstream, e.g. in debug-int.lisp, has implicit dependencies
4570 on the unsignedness of integer representation of machine words,
4571 and that caused the system to bomb out with infinite regress
4572 when trying to recover from type errors involving signed values,
4573 e.g. (BUTLAST '(1 2 3) -1).)
4574 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4575 (This was one of the bugs Peter Van Eynde reported back in July.)
4576 * The system now uses code inspired by Colin Walters' O(N)
4577 implementation of MAP (from the cmucl-imp@cons.org mailing
4578 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4579 inline the MAP operation, and there is more than one
4580 sequence argument to the MAP call (so that it can't just
4581 do ETYPECASE once and for all based on the type of the
4582 single sequence argument). (The old non-inline implementation
4583 of the general M-argument sequence-of-length-N case required
4584 O(M*N*N) time when any of the sequence arguments were LISTs.)
4585 * The QUIT :UNIX-CODE keyword argument has been renamed to
4586 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4587 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4589 * Raymond Wiker's patches to port dynamic loading from Linux to
4590 FreeBSD have been added.
4591 * The BUGS file is now more nearly up to date, thanks in large part
4592 to Martin Atzmueller's review of it.
4593 * The debugger now flushes standard output streams before it begins
4594 its output ("debugger invoked" and so forth).
4595 * The core version number and fasl file version number have both
4596 been incremented, because of incompatible changes in the layout
4598 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4599 used to not be called for a saved Lisp image.)
4600 * Martin Atzmueller's version of a patch to fix a compiler crash,
4601 as posted on sbcl-devel 13 September 2000, has been installed.
4602 * Instead of installing Martin Atzmueller's patch for the
4603 compiler transform for SUBSEQ, I deleted the compiler transform,
4604 and transforms for some similar consing operations.
4605 * A bug in signal handling which kept TRACE from working on OpenBSD
4607 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4608 away by constant folding
4609 * The system now defines its address space constants in one place
4610 (in the Lisp sources), and propagates them automatically elsewhere
4611 (through GENESIS and the sbcl.h file). Therefore, patching the
4612 address map is less unnecessarily tedious and error-prone. The
4613 Lisp names of address space constants have also been systematized.
4614 * CVS tags like dollar-Header-dollar have been removed from
4615 the sources, because they have never saved me trouble and
4616 they've been source of trouble working with patches and other
4617 diff-related operations.
4618 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4619 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4621 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4623 * The system has been ported to OpenBSD.
4624 * The system now compiles with a simple "sh make.sh" on the systems
4625 that it's supported on. I.e., now you no longer need to tweak
4626 text in the target-features.lisp-expr and symlinks in src/runtime/
4627 by hand, the make.sh takes care of it for you.
4628 * The system is no longer so grossly inefficient when compiling code
4629 involving vectors implemented as general (not simple) vectors (VECTOR T),
4630 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4631 which dares to use the various sequence functions on non-simple
4632 vectors, takes less of a performance hit.
4633 * There is now a primitive type predicate VECTOR-T-P
4634 to test for the (VECTOR T) type, so that e.g.
4635 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4636 can now be compiled with some semblance of efficiency. (The old code
4637 turned the type declaration into a full call to %TYPEP at runtime!)
4638 * AREF on (VECTOR T) is still not fast, since it's still compiled
4639 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4640 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4641 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4642 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4643 without making full calls to SUBTYPEP at runtime.
4644 (Some analogous efficiency issues for non-simple vectors specialized to
4645 element types other than T, or for non-simple multidimensional arrays,
4646 have not been addressed. They could almost certainly be handled the
4647 same way if anyone is motivated to do so.)
4648 * The changes in array handling break binary compatibility, so
4649 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4650 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4651 instead of (VALUES T T).
4652 * By following the instructions that Dan Barlow posted to sbcl-devel
4653 on 2 July 2000, I was able to enable primitive dynamic object
4654 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4655 functionality is not implemented (since it calls ld to resolve
4656 library references automatically, requiring RUN-PROGRAM for its
4657 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4658 try to resolve library references) is now supported.
4659 * The system now flushes the standard output streams when it terminates,
4660 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4661 them at several other probably-convenient times, e.g. in each pass of
4662 the toplevel read-eval-print loop, and after evaluating a form given
4663 as an "--eval" command-line option. (These changes were motivated by a
4664 discussion of stream flushing issues on cmucl-imp in August 2000.)
4665 * The source transform for TYPEP of array types no longer assumes
4666 that an array whose element type is a not-yet-defined type
4667 is implemented as an array of T, but instead punts, so that the
4668 type will be interpreted at runtime.
4669 * There is now some support for cross-compiling in make.sh: each of
4670 the phases of make.sh has its own script. (This should be transparent
4671 to people doing ordinary, non-cross-compile builds.)
4672 * Since my laptop doesn't have hundreds of megabytes of memory like
4673 my desktop machine, I became more motivated to do some items on
4674 my to-do list in order to reduce the size of the system a little:
4675 ** Arrange for various needed-only-at-cold-init things to be
4676 uninterned after cold init. To support this, those things have
4677 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4678 symbols with such names are now uninterned after cold init).
4679 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4680 fasl files for cold load.
4681 ** Remove the old compiler structure pooling code (which used to
4682 be conditional on the target feature :SB-ALLOC) completely.
4683 ** Redo the representation of some data in cold init to be more compact.
4684 (I also looked into supporting byte compiled code at bootstrap time,
4685 which would probably reduce the size of the system a lot, but that
4686 looked too complicated, so I punted for now.)
4687 * The maximum signal nesting depth in the src/runtime/ support code has
4688 been reduced from 4096 to 256. (I don't know any reason for the very
4689 large old value. If the new smaller value turns out to break something,
4690 I'll probably just bump it back up.)
4691 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4693 * Many, many bugs reported by Peter Van Eynde have been added to
4694 the BUGS list; some have even been fixed.
4695 * While enabling dynamic object file loading, I tried to make the
4696 code easier to understand, renaming various functions and variables
4697 with less ambiguous names, and changing some function calling
4698 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4699 * While trying to figure out how to do the OpenBSD port, I tried to
4700 clean up some of the code in src/runtime/. In particular, I dropped
4701 support for non-POSIX signal handling, added various comments,
4702 tweaked the code to reduce the number of compilation warnings, and
4703 renamed some files to increase consistency.
4704 * To support the new automatic configuration functionality in make.sh,
4705 the source file target-features.lisp-expr has been replaced with the
4706 source file base-target-features.lisp-expr and the machine-generated
4707 file local-target-features.lisp-expr.
4708 * fixed a stupid quoting error in make.sh so that using CMU CL
4709 "lisp -batch" as cross-compilation host works again
4711 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4713 * DESCRIBE no longer tries to call itself recursively to describe
4714 bound/fbound values, so that it no longer fails on symbols which are
4715 bound to themselves (like keywords, T, and NIL).
4716 * DESCRIBE now works on generic functions.
4717 * The printer now prints less-screwed-up representations of closures
4718 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4719 * A private symbol is used instead of the :EMPTY keyword previously
4720 used to mark empty slots in hash tables. Thus
4721 (DEFVAR *HT* (MAKE-HASH-TABLE))
4722 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4723 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4724 now does what ANSI says that it should. (You can still get
4725 similar noncompliant behavior if bang on the hash table
4726 implementation with all the symbols you get back from
4727 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4728 This breaks binary compatibility, since tests for equality to
4729 :EMPTY are wired into things like the macroexpansion of
4730 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4732 * There's now a minimal placeholder implementation for CL:STEP,
4733 as required by ANSI.
4734 * An obscure bug in the interaction of the normal compiler, the byte
4735 compiler, inlining, and structure predicates has been patched
4736 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4737 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4739 * Missing ordinary arguments in a macro call are now detected even
4740 when the macro lambda list contains &KEY or &REST.
4741 * The debugger no longer complains about encountering the top of the
4742 stack when you type "FRAME 0" to explicitly instruct it to go to
4743 the top of the stack. And it now prints the frame you request even
4744 if it's the current frame (instead of saying "You are here.").
4745 * As specified by ANSI, the system now always prints keywords
4746 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4748 * The default initial SIZE of HASH-TABLEs is now smaller.
4749 * Type information from CLOS class dispatch is now propagated
4750 into DEFMETHOD bodies, so that e.g.
4751 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4753 is now basically equivalent to
4754 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4755 (DECLARE (TYPE SINGLE-FLOAT X))
4757 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4758 operation, without having to do run-time type dispatch.
4759 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4760 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4761 * Since I don't seem to be making big file reorganizations very often
4762 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4763 is overflowing my ability to conveniently back them up), I've finally
4764 checked the system into CVS. (The CVS repository is on my home system,
4765 not at SourceForge -- putting it on SourceForge might come later.)
4766 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4767 high-level GC-NOTIFY-FOO functions send their output. (There's
4768 still very little control of where low-level verbose GC functions
4769 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4770 less than it used to -- the GC-NOTIFY-FOO functions are now under
4771 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4772 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4773 in only one place in the source code, and propagates it automatically
4774 everywhere that it's needed. Thus e.g. when I bump the version from
4775 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4776 * The C source files now include boilerplate legalese and documentation
4777 at the head of each file (just as the Lisp source files already did).
4778 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4779 to his page will be replaced with a link to his new CLiki service.
4781 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4783 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4784 * The build process now looks for GNU make under the default name "gmake",
4785 instead of "make" as it used to. If GNU make is not available as "gmake"
4786 on your system, you can change this default behavior by setting the
4787 GNUMAKE environment variable.
4788 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4789 can build without error under CMU CL.
4791 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4793 * There is now a partial SBCL user manual (with some new text and some
4794 text cribbed from the CMU CL manual).
4795 * The beginnings of a profiler have been added (starting with the
4796 CMU CL profiler and simplifying and cleaning up). Eventually the
4797 main interface should be through the TRACE macro, but for now,
4798 it's still accessed through vaguely CMU-CL-style functions and macros
4799 exported from the package SB-PROFILE.
4800 * Some problems left over from porting CMU CL to the new
4801 cross-compilation bootstrap process have been cleaned up:
4802 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4803 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4804 part when getting rid of the compiler *BACKEND* stuff.)
4805 ** The value of *NULL-TYPE* was screwed up, because it was
4806 being initialized before the type system knew the final
4807 definition of the 'NULL type. This screwed up several key
4808 optimizations in the compiler, causing inefficiency in all sorts
4809 of places. (I found it because I wanted to understand why
4810 GET-INTERNAL-RUN-TIME was consing.)
4811 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4812 PROCLAIM FTYPE information. Unfortunately this broke binary
4813 compatibility again, since now the forms output by DEFGENERIC
4814 to refer to functions which didn't exist in 0.6.3.
4815 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4816 can use the new (as of 0.6.3) transform for SOME into MAP into
4818 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4819 values to (INTEGER 0 1000000), so that the time code will no longer
4820 occasionally get blown up by Linux returning 1000000 microseconds
4821 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4822 its output conform to the ANSI spec. (Alas, this makes its output
4823 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4824 * A full call to MAP NIL with a single sequence argument no longer conses.
4825 * fixes to problems pointed out by Martin Atzmueller:
4826 * The manual page no longer talks about multiprocessing as though
4827 it were currently supported.
4828 * The ILISP support patches have been removed from the distribution,
4829 because as of version 5.10.1, ILISP now supports SBCL without us
4830 having to maintain patches.
4831 * added a modified version of Raymond Toy's recent CMU CL patch for
4832 EQUALP comparison of HASH-TABLE
4834 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4836 * The system still can't cross-compile itself with
4837 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4838 entails), but at least it can compile more of itself that way
4839 than it used to be able to, and various buglets which were uncovered
4840 by trying to cross-compile itself that way have now been fixed.
4841 * This release breaks binary compatibility again. This time
4842 at least I've incremented the FASL file format version to 2, so that the
4843 problem can be detected reliably instead of just causing weird errors.
4844 * various new style warnings:
4845 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4846 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4847 ** using the lexical binding of a variable named in the *FOO* style
4848 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4849 as specified by ANSI.
4850 * *RANDOM-STATE* is no longer automatically initialized from
4851 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4852 default behavior of the system is to repeat its behavior every time
4853 it's run. If you'd like to change this behavior, you can always
4854 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4855 old convention there was no comparably easy way to get the system to
4856 repeat its behavior every time it was run.
4857 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4858 FTYPE declarations has been removed, in favor of their ANSI
4859 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4860 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4861 the types of their sequence arguments can be determined at compile time.
4862 This is done through a new open code expansion for MAP which eliminates
4863 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4864 argument types can be determined at compile time.
4865 * The optimizer now transforms COERCE into an identity operation when it
4866 can prove that the coerced object is already of the correct type. (This
4867 can be a win for machine generated code, including the output of other
4868 optimization transforms, such as the MAP transform above.)
4869 * Credit information has been moved from source file headers into CREDITS.
4870 * Source file headers have been made more standard.
4871 * The CASE macro now compiles without complaining even when it has
4874 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4876 * (Note that the way that the PCL macroexpansions were rewritten
4877 to accommodate the change in DEFGENERIC below breaks binary
4878 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4879 not run under sbcl-0.6.2. Once we get out of alpha releases,
4880 i.e. hit release 1.0.0, we'll probably try to maintain binary
4881 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4882 and sbcl-1.4.4. Until then, however, it might be fairly common
4883 for maintenance releases to break binary compatibility.)
4884 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4886 * The reporting of unhandled signals has been changed to print some
4887 explanatory text as well as the report form. (Previously only
4888 the report form was printed.)
4889 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4890 it defines, so that the compiler no longer issues undefined function
4891 warnings for compiled-but-not-yet-loaded generic functions.
4892 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4893 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4894 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4895 section of the ANSI spec are supported.
4896 * The "" nickname for the "KEYWORD" package has been removed.
4897 The reader still handles symbol tokens which begin with a package marker
4898 as keywords, but it doesn't expose its mechanism for doing so in the
4899 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4900 * The system now issues STYLE-WARNINGs for contradictory TYPE
4901 proclamations. (Warnings for contradictory FTYPE proclamations would
4902 be nice too, but those can't be done usefully unless the type system
4903 is made smarter about FUNCTION types.)
4904 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4905 systematized, so that "*target-*.lisp is supposed to exist only on the
4906 target and imply that there's a related file which exists on the
4907 host, and *host-*.lisp is supposed to exist only on the host and imply
4908 that there's a related file which exists on the target. This involves a
4909 lot of renaming. Hopefully the acute confusion caused by the renaming
4910 will be justified by the reduction in chronic confusion..
4911 ** runtime-type.lisp -> early-target-type.lisp
4912 ** target-type.lisp -> late-target-type.lisp
4913 ** early-host-format.lisp -> early-format.lisp
4914 ** late-host-format.lisp -> late-format.lisp
4915 ** host-error.lisp -> misc-error.lisp
4916 ** early-error.lisp -> early-target-error.lisp
4917 ** late-error.lisp -> late-target-error.lisp
4918 ** host-defboot.lisp -> early-defboot.lisp
4919 ** code/misc.lisp -> code/target-misc.lisp
4920 ** code/host-misc.lisp -> code/misc.lisp
4921 ** code/numbers.lisp -> code/target-numbers.lisp
4922 ** code/early-numbers.lisp -> numbers.lisp
4923 ** early-host-type.lisp -> early-type.lisp
4924 ** late-host-type.lisp -> late-type.lisp
4925 ** host-typep.lisp -> typep.lisp
4926 ** load.lisp -> target-load.lisp
4927 ** host-load.lisp -> load.lisp
4928 ** host-disassem.lisp -> disassem.lisp
4929 ** host-insts.lisp -> insts.lisp
4930 ** byte-comp.lisp -> target-byte-comp.lisp
4931 ** host-byte-comp.lisp -> byte-comp.lisp
4932 ** host-signal.lisp -> signal.lisp
4933 ** host-defstruct.lisp -> defstruct.lisp
4934 ** late-target-type.lisp -> deftypes-for-target.lisp
4935 Furthermore, several other previously target-only files foo.lisp (e.g.
4936 hash-table.lisp and random.lisp) have been split into a target-and-host
4937 foo.lisp file and a target-only target-foo.lisp file, with their key type
4938 definitions in the target-and-host part, so that the cross-compiler will
4939 know more about target types.
4940 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4941 gone away. In their place are various *BACKEND-FOO* variables
4942 corresponding to the slots of the old structure.
4943 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4944 into the target SBCL has been fixed.
4945 * The chill.lisp system for loading cold code into a running SBCL
4947 * Support for the CMU CL "scavenger hook" extension has been removed.
4948 (It was undocumented and unused in the CMU CL sources that SBCL was
4949 derived from, and stale in sbcl-0.6.1.)
4950 * Various errors in the cross-compiler type system were detected
4951 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4952 (enabling various consistency checks). Many of them were fixed,
4953 but some hard problems remain, so the compiler is back to
4954 running without *TYPE-SYSTEM-INITIALIZED* for now.
4955 * As part of the cross-compiler type system cleanup, I implemented
4956 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4957 * I have started adding UNCROSS calls throughout the type system
4958 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4959 on cross-compiler input files will be able to go away, and various
4961 * CONSTANTP now returns true for quoted forms (as explicitly required
4964 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4966 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4967 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4968 (MAKE-LIST -1). In the longer term, it ought to become true
4969 that declarations are assertions even at SAFETY 1. For now, it's not
4970 quite true even at SAFETY 3, but it's at least more nearly true..
4971 (Note that this change seems to increases the size of the system by
4972 O(5%) and to decrease the speed of the compiler by 20% or more.)
4973 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4974 for the problem of printing dozens of lines of distracting information
4975 about low-level system machinery as part of the top stack frame
4976 on entry to the debugger when an undefined function was called.
4977 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4978 is not reset to COMMON-LISP-USER.
4979 * Compilation of stuff related to dyncount.lisp has been made conditional
4980 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4981 smaller. The various dyncount-related symbols have been moved into
4982 a new "SB-DYNCOUNT" package.
4983 * tty-inspect.lisp has been renamed to inspect.lisp.
4984 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4985 feature has gone away. (When we eventually port to other flavors of
4986 libc and/or Unix, we'll try to make the differences between flavors
4987 invisible at the user level.)
4988 * Various other *FEATURES* tags, and/or their associated conditionals,
4989 have been removed if obsolescent, or given better documentation, or
4990 sometimes given more-mnemonic names.
4992 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4994 * tidied up "make.sh" script
4995 * tidied up system directory structure
4996 * better "clean.sh" behavior
4997 * added doc/FOR-CMUCL-DEVELOPERS
4998 * many many small tweaks to output format, e.g. removing possibly-confusing
4999 trailing #\. character in DESCRIBE-INSTANCE
5000 * (EQUALP #\A 'A) no longer signals an error.
5001 * new hashing code, including EQUALP hashing
5002 * tidied up Lisp initialization and toplevel
5003 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5004 * command line argument processing
5005 * added POSIX-GETENV function to deal with Unix-ish environment variables
5006 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5007 terminating SBCL on EOF
5008 * non-verbose GC by default
5009 * There is no more "sbcl" shell script; the sbcl file is now the C
5010 runtime executable (just like CMU CL).
5011 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5013 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5014 symbol and package information as Lisp native symbol and package objects
5015 instead of strings naming symbols and strings naming packages. This way,
5016 whenever packages are renamed (as in warm init), debug information is
5017 transformed along with everything else.
5018 * tweaked the optimization policy declarations which control the building
5019 of SBCL itself. Now, among other things, the system no longer saves
5020 source location debugging information. (This helps two problems at once
5021 by reducing SBCL size and by keeping SBCL from trying to look for its
5022 sources -- which may not exist -- when reporting errors.)
5023 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5024 debugging and testing purposes
5025 * cleaned up printing, making the printer call PRINT-OBJECT for
5026 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5027 methods, giving nearly-ANSI behavior
5028 * converted almost all special variables to use *FOO* naming convention
5029 * deleted PARSE-TIME functionality, since it can be done portably
5030 * moved some files out of cold init into warm init
5031 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5033 * regularized formatting of source files
5034 * added an install.sh script
5035 * fixed ridiculous memory usage of cross-compiler by making
5036 compiler/alloc.lisp not try to do pooling unless it can hook
5037 itself into the GC of the cross-compilation host. Now the system
5038 builds nicely on my old laptop.
5039 * added :SB-ALLOC in target-features.lisp-expr
5040 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5041 was not implemented)
5042 * re-did condition handling and note reporting in the compiler. Notes
5043 are no longer handled by signalling conditions. Style warnings
5044 and warnings are handled more correctly and reported in such a way
5045 that it's easy to find one or the other in your output (so that you
5046 can e.g. figure out which of many problems caused COMPILE-FILE to
5048 * changed the severity of several compiler warnings from full WARNING
5049 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5050 compiler note reporting so that it doesn't use the condition system
5051 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5052 COMPILE-FILE command)
5053 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5054 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5055 no longer supported, since their implementation was deeply intertwingled
5056 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5057 * removed bogus "support" for compiler macros named (SETF FOO), and
5058 removed the compiler macro for SETF INFO (but only after making a fool
5059 of myself on the cmucl-imp mailing list by posting a bogus patch for
5060 DEFINE-COMPILER-MACRO..)
5061 * Compiled files containing forms which have side effects on the Lisp
5062 reader (such as DEFPACKAGE forms) are now handled more correctly.
5063 (Compiler queuing of top level lambdas has been suppressed by setting
5064 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5065 * deleted various currently-unused source files, e.g. gengc.lisp. They
5066 may be added back at some point e.g. when porting to other architectures,
5067 but until they are it's distracting to distribute them and to try to
5069 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5070 there were so many of them they're just distractions, and UNCROSS is
5071 known to be able to handle the current sources
5072 * moved PROFILE functionality into TRACE, so that it will be clear
5073 how the wrapping and unwrapping of functions when you profile them
5074 interacts with the wrapping and unwrapping of functions when you
5075 trace them. (Actually, the functionality isn't there yet, but at least
5076 the interface specification is there. Hopefully, the functionality will
5077 arrive with some future maintenance release.)
5078 * removed host-oops.lisp
5079 * changed signature of QUIT function to allow UNIX-CODE argument
5080 * fixed READ-SEQUENCE bug
5081 * tweaked verbose GC output so that it looks more like the progress
5082 output that ANSI specifies for functions like LOAD
5083 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5084 * added <http://sbcl.sourceforge.com> to the banner information printed by