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)
17 * bug fix: fix ERROR leaking memory (reported by David Thompson)
19 changes in sbcl-1.0.26 relative to 1.0.25:
20 * incompatible change: an interruption (be it a function passed to
21 INTERRUPT-THREAD or a timer function) runs in an environment where
22 interrupts can be enabled. The interruption can use
23 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
24 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
25 potentially running out of stack. Keep in mind that in the absence
26 of WITHOUT-INTERRUPTS some potentially blocking operation such as
27 acquiring a lock can enable interrupts.
28 * incompatible change: GC-OFF and GC-ON are removed, as they were
29 always unsafe. Use WITHOUT-GCING instead.
30 * new feature: runtime option --disable-ldb
31 * new feature: runtime option --lose-on-corruption to die at the
32 slightest hint of possibly non-recoverable errors: running out of
33 memory, stack, alien stack, binding stack, encountering a memory
34 fault, etc. In the absence of --lose-on-corruption a warning is
36 * enhancement: detect binding stack exhaustion
37 * enhancement: detect alien stack exhaustion on x86/x86-64
38 * improvement: generally more stable and reliable interrupt handling
39 * improvement: there is a per thread interruption queue,
40 interruptions are executed in order of arrival
41 * improvement: a repeating timer reschedules itself when the it has
42 finished, but expiration times are spaced equally. If an
43 expiration time is in the past it will trigger after a short grace
44 period that may give a chance to other things to run.
45 * optimization: slightly faster gc on multithreaded builds
46 * optimization: faster WITHOUT-GCING
47 * bug fix: when JOIN-THREAD signals an error, do it when not holding
48 important locks so that the debugger/handler doesn't produce
49 recursive errors or deadlock.
50 * bug fix: real-time signals are not used anymore, so no more
51 hanging when the system wide real-time signal queue gets full.
52 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
54 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
55 * bug fix: fix random memory faults related to interrupts on alpha
56 * bug fix: fix random memory faults related to interrupts on sparc
57 * bug fix: fix deadlocks related to starting threads
58 * bug fix: fix deadlines on locks on futex platforms
59 * bug fix: restore errno in signal handlers
60 * bug fix: fix deadlocks related to hash tables
61 * bug fix: fix deadlocks in pcl
63 changes in sbcl-1.0.25 relative to 1.0.24:
64 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
65 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
66 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
67 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
68 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
69 better name and does not return values so stale on multiprocessor systems.
70 Also, HOLDING-MUTEX-P was added for about the only sane usage of
72 * improvement: unithread builds keep track of MUTEX-VALUE.
73 * improvement: reading from a TWO-WAY-STREAM does not touch the output
74 stream anymore making it thread safe to have a concurrent reader and
75 a writer, for instance, in a pipe.
76 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
77 arguments, making compiler-macros for SETF-functions able to inspect
78 their constant arguments.
79 * improvement: COMPILE-FILE reports times with millisecond accuracy
80 (thanks to Luis Oliveira)
81 * optimization: CHAR-CODE type derivation has been improved, making
82 TYPEP elimination on subtypes of CHARACTER work better. (reported
83 by Tobias Rittweiler, patch by Paul Khuong)
84 * bug fix: setting alien structure fields of type struct by value now
85 computes the right offset for the memory copy.
86 * bug fix: compilation problem involving inlined calls to aliens with
87 result type VOID. (reported by Ken Olum)
88 * bug fix: #235a; sequential inline expasion in different policies no
89 longer reuses the functional from the previous expansion site.
90 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
91 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
92 * bug fix: named ENUMs in multiply-referenced alien record types no longer
93 cause an implied type redefinition. Regression from 1.0.21.29.
94 * bug fix: sign-extension of small signed return values now works with C
95 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
96 * improvements to the Windows port:
97 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
98 Westfold, thanks to Rudi Schlatte)
100 changes in sbcl-1.0.24 relative to 1.0.23:
101 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
102 vector of a multidimensional SIMPLE-ARRAY.
103 * new feature: the system now signals a continuable error if standard
104 readtable modification is attempted. (thanks to Tobias Rittweiler)
105 * new feature: DIRECTORY has been extended with a non-standard keyword
106 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
107 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
108 has been robustified and documented better. (thanks to Robert Goldman)
109 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
110 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
111 * tradeoff: constant FORMAT control strings are now compiled unless
112 SPACE > SPEED (previously only when SPEED > SPACE.)
113 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
115 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
117 * bug fix: lexical type declarations are now correctly reported by
118 SB-CLTL2. (reported by Larry D'Anna)
119 * bug fix: STRING-TO-OCTETS did not handle :START properly when
120 using UTF-8 as external format. (thanks to Luis Oliveira)
121 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
122 are signalled correctly. (thanks to Stas Boukarev)
123 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
124 designator. (thanks to Tobias Rittweiler)
125 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
126 designator, and returns T instead of the function. (thanks to
128 * bug fix: direct superclasses of STANDARD-CLASS and
129 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
130 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
131 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
132 when no class with the named slot yet exists no longer causes a
133 compile-time style-warning.
134 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
135 in safe code. (reported by Didier Verna)
136 * bug fix: #430; nested structure constructors can stack allocate.
137 * bug fix: on some 64-bit platforms dynamic space size was truncated
138 to #xffff0000 bytes. (reported by Benjamin Lambert)
139 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
140 REPL unusable. (reported by Daniel Herring)
141 * bug fix: return values of READ-SEQUENCE did not take :START into
142 account on file streams, regressions since 1.0.12.22. (reported by
143 Thomas Russ, patch by Paul Khuong)
144 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
145 method specializer used to confuse permuation vector optimization.
146 * bug fix: system inserted bogus implicit type declarations for local
147 special variables in DEFMETHOD bodies.
148 * bug fix: #354; duplicated frames in backtraces due to
149 non-tail-call-optimized XEPs to functions with return type NIL
151 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
152 initializes structure object slots according to DEFSTRUCT initforms,
153 and DEFSTRUCT forms :INCLUDEind structure classes defined using
154 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
155 (reported by Bruno Haible and Stephen Wilson)
156 * bug fix: #395; fill-pointer output streams used now support
157 element-type BASE-CHAR as well.
158 * bug fix: compiler error when attempting to derive return value of
159 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
161 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
163 * bug fix: Mac OS X binaries should now be portable between Leopard
166 changes in sbcl-1.0.23 relative to 1.0.22:
167 * enhancement: when disassembling method functions, disassembly
168 for the associated fast function is also produced.
169 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
171 * optimization: printing with *PRINT-PRETTY* true is now more
172 efficient as long as the object being printed doesn't require
173 special handling by the pretty printer.
174 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
175 now interact correctly with type declarations.
176 * partial bug fix: PCL detects infinite recursion during wrapper
177 validation. (thanks to Attila Lendvai)
178 * bug fix: #426; nested function calls are inlined properly.
179 Previously if FOO was an inline function, in calls of the form
180 (FOO (FOO ...)) the outer call was not inlined.
181 * bug fix: long long arguments passed in registers to alien
182 functions on PowerPC Linux and NetBSD are now handled correctly
183 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
185 changes in sbcl-1.0.22 relative to 1.0.21:
186 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
187 for the shared object in the current directory, but passes the native
188 namestring of the designated pathname to the operation system's shared
189 object loading function as-it.
190 * minor incompatible change: --disable-debugger toplevel option now takes
191 effect before processing of initialization files and --eval or --load
193 * new feature: new commandline argument: --script, which supports
194 shebang lines. See documentation for details. (based on work by
196 * new feature: SAVE-LISP-AND-DIE can save current values of
197 --dynamic-space-size and --control-stack-size in the executable core,
198 causing it to skip normal runtime option processing. See documentation
199 for details. (thanks to Zach Beane)
200 * enhancement: inoccous calls to EVAL or generic functions dispatching
201 on subclasses of eg. STREAM no longer cause compiler notes to appear.
202 * enhancement: the system no longer resignals errors from --load and
203 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
204 associated with the original error to be lost. (thanks to Ariel
206 * enhancement: :DONT-SAVE keyword argument has been added to
207 LOAD-SHARED-OBJECT for controlling interaction with
208 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
209 the effects of an earlier LOAD-SHARED-OBJECT call.
210 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
211 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
212 provided. (reported by Cedric St-Jean)
213 * bug fix: circularity handling in the reader did not treat raw
214 structure slots correctly. (reported by Cedric St-Jean)
215 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
216 file descriptors when there were none.
217 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
218 was broken. (reporteed by Willem Broekema)
219 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
220 pathnames without a directory.
221 * bug fix: COMPILE-FILE did not warn about undefined variable
222 references at toplevel, and LOAD of the resulting fasl did
224 * bug fix: functions with non-required arguments used to end up with
225 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
226 * bug fix: redefining a function with non-required arguments didn't
227 update the system's knowledge about its call signature properly.
228 * bug fix: fixed #431; incompatible alien record type redefinitions
229 are detected and handled. (reported by Neil Haven)
230 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
231 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
233 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
234 after alien stack frames.
235 * bug fix: strings passed to mk{s,d}temp lacked null termination
237 changes in sbcl-1.0.21 relative to 1.0.20:
238 * new feature: the compiler is able to track the effective type of a
239 generic function across method addition and removal.
240 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
241 appropriately typed structure slots without locking.
242 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
243 information like those gathered by TIME using a programming-friendly
245 * new feature: TIME reports time taken even if the form performs a
246 non-local transfer of control.
247 * enhancement: reduced conservativism on GENCGC platforms: on
248 average 45% less pages pinned (measured from SBCL self build).
249 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
250 mutate constant symbols or violate declaimed type of the symbol.
251 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
252 sleeping on the semaphore in heavily contested semaphores.
253 * bug fix: semaphores and condition variables were not interrupt
255 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
256 owned by other threads anymore.
257 * bug fix: FIND on lists called KEY outside the specified
258 subsequence. (reported by budden)
259 * bug fix: LOG doesn't use single-float intermediate results when
260 given mixed integer and double-float arguments, leading to better
261 precision. (reported by Bob Felts)
262 * bug fix: LOG with base zero returned values of inconsistent type.
264 changes in sbcl-1.0.20 relative to 1.0.19:
265 * minor incompatible change: OPTIMIZE qualities
266 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
267 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
268 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
269 * documentation: some slot access efficiency guidelines have been
270 added to the user manual.
271 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
272 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
273 as ASSOC and MEMEBER.
274 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
275 to more efficient EQ-comparison versions more often.
276 * optimization: enhanced derivation of DOLIST iteration variable type
278 * optimization: constant folding of simple (LIST ...) forms as DOLIST
280 * optimization: runtime lookup of function definitions can be
281 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
282 * optimization: compiler is able to derive the return type of
283 (AREF (THE STRING X) Y) as being CHARACTER.
284 * optimization: CLRHASH on empty hash-tables no longer does pointless
285 work. (thanks to Alec Berryman)
286 * optimization: REPLACE deftransforms don't punt when :START1 or
288 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
289 no longer emits a code deletion note. (reported by Andrew
291 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
292 functions with an odd number of &OPTIONAL arguments, a &REST
293 argument, and one or more &KEY arguments at the call site.
294 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
295 arguments appeared at call sites as well.
296 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
297 READ-LINE &co can no longer read from them afterwards. (reported
299 * bug fix: fixed #427: unused local aliens no longer cause compiler
300 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
302 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
303 alien-stack leakage. (reported by Andy Hefner)
304 * bug fix: PROGV signals an error when an attempt to violate declared
305 type of a variable or bind a constant is made.
306 * bug fix: SET signals an error when an attempt to violate declared
307 type of a variable is made.
308 * bug fix: restart computation during the execution of a restart
309 test function no longer causes infinite recursion. (reported by
311 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
312 signals an error. (thanks to Pierre Mai)
313 * bug fix: EXPORT left symbol unexported in conflict situations.
314 (thanks to Michael Weber)
315 * bug fix: correctly handle name conflicts involving CL:NIL.
316 (thanks to Michael Weber)
317 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
318 conflicts arising from USEing package with conflicting symbols
319 correctly. (thanks to Michael Weber)
321 changes in sbcl-1.0.19 relative to 1.0.18:
322 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
323 warnings that go otherwise unhandled will be muffled if they are
324 of the type that's the value of this variable.
325 * optimization: stack allocation is slightly more efficient on x86
327 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
328 where a raw slot always is initialized using the initform whose
329 type is not know sufficiently well a compile-time are now compiled
330 correctly. (reported by John Morrison)
331 * bug fix: compiler no longer makes erronous assumptions in the
332 presence of non-foldable SATISFIES types.
333 * bug fix: stack analysis missed cleanups of dynamic-extent
334 arguments in non-let-converted calls to local functions.
335 * improvements to the Windows port:
336 ** adjusted address spaces for building on both Win32 and
337 Win64. (thanks for John Connors)
338 * fixed some bugs revealed by Paul Dietz' test suite:
339 ** interval arithmetic during type derivation used inexact integer
340 to single-float coercions.
341 ** arithmetic operations involving large integers and single
342 floats give the same results in compiled and interpreted code.
343 ** deriving the result type of COERCE no longer signals an error
344 if the derived type of the second argument is a MEMBER type
345 containing invalid type specifiers.
346 ** ADJOIN with constant NIL as second argument works correctly.
347 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
349 changes in sbcl-1.0.18 relative to 1.0.17:
350 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
351 profiles only the current thread.
352 * minor incompatible change: changes to SYMBOL-VALUE of constants
353 defined with DEFCONSTANT now signal an error.
354 * enhancement: SB-SPROF now has support for wallclock profiling,
355 and is also able to profile specific threads. REPORT output
356 has also additional sorting options.
357 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
359 * optimization: structure allocation has been improved
360 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
361 ** out of line constructors are ~10% faster.
362 ** inline constructors are ~15% faster.
363 ** inline constructors are capable of dynamic extent allocation
364 (generally on x86 and x86-64, in some cases on other platforms
366 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
368 * optimization: file compiler is now able to coalesce non-circular
369 lists, non-base strings, and bit-vectors. Additionally, constants
370 are never referenced through SYMBOL-VALUE at runtime.
371 * optimization: code defining methods on PRINT-OBJECT (and other
372 generic functions in the COMMON-LISP package) now loads faster.
373 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
375 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
376 fasl is now deleted. (reported by Attila Lendvai)
377 * bug fix: READ-LINE always returned NIL for the last line in files.
378 (reported by Yoshinori Tahara)
379 * bug fix: more accurate disassembly annotations of foreign function
380 calls. (thanks to Andy Hefner)
381 * bug fix: trimming non-simple strings and non-string string
382 designators when the there is nothing to trim works properly.
383 (thanks to James Knight)
384 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
385 (thanks to Travis Cross)
386 * fixed some bugs revealed by Paul Dietz' test suite:
387 ** NIL is a valid function name (regression at 1.0.13.38)
388 ** FILL on lists was missing its return value (regression at 1.0.12.27)
389 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
390 fill pointers properly (regression at 1.0.12.23)
391 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
392 displacement indices properly (regression at 1.0.12.23)
394 changes in sbcl-1.0.17 relative to 1.0.16:
395 * temporary regression: user code can no longer allocate closure
396 variable storage on stack, due to bug 419 without explicitly
397 requesting it. Please consult sbcl-devel for advice if you need to
398 use this feature in the meanwhile.
399 * new feature: runtime argument --control-stack-size can be used to
400 adjust thread default control stack size.
401 * enhancement: improved TIME output
402 ** all times are reported using the measured accuracy (milliseconds
403 for real and GC times, microseconds for everything else.)
404 ** processor cycle counts on x86 and x86-64.
405 ** interpreted forms are counted for both evaluator modes.
406 ** number of lambdas converted by the compiler is reported.
407 ** CPU percentage report (computed from real and total run time.)
408 ** more comprehensive run time reporting, using a condenced format
409 ** interperted form, lambda, and page fault counts are omitted
411 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
413 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
414 * optimization: two argument forms of LAST are upto ~10% faster
415 in normal SPEED policies.
416 * optimization: NCONC no longer needs to heap cons its &REST list
417 in normal SPEED policies.
418 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
420 * bug fix: UNION and NUNION work with :TEST-NOT once more,
421 regression since 1.0.9.1. (thanks to Eric Marsden)
422 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
423 since 1.0.15.36. (thanks to Paul Khuong)
424 * bug fix: LAST when always returned the whole list when given a bignum
425 as the second argument.
426 * bug fix: dynamic extent allocation of nested lists and vectors
427 could leak to otherwise accessible parts.
428 * bug fix: invalid optimization of heap-allocated alien variable
430 * bug fix: fasl header checking is less vulnerable to different
431 platform word lengths.
432 * bug fix: more correct assembler syntax for GNU binutils
433 2.18.50.0.4 support. (thanks to Marijn Schouten)
434 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
436 * bug fix: the fopcompiler can handle LOCALLY forms (with no
437 declarations) successfully. (reported by Attila Lendvai)
439 changes in sbcl-1.0.16 relative to 1.0.15:
440 * minor incompatible change: revert the changes to sb-posix's error
441 signaling added in 1.0.14.
442 * minor incompatible change: change PROBE-FILE back to returning
443 NIL whenever we can't get a truename, as was the case before 1.0.14.
444 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
445 inherits from ERROR instead of just CONDITION.
446 * new feature: SB-INTROSPECT can provide source locations for instances
447 as well. (thanks to Tobias Rittweiler)
448 * optimization: binding special variables now generates smaller code
449 on threaded platforms.
450 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
451 and cases where no :TEST is given but the compiler can infer that
452 the element to search is of type (OR FIXNUM (NOT NUMBER)).
453 * optimization: better LOGNOT on fixnums.
454 * optimization: modular arithmetic for a particular requested width
455 is implemented using a tagged representation unless a better
456 representation is available.
457 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
458 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
459 for wellformedness and returns a specialized vector. (reported by
460 Francois-Rene Rideau)
461 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
462 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
463 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
464 unscheduling timer at the same time as another timer fires.
465 * bug fix: don't reschedule timers for dead threads.
466 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
467 * bug fix: copying output from RUN-PROGRAM to a stream signalled
468 bogus errors if select() was interrupted.
469 * enhancement: add support for fcntl's struct flock to SB-POSIX.
471 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
472 * enhancement: cleaner backtraces for interactive interrupts, as
473 well as other cases where the interesting frames used to be
474 obscured by interrupt handling frames.
475 * enhancement: untracing a whole package using (UNTRACE "FOO") is
476 now supported, and tracing a whole package using (TRACE "FOO") now
477 traces SETF-functions as well.
478 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
479 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
480 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
481 * unreadably printed representation of hash-tables now includes
483 * bug fix: partially fixed #188: type propagation from assignments
484 is now more efficient.
485 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
486 'DOUBLE-FLOAT) are not flushable.
487 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
488 full address of the object, and none of the tag bits.
489 * bug fix: readably printing hash-tables now respects other printer
490 control variables. (reported by Cedric St-Jean)
491 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
492 keyword with MAKE-HASH-TABLE.
493 * bug fix: export SB-POSIX:MKSTEMP.
494 * bug fix: SORT was not interrupt safe.
495 * bug fix: XREF accounts for the last node of each basic-block as
497 * bug fix: MAKE-INSTANCE optimizations interacted badly with
498 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
499 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
500 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
501 create obsolete instances in the case of redefinition or
502 obsoletion of a superclass. (thanks to Andy Hefner)
503 * bug fix: Support for the Alpha architecture has been revived; it had
504 suffered somewhat from lack of maintenance since sbcl-1.0.
505 * improvements to the Windows port:
506 ** The system detects the codepage to use at startup, instead of
507 using the value from the saved core. (thanks to Kei Suzuki)
509 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
510 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
511 (see documentation for details.)
512 * revived support for OpenBSD (contributed by Josh Elsasser)
513 * partially fixed bug #108: ROOM no longer suffers from occasional
514 (AVER (SAP= CURRENT END)) failures.
515 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
516 bodies no longer cause a WARNING to be signalled. (reported by
518 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
519 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
520 the bug in TRUNCATE.) (reported by Michael Weber)
521 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
522 no samples. (reported by Andy Hefner)
523 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
524 no longer appear as (NIL ...) frames in backtraces.
525 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
526 is now actually exported from SB-EXT as documented. (reported by
528 * bug fix: sb-aclrepl now correctly understands how to inspect
529 single-floats on 64-bit platforms where single-floats are not boxed.
530 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
531 yet been finalized. (reported by Levente Meszaros)
532 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
533 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
534 SBCL-specific optimize qualities.
536 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
537 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
538 an executable in the search path, and does so in the child
539 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
540 been removed; it can be found in the CVS history, for somebody who
541 needs that search behavior (see the manual).
542 * minor incompatible change: compiler policy re. weakening type
543 checks has changed: now type checks are weakened only if SAFETY < 2
545 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
546 unparsing of directory pathnames as files. Analogously,
547 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
548 filename to parse into a directory pathname.
549 * enhancement: implicit generic function creation now signals a
550 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
551 users can bind handlers for and muffle around calls to LOAD.
552 * enhancement: RUN-PROGRAM allows unicode arguments and environments
553 to be used (using the default stream external format), and allows
554 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
555 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
556 strings and vectors whose element-type or simplicity is not fully
557 known at compile-time.
558 * optimization: STRING-TRIM and related functions no longer allocate
559 a new string when no trimming needs to be performed. These functions
560 are also faster than before when the input string has been declared
562 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
563 * optimization: READ-LINE is significantly faster for files containing
565 * optimization: non-open coded uses of character comparison operators
566 (e.g. char=) no longer cons when called with more than one parameter
567 on platforms supporting dynamic-extent allocation.
568 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
569 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
570 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
572 * bug fix: some sequence functions elided bounds checking when
574 * bug fix: too liberal weakening of union-type checks when SPEED >
576 * bug fix: more bogus fixnum declarations in ROOM implementation
579 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
580 * new feature: MAKE-HASH-TABLE now experimentally accepts a
581 :SYNCHRONIZED argument, which makes the hash-table safe for
582 concurrent accesses (but not iteration.) See also:
583 SB-EXT:WITH-LOCKED-HASH-TABLE, and
584 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
585 * optimization: CONCATENATE on strings is an order of magnitude faster
586 in code compiled with (> SPEED SPACE).
587 * optimization: SUBSEQ is ~50% faster on lists.
588 * bug fix: bug 417 fixed -- source location reporting is now more robust.
589 * bug fix: SUBSEQ on a list will now correctly signal an error if if
590 END is smaller then START.
591 * bug fix: SB-PROFILE will no longer report extra consing for nested
592 calls to profiled functions.
593 * bug fix: ROOM implementation had bogus fixnum declarations which
594 could cause type-errors when calling ROOM on large images.
595 * bug fix: if file compilation is aborted, the partial fasl is now
596 deleted, and COMPILE-FILE returns NIL as the primary value.
597 * bug fix: number of thread safety issues relating to SBCL's internal
598 hash-table usage have been fixed.
599 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
600 be returned from its body when the values were being returned
601 using unknown-values return convection and the W-P-O was wrapped
602 inside an UNWIND-PROTECT.
603 * bug fix: sb-posix should now compile again under Windows, enabling
606 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
607 * incompatible change: hash-table accessor functions are no longer
608 automatically protected by locks. Concurrent accesses on the same hash-table
609 from multiple threads can give inconsistent results or even corrupt the
610 hash-table completely. Multi-threaded applications should do their own
611 locking at the correct granularity. In the current implementation it is
612 still safe to have multiple readers access the same table, but it's not
613 guaranteed that this property will be maintained in future releases.
614 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
615 and will signal an error at runtime.
616 * enhancement: SB-THREAD package now exports a semaphore interface.
617 * enhancement: CONS can now stack-allocate on x86 and
618 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
620 * enhancement: nested lists can now be stack allocated on
621 platforms providing stack allocation support.
622 * enhancement: dynamic-extent support has been extended to support
623 cases where there are multiple possible sources for the stack
625 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
626 if the mutex is uncontested on Linux.
627 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
628 as specified by the CLHS. (thanks to Tobias Rittweiler)
629 * bug fix: NaN comparison now works on x86-64.
630 * bug fix: CLOSE :ABORT T on a stream with pending output now
632 * bug fix: instances of non-standard metaclasses using standard
633 instance structure protocol sometimes missed the slot type checks
635 * bug fix: known functions can be inlined in high-debug code.
637 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
638 * minor incompatible change: the MSI installer on Windows no longer
639 associates .lisp and .fasl files with the installed SBCL.
640 * minor incompatible change: :UNIX is no longer present in *FEATURES*
641 on Windows. (thanks to Luis Oliviera)
642 * new feature: SB-CLTL2 contrib module now implements
643 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
644 * optimization: scavenging weak pointers is now more efficient,
645 requiring O(1) instead of O(N) per weak pointer to identify
646 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
647 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
648 method bodies using constant slot-names when the first argument is
649 a specializer parameter for the method.
650 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
651 names, when the class of the instance is a direct instance
652 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
653 as slow as the constant slot-name case.)
654 * optimization: member type construction is now O(N) instead
656 * optimization: UNION and NUNION are now O(N+M) for large
657 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
659 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
660 now more readable in environments like Slime which display it.
661 (thanks to Tobias C. Rittweiler)
662 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
663 was able to infer the correct argument type for the object on which
664 the CAS operation was being performed.
665 * bug fix: copy propagation interfered with parallel assignment
666 semantics in local calls. (reported by Paul Khuong)
667 * bug fix: the signed modular fixnum shift compiled to wrong code on
668 x86 and x86-64. (spotted by a slight modification to some of
671 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
672 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
673 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
674 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
676 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
677 and generic function now print the object identity for anonymous
678 instances. (thanks to Attila Lendvai)
679 * enhancement: as an extension to MOP, SBCL now supports SETF of
680 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
681 (thanks to Attila Lendvai)
682 * enhancement: sb-cover has an alternate annotation mode which puts
683 the form annotations at the car of the form, instead of annotating
684 it completely. The benefit of this mode is that it shows better
685 which forms were instrumented by the compiler.
686 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
687 by printer control variables. (reported by Dan Corkill)
688 * bug fix: system leaked memory when delayed output was performed by
689 the OS in smaller chunks then expected. (thanks to David Smith)
690 * bug fix: system leaked memory when file streams were not closed
692 * bug fix: large objects written to slow streams that were modified
693 after the write could end up with the modified state written to
694 the underlying file descriptor.
695 * bug fix: multiple threads operating in parallel on the same stream
696 could cause buffer-overflows.
697 * bug fix: source location information is stored correctly
698 (broken since 1.0.6). This bug would generally show up as the
699 Slime debugger highlighting the wrong form when the sldb-show-source
701 * bug fix: Branch forms are again annotated as branches in the sb-cover
703 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
704 on x86-64. (reported by Gregory Vanuxem)
705 * bug fix: EQUALP could return wrong results for structures with raw
706 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
707 word). (reported by Vjacheslav Fyodorov)
708 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
711 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
712 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
713 atomic compare-and-swap operations on threaded platforms.
714 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
715 allows assigning a global minimum value to optimization qualities
716 (overriding proclamations and declarations).
717 * enhancement: closed over variables can be stack-allocated on x86
719 * performance bug fix: GETHASH and (SETF GETHASH) are once again
721 * optimization: slot definition lookup is now O(1). This speeds up
722 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
723 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
724 * optimization: ASSOC and MEMBER can now be open-coded for all
725 combinations of keyword arguments when second argument is constant
726 and SPEED >= SPACE. In other cases a specialized version is
728 * bug fix: using obsoleted structure instances with TYPEP and
729 generic functions now signals a sensible error.
730 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
731 (reported by Kristoffer Kvello)
732 * bug fix: backtrace construction is now more careful when making
733 lisp-objects from pointers on the stack, to avoid creating bogus
734 objects that can be seen by the GC.
735 * bug fix: defaulting of values in contexts expecting more than 7
736 variables now works on x86-64. (reported by Christopher Laux)
737 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
739 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
740 as the property-list of a symbol.
741 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
742 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
743 situations requested, are once again file-compileable. (reported
746 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
747 * MOP improvement: support for user-defined subclasses of
748 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
749 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
750 the expansion of DEFMETHOD, is responsible for generating a form
751 which creates a list of specializers when evaluated. Additional
752 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
753 debugging and introspective support.
754 * minor incompatible change: the (unsupported) spinlock interface
755 has changed: free spinlock now has the value NIL, and a held spinlock
756 has the owning thread as its value.
757 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
758 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
760 * enhancement: name of a socket-stream is now "a socket" instead of
762 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
763 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
764 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
765 for generic functions. (Thanks to Tobias C. Rittweiler)
766 * optimization: bignum printing speed has been improved by 20-40%
767 (depending on the bignum size.)
768 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
770 * bug fix: the cache used by the CLOS to store precomputed effective
771 methods, slot offsets, and constant return values is now thread and
773 * bug fix: generic function dispatch function updating is now thread
774 and interrupt safe (in the sense that the known issues have been
776 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
777 * bug fix: interning EQL-specializers is now thread and interrupt safe.
778 * bug fix: asdf systems with dependencies to the SB-POSIX or
779 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
780 * bug fix: interrupt safety of applicable method computation has been
783 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
784 * new contrib: sb-cover, an experimental code coverage tool, is included
786 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
787 significantly faster.
788 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
789 produce more efficient code on x86-64 if the compiler can determine
790 that (+ INDEX <constant>) does not require a bounds check and FOO
791 has an element type at least 8 bits wide.
792 * enhancement: a new, experimental synchronous timeout facility is
793 provided. Refer to SB-SYS:WITH-DEADLINE for details.
794 * enhancement: when a symbol name conflict error arises, the
795 conflicting symbols are always printed with a package prefix.
796 (thanks to Kevin Reid)
797 * enhancement: stepping is now once again supported on the SPARC and
798 MIPS platforms. (It is also now more likely to work on CheneyGC
800 * enhancement: sb-sprof can now also track and report accurate call
802 * bug fixes: the treatment of non-standard subclasses of
803 SB-MOP:SPECIALIZER is more correct.
804 * incompatible change: PURIFY no longer copies the data from the
805 dynamic space into the static and read-only spaces on platforms
806 that use the generational garbage collector
807 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
809 * bug fix: GC race condition occasionally resulting in crashes with
810 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
812 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
814 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
815 a STYLE-WARNING to be signalled (regression from 1.0.4.)
816 * bug fix: an asynchronous interrupt could previously leave the
817 system running with GC inhibited.
818 * bug fix: a DECLARE form evaluated at top-level now causes an error
819 rather than silently (or verbosely) returning NIL.
820 * bug fix: trying to dissassemble functions compiled at high DEBUG could
821 result in a error being signalled due to source form lookup errors.
822 (reported by Peter Graves)
824 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
825 * incompatible change: removed writer methods for host-ent-name,
826 host-ent-addresses -- changing the values did not update the DNS
828 * minor incompatible change: changed experimental JOIN-THREAD interface
829 * documentation: the manual now lists reader and writer methods
830 in class slot documentation sections. (thanks to Richard M Kreuter)
831 * documentation: unwinding from asyncronous events has been
832 documented as unsafe.
833 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
834 in multithreaded application code.
835 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
836 platforms. (thanks to James Anderson for the optimization hint)
837 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
839 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
840 variants no longer cons.
841 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
842 their NOT- variants no longer cons.
843 * optimization: EQUAL hash tables no longer use SXHASH for objects
844 of all data types, but instead use an EQL hash for types for which
845 EQUAL is the same as EQL.
846 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
847 are significantly faster.
848 * optimization: new STRING-OUTPUT-STREAM implementation is much
849 faster for large outputs and conses 30% less on average.
850 * enhancement: XREF information is now collected for references made
851 to global variables using SYMBOL-VALUE and a constant argument.
852 * enhancement: SIGINT now causes a specific condition
853 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
854 * bug fix: FILE-POSITION used to signal an error in some cases where
855 ANSI requires it to return NIL.
856 * bug fix: ADJUST-ARRAY is now interrupt-safe.
857 * bug fix: adding and removing fd-handlers is now interrupt-safe.
858 * bug fix: inlined calls to C now ensure 16byte stack alignment on
860 * bug fix: bad type declaration in the CLOS implementation has
861 been fixed. (reported by James Anderson)
862 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
863 (thanks to Lutz Euler)
864 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
865 required by the CLHS. (reported by Samium Gromoff)
866 * bug fix: dead unbound variable references now signal an error.
867 (reported by Marco Monteiro)
868 * bug fix: / with an unused value was being deleted in safe code.
869 (thanks to Marco Monteiro and Kevin Reid)
870 * bug fix: number of characters that can be written onto a single
871 line in a file is unlimited.
872 * bug fix: some GC deadlocks caused by asynchronous interrupts have
873 been fixed by inhibiting interrupts for when GC is disabled.
874 * bug fix: some interrupt safety issues with GETHASH, (SETF
875 GETHASH), CLRHASH and REMHASH have been fixed.
876 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
877 specifier no longer causes infinite recursion.
878 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
879 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
880 * bug fix: an &environment argument with the correct variable information
881 is passed to macros that are expanded during byte compilation.
882 (reported by Samium Gromoff)
883 * bug fix: a logic error was causing the hash function for bignums to
884 have excessive amounts of collisions. (reported by Faré Rideau)
885 * bug fix: modifying the contents of an array could change the return
886 value of SXHASH on that array, which is only allowed for strings
887 and bit vectors. (bug introduced in 0.9.16)
888 * improvement: the x86-64/darwin port now passes all tests (except
889 for the debugger tests) but should still be considered
890 experimental until this is fixed.
891 * improvement: a style-warning is signaled for CASE (etc) clauses with
892 duplicate keys. (thanks to Kevin Reid)
893 * improvement: macos/x86 and macos/x86-64 now use mach exception
894 handlers for dealing with illegal instructions (for trapping and
895 error handling) and memory protection violations (for GC).
897 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
898 * new platform: experimental support for x86-64/darwin (MacOS).
899 * incompatible change: the thread-safe (on most platforms) getaddrinfo
900 and getnameinfo sockets functions are used instead of gethostbyaddr
901 and gethostbyname, on platforms where the newer functions are available.
902 As a result, the ALIASES field of HOST-ENT will always be NIL on these
904 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
905 don't print the startup banner, but behave as if --noinform was passed
906 as a command line argument. (thanks to Kevin Reid)
907 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
908 * optimization: code using alien values with undeclared types is much faster.
909 * optimization: the compiler is now able to open code SEARCH in more cases.
910 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
911 * bug fix: using standardized COMMON-LISP special variables as loop
912 variables no longer signals bogus package lock violations. (reported
914 * bug fix: declaring local loop variables to be of a range-limited type
915 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
916 (reported by Andras Simon)
917 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
918 bugs remain on x86-64.)
919 * bug fix: the #= and ## reader macros now interact reasonably with
920 funcallable instances.
921 * bug fix: type-checks for function arguments were compiled using the
922 compiler policy settings of the wrong lexical environment (in 1.0.2
924 * bug fix: SHADOW accepts characters as string designators, as required
925 by the spec (thanks to Eric Marsden)
926 * bug fix: fixed GC safety issues when foreign functions are called with
927 non-base strings as arguments
928 * bug fix: more consistent error output for fatal-compiler-warnings, like
930 * bug fix: functions with proper names were showing up as NIL in
931 backtraces (reported by Edi Weitz, regression in 1.0.2)
933 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
934 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
935 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
936 produce more efficient code on the x86 if the compiler can determine
937 that (+ INDEX <constant>) does not require a bounds check and FOO
938 has an element type at least 8 bits wide.
939 * bug fix: references to undefined variables are handled the same way
940 in toplevel forms as in the normal compiler
941 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
943 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
944 (thanks to Jon Buller)
945 * improvement: faster compilation times for complex functions
946 * improvement: added readlink support to SB-POSIX (thanks to Richard
949 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
950 * improvement: experimental support for mach exception handling on
951 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
953 * improvement: support for GBK external format.
954 (thanks to Chun Tian (binghe))
955 * improvement: the debugger now displays variables that have been closed
956 over, in code compiled with (DEBUG 2) or higher
957 * improvement: support for executable cores on NetBSD (thanks to
959 * new feature: added a RESTART-FRAME debugger command
960 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
961 be used to provide an implementation for FILE-POSITION on Gray streams
962 (thanks to Eric Marsden)
963 * new feature: add syslog, openlog and closelog support to SB-POSIX
964 (thanks to Richard Kreuter)
965 * optimization: the function call overhead in code compiled with
966 a high DEBUG optimization setting is significantly
967 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
968 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
969 (thanks to Tony Martinez)
970 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
971 evaluated once (reported by Kevin Reid)
972 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
973 works again (thanks to Kilian Sprotte)
974 * bug fix: an error is signaled for tagbodies with duplicate tags
975 (thanks to Stephen Wilson)
976 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
977 * bug fix: Win32 port can now handle foreign code unwinding Lisp
978 stack frames from alien callbacks.
979 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
980 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
981 65k pages (thanks to David Woodhouse)
982 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
984 changes in sbcl-1.0.1 relative to sbcl-1.0:
985 * new platform: FreeBSD/x86-64, including support for threading.
986 * new feature: the compiler stores cross-referencing information
987 about function calls (who-calls), macroexpansion (who-macroexpands)
988 and special variables (who-binds, who-sets, who-references) for code
989 compiled with (< SPACE 3). This information is available through the
990 sb-introspect contrib.
991 * new feature: users may subclass SEQUENCE, and have instances of
992 these classes interoperate with standard Common Lisp functions if
993 a number of methods are defined. (This feature is experimental
994 and the interface subject to change based on feedback from SBCL
995 users and the general community)
996 * improvement: sb-sprof traces call stacks to an arbitrary depth on
997 x86/x86-64, rather than the previous fixed depth of 8
998 * improvement: another pthread back-end of mutex "pthread-futex".
999 * bug fix: non-ascii command-line arguments are processed correctly
1000 (thanks to Yaroslav Kavenchuk)
1001 * bug fix: non-required arguments were not passed correctly when a method
1002 defined using DEFMETHOD was called from a mop-generated method using
1003 CALL-NEXT-METHOD (reported by Pascal Costanza)
1004 * bug fix: recursion is now permitted in accessors through
1005 SLOT-UNBOUND. (reported by Pascal Costanza)
1006 * bug fix: an error was signaled at startup if the HOME environment
1007 variable was defined, but had an empty value (reported by Peter Van Eynde)
1008 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1009 signal a PROGRAM-ERROR, not a BUG.
1010 * bug fix: SB-POSIX:READDIR works when built with large file support.
1011 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1013 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1014 for better automatic indentation support. (thanks to Matt Pillsbury)
1015 * optimization: loading generic functions no longer takes O(n^2) time,
1016 proportional to the amount of methods in the generic function
1017 (reported by Todd Sabin and Jeremy Brown)
1018 * optimization: the FIND and POSITION family of sequence functions
1019 are significantly faster on arrays whose element types have been
1021 * improvements to the Windows port:
1022 ** Intermittent heap corruption problems have been fixed. (thanks
1023 to Alastair Bridgewater)
1024 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1026 ** Lisp is able to unwind foreign exception frames from alien
1027 callbacks. (thanks to Alastair Bridgewater)
1029 changes in sbcl-1.0 relative to sbcl-0.9.18:
1030 * improvement: experimental support for threading on FreeBSD/x86.
1031 (thanks to NIIMI Satoshi)
1032 * improvement: runtime option --dynamic-space-size can be used
1033 to set the size of the dynamic space reserved on startup.
1034 * improvement: floating point modes in effect are now saved in
1035 core, and restored on startup.
1036 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1037 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1038 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1039 threads are not running after *SAVE-HOOKS* have run.
1040 * improvement: writes to CLOS instance slots are type-checked in code
1041 compiled with (SAFETY 3)
1042 * improvement: floating-point exception handling on FreeBSD (thanks to
1044 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1045 (thanks to Zach Beane)
1046 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1048 * improvement: added support for the Shift-JIS external format.
1049 (contributed by NIIMI Satoshi)
1050 * improvement: callbacks are supported on Linux/PPC. (thanks to
1052 * bug fix: compiler bug triggered by a (non-standard) VALUES
1053 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1054 * bug fix: file compiler no longer confuses validated and already
1055 dumped structurres. (reported by Kaersten Poeck)
1056 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1057 fill-pointer signals a type-error as required. (thanks to
1059 * bug fix: disassemly of funcallable instances works.
1060 * bug fix: single stepping on PPC.
1061 * bug fix: fix thread-safety problems in the type system (generally
1062 manifesting as nonsensical errors like "STRING is a bad type specifier
1063 for sequences" or "The value 1 is not of type FIXNUM")
1064 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1065 (reported by Josip Gracin)
1066 * bug fix: an error is signaled for attempts to displace arrays with
1067 incompatible element types (thanks to Mario Mommer)
1068 * bug fix: more correct handling of wide characters in debug info
1069 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1070 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1071 and don't cause extra consing
1072 * optimization: MAP and MAP-INTO are significantly faster on vectors
1073 whose elements types have been declared.
1074 * Improvements to SB-SPROF:
1075 ** Support for allocation profiling
1076 ** Reduced profiling overhead, especially for long profiling runs
1077 * Improvements to the Windows port:
1078 ** floating point exceptions are now reported correctly.
1079 ** stack exhaustion detection works partially.
1080 ** more accurate GET-INTERNAL-REAL-TIME.
1081 ** better RUN-PROGRAM behaviour in case of errors in the child
1083 ** PROBE-FILE now simplifies pathnames correctly.
1084 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1085 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1086 1st 1970 (although time timezone data will be unreliable).
1087 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1089 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1090 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1091 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1093 * bug fix: two potential GC deadlocks affecting threaded builds.
1094 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1095 returns T (reported by Anton Kazennikov)
1096 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1097 with non-variable places
1098 * bug fix: remove a race condition in the setting of
1099 funcallable-instance functions, this should make threaded CLOS
1100 code more stable against memory faults.
1101 * bug fix: corruption of specials when unbinding is interrupted by an
1102 asynchronous unwind (reported by Hannu Koivisto)
1103 * improvement: the debugger will now also display local variables that
1104 are only used once, for code compiled with a DEBUG optimization quality
1107 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1108 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1109 * incompatible change: External-format support for FFI calls. The
1110 SB-ALIEN:C-STRING no longer implies an ASCII
1111 external-format. Instead the string is subject to external-format
1112 conversion. Additionally return values of type SB-ALIEN:C-STRING
1113 are by default of type (SIMPLE-ARRAY CHARACTER), not
1114 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1115 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1116 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1117 * incompatible change: SB-EXT package no longer contains the
1118 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1119 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1121 * incompatible change: the single-stepper is no longer available
1122 on Alpha, Mips and Sparc platforms.
1123 * minor incompatible change: the direct superclasses of
1124 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1125 not (STANDARD-OBJECT FUNCTION). This makes the
1126 class-precedence-lists of GENERIC-FUNCTION and
1127 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1129 * minor incompatible change: the default stream external format on
1130 non-unicode SBCL is no longer detected from the locale, but is always
1132 * new feature: Add a version of evaluator that uses an interpreter instead
1133 of the compiler. EVAL still uses the compiler by default, to switch it
1134 to use the interpreter, set the value of the variable
1135 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1136 * minor incompatible change: the single-stepper REPL has been merged
1137 with the normal debugger (see the "Stepping" heading of the debugger help
1138 for more details). The debugger command STEP will no longer switch
1139 to the single-stepper REPL.
1140 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1141 for a type now works.
1142 * bug fix: loading of default sysinit file works. (thanks to Leonid
1144 * bug fix: better detection of circularities in the file-compiler.
1145 (reported by Marco Monteiro)
1146 * bug fix: the CL pathname functions now work with files that have
1147 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1148 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1149 systems with Unix98 pty semantics.
1150 * bug fix: ASDF-INSTALL will now work with bsd tar.
1151 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1153 * bug fix: timers expiring in dead threads no longer cause a
1154 type-error (reported by Paul "Nonny Mouse").
1155 * bug fix: thanks to more lightweight single-stepper instrumentation,
1156 code compiled with (DEBUG 3) will compile and execute significantly faster,
1157 and will have more accurate type-inferencing than before
1158 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1159 whose bindings are modified
1160 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1161 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1162 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1163 CL:SOFTWARE-VERSION with proper version
1165 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1166 * feature: implemented the READER-METHOD-CLASS and
1167 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1168 as specified by AMOP.
1169 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1171 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1172 (thanks to Lutz Euler)
1173 * optimization: hashing of general arrays and vectors has been
1174 improved. (reported by Any Fingerhut)
1175 * enhancement: SB-INTROSPECT is now able to find definitions of
1176 profiled functions. (thanks to Troels Henriksen)
1177 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1178 as well, allowing compiler-macros for SETF-functions to expand.
1179 * enhancement: step-instrumentation no longer wraps known
1180 single-value functions with multiple-value context, allowing
1181 better type inference.
1182 * fixed bug #337: use of MAKE-METHOD in method combination now works
1183 even in the presence of user-defined method classes. (reported by
1184 Bruno Haible and Pascal Costanza)
1185 * fixed bug #339(c): if there are applicable methods not part of any
1186 long-form method-combination group, call INVALID-METHOD-ERROR.
1187 (reported by Bruno Haible)
1188 * fixed bug #361: the :FUNCTION initarg in the protocol for
1189 initialization of methods can now be used to override
1190 internally-produced optimized functions. (reported by Bruno
1192 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1193 system-provided lambda expression no longer cause warnings about
1194 unbound #:|pv-table| symbols.
1195 * bug fix: improved the handling of type declarations and the
1196 detection of violations for keyword arguments with non-constant
1198 * bug fix: potentially erronous calls to PATHNAME and
1199 MERGE-PATHNAMES were being flushed in some cases.
1200 (reported by Richard Kreuter)
1201 * bug fix: compiled calls to TYPEP were mishandling obsolete
1202 instances. (reported by James Bielman and Attila Lendvai)
1203 * bug fix: format strings with ~> without matching ~< no longer
1204 trigger an AVER, but signal an understandable error instead.
1205 (reported by Antonio Martinez)
1206 * bug fix: specifying an output-file without a pathname-type for
1207 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1208 of input-file instead of "fasl". (reported by Robert Dodier)
1209 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1210 (reported by James Y Knight).
1211 * bug fix: compiler-macros-function did not consider the environment
1212 argument for shadowing by local functions.
1213 * bug fix: compiler-macros expansion was inhibited by local INLINE
1215 * bug fix: inline expansions of known functions were subject to
1216 step-instrumentation in high debug policies, leading to problems
1217 with type-inference.
1218 * bug fix: compiler failed to differentiate between different CONS
1219 types in some cases.
1220 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1221 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1222 * bug fix: #368: incorrect use of expressed vs. upgraded array
1224 * bug fix: #306a: more precise unions of array types.
1225 * thread-safety improvements:
1226 ** CONDITION-WAIT could return early on Linux, if the thread was
1227 interrupted and subsequently continued with SIGCONT.
1228 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1230 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1231 * added support for the ucs-2 external format. (contributed by Ivan
1233 * minor incompatible change: pretty printing of objects of type
1234 (cons symbol) is, in the default pprint-dispatch-table, now
1235 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1237 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1238 called later in a class's lifetime, possibly as late as when the
1239 first instance of the class is created. Previously,
1240 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1241 class became finalizeable.
1242 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1243 for multibyte external-format streams. (thanks to "vbzoli")
1244 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1245 with :ALLOCATION :CLASS inherited from superclasses of the
1247 * fixed bug: anonymous classes can now be created using the :NAME
1248 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1249 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1250 * fixed bug: core-files saved with :EXECUTABLE T can again be
1251 executed when SBCL_HOME isn't set. (reported by James Knight)
1252 * fixed bug: toplevel LOCALLY forms with declarations could
1253 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1254 * fixed bug: printing from several different threads using different
1255 values of *print-case* could cause invalid output, due to
1256 some internal special variables of the printer not being bound
1257 thread-locally (reported by Max Mikhanosha)
1258 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1259 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1260 (reported by Pascal Costanza)
1261 * fixed bug: FINALIZE-INHERITANCE is called from
1262 REINITIALIZE-INSTANCE on classes when the class has previously
1263 been finalized, as required by AMOP.
1264 * minor code generation optimizations:
1265 ** better register allocation in CLOS dispatching functions
1266 ** overflow detection when coercing signed bytes to fixnums on x86-64
1267 ** is now implemented with one IMUL instruction instead of three shifts
1268 ** more efficient bit-vector access on x86 and x86-64
1269 ** more efficient access to raw structure slots on x86 and x86-64
1270 * fixed some bugs revealed by Paul Dietz' test suite:
1271 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1272 return its argument.
1274 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1275 * feature: thread support on Solaris/x86, and experimental thread support
1277 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1279 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1280 default method for SLOT-UNBOUND.
1281 * minor incompatible change: prevent the user from specializing the
1282 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1283 somewhat counter to the intent of the protocol, I (CSR) think, and
1284 additionally it just doesn't work in SBCL as currently
1285 implemented, thanks to optimizations (that are always valid for
1286 the other three SLOT-VALUEish functions, but not for the setter).
1287 * minor incompatible change: the :SB-LDB feature is now enabled by
1288 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1289 the low-level debugger.
1290 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1291 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1292 down to an SB-SYS:FD-STREAM, allowing direct communication in
1293 more cases. Temporary files and pipes are still used as a fallback
1295 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1296 on the non-reentrant C functions localtime(3) and gmtime(3).
1297 * bug fix: global optimization policy was not visible in LOCALLY and
1299 * bug fix: class objects can be used as specializers in methods.
1300 (reported by Pascal Costanza)
1301 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1302 with a type error. (reported by blitz_ on #lisp)
1303 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1304 failed with a type error. (reported by Pascal Bourguignon)
1305 * bug fix: merging pathnames against defaults with :DIRECTORY
1306 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1307 (reported by James Y Knight)
1308 * bug fix: saving large (>2GB) cores on x86-64 now works
1309 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1310 constant index and a (simple-array (signed-byte 32)) array
1311 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1312 error (patch by Robert J. Macomber)
1313 * feature: TIME now displays the amount of run-time spent in GC
1314 * bug fix: The debugger now does a better job of respecting
1315 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1316 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1317 workaround for bug 403.)
1318 * bug fix: floating point exception handling now works on FreeBSD
1319 (thanks to NIIMI Satoshi)
1320 * fixed some bugs revealed by Paul Dietz' test suite:
1321 ** MISC.641: LET-conversion were not supposed to work in late
1323 * improvements to the Win32/x86 port:
1324 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1325 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1327 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1328 * new feature: source path information is generated for macro-expansion
1329 errors for use in IDE's like Slime (thanks to Helmut Eller)
1330 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1331 * bug fix: compiler does not loop forever on an invalid type in
1333 * improvement: compilation of most CLOS applications is significantly
1335 * optimization: added a limited bytecode compiler for simple toplevel
1336 forms, speeding up compilation and FASL loading
1337 * bug fix: the statistical profiler now properly distinguishes anonymous
1340 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1341 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1342 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1343 into subclasses of the finalized class. Now user calls to
1344 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1345 system return before any subclasses are finalized.
1346 * minor incompatible change: The reader no longer ignores errors
1347 regarding non-existent packages in #+ and #- feature tests.
1348 * new feature: command line options --no-sysinit, --no-userinit to
1349 inhibit loading the corresponding init files
1350 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1351 for sending data through UDP sockets (thanks to François-René Rideau)
1352 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1353 error is signalled instead (thanks to François-René Rideau)
1354 * minor incompatible change: Error signalling behaviour of lexical
1355 operations violating package locks has changed slightly. Refer to
1356 documentation on package locks for details.
1357 * enhancement: EVAL can process IF-expressions without resorting to the
1359 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1360 constant-folded properly due to missing out-of-line functions.
1361 (reported by Mika Pihlajamäki)
1362 * bug fix: LISTEN sometimes returned T even in cases where no data was
1363 immediately available from the stream
1364 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1365 were reversed. (reported by Levente Mészáros)
1366 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1367 when given a vector displaced to an adjustable array. (reported
1369 * bug fix: derivation of float boundaries from numbers outside the
1370 appropriate float range (reported by John Wiseman)
1371 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1373 * bug fix: precomputing cache entries for generic functions with
1374 some subclasses of specializers as yet invalid does not attempt to
1375 fill a cache line with a negative offset. (reported by Levente
1377 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1378 allow condition class objects as arguments to DOCUMENTATION and
1379 (SETF DOCUMENTATION); only find and set documentation for
1380 structure names for the STRUCTURE doc-type. (suggested by Gary
1382 * improvements to the Win32/x86 port:
1383 ** (user-homedir-pathname) and default initialization file
1384 locations now know about the user's "Documents and Settings"
1385 directory (thanks to Yaroslav Kavenchuk)
1386 ** run-program is implemented (thanks to Mike Thomas)
1387 ** sockets support (thanks to Timothy Ritchey)
1388 ** better backtrace support (thanks to Alastair Bridgewater)
1389 ** sb-grovel supported
1390 ** asdf-install and sb-posix work somewhat
1391 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1392 * improvements to the Solaris/x86 port:
1393 ** works on Solaris 11/Solaris Express
1394 ** floating-point exception handling support
1395 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1396 * fixed some bugs revealed by Paul Dietz' test suite:
1397 ** REMOVE-METHOD returns its generic function argument even when
1398 the method is not one of the generic functions' methods.
1399 ** objects declared as MEMBER types can be admissible arguments to
1400 structure accessors.
1401 ** printing characters should simply be printed by the FORMAT ~:C
1403 ** compiler failure when compiling functions with hairy constant
1404 defaults for optional parameters.
1405 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1406 function, which is already optimized.
1408 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1409 * new platform: experimental support for SBCL x86/Darwin, including
1410 MacOS X 10.4.5 on Intel.
1411 * new feature: Unicode character names are now known to the system
1412 (through CHAR-NAME and NAME-CHAR).
1413 * new feature: the filesystem location of SBCL's core file is
1414 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1415 * minor incompatible change: the contrib modules SB-POSIX and
1416 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1417 this change is to make it easier to distribute
1418 location-independent binaries.
1419 * bug fix: as implied by AMOP, standardized classes no longer have
1420 slots named by external symbols of public packages. (reported by
1422 * optimization: calling functions via constant symbols -- as in
1423 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1424 via the function object as in (FUNCALL #'FOO).
1425 * optimization: on x86 and x86-64, the calling convention for the
1426 case of dealing with an unknown number of values has been altered
1427 to be friendlier to the prediction heuristics implemented,
1428 particularly with reference to CALL and RET pairing. (thanks to
1429 Alastair Bridgewater)
1430 * optimization: on x86, the code for access to thread-local symbol
1431 values and binding/unbinding of thread-local symbols is smaller.
1432 (thanks to Alastair Bridgewater)
1433 * enhancement: CONSTANTP is now able to determine constantness of
1434 more complex forms, including calls to constant-foldable standardized
1435 functions and some special forms beyond QUOTE.
1436 * fixed bug: occasional GC crashes on Solaris/x86
1437 * optimization: x86-64 supports stack allocation of results of simple
1438 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1439 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1440 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1441 (thanks to James Knight)
1442 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1443 by the standard (thanks to Stephen Compall)
1445 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1446 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1447 be used for bundling the runtime and the core file into one
1448 executable binary. This feature is not currently supported on all SBCL
1449 platforms. (thanks to James Bielman and NIIMI Satoshi)
1450 * new feature: a generational or ephemeral garbage collector is now
1451 the default on the PowerPC platform (both Linux and Darwin). The
1452 old Cheney (stop and copy) collector is a build-time option.
1453 (thanks to Cyrus Harmon, after Raymond Toy)
1454 * minor incompatible change: the method by which SBCL finds its
1455 contributed modules has changed; it no longer relies on symbolic
1456 links from an $SBCL_HOME/systems directory, but searches directly
1457 in the subdirectories of $SBCL_HOME.
1458 * enhancement: the dynamic heap size on the Linux/PPC platform is
1459 markedly larger, even using the older Cheney garbage collector.
1460 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1461 certain complicated string types. (reported by Gary King)
1462 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1463 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1464 * fixed bug: DOCUMENTATION on structures no longer signals an
1465 error if no structure type of the right name exists. (reported by
1467 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1468 applicable non-standard primary or :AROUND method on
1469 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1470 no longer cause unbound variable errors. (reported by Kevin Reid)
1471 * fixed bug: metacircle resolution in cases where methods have slots
1472 added before the slots from STANDARD-METHOD. (reported by Jean
1474 * fixed bug: the Power PC assembly code for calling into lisp
1475 sometimes computed its return address wrongly (depending on the
1476 15th bit of the address, fixed at link time). This used to
1477 manifest itself in a segmentation violation while building PCL.
1478 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1479 many others over the years)
1480 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1481 the stack pointer (treated as a signed integer) is negative.
1482 (thanks to Peter van Eynde)
1484 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1485 * new platform: experimental support for the Windows operating
1486 system has been added. (thanks to Alastair Bridgewater)
1487 * fixed several bugs in and robustified the PPC FFI (including
1488 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1489 * bug fix: allow non-simple string symbol names (reported by Paul
1491 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1493 * bug fix: saving a core corrupted callbacks on x86/x86-64
1494 * bug fix: closed a loophole in metacircularity detection and
1495 grounding in the PCL implementation of CLOS.
1496 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1497 (reported by tomppa on #lisp)
1498 * bug fix: clear the direction flag on Lisp -> C transitions, as
1499 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1500 (reported by Andrej Grozin and Hendrik Maryns)
1501 * optimization: major improvements to GC efficiency on GENCGC platforms
1502 * optimization: faster implementation of EQUAL
1503 * optimization: emit more efficient opcodes for some common
1504 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1505 * optimization: several other minor code-generation improvements on x86-64
1506 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1508 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1509 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1510 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1511 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1512 by ANSI as a generic function, and for consistency (SETF
1513 GENERIC-FUNCTION-NAME) is treated likewise.
1514 * fixed bug #233.b: make constraint propagation notice when a variable
1515 value is changed after it is referenced but before it is used
1516 * fixed bug #296: no more arbitrary behaviour from filenames with
1517 odd characters as --load arguments. (reported by Adam Warner)
1518 * fixed bug #390: :CHARACTER-SET pathname components now work as
1519 expected. (reported by Tim Daly Jr)
1520 * fixed bug #391: complicated :TYPE intersections in slot
1521 definitions no longer cause an error in PCL internals.
1522 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1523 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1524 returning the number of octets which would be written to the
1525 file-stream. (thanks to Robert J. Macomber)
1526 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1527 arguments to the internal stream functions. (thanks to David
1529 * bug fix: stack exhaustion is now handled on the x86/SunOS
1530 platform. (thanks to Daisuke Homma)
1531 * bug fix: more accurate ROOM results on GENCGC platforms
1532 * optimization: improved type inference for arithmetic-for
1533 index variables in LOOP
1534 * optimization: faster floating-point SQRT on x86-64
1535 * fixed some bugs revealed by Paul Dietz' test suite:
1536 ** DOCUMENTATION returns NIL instead of "" for method combinations
1537 that don't have a docstring
1539 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1540 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1541 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1542 therefore are not customizeable by user code (as seems to be at
1543 least permitted and maybe required by AMOP). As a consolation,
1544 however, the SBCL implementation of these functions now calls
1545 REINITIALIZE-INSTANCE as specified by AMOP.
1546 * bug fix: slot-definition documentation is propagated as per ANSI
1547 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1548 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1549 argument, as do eslotd initialization methods. (from Pascal
1550 Costanza's "Closer" project)
1551 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1552 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1554 * bug fix: it is now possible to have more than one subclass of
1555 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1556 (reported by Bruno Haible, Pascal Costanza and others)
1557 * bug fix: the dependent update protocol now works for generic
1558 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1559 and Pascal Costanza)
1560 * bug fix: condition-class instances corresponding to
1561 DEFINE-CONDITION forms are now created eagerly. (reported by
1562 Kalle Olavi Niemitalo on comp.lang.lisp)
1563 * bug fix: floating point printing is more accurate in some
1564 circumstances. (thanks to Simon Alexander)
1565 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1566 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1567 * bug fix: callbacks on OS X now preserve stack-alignment.
1568 (thanks to Cyrus Harmon)
1569 * enhancement: the x86-64 disassembler is much better at
1570 disassembling SSE instructions. (thanks to Lutz Euler)
1571 * enhancement: CHECK-TYPE on a variable now allows type-inference
1572 (thanks to Vincent Arkesteijn)
1573 * optimization: improved performance of EUC-JP external format.
1574 (thanks to NIIMI Satoshi)
1575 * optimization: performance improvements to IO on file streams of
1576 :ELEMENT-TYPE CHARACTER
1577 * optimization: much faster memory allocation on x86-64
1578 * optimization: faster garbage collections (latency and throughput)
1580 * optimization: faster bignum subtraction and division on x86 and x86-64
1581 (thanks to Lutz Euler)
1582 * optimization: more accurate type-inference for loops that use a
1583 floating point index variable or a negative step.
1585 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1586 * bug fix: add a workaround to SBCL looping infinitely at startup on
1587 Linux kernels with apparently buggy implementations of personality().
1588 (thanks to Svein Ove Aas)
1589 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1590 (thanks to David Lichteblau)
1591 * bug fix: local bindings shadow global symbol macros for
1592 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1593 * optimization: non-open coded uses of numeric comparison operators
1594 (e.g. >) no longer cons when called with more than one parameter
1595 on platforms supporting dynamic-extent allocation.
1596 * enhancement: saving cores with foreign code loaded is now supported
1597 on MIPS/Linux in addition to the previously supported platforms.
1598 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1599 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1600 * bug fix: division by zero in sb-sprof when no samples were collected
1601 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1602 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1603 (thanks to Alexander Kjeldaas)
1604 * new feature: ignoring the return values of destructive functions
1605 like NREVERSE signals a compile-time style-warning.
1606 (thanks to Kevin Reid)
1607 * new feature: the alignment of alien structure fields can be
1608 explicitly specified. (thanks to Cyrus Harmon)
1609 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1610 and dump core on SIGQUIT
1612 ** incompatible change: threads do not inherit values of specials
1613 from their parents (see manual)
1614 ** bug fix: threads stacks belonging to dead threads are freed by the
1615 next exiting thread, no need to gc to collect thread stacks anymore
1616 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1617 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1618 ** bug fix: dynamic variable and thread start related gc lossage
1619 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1621 * fixed some bugs revealed by Paul Dietz' test suite:
1622 ** SUBTYPEP is slightly more accurate on heinously complicated
1623 CONS types where some of the members have uncertain (in the
1624 NIL, NIL sense) type relationships to each other.
1626 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1627 no longer purified unless :PURIFY T is explicitly specified.
1628 ** Non-purified cores are significantly smaller than before
1630 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1631 * new feature: timers based on Zach Beane's excellent timer package
1632 * added support for the following external formats: koi8-u,
1633 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1634 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1635 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1636 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1637 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1638 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1639 * incompatible change: a threaded SBCL will no longer revert to
1640 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1641 * bug fix: interrupts are disabled until startup is complete; no
1642 more sigsegvs when receiving a signal to soon
1643 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1645 * bug fix: add a workaround for the memory randomization features in
1646 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1647 workaround will only be in effect on systems with the proc filesystem
1649 * bug fix: printing objects of type HASH-TABLE signals a
1650 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1652 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1653 have been running for over 50 days. (reported by Gilbert Baumann)
1654 * bug fix: the logic for getting names of functions gets less
1655 confused when confronded with alternate-metaclass
1656 funcallable-instances. (reported by Cyrus Harmon)
1657 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1658 consistent, even on internal alternate-metaclass objects.
1659 * bug fix: flush closure information collected by physical
1660 environment analysis prepass before the main pass. (bug reported
1662 * bug fix: compiler pack phase does not modify a hashtable, which is
1663 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1664 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1665 may be expanded. (reported by Alexander <malishev> on c.l.l)
1666 * bug fix: no more style-warnings when compiling code using LAST
1667 under high-SPEED low-SPACE settings. (reported by David Wallin
1669 * fixed some incompatibilities between SBCL's MOP and the MOP
1671 ** the METAOBJECT class is now implemented;
1672 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1673 STANDARD-OBJECT, as required;
1674 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1675 now compatible (as required by VALIDATE-SUPERCLASS); there
1676 remains a consistency requirement over the presence or absence
1677 of the FUNCTION class in the superclass list of finalized
1678 classes; see the manual for more details;
1679 ** the :AROUND method for COMPUTE-SLOTS on
1680 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1681 requested slot ordering.
1683 ** bug fix: parent thread now can be gc'ed even with a live
1685 ** bug fix: binding a special with PROGV to no value is not
1686 the same as that symbol not having been bound (thanks to
1688 ** bug fix: binding specials is thread safe (thanks to
1690 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1691 and SB-SPROF do not die with 'no handler for signal XX in
1692 interrupt_handle_now(..)' anymore
1693 ** bug fix: WITH-TIMEOUT works with multiple running threads
1694 * fixed some bugs revealed by Paul Dietz' test suite:
1695 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1696 the :method-class keyword argument.
1698 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1699 * new port: the Solaris operating system on x86 processors is now
1700 mostly supported, though some rough edges in the environment
1701 remain. (thanks to Daisuke Homma)
1702 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1703 space for its heap. (thanks to Thiemo Seufer)
1704 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1705 closures. (thanks to Thiemo Seufer)
1706 * minor incompatible change: eof selects abort in the debugger.
1707 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1708 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1710 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1711 deprecated, and will go away in a future revision of SBCL.
1712 * minor incompatible change: GC-ON and GC-OFF are no longer
1713 implemented with a counter, it does not matter how many times gc
1714 is switched on or off
1715 * bug fix: discriminating functions for generic function classes
1716 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1717 make invalid assumptions about method precedence order. (reported
1719 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1720 platforms. (thanks to Thiemo Seufer)
1721 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1722 * bug fix: correct pathname printing with printer escaping is on.
1723 (thanks to Kevin Reid)
1724 * bug fix: complex VOP definitions in "user-space" no longer trigger
1725 package locks. (reported by Zach Beane)
1726 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1727 for funcallable-instance functions, and is no different from
1729 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1730 characters in the current readtable. (reported by Nicholas Neuss)
1731 * bug fix: SUBTYPEP on various CONS types returns more a more
1732 accurate acknowledgment of its certainty.
1733 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1734 lists in some cases. This partially fixes bug 384.
1735 * flush all standard streams before prompting in the REPL and the
1737 * bug fix: signal handling and triggering gc do not conflict
1738 directly anymore, in particular a high frequency sb-sprof does
1739 not prevent gc from running
1740 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1741 approximation for timezone and DST information between the
1742 universal time epoch and the smallest negative 32-bit time_t.
1743 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1744 year 1899 so long as the encoded time is non-negative.
1745 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1746 the correct answer for non-constant rotate amounts when performing
1747 an inline 32-bit rotation.
1749 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1750 there is only one thread in the session
1751 ** bug fix: memory leak for streams created in one thread and
1752 written to in another
1753 ** bug fix: lockup when compiled with gcc4
1754 ** bug fix: race that allows the gc to be triggered when gc is
1756 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1757 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1759 ** bug fix: finished threads are gc'ed properly
1760 * fixed some bugs revealed by Paul Dietz' test suite:
1761 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1762 has arguments with a different set of applicable methods from
1763 the orignal arguments.
1764 ** The type error thrown by MAP now has a correct expected-type
1766 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1767 name a compiled function.
1768 ** Three MISC tests where a large bignum was improperly coerced to
1769 a float that couldn't represent that bignum during type
1770 derivation were fixed.
1771 ** SUBTYPEP can now handle types involving the negation of a
1772 list-form FUNCTION type.
1773 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1774 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1775 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1777 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1778 * New feature: Experimental support for bivalent streams: streams
1779 opened with :element-type :default now allow character and binary
1780 (unsigned-byte 8) I/O
1781 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1782 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1784 * bug fix: no more highly sporadic "couldn't check whether ~S is
1785 readable" when reading a stream and an interrupt hits in the middle
1786 of a select system call
1787 * compiler better recognizes complex arrays (reported by Eduardo
1789 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1790 for variables on SBCL built with linkage-tables. (reported by Luis
1792 * various error reporting improvements.
1793 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1794 (Thanks to Thiemo Seufer)
1795 * bug fix: nested reader invokations work correctly
1796 * bug fix: it is possible to have simultaneous references to foreign
1797 code and foreign data with the same name.
1799 ** added x86-64 support
1800 ** incompatible change: the threading api now works with thread
1801 objects instead of thread ids
1802 ** bug fix: threads are protected from signals and interruption when
1803 starting up or going down
1804 ** bug fix: a race where an exiting thread could lose its stack to gc
1805 ** bug fix: don't halt on infinite error in threads if possible
1806 ** fixed numerous gc deadlocks introduced in the pthread merge
1807 ** bug fix: fixed thread safety issues in read and print
1808 ** bug fix: debugger doesn't hang on session lock if interrupted at
1809 an inappropriate moment
1810 ** bug fix: run-program is now thread safe(r)
1811 ** bug fix: inner with-recursive-lock no longer releases the mutex
1812 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1813 * fixed some bugs revealed by Paul Dietz' test suite:
1814 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1815 EXPECTED-TYPE slots filled.
1816 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1818 ** MAKE-CONDITION accepts classes as type-designators.
1819 ** COMPILE may never return NIL.
1820 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1821 range before calling Unix time functions
1823 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1824 * numerous signal handling fixes to increase stability
1825 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1826 * minor incompatible change: we now correctly canonize default
1827 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1828 as per the MOP, rather than the historical (INITARG INITFUNCTION
1829 INITFORM). (reported by Bruno Haible)
1830 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1831 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1832 for more information.
1833 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1834 pathname is a directory pathname.
1835 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1836 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1838 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1839 :METACLASS argument in addition to a class name. (reported by
1840 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1841 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1842 (reported by Bruno Haible; patch from Gerd Moellmann)
1843 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1845 * bug fix: more cleanups to the floating point exception handling on
1846 x86-64 (thanks to James Knight)
1847 * bug fix: the compiler does not try to constant fold calls of
1848 COERCE and BIT-* functions when they return freshly constructed
1849 objects. (reported by Nikodemus Siivola and Paul Dietz)
1850 * optimization: improved the MIPS versions of generic (in the
1851 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1852 * optimization: direct conversion of (unsigned-byte 32) to floats on
1853 the PowerPC platform.
1854 * optimization: structure instances with raw slots now use less
1855 memory, and probably show better memory locality. (thanks to
1857 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1858 are treated as requests for stack allocation on the x86-64,
1859 Alpha-32, PPC and SPARC platforms.
1860 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1861 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1862 (thanks to Tony Martinez)
1864 ** gcing a dead thread can no longer lead to lockups
1865 ** threads block signals until they are set up properly
1866 ** errno is no longer shared by threads
1867 ** interrupt-thread restores the eflags register on x86
1868 ** fixed some lockups due to gc/thread interaction
1869 * fixed some bugs revealed by Paul Dietz' test suite:
1870 ** invalid dotted lists no longer raise a read error when
1871 *READ-SUPPRESS* is T
1872 ** signal an error if a symbol that names a declaration is used
1873 as the name of a type, or vice versa
1874 ** allow using the (declare (typespec var*)) abbreviation for
1875 (declare (type typespec var*)) with all type specifiers
1876 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1877 ** always return NIL from PPRINT-POP when OBJECT is NIL
1878 ** don't signal errors when pretty-printing malformed LABELS,
1879 FLET or MACROLET forms
1880 ** declarations in a DOLIST body are also in scope for the
1882 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1884 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1887 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1888 * fixed cross-compiler leakages that prevented building a 32-bit
1889 target with a 64-bit host compiler.
1890 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1891 opened with :IF-EXISTS :OVERWRITE.
1892 * fixed bug 281: error for an invalid qualifier in a short-form method
1893 combination method is not signalled until the faulty method is called.
1894 * bug fix: iteration variable type inferrer failed to deal with open
1895 intervals. (reported by Alan Shields)
1896 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1897 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1898 or more runtime options were provided to the sbcl binary.
1899 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1900 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1901 * optimization: there's now a fast-path for fixnum arguments in the
1902 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1903 * optimization: the code generated on x86-64 is more compact thanks
1904 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1905 * bug fix: floating-point exception handling now partly works on
1906 x86-64. (Thanks to James Knight)
1907 * improvement to the MIPS backend from Thiemo Seufer: C-style
1908 64-bit long long arguments and return values to alien functions
1910 * fixed some bugs revealed by Paul Dietz' test suite:
1911 ** the type-error signalled from WARN has a filled-in DATUM slot.
1912 ** the type-error required when a stream is not associated with
1913 a file has the stream as its datum.
1914 ** type-errors on single-floats on x86-64 no longer have
1915 :INVALID-OBJECT as the datum
1916 ** the type-errors signalled for invalid function names now have
1917 a correct expected type
1918 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1919 for structures defined with a :TYPE. Documentation strings for
1920 typed structures are no longer immediately discarded
1921 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1922 broadcast streams and synonym streams. FILE-LENGTH now also works
1923 on broadcast streams.
1925 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1926 * incompatible change: the --noprogrammer option, deprecated since
1927 version 0.7.5, has been removed. Please use the equivalent
1928 --disable-debugger option instead.
1929 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1931 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1932 inoperational for a while now) has been completely removed.
1933 * null lexical environments are now printed as #<NULL-LEXENV>,
1934 significantly reducing the amount of clutter in typical backtraces.
1935 * documentation on weak pointers, finalization, and after GC hooks
1936 has been added to the manual.
1937 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1938 as other specialized array types, is much faster. SUBSEQ and
1939 COPY-SEQ on such arrays have also been sped up.
1940 * optimization: EQL is now more efficient when at least other argument
1941 is known to be of type (OR FIXNUM (NOT NUMBER)).
1942 * fixed bug: the runtime is now less vulnerable to changes in the
1943 size of the SBCL object on OS X, and virtual memory is reserved for
1944 all spaces that need to be at a fixed address.
1945 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1946 * fixed bug: finalizers and after GC hooks that cause consing are now
1948 * fixed bug: compiler error messages and summaries are now printed to
1949 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1950 * fixed inference of the upper bound of an iteration variable.
1951 (reported by Rajat Datta).
1952 * fixed bug 211e: calling local functions with duplicated constant
1953 keyword argument no longer causes a bogus style warning about an
1955 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1956 ftype declaration to be disregarded. (reported by Dave Roberts)
1957 * fixed bug 373: caused by erronous compilation of references to alien
1958 variables in the runtime on ppc/darwin.
1959 * fixed bug 376: CONJUGATE type deriver.
1960 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1961 (reported by Baughn on #lisp)
1962 * a cross-compiler bug on non-x86 platforms has been identified and
1963 fixed. (thanks to Bruno Haible)
1964 * improvements to the MIPS runtime code for increased stability.
1965 (thanks to Thiemo Seufer)
1966 * increased the maximimum compact environment size to allow
1967 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1968 * partial workaround for bug 135: don't name the function we're
1969 calling for hairy cases of EVAL, so as not to accumulate one environment
1970 entry per call to EVAL. (reported by Kevin Reid)
1971 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1972 * optimization: type testing for non-vector arrays should be faster.
1973 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1975 * bug fix: setting 31st element of a bit vector to zero did not work
1977 * bug fix: redefining a class definition which failed due to a
1978 previous accessor / function clash now works (but see BUGS entry
1979 #380 for more problems in this area). (thanks to Zach Beane)
1980 * the long form of DEFINE-METHOD-COMBINATION disables method group
1981 checking when given a single method group with pattern *. (thanks
1983 * on x86 compiler supports stack allocation of results of simple
1984 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1985 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1986 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1987 * fixed some bugs related to Unicode integration:
1988 ** the restarts for recovering from input and output encoding
1989 errors only appear when there is in fact such an error to
1991 * fixed some bugs revealed by Paul Dietz' test suite:
1992 ** MISC.549 and similar: late transformation of unsafe type
1993 assertions into derived types caused unexpected code
1995 ** SCALE-FLOAT type deriver is less wrong.
1996 ** type derivers for EXP, LOG and similar functions compute result
1997 types for complex arguments better.
1998 ** (MISC.563) CONJUGATE type deriver works for very restricted
2000 ** out-of-line type testers for character strings are available.
2001 ** EQUAL compiler transform understands specialness of objects
2003 ** accessing double-floats stored on the stack now works on x86-64.
2004 ** debugger internals could sometimes create invalid lispobjs,
2005 resulting in GC crashes.
2006 ** MISC.548: type check weakening can convert required type into
2008 ** initialization forms for bindings are not in scope of free special
2011 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2012 * incompatible change: thread support for non-NPTL systems has
2013 been removed - locking is buggy and unreliable. A threaded
2014 SBCL build will now warn at startup and refuse to create
2015 new threads, unless futex support is detected in the kernel
2016 * incompatible change: the top level REPL now has only an ABORT
2017 restart associated with it, not TOPLEVEL and ABORT as it used to.
2018 TOP and TOPLEVEL are now available as debugger commands for
2019 returning to the top level.
2020 * incompatible change: forms evaluated in the REPL now use the
2021 global optimization policy.
2022 * incompatible change: user- and system-initialization files are
2023 no longer processed with LOAD, but by READ and EVAL; hence the
2024 global optimization policy, startup package, readtable, etc,
2026 * The .fasl file format number has been incremented because of
2027 various incompatible changes.
2028 * internal entry point details and argument counts no longer appear
2029 in backtraces unless explicitly requested by setting
2030 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2031 * built-in and standard functions no longer have names like "top
2032 level local call to FOO".
2033 * fixed bug 32: functions defined in non-null lexical environments
2034 now have more legible printed representation
2035 * fixed bug 33: functions defined in non-null lexical environemnts
2036 are now more amenable to inspection by INSPECT.
2037 * workaround for bug 354: XEPs no longer appear in backtraces unless
2038 explicitly requested.
2039 * fixed bug: receiving the signal which results from attempting to
2040 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2041 SIGBUS on other *BSDs) on architectures where the C stack is also
2042 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2043 to debugger confusion or wild instability. (reported by Cheuksan
2045 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2046 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2047 (reported by Lutz Euler)
2048 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2049 compiled code. (reported by Paul Dietz)
2050 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2051 specialized parameter to a method to something that is not TYPEP
2052 the specializer is now possible.
2053 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2054 face of package deletion.
2055 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2056 pathname. (reported by David Tolpin on c.l.l)
2057 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2058 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2059 STRING-STREAM and FILE-STREAM, respectively.
2060 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2061 than previously; in addition, its test suite is now run on build.
2062 (thanks to Luke Gorrie)
2063 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2064 * optimization: added a immediate representation for single-floats
2066 * fixed some bugs revealed by Paul Dietz' test suite:
2067 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2068 ** attempting to create a package with a colliding nickname causes
2069 correctable errors to be signalled.
2070 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2071 ** several x86-64 backend bugs related to sign-extension of immediate
2074 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2075 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2076 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2078 * fixed loading of multiply forward-referenced layouts.
2079 (thanks to Cheuksan Wang)
2080 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2081 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2082 (see COMPILE-FILE documentation for details of :PRINT option)
2083 * fixed bugs 19 and 317: fixed-format floating point printing is
2084 more accurate. This also fixes a bug reported by Adam Warner
2085 related to the ~@F format directive.
2086 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2088 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2089 dispatch table if the from-char is a dispatch macro character.
2090 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2091 by the wordsize no longer ignore the last word. (reported by Lutz
2093 * fixed bug in type checking of dynamic-extent variables. (reported
2095 * optimization: sequence traversal functions use their freedom to
2096 coerce function designators to functions.
2097 * optimization: code with many calls to CLOS methods specialized on
2098 CLOS classes has had redundant type checks removed.
2099 * optimization: added declarations to speed up operations that access
2100 the internal character database (for example STRING-UPCASE)
2101 * optimization: comparison operations between floats and sufficiently small
2102 fixnums no longer create extra rationals
2103 * fixed some bugs related to Unicode integration:
2104 ** portions of multibyte characters at the end of buffers for
2105 character-based file input are correctly transferred to the
2106 start of the buffer at the next read.
2107 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2108 passing it through to OPEN.
2109 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2110 argument given to internal calls to OPEN.
2111 ** the built-in comment readers (introduced by character sequences
2112 ";" and "#|") are more forgiving to encoding errors; they will
2113 STYLE-WARN and then attempt to resync the stream at a character
2114 boundary. (thanks to Teemu Kalvas)
2115 * fixed some bugs in the x86-64 port:
2116 ** Negative short int return values from c-calls are sign-extended
2118 ** The stack is aligned to 16-bytes for c-calls, as required by
2119 the ABI. (thanks to Cheuksan Wang)
2120 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2121 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2122 for enums. (thanks to Vincent Arkesteijn)
2123 ** Multiple small optimizations and bugfixes for floating point
2125 * fixed some bugs revealed by Paul Dietz' test suite:
2126 ** Space, Tab, Linefeed, Return and Page have the invalid
2127 secondary constituent character trait.
2128 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2130 ** WITH-INPUT-FROM-STRING should only update the index place on
2132 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2133 works more reliably.
2134 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2135 with one-dimensional arrays with fill pointers.
2136 ** TYPE-OF failed on a complex with an integer realpart and a
2138 ** compiler failure during type inference for the code of form
2139 (IF (EQL X (THE ...)) ...) (MISC.535).
2141 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2142 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2143 platform. The port supports 61-bit fixnums, large memory spaces
2144 and reloading shared object files.
2145 * enhancement: saving cores with foreign code loaded is now
2146 supported on ppc/Darwin in addition to the previously supported
2148 * enhancement: the statistical profiler now walks deeper into the
2149 call stack for more meaningful call-graphs and accrued time
2150 reports (x86/x86-64 only). It also now reports time spent in
2152 * enhancement: it is now possible to trace most individual methods
2153 of a generic function in addition to tracing the generic function
2155 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2157 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2158 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2159 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2160 produces an error. (thanks to Vincent Arkesteijn)
2161 * bug fix: NAMESTRING on pathnames with :WILD components in their
2162 directories works correctly. (thanks to Artem V. Andreev)
2163 * fixed bug 125: compiler preserves identity of closures. (reported
2165 * bug fixed: functions with &REST arguments sometimes failed with
2166 "Undefined function" when compiled with (DEBUG 3). (reported by
2168 * bug fix: overflow during compiling of setting element of a bit
2169 vector with constant index and value. (reported by Timmy Douglas)
2170 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2171 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2172 * on x86 compiler partially supports stack allocation of dynamic-extent
2174 * GO and RETURN-FROM do not check the extent of their exit points
2175 when compiled with SAFETY 0.
2176 * fixed some bugs related to Unicode integration:
2177 ** encoding and decoding errors are now much more robustly
2178 handled; it should now be possible to recover even from invalid
2179 input or output to the terminal. (thanks to Teemu Kalvas)
2180 ** provided a first cut at implementing STRING-TO-OCTETS and
2181 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2182 ** altered the SB-MD5 contributed module slightly, changing the
2183 interface just enough for it to be supportable for builds where
2184 lisp characters are not eight bits.
2185 * fixed some bugs revealed by Paul Dietz' test suite:
2186 ** the FORMATTER-generated functions for ~V[ conditionals require
2187 the correct number of arguments.
2188 ** READ-FROM-STRING returns the mandated second value when applied
2189 to displaced strings.
2190 ** the #\Rubout and #\Backspace characters are treated as invalid
2191 constituent characters by the tokenizer.
2193 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2194 * new feature: reloading changed shared object files with
2195 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2196 * new feature: references to foreign variables and functions
2197 can now be compiled and loaded before the corresponding shared
2198 object file is loaded, as long as the foreign definitions are
2199 available at runtime.
2200 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2201 supported. (thanks to Dan Debertin)
2202 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2203 just keywords. (thanks to Vincent Arkesteijn)
2204 * fixed bug #331: structure-class instances corresponding to
2205 DEFSTRUCT forms are now created eagerly.
2206 * fixed bug #345: backtraces from calls to undefined functions work
2207 on x86 as well. Related bug #61 is now also partially fixed on x86
2208 (backtraces from throws to unknown catch tags.)
2209 * bug fix: lambda-list parsing is now stricter vrt. order and number
2210 of lambda-list keywords.
2211 * bug fix: as specified by AMOP, an error is signalled if a
2212 class-option appears multiple times in a DEFCLASS form. (reported
2214 * bug fix: RANDOM can be compiled when the compiler derives the type
2215 of its numeric argument as a disjoint set of small integers.
2216 (reported by Paul Dietz)
2217 * bug fix: starting a core saved with shared objects loaded when
2218 those objects are not available no longer causes threaded SBCL to
2219 hang. (reported by Sean Ross)
2220 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2222 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2223 argument, which must always be NIL. (reported by Kalle Niemitalo)
2224 * bug fix: printing 1.0d+23 no longer results in an error.
2225 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2226 * bug fix: structure slot setters preserve evaluation order. (thanks
2228 * fixed some bugs related to Unicode integration:
2229 ** RUN-PROGRAM can allow its child to take input from a Lisp
2230 stream. (reported by Stefan Scholl)
2231 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2232 locales. (reported by Ken Causey)
2233 * fixed some bugs revealed by Paul Dietz' test suite:
2234 ** INCF, DECF and REMF evaluate their place form as specified in
2236 ** FORMATTER expands ~{ ~} iteration directives with V or #
2237 parameters correctly.
2238 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2239 consequent uses no arguments correctly.
2240 ** the system has a partial understanding of the (COMPLEX RATIO)
2242 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2245 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2246 * new feature: a build-time option (controlled by the :SB-UNICODE
2247 keyword feature, enabled by default) for building the system with
2248 support for the entire 21-bit character space defined by the
2250 * new feature: the system now has rudimentary external-format
2251 support; the primary user-visible change at this time is that
2252 characters with the high bit set (such as lower-case-e-acute) will
2253 print correctly to a terminal in a UTF-8 environment.
2254 * minor incompatible change: BASE-CHAR no longer names a class;
2255 however, CHARACTER continues to do so, as required by ANSI.
2256 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2257 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2258 has been moved to the SB-EXT package (temporarily re-exported from
2260 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2261 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2263 * fixed bug #351: better error-handlind and reporting for malformed
2265 * fixed bug #350: bignum-printing is now more memory-efficient,
2266 allowing printing of very large bignums, eg. (expt 2 10000000).
2267 (reported by Bruno Haible)
2268 * fixed bug #302: better primitive-type selection for intersection
2270 * fixed bug #308: non-graphic characters now all have names, as
2271 required. (reported by Bruno Haible)
2272 * bug fix: redefining a class with different superclasses now correctly
2273 removes it from the direct-subclasses of its previous superclasses.
2274 (reported by David Morse)
2275 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2276 new value now works. (reported by Bruno Haible)
2277 * bug fix: correct canonicalization of multiple non-standard slot
2278 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2279 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2280 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2281 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2282 now exists, an signals an error.
2283 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2284 during macroexpansion for non-iteration variables as well. (reported
2285 by Bruno Haible for CMUCL)
2286 * bug fix: Cyclic structures and unprintable objects in compiler
2287 messages no longer cause errors. (reported by Bruno Haible)
2288 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2289 and READ-FROM-STRING all now return a primary value of NIL if
2290 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2291 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2292 (reported by Bruno Haible for CMUCL)
2293 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2294 do not designate a condition. (reported by Bruno Haible for
2296 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2297 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2298 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2299 permitting resolution in favour of any of the conflicting symbols.
2300 (reported by Bruno Haible for CMUCL)
2301 * bug fix: EQUAL compiler optimizations is less aggressive on
2302 strings which can potentially compare true despite having distinct
2303 specialized array element types.
2304 * bug fix: unit enumerations can be defined without dividing by
2305 zero. (thanks to Vincent Arkesteijn)
2306 * FORMAT compile-time argument count checking has been enhanced.
2307 (report from Bruno Haible for CMUCL)
2308 * a partial workaround for the bug 262: the compiler does not try to
2309 inline-expand a local function doing RETURN-FROM from a deleted
2310 BLOCK. (thanks to Peter Denno for the bug report and to David
2311 Wragg for the simple test case)
2312 * fixed some bugs revealed by Paul Dietz' test suite:
2313 ** RENAME-PACKAGE allows all package designators as new package
2315 ** constraint propagation and lambda variable substitution are
2316 more cautious in dealing with partially deleted code.
2317 ** compiler sometimes lost reoptimization passes.
2318 ** CERROR, given a condition as condition designator, passes its
2319 remaining arguments to the continue format control without
2321 ** Case-altering FORMAT directives work correctly on non-ASCII
2323 ** The REAL type specifier handles bounds outside the floating
2324 point ranges without signalling FLOATING-POINT-OVERFLOW.
2325 ** Functions with IR1-transformations can create intercomponent
2326 references to global functions.
2327 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2329 ** FORMAT ~R treats a nil value for its first parameter correctly.
2331 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2332 * enhancement: saving cores with foreign code loaded is now
2333 supported on x86/NetBSD and sparc/Linux in addition to the previously
2334 supported platforms.
2335 * bug fix: on some platforms repeated installations caused multiple
2336 copies of HTML documentation to be installed -- should not happen
2337 any more. (reported by Stefan Scholl)
2338 * bug fix: parsing self-recursive alien record types multiple times
2339 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2340 original patch by Helmut Eller for CMUCL)
2341 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2342 (thanks to Richard Kreuter)
2343 * bug fix: defining classes whose accessors are methods on existing
2344 generic functions in other (locked) packages no longer signals
2345 bogus package lock violations. (reported by François-René Rideau)
2346 * bug fix: special variables as DEFMETHOD parameters no longer have
2347 associated bogus type declarations. (reported by David Wragg and
2349 * bug fix: read-write consistency on streams of element-type
2350 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2351 * bug fix: redefiniton of the only method of a generic function with
2352 no DEFGENERIC no longer emits a full WARNING. In addition,
2353 redefinition of generic functions with no DEFGENERIC to an
2354 incompatible lambda list now signals an error. (thanks to Zach
2356 * bug fix: DEFGENERIC now works even when there's a function of the
2357 same name in an enclosing lexical environment. (thanks to Zach
2359 * fixed compiler failure, caused by instrumenting code during
2360 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2361 * optimization: added loop analysis and improved register allocation
2362 * fixed some bugs revealed by Paul Dietz' test suite:
2363 ** POSITION on displaced vectors with non-zero displacement
2364 returns the right answer.
2365 ** (SIMPLE-STRING) is a valid type specifier for sequence
2367 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2369 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2370 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2372 ** The justification version of the FORMAT ~< directive treats
2373 non-zero minpad parameter correctly.
2375 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2376 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2377 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2378 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2379 the supported interface.
2380 * new feature: Single-stepping of code compiled with DEBUG 2 or
2381 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2382 * new feature: saving cores with foreign code loaded is now
2383 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2384 Timothy Moore's work for CMUCL)
2385 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2386 parameters to * instead of NIL if no initform is supplied.
2387 (reported by Johan Bockgaard)
2388 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2389 correctly when FUNCALL appears as the car of the form. Note:
2390 despite this FUNCALL forms are not currently subject to
2391 compiler-macro expansion. (port of Raymond Toy's fix for the
2392 same from CMUCL, reported by Johan Bockgaard)
2393 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2394 (thanks for Teemu Kalvas)
2395 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2396 case instead of terminating the iteration (thanks for Julian
2397 Squires, Sean Champ and Raymond Toy)
2398 * bug fix: incorrect expansion of defgeneric that caused a style
2399 warning. (thanks for Zach Beane)
2400 * bug fix: slot accessor effective method computation works properly
2401 for classes with multiple non-standard applicable methods on
2402 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2403 * on x86 compiler supports stack allocation of results of LIST and
2404 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2405 CMUCL implementation by Gerd Moellmann)
2406 * fixed some bugs revealed by Paul Dietz' test suite:
2407 ** FORMAT strings with both the ~<~:;~> form of the justification
2408 directive and pretty-printing directives cause an error.
2410 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2411 * incompatible change: the internal functions
2412 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2413 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2414 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2415 instead of the old functions.
2416 * new feature: on platforms where "dladdr" is available foreign
2417 function names now appear in backtraces. (based on Helmut Eller's
2419 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2420 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2422 * documentation: ASDF manual is now distributed with SBCL.
2423 * bug fix: SBCL can now load its contributed modules with REQUIRE
2424 even if the system-provided entries have been removed from
2425 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2427 * bug fix: backtraces involving undefined functions or assembly
2428 routines are more informative. (thanks to Brian Downing)
2429 * bug fix: mutually referent alien structures now work correctly.
2430 (reported by Rick Taube)
2431 * bug fix: structures defined by WITH-ALIEN can be referred to
2432 within other definitions in the same WITH-ALIEN.
2433 * bug fix: division operators (MOD, TRUNCATE and the like) with
2434 constant zero divisors and integer dividends no longer generate
2436 * bug fix: provide default methods for INPUT-STREAM-P and
2437 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2438 * bug fix: improve the detection and resolution of MOP metacycles
2439 (where effective method computation depends on the generic function
2440 having its effective method computed). (reported by Bruno Haible)
2441 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2442 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2443 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2444 * bug fix: pathnames with relative directory components can be
2445 represented relative to default pathnames.
2446 * optimization: in taking the GCD of bignums, reduce the two bignums
2447 to approximately the same size (using Euclid's algorithm) before
2448 applying the more sophisticated binary GCD. (thanks to Juho
2450 * optimization: COUNT on bitvectors now operates word-at-a-time.
2451 * optimization: ASH with a positive, but not necessarily constant,
2452 (leftwards) shift, when in a modular context, is compiled to a
2454 * fixed some bugs revealed by Paul Dietz' test suite:
2455 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2456 if the corresponding argument is NIL.
2457 ** FORMAT directives accepting a minpad parameter treat negative
2459 ** literal commas as character parameters to FORMAT directives are
2461 ** literal spaces directly after ~<Newline> directives within a
2462 format-logical-block (~:< ~@:>) do not induce :FILL-style
2463 conditional newlines.
2464 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2465 as specified: it no longer includes conditional newlines).
2466 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2468 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2469 :PER-LINE-PREFIX argument does not evaluate to a string.
2471 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2472 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2473 the manual for details; for now, package locks can be disabled by
2474 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2475 if no major problems are found then it is likely that they will be
2476 compiled in unconditionally.
2477 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2478 unsupported operators on all platforms. To load a shared library
2479 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2480 object file, link it into a shared library outside of SBCL and
2481 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2483 * bug workaround: ROOM T and the GC were not working together
2484 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2485 reporting facility aren't preserved. That reporting has been
2486 disabled, so now until and unless someone figures out how to make it
2487 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2488 * minor incompatible change: as threatened around sbcl-0.8.0, the
2489 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2490 The symbols are also exported from SB-PCL for backwards
2491 compatibility, but more so than before SB-PCL should be treated as
2492 an implementation-internal package.
2493 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2495 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2496 It is known as of this release not to work on the Alpha, however.
2497 * fixed bug #167: errors signalled due to illegal syntax in method
2498 bodies are now more legible.
2499 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2500 designators and can hence be used with TYPEP.
2501 * fixed bug #333: CHECK-TYPE now ensures that the type error
2502 signalled, if any, has the right object to be accessed by
2503 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2504 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2505 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2507 * fixed bug #334: programmatic addition of slots using specialized
2508 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2509 and :ALLOCATION :CLASS slots.
2510 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2511 float exponents. (rereported by Peter Seibel)
2512 * fixed a bug: #\Space (and other whitespace characters) are no
2513 longer considered to be macro characters in standard syntax by
2514 GET-MACRO-CHARACTER.
2515 * fixed bug: initialization of condition class metaobjects no longer
2516 causes an instance of the condition to be created. (reported by
2518 * fixed bug: it is now possible to have slots such that
2519 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2520 description is neither :INSTANCE nor :CLASS.
2521 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2522 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2523 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2524 non-standard SLOT-DEFINITION-ALLOCATION values.
2525 * fixed bugs: various race conditions handling exiting threads.
2526 CL-PPCRE's thread torture test now passes.
2527 * fixed arguably-a-bug: GC time proportional to number of threads in
2528 system even when most of them are idle
2529 * optimization: improved performance of BIT and SBIT on bit-vectors.
2530 * .fasl file incompatibility: The fasl file version number has
2531 been incremented because of changes associated with package locks.
2533 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2534 * minor incompatible change: the system no longer provides
2535 optimization hints (conditions which are TYPEP
2536 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2537 mode; these hints are still emitted when higher SPEED optimization
2539 * new contrib module: a sampling profiler (profiling by statistical
2540 sampling, rather than by instrumenting functions) is available as
2541 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2542 * the behaviour of the standard function ED is now customizeable by
2543 third parties through a hook variable: see ED's documentation
2544 string for information on the protocol.
2545 * the compiler no longer emits efficiency notes for (FUNCALL X)
2546 when the type of X is uncertain under default optimization
2548 * fixed bug 276: mutating a binding of a specialized parameter to a
2549 method to something that is not TYPEP the specializer is now
2551 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2552 exhibit the required behaviour on the x86 platform. (thanks to
2553 Peter van Eynde, Eric Marsden and Bruno Haible)
2554 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2555 even for difficult arguments. (reported by Peter Graves)
2556 * fixed bug 141a: the backquote printer now descends quoted
2558 * fixed another bug in backquote printing: no more destructive
2559 modification of the form's list structure. (reported by Brian
2561 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2562 move between its address being taken and the call to
2563 interrupt_thread, fixing a crashing race condition.
2564 * the SB-POSIX contrib implementation has been adjusted so that it
2565 no longer exhibits ridiculously poor performance when constructing
2566 instances corresponding to C structs.
2568 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2569 * minor incompatible change: the sb-grovel contrib now treats C
2570 structures as alien (in the sense of SB-ALIEN) objects rather than
2571 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2572 has implications for memory management of client code
2573 (sb-grovel-returned objects must now be manually managed) and for
2574 type safety (alien objects now have full types).
2575 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2576 used to control emission of compiler diagnostics, rather than the
2577 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2578 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2579 quality should be considered deprecated.
2580 * install.sh now installs the user manual as well
2581 * (not quite a new documentable feature, but worth considering in
2582 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2583 beginnings of a semantically meaningful condition hierarchy is
2584 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2586 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2587 designator as the defaults argument.
2588 * fixed bug: Displaced arrays whose displaced-to array has become
2589 too small now cause ARRAY-DIMENSION to signal an error, providing
2590 for safer bounds-checking. (reported by Bruno Haible)
2591 * fixed bug: DEFCLASS slot definitions with identical :READER and
2592 :WRITER names now signal a reasonable error. (reported by Thomas
2594 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2596 * fixed bug: Invalid slot specification errors now print correctly.
2597 (thanks to Zach Beane)
2598 * fixed bug 320: Shared to local slot value transfers in class
2599 redefinitions now happen corrently from superclasses as
2600 well. (reported by Bruno Haible)
2601 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2603 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2604 improper lists correctly. (reported by Bruno Haible)
2605 * fixed bug 313: source-transform for <fun-name> was erroneously
2606 applied to a call of a value of a variable with name <fun-name>.
2607 (reported by Antonio Menezes Leitao)
2608 * fixed bug 307: The obsolete instance protocol ensures that
2609 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2610 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2611 remove dead unknown-values globs from the middle of the stack.
2612 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2613 easier to resume long computations after using *BREAK-ON-SIGNALS*
2614 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2615 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2616 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2617 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2618 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2619 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2621 * fixed a (fairly theoretical) bug in string printing: if
2622 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2623 does not have array-element-type equal to the most general string
2625 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2626 function name. (thanks to Nikodemus Siivola)
2627 * fixed bug: compile-time format string checker failed on
2628 non-closed ~{. (reported by Thomas F Burdick)
2629 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2630 don't warn on legal loop code involving more than one aggregate
2631 boolean. (thanks to Nikodemus Siivola)
2632 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2633 in keyword position to MAKE-PACKAGE was not being reported as
2635 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2636 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2638 * fixed bugs 280 and 312: the checking for multiple definitions in a
2639 file is less likely to become confused by uses of inline
2641 * fixed bug: the #S reader macro performs the keyword coercion
2642 specified for slot names. (reported by Kalle Niemitalo)
2643 * fixed bug: lambda lists may contain symbols whose names start with
2644 & but are not lambda-list-keywords; their occurrence triggers a
2646 * fixed bug 321: define-method-combination argument lambda lists do
2647 not cause invalid code to be generated when &optional or &aux
2648 variables have default values. (reported by Bruno Haible)
2649 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2650 have CLOS classes; this appears also to have fixed reported
2651 instability in stack exhaustion detection.
2652 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2653 MUFFLE-WARNING when no associated restart is present is now
2655 * optimization: rearranged the expansion of various defining macros
2656 so that each expands into only one top-level form in a
2657 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2659 * optimization: used a previously little-used slot in symbols to
2660 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2662 * fixed some bugs revealed by Paul Dietz' test suite:
2663 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2665 ** Fixed an optimization bug related to inheritance of initforms
2666 from local to shared slots.
2667 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2668 ** CAST optimizer forgot to flush argument derived type.
2669 ** print/read consistency on floats is now orders of magnitude
2670 more likely. (thanks also to Bruno Haible for a similar report
2672 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2673 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2674 for objects of type REAL. Make it so.
2675 ** SXHASH is specified (infelicitously) to respect similarity,
2676 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2677 it so. (thanks to Markus Ziegler)
2678 ** on the Alpha, the compiler succeeds in compiling functions
2679 returning a known number of arguments greater than 63.
2680 ** fixed handling of invalid NIL arguments in keyword position
2682 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2684 ** provided a readably-printable representation for RANDOM-STATE
2686 ** ensured that pathnames, where they have a namestring, always
2687 print using #P"..." syntax.
2689 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2690 * Support for the forthcoming 2.0 version of the NetBSD kernel
2691 running on x86 hardware has been added. (thanks to Perry
2692 E. Metzger most immediately, and others for their past work)
2693 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2694 current release version; SBCL's OpenBSD support had been broken
2695 since about the time of OpenBSD's switch to ELF binary format.)
2696 * [placeholder for DX summary]
2697 ** user code with &REST lists declared dynamic-extent, under high
2698 speed or space and low safety and debug optimization policy.
2699 * The manual has been converted to Texinfo format and the debugger
2700 chapter from the cmucl manual has been added.
2701 * A facility has been added to extract documentation strings from
2702 sbcl and store them as Texinfo-formatted snippets for inclusion in
2703 the manual (via Texinfo's @include directive)
2704 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2705 the test case to Dave Roberts)
2706 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2707 pointers. (reported by Sean Ross)
2708 * bug fix: PROFILE output is printed nicely even for large numerical
2709 values. (thanks to Zach Beane)
2710 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2711 greater than 32 handle EOF correctly.
2712 * bug fix: on X86 an immediate argument of the IMUL instruction is
2713 correctly printed in disassembly. (thanks to Lutz Euler)
2714 * bug fix: class slots in redefined classes preserve their old
2715 values. (thanks to Bruno Haible and Nikodemus Siivola)
2716 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2717 succeeds. (reported by Marco Baringer)
2718 * fixed some bugs revealed by Paul Dietz' test suite:
2719 ** READ-SEQUENCE now works on ECHO-STREAMs.
2720 ** RATIONALIZE works more according to its specification. (thanks
2723 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2724 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2725 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2726 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2727 should matter to you only if you rebind the printer control
2728 variables and then find you want different bindings in the
2729 debugger than in the ordinary execution of your program.)
2730 * The runtime build system has been tweaked to support building
2731 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2732 assembler and linker. (thanks to Nikodemus Siivola)
2733 * Unbound, undefined, undeclared variables now trigger full
2734 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2735 more often programmer error than deliberate exploitation of undefined
2737 * optimization: the hash algorithm for strings has changed to one
2738 that is less vulnerable to spurious collisions. (thanks to Juho
2740 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2741 needless bounds checking. (thanks to Juho Snellman)
2742 * optimization: implemented multiplication as a modular
2743 (UNSIGNED-BYTE 32) operation on the PPC backend.
2744 * fixed some bugs revealed by Paul Dietz' test suite:
2745 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2747 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2748 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2749 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2751 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2753 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2755 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2756 their output stream on EOF from read.
2757 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2758 have been read to end-of-file.
2759 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2761 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2762 description of determination of which consecutive characters
2764 ** Printing the "Space" character with escaping on now yields "#\\ ",
2765 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2766 ** Reading floating-point numbers with *READ-BASE* set to a number
2767 less than 10 works correctly.
2768 ** Reading floating-point numbers with *READ-BASE* set to a number
2769 more than 10 works correctly.
2770 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2771 the readtable currently in effect.
2773 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2774 * minor incompatible change: parsing of namestrings on a physical
2775 (Unix) host has changed; numbers after the final #\. in a
2776 namestring are no longer interpreted as a version field. This is
2777 intented to be largely invisible to the user, except that the
2778 meaning of the namestring "*.*.*" has changed: it now refers to a
2779 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2780 should usually be replaced by
2781 (make-pathname :name :wild :type :wild :version :wild)
2782 with the added benefit that this is more likely to be portable.
2783 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2784 option to OPEN now signals an error if the file being opened
2785 exists; this may have an impact on existing code.
2786 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2787 signals received, on the PowerPC platforms (both Linux and
2788 Darwin). (thanks to Pierre Mai for pointing out the location of
2790 * several fixes on OS X: The system now builds and runs cleanly on
2791 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2792 (thanks to Brian Mastenbrook)
2793 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2794 second-resolution: integer multiples of 1/3600 between -24 and 24.
2795 (thanks to Vincent Arkesteijn)
2796 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2797 type when called with 1 argument; PEEK-CHAR checked type of
2798 PEEK-TYPE only after having read first character from a
2799 stream. (reported by Peter Graves)
2800 * bug fix: the garbage collector now has much better locality
2801 behaviour, and in particular no longer treats all memory as being
2802 exhausted when many small objects point to each other in a deeply
2804 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2805 recognized as being TYPEP their class.
2806 * bug fix: the PUSHNEW documentation string has been corrected.
2807 (thanks to Vincent Arkesteijn)
2808 * bug fix: defaulting of the value for the last of an atypically
2809 large number of multiple values being bound was not being
2810 performed correctly on the Alpha or PPC platforms
2811 * optimization: implemented multiplication as a modular
2812 (UNSIGNED-BYTE 32) operation on the x86 backend.
2813 * optimization: SEARCH on simple-base-strings can now be open-coded.
2814 (see also contrib/compiler-extras.lisp for inspiration for
2815 teaching the compiler about the Boyer-Moore algorithm).
2816 * value, returned by MAX (and MIN) called with several EQUALP, but
2817 not EQL, arguments now does not depend on compiler settings.
2818 * fixed some bugs revealed by Paul Dietz' test suite:
2819 ** in stack analysis liveness information is propagated from
2820 non-local entry points.
2821 ** pathwise CAST removing failed when the CAST node did not start
2823 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2824 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2826 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2827 host is already defined.
2828 ** RENAME-FILE works on streams instead of signalling an internal
2830 ** PEEK-CHAR uses the current readtable when determining whether
2831 or not a character is whitespace.
2832 ** MERGE-PATHNAMES handles the case when the pathname does not
2833 specify a name while the default-pathname specifies a version.
2834 ** Pathnames now stand a better chance of respecting print/read
2836 ** Attempting to use standardized file system operators with a
2837 pathname with invalid :DIRECTORY components signals a
2839 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2841 ** DIRECTORY on logical pathnames is more correct.
2842 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2843 signal a TYPE-ERROR if their argument is not a stream.
2844 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2845 designator argument does not designate a stream.
2846 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2847 examining the synonym.
2848 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2850 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2851 element-type larger than ([UN]SIGNED-BYTE 32).
2853 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2854 * When built with the :SB-FUTEX feature, threaded builds now take
2855 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2856 for faster/more reliable mutex and condition variable support.
2857 * Incompatible change (but one you probably shouldn't have been using
2858 anyway): the interface and code for arbitrating between multiple
2859 threads in the same user session has been redesigned.
2860 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2861 error when called without an explicit environment argument.
2862 (thanks to Nikodemus Siivola)
2863 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2864 confused about their position. (thanks to Adam Warner and Gerd
2866 * bug fix: Pretty printing backquoted forms with unquotations in the
2867 argument list position of various code constructs such as LAMBDA
2868 now works correctly. (reported by Paul Dietz)
2869 * bug fix: Pretty printing unquotations no longer loses all
2870 stream position information.
2871 * optimization: performance of string output streams is now less
2872 poor for multiple small sequence writes.
2873 * optimization: performance of CSUBTYPEP in the presence of complex
2874 expressions involving CONS and NOT many times has been improved.
2875 (reported by Paul Dietz)
2876 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2878 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2879 streams for the REPL, for the debugger, and so on. (thanks to
2881 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2882 * fixed some bugs revealed by Paul Dietz' test suite:
2883 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2884 is ignored if it is too small, rather than propagating through
2885 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2886 ** extremely complex negations of CONS types were not being
2887 sufficiently canonicalized, leading to inconsistencies in
2889 ** VALUES tranformer lost derived type.
2891 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2892 * fixed a bootstrapping bug: the build process no longer assumes
2893 that the various BOOLE-related constants have the same value in
2894 host and target lisps. (noted by Paul Dietz' test suite on an
2895 SBCL binary built from CLISP)
2896 * The system can now be dynamically linked on the MIPS platform,
2897 which enables dynamic loading of foreign code from Lisp. (thanks
2898 to Ralf Baechle for discussions on the MIPS ABI)
2899 * The system now records debugging information for its own source
2900 files in a filesystem-position-independent manner, relative to
2901 the "SYS" logical host.
2902 * fixed a compiler bug: MV-LET convertion did not check references
2903 to the "max args" entry point. (reported by Brian Downing)
2904 * tweaked disassembly notes to be less confident about proclaiming
2905 some instruction as an LRA. (thanks to Brian Downing)
2906 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2907 now each have their own history, command character, and other
2908 characteristics. (thanks to David Lichteblau)
2909 * fixed some bugs revealed by Paul Dietz' test suite:
2910 ** compiler failure in compiling LOGAND expressions including a
2912 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2913 shift greater than 32.
2914 ** FLUSH-DEST did not mark blocks for type check regeneration.
2915 ** HANDLER-CASE failed to accept declarations in handler clauses
2916 in some circumstances.
2918 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2919 * New code in contrib/sb-introspect (still probably not entirely
2920 stable yet) provides some support for smart Lisp development
2921 environments like SLIME.
2922 * The conditions signalled for errors occurring when loading .fasl
2923 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2924 in a way which should help ASDF recover gracefully.
2925 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2926 stuff has been cleaned up. If you code contrib/ stuff, this might
2927 affect you, and you can look at contrib/README, contrib/STANDARDS,
2928 and/or the 0.8.4.27 diff to check.
2929 * In full calls the compiler now does not generate checks for declared
2930 argument types for all arguments.
2931 * various threading fixes
2932 ** and some experimental patches which didn't make it into
2933 the main tree for this release, but which are shipped in
2934 contrib/experimental-thread.patch as a possible fix for some
2935 failures (deadlock, spinning...) in GC-intensive multithreaded
2937 * fixed PPC build problem (source code incompatibility of different
2938 library versions): added offsetof() hackery which attempts to divine
2939 where glibc maintainers put uc_mcontext today
2940 * fixed bug 282: compiler does not trust type assertions while passing
2941 arguments to a full call.
2942 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2943 &OPTIONAL VALUES type parameter.
2944 * fix bug 214: algorithm for noting rejected templates is now more
2945 similar to that of template seletion. (also reported by rydis on
2947 * fixed bug 141b: printing backquoted information readably and prettily
2948 inserts a space where necessary.
2949 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2950 (CHARACTER 10) are now reported as errors, rather than propagated
2951 as unknown types. (reported by piso on #lisp)
2952 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2953 with values NIL and :ERROR. (thanks to Milan Zamazal)
2954 * fixed bug 191c: CLOS now does proper keyword argument checking as
2955 described in CLHS 7.6.5 and 7.6.5.1.
2956 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2957 longer raise an error; further, using a list as a for-as-arithmetic
2958 counter now raises a meaningful error.
2959 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2960 now understood by sequence creation functions such as MAKE-SEQUENCE
2962 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2963 a STRING-INPUT-STREAM.
2964 * compiler enhancement: SIGNUM is now better able to derive the type
2966 * type declarations inside WITH-SLOTS are checked. (reported by
2968 * fixed some bugs revealed by Paul Dietz' test suite:
2969 ** incorrect optimization of TRUNCATE for a positive first
2970 argument and negative second.
2971 ** compiler failure in let-convertion during flushing dead code.
2972 ** compiler failure while deriving type of TRUNCATE on an
2973 interval, containing 0.
2974 ** ASH of a negative bignum by a negative bignum count now returns
2976 ** intersection of CONS types now canonicalizes properly, fixing
2977 inconsistencies in SUBTYPEP.
2979 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2980 * incompatible change: The --disable-debugger command line
2981 option now clobbers the debugger at a more fundamental
2982 level, by redefining #'INVOKE-DEBUGGER instead of by
2983 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2984 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2985 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2986 enter the debugger REPL and then suffer recursive errors
2987 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2988 while under the new system, BREAK in --disable-debugger mode
2989 terminates the system just as an unhandled error would.
2990 * fixed compiler performance when processing loops with a step >1;
2991 * bug fix: DOCUMENTATION now retrieves generic function
2992 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2993 support has been systematized, and now supports the methods
2994 specified by ANSI, along with a default method and a method for
2995 slot documentation. (reported by Nathan Froyd)
2996 * bug fix: effective methods associated with a generic function are
2997 no longer cached over a change of that generic function's method
2998 combination. (reported by Andreas Fuchs)
2999 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3000 names a non-existent directory. (thanks to Andreas Fuchs)
3001 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3002 the PPC platform, lead to stack corruption; this has been fixed.
3003 (reported by Rainer Joswig)
3004 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3005 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3006 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3007 arithmetic in general, is now much more reliable on the PPC
3009 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3010 platform now returns the right answer.
3011 * optimization: restored some effective method precomputation in
3012 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3013 precomputation is now tunable.
3014 * optimization: compiler-internal data structure use has been
3015 reviewed, and changes have been made that should improve the
3016 performance of the compiler by about 20%.
3017 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3018 simple-base-strings and simple-bit-vectors is improved.
3019 * optimization: the optimization of 32-bit logical and arithmetic
3020 functions introduced in version 0.8.3 on the x86 has been
3021 implemented on the mips, ppc and sparc platforms; an
3022 implementation of the same facility, but for 64-bit arithmetic,
3023 has been added for the alpha.
3024 * microoptimization: the compiler is better able to make use of the
3025 x86 LEA instruction for multiplication by constants.
3026 * bug fix: in some situations compiler did not report usage of
3027 generic arithmetic in (SPEED 3) policy.
3028 * bug 145b fix: compiler used wrong type specifier while converting
3029 MEMBER-types to numeric.
3030 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3032 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3033 index argument now works properly on the Alpha platform.
3034 * bug fix: floating point exception treatment on the Alpha platform
3036 * bug fix: FILE-POSITION works much better on string input and
3037 output streams. (thanks to Nikodemus Siivola)
3038 * bug fix: many threading/garbage collection symptoms sorted.
3039 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3040 might be pseudo-atomic.
3041 * internal change: Stopping for GC is now done with signals not
3042 ptrace. GC is now done in whichever thread wanted it, instead of
3044 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3046 * bug fix: result form in DO is not contained in the implicit
3048 * incompatible change: ICR structure is changed; the value part of
3049 CONTINUATION is now called LVAR; corresponding functions are
3050 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3051 * added type deriver for ISQRT (thanks to Robert E. Brown).
3052 * bug fix: better support for loading from the command line when an
3053 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3055 * fixed some bugs revealed by Paul Dietz' test suite:
3056 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3057 ** ROUND and FROUND now give the right answer when given very
3058 small float arguments.
3059 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3061 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3062 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3063 ** LCM should return a non-negative integer.
3064 ** PARSE-INTEGER returned the index of a terminator instead of the
3065 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3066 ** PARSE-INTEGER returned an incorrect index being applied to a
3068 ** LCM with two arguments of 0 returns 0 rather than signalling
3070 ** unsigned addition of a 32-bit constant with the high bit set no
3071 longer causes an internal compiler error on the x86.
3072 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3073 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3074 with negative last argument.
3075 ** byte specifiers with zero size and position no longer cause
3076 an error during type derivation.
3077 ** bignum multiplication on the Alpha platform now returns the
3079 * porting: The system now builds on SuSE AMD64, although it still
3080 generates a 32-bit binary.
3081 * .fasl file incompatibility: The fasl file version number has
3082 been incremented (because of the changes to internal compiler
3083 data structures referred to above).
3085 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3086 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3087 more accurately, on the Darwin kernel running on PowerPC hardware.
3088 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3089 * Compiler code deletion notes now signal a condition of type
3090 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3091 an associated MUFFLE-WARNING restart.
3092 * The compiler now performs limited argument count validation of
3093 constant format strings in FORMAT, and where appropriate in ERROR,
3094 CERROR and WARN. (thanks to Gerd Moellmann)
3095 * New ASDF-INSTALL contrib can be used for automatic download and
3096 installation of third-party Lisp code from CCLAN or other sites
3098 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3099 which forces another thread to execute a function supplied by the
3101 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3102 now accept and act upon their :ELEMENT-TYPE keyword argument.
3103 (reported by Martin Atzmueller, Edi Weitz)
3104 * bug fix: FILE-POSITION now accepts position designators up to
3105 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3106 is the greater. (thanks to Patrik Nordebo)
3107 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3108 Kalle Olavi Niemitalo)
3109 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3110 printable. (reported by Eric Marsden)
3111 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3112 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3113 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3114 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3115 (thanks to Stig Sandoe)
3116 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3117 UNREAD-CHAR now correctly includes the unread character in the
3118 target sequence. (thanks to Gerd Moellmann)
3119 * bug fix in threaded builds: the system can now be suspended and
3120 resumed by shell job control with minimal disruption.
3121 * bug fixes in times and timezones >2038AD
3122 * better handling of "where is GNU make?" problem in build scripts
3123 (thanks to Nikodemus Siivola)
3124 * new optimization: inside a named function any reference to a
3125 function with the same name is considered to be a self-reference;
3126 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3127 optimization quality.
3128 * new optimization on x86: logical functions and + now have
3129 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3130 used when the result is truncated to 32 bits.
3131 * VALUES declaration is partially enabled.
3132 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3133 * fixed some bugs revealed by Paul Dietz' test suite:
3134 ** The system now obeys the constraint imposed by
3135 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3136 types form a lattice under type intersection.
3137 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3138 ** ASSOC now ignores NIL elements in an alist.
3139 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3140 and (1+ MOST-POSITIVE-FIXNUM) answers.
3141 ** The addition of a method with invalid qualifiers to a generic
3142 function does not cause an error to be signalled immediately;
3143 a warning is signalled, and the error is generated only on
3144 calling the generic function.
3145 * changed .fasl file version number, in part to add type codes for
3146 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3147 obscure ANSI requirements
3149 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3150 * fixed bug 148: failure to inline-expand a local function left
3151 garbage, confusing the compiler.
3152 * fixed bugs 3cd: structure slot readers perform type check if the
3153 slot can have an invalid value (i.e. it is either not initialized
3154 or can be written with a less specific slot writer).
3155 * bug fix: the compiler now traps array references to elements off
3156 the end of an array; previously, the bounds checking in some
3157 circumstances could go off-by-one.
3158 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3160 * type declarations for array element types now obey the description
3161 on the CLHS page "Declaration TYPE", as per discussions on
3162 sbcl-help around 2003-05-08. This means that a declaration
3163 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3164 declaration, all references to BAR will be asserted or assumed
3165 (with THE, so dependent on compiler policy) to involve objects of
3166 type FOO. Note that no such declaration is implied in
3167 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3168 * declared types of functions from the "Conditions"
3169 chapter. (reported by Paul Dietz)
3170 * bug fix: CERROR accepts a function as its first argument.
3171 * bug fix: NTH an NTHCDR accept a bignum as index
3172 arguments. (reported by Adam Warner)
3173 * optimization: character compare routines now optimize comparing
3174 against a constant character. (reported by Gilbert Baumann)
3175 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3176 argument now works properly on the MIPS platform.
3177 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3178 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3179 declared to have a functional type, does not cause an error
3181 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3182 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3183 * fixed bug: if last continuation of a deleted block has a
3184 destination, this destination should be deleted too. (reported by
3186 * fixed a bug in the bootstrap process: the host compiler's values
3187 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3188 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3189 test case from Patrik Nordebo)
3190 * improved the ability of the disassembler on the PPC platform to
3191 provide helpful disassembly notes.
3192 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3193 the class in more cases than previously.
3194 * bug fix: FILE-POSITION now understands :START and :END for
3195 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3196 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3197 8) data. (thanks to David Lichteblau)
3198 * bug fix: it is possible to add a method to a generic function
3199 without lambda list.
3200 * bug fix: reader failed to signal END-OF-FILE inside an
3201 object representation. (reported by Nikodemus Siivola)
3202 * fixed some bugs revealed by Paul Dietz' test suite:
3203 ** LAST and [N]BUTLAST should accept a bignum.
3204 ** condition slot accessors are methods.
3205 ** (VECTOR NIL) is a subtype of STRING.
3207 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3208 * minor incompatible change: some nonsensical specialized lambda
3209 lists (used in DEFMETHOD) which were previously ignored now signal
3211 * minor incompatible change: the system is now aware of the types of
3212 variables in the COMMON-LISP package, and will signal errors for
3213 most violations of these type constraints (where previously they
3214 were silently accepted).
3215 * minor incompatible change: COMPILE-FILE now uses the freedom
3216 afforded (ANSI 3.2.2.3) to use derived function types for
3217 functions defined in the same file. This also permits the system
3218 to warn on static type mismatches and function
3219 redefinition. (Currently it does not work with high DEBUG level.)
3220 * minor incompatible change: VALUES declaration is disabled.
3221 * When issuing notes, the compiler now signals a condition of type
3222 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3223 restart for use in user handlers. It is expected that the
3224 COMPILER-NOTE condition will eventually become a condition
3225 supertype to a hierarchy of note types, which will then be
3226 handleable in a similar fashion. However, at the moment, no such
3227 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3228 but it's an implementation detail, not a classification for the
3230 * Changes in type checking closed the following bugs:
3231 ** type checking of unused values (192b, 194d, 203);
3232 ** template selection based on unsafe type assertions (192c, 236);
3233 ** type checking in branches (194bc).
3234 * A short form of VALUES type specifier has ANSI meaning (it has
3235 increased the number of situations when SBCL cannot perform type
3237 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3238 keywords or constants is permissible.
3239 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3240 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3241 classes. (thanks to Antonio Martinez)
3242 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3243 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3244 argument to a pretty-print stream. Instead, it leaves any such
3245 operations to DESCRIBE-OBJECT methods.
3246 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3248 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3249 lambda lists. (thanks to David Lichteblau)
3250 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3251 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3252 (thanks to Gerd Moellmann)
3253 * fixed bug 246: increased compilation speed of long
3254 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3256 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3258 * DEFCONSTANT now throws a condition of type
3259 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3260 constant to a non-EQL value; CONTINUE and ABORT restarts
3261 respectively change and preserve the value.
3262 * fixed bug 63: The code walker, part of the implementation of CLOS,
3263 is now better at handling symbol macros.
3264 * bug fix: There is no longer an internal implementation type named
3265 CL:LENGTH. (reported by Raymond Toy)
3266 * bug fix: In macro-like defining macros/special operators the
3267 implicit block does not enclose the lambda list.
3268 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3269 enclosed in parentheses) are not suitable as type specifiers, and
3270 their use properly signals an error now.
3271 * bug fix: An argument count mismatch for a type specifier in code
3272 being compiled no longer causes an unhandled error at compile
3273 time, but signals a compile-time warning.
3274 * fixed simple vector readable printing
3275 * bug fix: DESCRIBE takes more care over whether the class
3276 precedence list slot of a class is bound before accessing it.
3277 (reported by Markus Krummenacker)
3278 * bug fix: FORMATTER can successfully compile pretty-printer format
3279 strings which use variants of the ~* directive inside.
3280 * bug fix: SEARCH now applies its TEST predicate to the elements of
3281 the arguments in the correct order. (thanks to Wolfhard Buss)
3282 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3283 in type inference. (thanks to Robert E. Brown)
3284 * bug fix: Reading in symbols with an explicit package name of ""
3285 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3286 * fixed some bugs revealed by Paul Dietz' test suite:
3287 ** NIL is now allowed as a structure slot name.
3288 ** Arbitrary numbers, not just REALs, are allowed in certain
3289 circumstances in LOOP for-as-arithmetic clauses.
3290 ** Multiple class redefinitions before slot access no longer
3291 causes a type error.
3292 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3293 association between the name and a class.
3294 ** Generic functions with non-standard method-combination and over
3295 five methods all of which return constants no longer return NIL
3296 after the first few invocations. (thanks to Gerd Moellmann)
3297 ** CALL-NEXT-METHOD with no arguments now passes the original
3298 values of the arguments, even in the presence of assignment.
3299 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3301 ** DISASSEMBLE works with closures and funcallable instances.
3302 ** ADD-METHOD now returns the generic function, not the new method.
3303 ** FIND-METHOD signals an error if the lengths of the specializers
3304 is incompatible with the generic function, even if the ERRORP
3306 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3307 which its argument is a member.
3308 ** DEFCLASS only redefines the class named by its class-name
3309 argument if that name is the proper name of the class;
3310 otherwise, it creates a new class.
3311 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3312 of the UNBOUND-SLOT condition to the name of the slot.
3313 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3314 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3315 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3316 SLOT-MAKUNBOUND in the specified fashion.
3318 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3319 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3320 cross-compilation host. As a consequence, we can now bootstrap our
3321 way up to SBCL starting with a bare gcc toolchain and human-readable
3322 source code (first the source to CLISP, then the source to SBCL).
3323 * A contributed module containing a partial implementation of the
3324 simple-streams interface has been included. (thanks to Rudi
3326 * A contributed module implementing the RFC1321 Message Digest
3327 Algorithm, known as MD5, has been included.
3328 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3329 no longer has any effect, as the code controlled by this feature
3330 has been deleted. (As far as we know, no-one has ever built using
3331 this feature, and its semantics were confused in any case).
3332 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3333 work on conditions (as required by the ANSI specification),
3334 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3335 expected behaviour on conditions. Users should note, however,
3336 that such behaviour is not required by the ANSI specification,
3337 and so use of this behaviour may render their code unportable.
3338 * fixed some bugs revealed by Paul Dietz' test suite:
3339 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3341 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3342 ** SXHASH on condition objects no longer returns NIL.
3343 ** :ALLOCATION :CLASS slots are better treated; their values are
3344 updated on class redefinition, and initforms inherited from
3345 superclasses are applied.
3346 ** REMOVE-METHOD returns its generic function argument even when
3347 no method was removed.
3348 ** SHARED-INITIALIZE now initializes the values of the requested
3349 slots, including those with :ALLOCATION :CLASS.
3350 ** ALLOCATE-INSTANCE now works on structure classes defined via
3351 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3353 ** SLOT-EXISTS-P now works on conditions, as well as structures
3355 ** MAKE-LOAD-FORM now has the required methods on
3356 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3357 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3358 keyword, and now implements the SLOT-NAMES argument.
3359 ** methods with &OPTIONAL arguments no longer allow too many
3360 arguments to be passed in the call without error.
3361 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3362 option is consistent with the required arguments of the generic
3363 function lambda list.
3364 * bug fix: REQUIRE accepts a string designator. (Thanks to
3366 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3367 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3368 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3369 * bug fix: The compiler now checks for duplicated variables in macro
3371 * bug fix: SETQ on globals returns the correct value.
3372 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3373 not a COMPILER-ERROR (followed by some other strange error on
3374 choosing the CONTINUE restart).
3375 * bug fix: make.sh and friends are now more consistent in the way that
3376 they look for GNU "make".
3378 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3379 * experimental native threads support (on x86 Linux >=2.4 only).
3380 This is not compiled in by default: you need to add :SB-THREAD to
3381 the target features. See the "Beyond ANSI" chapter of the manual
3383 * fix for longstanding nonANSIism: The old distinction between
3384 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3385 The return value from CL:FIND-CLASS is now a CLOS class, and
3386 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3387 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3388 * An interface to the MetaObject Protocol, as described in Kiczales,
3389 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3390 MIT Press, 1991, is available from the SB-MOP package.
3391 * incompatible change: the SB-PCL package should now be considered
3392 a private implementation detail, and no longer a semi-private MOP
3394 * minor incompatible change: due to rearrangement for threads, the
3395 control stack and binding stack are now allocated at arbitrary
3396 addresses instead of being hardcoded per-port. Users affected by
3397 this probably have to be doing advanced things with shared
3398 libraries, and will know who they are.
3399 * minor incompatible change: Previously, all --eval forms used were
3400 processed with READ before any of them were processed with EVAL.
3401 Now each --eval form is processed with both READ and EVAL before
3402 the next --eval form is processed. (Thus package operations like
3403 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3404 work as the user might reasonably expect.)
3405 * minor incompatible change: *STANDARD-INPUT* is now only an
3406 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3408 * minor incompatible change: Y-OR-N-P is now character-oriented,
3409 not line oriented. Also, YES-OR-NO-P now works without errors.
3410 (thanks to Antonio Martinez)
3411 * sb-aclrepl module improvements: an integrated inspector, added
3412 repl features, and a bug fix to :trace command.
3413 * Known functions, which cannot be open coded by the backend, are
3414 considered to be able to check types of their arguments. (fixing
3415 a bug report by Nathan J. Froyd)
3416 * fixed a bug in computing method discriminating functions: It is
3417 now possible to define methods specialized on classes which have
3418 forward-referenced superclasses. (thanks to Gerd Moellmann)
3419 * fixed evaluation order in optional entries (reported by Gilbert
3421 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3422 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3423 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3424 required-by-AMOP NAME argument, as well as CLASS and
3425 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3426 * fixed bug 20: DEFMETHOD can define methods using names that are
3427 not the proper names of classes to designate class specializers.
3428 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3429 Unix file descriptors, instead of blowing up. (thanks to Antonio
3431 * Garbage collection refactoring: user-visible change is that a
3432 call to the GC function during WITHOUT-GCING will not do garbage
3433 collection until the end of the WITHOUT-GCING. If you were doing
3434 this you were probably losing anyway.
3435 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3436 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3437 TYPEP the latter but not the former.
3438 * The compiler issues a full WARNING for calls to undefined functions
3439 with names from the CL package.
3440 * MAP-INTO for a vector destination is open coded. (reported by
3441 Brian Downing on c.l.l)
3442 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3443 documentation string.
3444 * fixed some bugs revealed by Paul Dietz' test suite:
3445 ** COPY-ALIST now signals an error if its argument is a dotted
3447 ** Condition slots are now accessed more correctly in the presence
3448 of multiple initargs for a given slot.
3449 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3450 exclude restarts of the same name associated with a different
3452 ** DEFCLASS of forward-referenced classes with another
3453 forward-referenced class in the superclasses list no longer
3455 ** Condition slots are now initialized once each, not multiple
3456 times. (thanks to Gerd Moellmann)
3457 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3459 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3460 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3461 arguments contain duplicated elements.
3462 ** RESTART-CASE understands local macros.
3463 ** RESTART-CASE associates exactly its own restarts with a condition.
3464 ** ENDP in safe mode checks its argument to be of type LIST.
3465 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3466 in question is unbound.
3467 ** Optimized MAKE-INSTANCE functions no longer cause internal
3468 assertion failures in the presence of duplicate initargs.
3469 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3470 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3471 propagated even in the ctor optimized implementation of
3473 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3475 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3476 * a better implementation of SXHASH on (simple) bit vectors,
3477 measured both in execution speed and in distribution of results
3478 over the positive fixnums, has been installed. Likewise, a better
3479 implementation of EQUAL for simple bit vectors is now available.
3480 * fixed CEILING optimization for a divisor of form 2^k.
3481 * fixed bug 240 (emitting extra style warnings "using the lexical
3482 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3483 by Antonio Martinez)
3484 * fixed SXHASH, giving different results for NIL depending on type
3485 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3486 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3487 variable. (found by Rolf Wester)
3488 * SBCL does not ignore type declarations for special
3489 variables. (reported by rif on c.l.l 2003-03-05)
3490 * some bug fixes in contrib/sb-aclrepl/
3491 * fixed some bugs revealed by Paul Dietz' test suite:
3492 ** a bug in the CONS type specifier, whereby the CAR and CDR
3493 types got intertwined, has been fixed;
3494 ** the type system is now able to reason about the interaction
3495 between INTEGER and RATIO types more completely;
3496 ** APPEND, [N]REVERSE and NRECONC check that those their
3497 arguments, which must be proper lists, are really so;
3498 ** An array specialized to be unable to hold elements has been
3499 implemented, as required -- yes, really -- by ANSI;
3500 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3501 on malformed property lists;
3503 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3504 * incompatible packaging change: in line with Unix convention,
3505 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3506 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3507 * REQUIRE and PROVIDE are now optionally capable of doing something
3508 useful. See the documentation string for REQUIRE.
3509 * infrastructure for a managed SBCL contrib system: contributed
3510 modules in this release include:
3511 ** the ASDF system definition facility;
3512 ** an interface to the BSD Sockets API;
3513 ** an ACL-like convenience interface to the repl;
3514 (thanks to Kevin Rosenberg)
3515 ** an implementation of ROTATE-BYTE, with efficient implementation
3517 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3518 gives a closer approximation to the right answer than previously.
3519 (thanks to Raymond Toy)
3520 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3521 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3522 optional environment arguments, as required by ANSI.
3523 * fixed bugs in other functions taking environment objects, allowing
3524 calls with an explicit NIL environment argument to be compiled
3526 * fixed bug 228: primary return values from
3527 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3528 COMPILE or FUNCTION.
3529 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3530 :TYPE will no longer signal errors on innocuous objects.
3531 * fixed bug 231b: SETQ is better at respecting type declarations in
3532 the lexical environment.
3533 * fixed a bug in DEFCLASS: classes named by symbols with no or
3534 unprintable packages can now be defined.
3535 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3536 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3537 * fixed compiler failure related to checking types of functions.
3538 (reported by Robert E. Brown)
3539 * the compiler is now much more consistent in its error-checking
3540 treatment of bounding index arguments to sequence functions: in
3541 (SAFETY 3) code, errors will be signalled in almost all cases if
3542 invalid sequence bounding indices are passed to functions defined
3543 by ANSI to operate on sequences.
3544 * fixed a bug in the build procedure: documentation of SBCL-specific
3545 packages is now preserved and available in the final Lisp image.
3546 * lifted FDEFINITION lookup out of loops in the implementation of
3547 many list operations. (thanks to Robert E. Brown)
3548 * fixed a bug in the reader: the #n# reader macro now works for
3549 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3550 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3551 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3552 E. Brown and Tony Martinez respectively)
3553 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3554 SLOT-BOUNDP: in particular, it is now able to optimize them much
3555 better, and is now not vulnerable to having packages renamed.
3556 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3557 always called when a slot is not present in an instance. (thanks
3559 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3560 longer signal an error on unreading a character following EOF on
3561 the previous constituent stream. (thanks to Tony Martinez)
3562 * fixed some bugs revealed by Paul Dietz' test suite:
3563 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3564 not just nonnegative fixnums;
3565 ** the logical bit-array operators such as BIT-AND now accept an
3566 explicit NIL for their "opt-arg" argument (to indicate a
3567 freshly-consed result bit-array);
3568 ** ELT now signals an error on an invalid sequence index in safe
3570 ** the type system is now cleverer about negations of numeric
3571 types, and consequently understands the BIGNUM and RATIO types
3573 ** the type system is now cleverer about the interaction between
3574 INTEGER and RATIO types: while bugs still remain, many more
3575 cases are accurately computed;
3576 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3577 if it is in the last clause;
3578 ** CONSTANTLY now correctly returns a side-effect-free function in
3580 ** DECLARE is no longer treated as a special-operator; in
3581 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3582 * incremented fasl file version number due to the change in the
3583 DEFSTRUCT-SLOT-DESCRIPTION structure.
3585 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3586 * minor incompatible change: code processed by the "interpreter" or
3587 EVAL now has a compilation optimization policy of (DEBUG 2)
3588 (changed from (DEBUG 1)) to improve debuggability of interactive
3589 development, and to allow the use of the debug RETURN command in
3591 * an experimental implementation of the RETURN command for the
3592 debugger has been included. (thanks to Frederik Kuivinen)
3593 * fixed bug 62: constraints were not propagated into a loop.
3594 * fixed bug in embedded calls of SORT (reported and investigated by
3596 * fixed some bugs revealed by Paul Dietz' test suite:
3597 ** printing and reading of arrays with some dimensions having
3598 length 0 (thanks to Gerd Moellmann);
3599 ** BOA constructor with &AUX argument without a default value does
3600 not cause a type error;
3601 ** CONSTANTP now returns true for all self-evaluating objects.
3603 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3604 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3605 accessors that are related by inheritance, as specified in the
3606 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3607 Valtteri Vuorikoski)
3608 * The compiler is now able to inline functions that were defined in
3609 a complex lexical environment (e.g. inside a MACROLET).
3610 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3612 * Support for the upcoming FreeBSD-5.0 release has been included.
3613 (thanks to Dag-Erling Smorgrav)
3614 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3615 effect when it is not in a toplevel context.
3616 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3617 stand a better chance of being correct. (thanks to Gerd
3619 * fixed bug in COERCE, which now signals an error on coercing a
3620 rational to a bounded real type which excludes the expected
3622 * The compiler is now able to derive types more accurately from the
3623 COERCE and COMPILE functions.
3624 * fixed bug 223: functional binding is considered to be constant
3625 only for symbols in the CL package.
3626 * fixed bug 231: SETQ did not check the type of a variable being set
3627 (reported by Robert E. Brown)
3628 * A new optimization for MAKE-INSTANCE has been included, fixing
3629 various bugs (including relating to :ALLOCATION :CLASS slots and
3630 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3632 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3633 ** As required by ANSI, LOOP now disallows anonymous collection
3634 clauses such as COLLECT I in conjunction with aggregate boolean
3635 clauses such as THEREIS (= I 1);
3636 ** LOOP now signals an error when any variable is reused in the
3637 same loop (including the potentially useful construct analogous
3638 to WITH A = 1 WITH A = (1+ A);
3639 ** IT is only a special loop symbol within the first clause of a
3640 conditional loop clause;
3641 ** LOOP with a typed iteration variable over a hashtable now
3642 signals a type error iff it should.
3643 * fixed some other bugs revealed by Paul Dietz' test suite:
3644 ** FILE-STREAM now names the class previously known as FD-STREAM;
3645 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3646 argument) no longer signals an error;
3647 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3648 of the slot symbol, rather than using the current package
3649 ((:CONC-NAME "") continues to intern the slot's name in the
3651 * incremented fasl file version number, because of the incompatible
3652 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3653 of the new implementation of DEFINE-COMPILER-MACRO.
3655 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3656 * Support for building SBCL for MIPS platforms running in
3657 little-endian mode has now been checked in, and basic
3658 functionality on said platforms verified.
3659 * minor incompatible change: PCL now records the pathname of a file
3660 in which methods and the like are defined, rather than its
3662 * minor incompatible change: TRUENAME now considers the truename of
3663 a file naming a directory to be the pathname with :DIRECTORY
3664 component indicating that directory.
3665 * minor incompatible change: a NAMED clause in the extended form of
3666 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3667 reason for the previous behaviour is unclear.
3668 * more systematization and improvement of CLOS and MOP conformance
3669 in PCL (thanks to Gerd Moellman and Pierre Mai):
3670 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3672 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3673 primary methods with no specializers;
3674 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3676 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3677 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3678 FORWARD-REFERENCED-CLASSes; error reporting on
3679 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3681 ** SXHASH on CLOS instances now uses a slot internal to the
3682 instance to return different numbers on distinct instances,
3683 while preserving the same return value through invocations of
3685 ** DEFMETHOD signals errors when methods with longer incongruent
3686 lambda lists are added to generic functions;
3687 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3688 CLASS, as specified in AMOP;
3689 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3690 on the order returned by the primary method for classes of
3691 class STANDARD-CLASS;
3692 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3693 * fixed some bugs shown by Paul Dietz' test suite:
3694 ** DOLIST puts its body in TAGBODY;
3695 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3697 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3698 value producing form;
3699 ** if more variables are given to PROGV than values, extra
3700 variables are bound and made to have no value;
3701 ** NSUBSTITUTE on list arguments gets the right answer with
3703 ** ELT signals an error of type TYPE-ERROR when the index argument
3704 is not a valid sequence index;
3705 ** LOOP signals (at macroexpansion time) an error of type
3706 PROGRAM-ERROR when duplicate variable names are found;
3707 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3708 ** FUNCALL of special-operators now cause an error of type
3710 ** PSETQ now works as required in the presence of side-effecting
3711 symbol-macro places;
3712 ** NCONC accepts any object as its last argument;
3713 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3715 ** loop-for-as-package does not require a package to be explicitely
3717 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3719 * fixed bug 166: compiler preserves "there is a way to go"
3720 invariant when deleting code.
3721 * fixed bug 172: macro lambda lists with required arguments after
3722 &REST arguments now cause an error to be signalled. (thanks to
3724 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3725 bug. (thanks to Matthew Danish)
3726 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3728 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3729 arguments are lexically rebound. (thanks to Gerd Moellmann and
3731 * fixed bug 194: error messages are now more informative when there
3732 is no primary method applicable in a call to a generic
3733 function. (thanks to Gerd Moellmann)
3734 * fixed bug in command line argument checking (thanks to Julian
3736 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3737 (thanks to Matthew Danish)
3738 * incremented fasl file version number, because of the
3739 SXHASH-related changes in the layout of CLOS data structures
3741 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3742 * minor incompatible change: The runtime (the Unix executable named
3743 "sbcl") is now much pickier about the .core files it will load.
3744 Essentially it now requires .core files to descend from the same
3745 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3746 as the runtime does. (The intent is to prevent the crashes which
3747 can occur, and which can even be reported as mysterious failures,
3748 when people patch the sources or change the build parameters
3749 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3750 sbcl and .core files.)
3751 * fixed bug: VALUES-LIST is no longer optimized away.
3752 * fixed bug 142: The FFI conversion of C string values to Lisp
3753 string values no longer conses excessively. (thanks to Nathan
3754 Froyd porting Raymond Toy's fix to CMU CL)
3755 * began to systematize and improve MOP conformance in PCL (thanks to
3756 Nathan Froyd, Gerd Moellman and Pierre Mai):
3757 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3759 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3760 ** FINALIZE-INHERITANCE is now called on class finalization;
3761 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3762 argument precedence order.
3763 * fixed bug 202: The compiler no longer fails on functions whose
3764 derived types contradict their declared type.
3765 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3766 so it can be non-toplevel.
3767 * The fasl file version number has changed (because of the new
3768 implementation of DEFMACRO).
3769 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3770 safe code, that any length requirement by their type-specifier
3771 argument is valid. The exceptions to this are described in bug
3773 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3774 functions now check the types of their inputs as required by ANSI.
3775 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3776 introduced symbol is DECLAREd to be SPECIAL.
3777 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3778 * fixed bug: :COUNT argument to sequence functions may be negative
3779 * fixed bug: body of DO-SYMBOLS may contain declarations
3780 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3781 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3782 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3783 symbol macro only once
3784 * fixed printing of call frame when argument list is unavailable
3785 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3786 * compiler no longer signals WARNING on unknown keyword
3789 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3790 * A beta-quality port to the mips architecture running Linux,
3791 based on the old CMUCL backend, has been made. It has been tested
3792 on a big-endian kernel, and works sufficiently well to be able to
3793 rebuild itself; it has not been tested in little-endian mode.
3794 * fixed an inconsistency between gencgc.c and purify.c which made
3795 dumping/loading .core files unreliable
3796 * fixed bug 120a: The compiler now deals correctly with IFs where
3797 the consequent is the same as the alternative, instead of
3798 misderiving the return type. (thanks to Alexey Dejneka)
3799 * fixed bug 113: Logical pathnames are now dumpable (the logical
3800 host is resolved at load-time, throwing an error if it is not
3802 * fixed bug 174: FORMAT's error message is slightly clearer when a
3803 non-printing character is used in a format directive.
3804 * fixed several bugs in compiler checking of type declarations, i.e.
3805 violations of the Python "declarations are assertions" principle
3806 (thanks to Alexey Dejneka)
3807 * fixed several bugs in PCL's error checking (thanks to Gerd
3809 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3811 * fixed bug in compilation of functions as first class values
3812 (thanks to Antonio Martinez-Shotton)
3813 * The compiler's handling TYPE-ERRORs which it can prove will
3814 inevitably happen at runtime has been cleaned up and corrected
3815 in several ways. (thanks to Alexey Dejneka)
3816 * improved argument type checking for various basic arithmetic
3817 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3818 lavished on them in the past that they can be compiled in many
3819 ways in different special cases
3820 * fixed bug 181: compiler checks validity of user supplied type
3822 * cleaned up code flushing in optimization: Function calls which
3823 should signal errors for safety purposes (e.g. which ANSI says
3824 should signal errors when their arguments are of incorrect type)
3825 are no longer optimized away.
3826 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3827 * incremented fasl file version number, because changes in the
3828 implementation of sequence functions like COERCE caused
3829 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3830 in old inline expansions) to become undefined. (Actually these
3831 changes were later undone, so we might very well be binary
3832 compatible with 0.7.7 after all, but leaving the version number
3833 incremented seemed like the simplest and most conservative
3836 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3837 * An alpha-quality port to the parisc architecture running Linux,
3838 based on the old CMUCL backend, has been made. This, even more so
3839 than the other backends, should be considered still a work in
3840 progress; known problems include that the Linux kernel in 64-bit
3841 mode does not propagate the correct sigcontext structure to
3842 userspace, and consequently SBCL on a parisc64 kernel will not
3844 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3845 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3846 suppressed.) Also now that the compiler is looking at declarations
3847 in the environment, it checks optimization declarations as well,
3848 and suppresses inlining when (> DEBUG SPEED).
3849 * More fixes have been made to treatment of floating point exception
3850 treatment and other Unix signals. In particular, floating point
3851 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3852 * The detection and handling of control stack exhaustion (infinite
3853 or very deeply nested recursion) has changed. Stack exhaustion
3854 detection is now done by write-protecting pages at the OS level
3855 and applies at all optimization settings; when found, a
3856 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3857 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3858 be caught using IGNORE-ERRORS.
3859 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3860 correctly when transforming recursive LABELS functions to LETs.
3861 (thanks to Alexey Dejneka)
3862 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3863 that are names of constants or global variables.
3864 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3865 alien routines with docstrings.
3866 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3867 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3869 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3870 (thanks to Eric Marsden)
3871 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3872 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3873 * Bug fix: The SPARC backend can now compile functions involving
3874 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3875 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3876 to a functions expecting a PATHNAME-DESIGNATOR.
3877 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3878 lambda lists. (thanks to Alexey Dejneka)
3879 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3880 functions so that EXPORTing the name of the class doesn't cause
3881 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3882 to Antonio Martinez for reporting this)
3883 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3884 constant-folded without failing an assertion. (thanks to Einar
3885 Floystad Dorum for reporting this)
3886 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3887 types are now dealt with more correctly by the compiler.
3888 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3889 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3890 behaviour analogous to RENAME-FILE. This puts its behaviour more
3891 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3893 * The fasl file version number has changed again. (because of the
3894 bug fix involving the names of PCL MAKE-INSTANCE functions)
3896 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3897 * bug fix: Floating point exceptions are treated much more
3898 consistently on the x86/Linux and PPC/Linux platforms.
3899 * Array initialization with :INITIAL-ELEMENT is now much faster for
3900 cases when the compiler cannot open code the array creation, but
3901 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3902 array accesses have also seen a speed increase.
3903 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3904 specified. (This was at the root of some bad interactions between
3905 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3907 * bug fix: Internal error arguments for undefined functions are now
3908 computed correctly on the PPC/Linux platform.
3909 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3910 Raymond Toy's patch for CMU CL)
3911 * Support for the Solaris 9 operating environment has been included
3912 (thanks to Daniel Merritt)
3913 * A very ugly but hopefully complete draft of the missing FFI chapter
3914 of the manual has been created by reformatting the corresponding
3915 CMU CL manual chapter into (currently very ugly and incoherent)
3916 DocBook and bringing it up to date for SBCL behavior. Thus, the
3917 manual is now essentially complete, at least by my extreme
3918 once-and-only-once standards, whereby it's acceptable to refer to
3919 the doc strings of SB-EXT functions as the primary documentation.
3920 * The fasl file version number has changed again, due to cleanup of
3921 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3922 is no longer a static symbol.)
3924 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3925 * SBCL now builds with OpenMCL (version 0.12) as the
3926 cross-compilation host; also, more progress has been made toward
3927 bootstrapping under CLISP.
3928 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3930 * bug 158 fixed: The compiler can now deal with integer loop
3931 increments different from 1; fixing this turned out also to fix
3933 * bug 169 fixed: no more bogus warnings about using lexical bindings
3934 despite the presence of perfectly good SPECIAL declarations (thanks
3935 to David Lichteblau)
3936 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3937 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3938 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3940 * bug 180 fixed: Method combination specifications no longer ignore
3941 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3942 * bug fix: Structure type predicate functions now check their argument
3943 count as they should.
3944 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3945 correctly. (thanks to Pierre Mai)
3946 * minor incompatible change: The --noprogrammer option is deprecated
3947 in favor of the new --disable-debugger option, which is very similar.
3948 (The major difference is that it takes effect at a slightly different
3949 time at startup, causing handling of errors in --sysinit and
3950 --userinit files will be affected differently.) The
3951 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3952 been added to allow this functionality to be controlled from ordinary
3953 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3954 maintainers, who might want to run non-interactive scripts to
3955 build SBCL cores which will later be used interactively.)
3956 * minor incompatible change: The LOAD function no longer, when given
3957 a wild pathname to load, loads all files matching that pathname.
3958 Instead, an error of type FILE-ERROR is signalled.
3960 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3961 * bug 147 fixed: The compiler preserves its block link/count
3962 invariants more correctly now so that it doesn't crash. (thanks
3964 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3966 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3967 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3968 no-value-for-this-character as NIL (as specified by ANSI).
3969 * HOST-NAMESTRING on physical pathnames now returns a string that is
3970 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3971 (thanks to Christophe Rhodes)
3972 * The Alpha port handles icache flushing more correctly. (thanks to
3974 * More progress has been made toward bootstrapping under CLISP. (thanks
3975 to Christophe Rhodes)
3976 * The fasl file format has changed again, because dynamic loading
3977 on OpenBSD (which has non-ELF object files) motivated some cleanups
3978 in the way that foreign symbols are transformed and passed around.
3979 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3980 is no longer treated as whitespace by the reader, but instead as
3981 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3982 |A
\7fB|, instead of A as it used to.
3984 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3985 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3986 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3987 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3988 this as of 0.7.1.45, but was left out of the previous news section
3989 (thanks to Dan Barlow)
3990 * SBCL now runs on the Solaris operating system on SPARC architectures
3991 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3992 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3993 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3994 * SPARC backend cleanups, allowing builds of cores optimized for V8
3995 and V9 SPARCS, and also emission of code targeted to a particular
3996 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3998 * SBCL is closer to bootstrapping under CLISP, thanks to various
3999 fixes by Christophe Rhodes.
4000 * The fasl file format has changed again, to allow the compiler's
4001 INFO database to support symbol macros.
4002 * The user manual (in doc/) is formatted into HTML more nicely.
4003 (thanks to coreythomas)
4004 * The system is smarter about SUBTYPEP relationships, especially
4005 those involving NOT types (including types such as ATOM which are
4006 represented internally using NOT types). Thus SUBTYPEP is less
4007 likely to return (VALUES NIL NIL) in general, and in particular
4008 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4010 * The fasl file format has changed again, because the internal
4011 representation of types now includes a new slot to support the new
4012 SUBTYPEP-of-NOT-types logic.
4013 * (not a change in the main branch of SBCL, but a related prototype
4014 which can hopefully be merged into the main branch of SBCL in the
4015 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4016 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4017 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4018 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4019 dispatch macro character combinations. (thanks to Alexey Dejneka)
4020 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4021 that superficially look like logical namestrings correctly.
4022 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4024 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4025 default supplied-p) for &optional and &key arguments. (thanks to
4028 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4029 * incompatible change: The compiler is now less aggressive about
4030 tail call optimization, doing it only when (> SPACE DEBUG) or
4031 (> SPEED DEBUG). (This is an incompatible change because there are
4032 programs which relied on the old CMU-CL-style behavior to optimize
4033 away their unbounded recursion which will now die of stack overflow.)
4034 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4035 for non-GENCGC systems has been increased to 20M (since that
4036 seems much closer to the likely performance optimum for modern
4037 systems than the old 4M value was)
4038 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4039 and *DEBUG-PRINT-LEVEL*
4040 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4041 of CMU CL's support for SPARC, and various endianness and other
4042 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4043 * new syntactic sugar for the Unix command line: --load foo.bar is now
4044 an alternate notation for --eval '(load "foo.bar")'.
4046 ** The system now detects stack overflow and handles it gracefully,
4047 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4048 optimization settings. (This is a good thing in general, and
4049 its introduction in this version should be particularly timely
4050 for anyone whose code fails because of suppression of tail
4052 ** The system now hunts for the C variable "environ" in a more
4053 devious way, to avoid segfaults when the C library version differs
4054 between compile time and run time. (thanks to Christophe Rhodes)
4055 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4056 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4057 ** The compiler no longer issues bogus style warnings for undefined
4058 classes in the same source file as the DEFCLASSes which defined
4059 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4061 ** fixes in CONDITION class precedence list for undefined function
4062 errors (thanks to Alexei Dejneka)
4063 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4064 correctly. (thanks to Dan Barlow)
4065 ** portability fixes aiming at bootstrapping under CLISP (thanks
4066 to Dave McDonald and Christophe Rhodes)
4067 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4068 ** fixes in type translation and and type inference (thanks to
4070 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4071 ** various fixes in the new ports (thanks to Dan Barlow)
4072 * several changes related to debugging:
4073 ** suppression of tail recursion, as noted above
4074 ** stack overflow detection, as noted above
4075 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4076 is now the default. (For some time encapsulation has been more
4077 reliable than the breakpoint-based :ENCAPSULATE NIL
4078 implementation, at least on X86 systems; and I just noticed that
4079 encapsulation also seems closer to the spirit of the ANSI
4082 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4084 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4085 up properly again. (There was a packaging bug in 0.7.0 which
4086 left their definitions in SB-SYS::LOAD-FOREIGN and
4087 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4088 vital for most things which interface to C-level interfaces,
4089 like extensions working with sockets or databases or
4090 Perl-compatible regexes or whatever, and the need to fix
4091 this bug is the main reason that 0.7.1 was released so
4093 ** DEFGENERIC is now choosier about the methods it redefines, so that
4094 reLOADing a previously-LOADed file containing DEFGENERICs does
4095 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4096 cycle now works as it should. (thanks to Alexey Dejneka)
4097 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4098 fixed by Christophe Rhodes. (He actually submitted this patch
4099 months ago, and I delayed until after 0.7.0.)
4100 ** Bug 111 (internal compiler confusion about runtime checks on
4101 FUNCTION types) was fixed by Alexey Dejneka.
4102 * Some internal cleanups (getting rid of variables which aren't
4103 needed now that the byte interpreter is gone) caused the fasl
4104 file format number to change again.
4106 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4107 * major incompatible change: The default fasl file extension, i.e. the
4108 default extension for files produced by COMPILE-FILE, has changed
4109 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4111 ** There are many changes in the implementation of the compiler.
4112 SBCL is now essentially a compiler-only implementation of ANSI
4113 Common Lisp. EVAL still "interprets" a few special cases, but
4114 almost all the interesting cases are handled by creating
4115 a LAMBDA expression, calling COMPILE on it, then calling
4116 FUNCALL on the result.
4117 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4118 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4119 Since the code is newer, there might still be some new bugs
4120 (though not as many as before Martin Atzmueller's fixes:-). But
4121 the new code is substantially simpler and clearer, and hopefully
4122 any remaining bugs will be simpler, less fundamental, and more
4123 fixable then the bugs in the old code.
4124 ** The revised compiler is still a little unsteady on its feet.
4126 *** The debugging information it produces (particularly the names
4127 of FUNCTION objects) is sometimes much less useful than what
4128 the old compiler produced.
4129 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4130 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4131 within a MACROLET) has been temporarily weakened.
4132 ** There are new compiler optimizations for various functions:
4133 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4134 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4135 *** the math functions TRUNCATE, FLOOR, and CEILING
4136 *** the function-of-all-trades COERCE
4137 Mostly these should be transparent, but there's one
4138 potentially-annoying problem (bug 117): when the compiler
4139 inline-expands a function and does type analysis on the result,
4140 it can create control paths which have type mismatches, and
4141 when it can't prove that those control paths aren't taken,
4142 it will issue WARNINGs about the type mismatches. This is
4143 a particular problem in practice for the new sequence functions.
4144 It's not clear how this should be fixed, and for now, a
4145 workaround is given in the entry for 117 in the BUGS file.
4146 ** (Because of the interaction between the two previous items --
4147 occasional inlining problems and new inline expansions -- some
4148 of the new sequence function optimizations won't really kick in
4149 completely until debugging information, and then inlining, are
4150 straightened out in some future version.)
4151 * minor incompatible changes:
4152 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4153 DIRECTORY no longer implicitly promotes NIL slots of its
4154 pathname argument to :WILD. In particular, when you ask for the
4155 contents of a directory (which you used to be able to do without
4156 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4157 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4158 ** changes in behavior that ANSI explicitly defines to be
4159 implementation dependent:
4160 *** The new compiler-only implementation still conforms with ANSI,
4161 but acts a little different than before. Besides the obvious
4162 changes in performance tradeoffs (that the cost per form passed
4163 to EVAL has gone up, and the cost per form executed by EVAL
4164 has gone down), the behavior of the system changes a little
4165 because there are no longer any interpreted function objects.
4166 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4167 e.g. doing COMPILE on the output of interactive DEFUN is
4169 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4171 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4172 has changed from (:CL) to NIL.
4173 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4174 e.g. #\Newline and #\Tab, don't have names specified in the
4175 ANSI Common Lisp standard, is now based on their ASCII symbolic
4176 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4177 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4178 are no longer used for output.
4179 ** changes in internal implementation constants:
4180 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4181 4 million. (If your application spends a lot of time GCing and
4182 you have a lot of RAM, you might want to experiment with
4183 increasing it even more.)
4184 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4185 However, almost all old code should still continue to work without
4186 immediate update, as SB-C-CALL is now a (deprecated) nickname
4188 ** Old operator names in the style DEF-FOO are now deprecated in
4189 favor of new corresponding names DEFINE-FOO, for consistency with
4190 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4191 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4192 internal symbols, but a few supported extensions like
4193 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4194 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4195 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4196 etc. as you get deeper into recursive calls to the debugger
4197 command loop, instead of the old "5]", "5]]", "5]]]"
4198 sequence. (I was motivated to do this when squabbles between
4199 ILISP and SBCL left me very deeply nested in the debugger. In the
4200 short term, this change will probably provoke more ILISP/SBCL
4201 squabbles, but hopefully it will be an improvement in the long run.)
4202 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4203 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4204 function, since declaiming return types involving aliens is
4205 (1) annoyingly messy to do by hand and (2) vital to efficient
4206 compilation of code which calls such functions.
4207 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4208 longer reexported by the SB-EXT package. They're solely useful
4209 for alien code, so it seems more logical that you should get
4210 them from the SB-ALIEN package, not in SB-EXT.
4211 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4212 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4213 features. Instead, the code that they used to control is always
4214 built into the system.
4215 * many other bug fixes
4216 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4217 advantage of the new EVAL-WHEN stuff and to clean them up in
4218 general, and they are now more ANSI-compliant in a number of
4219 ways. Martin Atzmueller is responsible for a lot of this.
4220 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4222 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4223 *** correct ERROR type for various file operations
4224 *** some fixes for Lisp streams
4225 *** DEFMETHOD syntax checking
4226 *** changing old weird representation of debug information as
4227 strings (which, among their other deficiencies, don't transform
4228 correctly when you rename packages, and don't change their
4229 print representation when you change things like *PACKAGE*
4230 and *PRINT-LENGTH*) to symbols and lists of symbols
4231 He also made several improvements and fixed several bugs in DESCRIBE.
4232 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4234 *** misbehavior of WRITE-STRING/WRITE-LINE
4235 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4236 and several other LOOP problems as well
4237 *** DIRECTORY when similar filenames are present
4238 *** DEFGENERIC with :METHOD options
4239 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4240 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4241 *** argument ordering in FIND with :TEST option
4242 *** mishandled package designator argument in APROPOS-LIST
4243 *** various problems in the backquote readmacro
4244 *** a bug in APROPOS
4245 *** probably some others that I'm not describing very well here,
4246 since the CVS log documents them by reference to sbcl-devel
4247 messages, and the SourceForge archives aren't working well.:-(
4248 ** Dan Barlow improved the Alpha port (and is making progress on the
4249 PPC port, for those of you who think different).
4250 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4251 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4252 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4253 continues to work on the SPARC port (for those of you in a position
4254 to look down upon our little PC-compatible boxes from a great height).
4255 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4256 and uses that copy, rather than the current dynamic value, when
4257 it's trying to decide whether to truncate output. Thus e.g.
4258 (let ((*print-lines* 50))
4259 (pprint-logical-block (stream nil)
4261 (let ((*print-lines* 8))
4262 (print (aref possiblybigthings i) stream)))))
4263 should now truncate the logical block only at 50 lines, instead of
4264 often truncating it at 8 lines, as it did before.
4265 * The doc/cmucl/ directory, containing old CMU CL documentation
4266 from the time of the fork, is no longer part of the base system.
4267 SourceForge has shut down its anonymous FTP service, and with it
4268 my original plan for distributing the old CMU CL documentation
4269 there. For now, if you need these files you can download an old
4270 SBCL source release and extract them from it.
4271 * The fasl file version number changed again, for dozens of reasons,
4272 some of which are apparent above.
4274 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4275 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4276 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4277 patch, so that DISASSEMBLE output is much nicer.
4278 * The code in the SB-PROFILE package now seems reasonably stable.
4279 I still haven't decided what the final interface should look like
4280 (I'd like PROFILE to interact cleanly with TRACE, since both
4281 facilities use function encapsulation) but if you have a need
4282 for profiling now, you can probably use it successfully with
4283 the current CMU-CL-style interface.
4284 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4285 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4286 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4287 link now returns the dangling link itself, and for similar
4288 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4289 link itself. (In these cases the old code signalled an error and
4290 looped endlessly, respectively.) Thus, DIRECTORY now works even
4291 in the presence of dangling and cyclic symbolic links.
4292 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4293 is now a supported extension again, since the consensus on
4294 sbcl-devel was that it can be useful for ordinary development
4295 work, not just for debugging SBCL itself.
4296 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4297 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4298 that currently-defined functions might be redefined later with
4299 different return types.
4300 * Hash tables can be printed readably, as inspired by CMU CL code
4301 of Eric Marsden and SBCL code of Martin Atzmueller.
4302 * better error handling in CLOS method combination, thanks to
4303 Martin Atzmueller porting Pierre Mai's CMU CL patches
4304 * more overflow fixes for >16Mbyte I/O buffers
4305 * A bug in READ has been fixed, so that now a single Ctrl-D
4306 character suffices to cause end-of-file on character streams.
4307 In particular, now you only need one Ctrl-D at the command
4308 line (not two) to exit SBCL.
4309 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4311 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4312 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4313 does the right thing.
4314 * The implementation of some type tests, especially for CONDITION
4315 types, is now tidier and maybe faster, due to CMU CL code
4316 originally by Douglas Crosher, ported by Martin Atzmueller.
4317 * Some math functions have been fixed, and there are new
4318 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4319 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4320 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4321 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4322 likely to be built into sbcl-0.7.0.)
4323 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4324 CL, ported to SBCL by Martin Atzmueller)
4325 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4326 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4327 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4328 * There's a new slam.sh hack to shorten the edit/compile/debug
4329 cycle for low-level changes to SBCL itself, and a new
4330 :SB-AFTER-XC-CORE target feature to control the generation of
4331 the after-xc.core file needed by slam.sh.
4332 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4333 COMPILE-FILE is no longer supported, so that now every function
4334 gets an entry point, so that block compilation looks a little
4335 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4336 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4337 returns the number of bytes consed since the system started,
4338 rather than the number consed since the first time the function
4339 was called. (The new definition parallels ANSI functions like
4340 CL:GET-INTERNAL-RUN-TIME.)
4341 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4342 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4343 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4344 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4345 * Fasl file version numbers are now independent of the target CPU,
4346 since historically most system changes which required version
4347 number changes have affected all CPUs equally. Similarly,
4348 the byte fasl file version is now equal to the ordinary
4351 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4352 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4353 is no longer recognized. I apologize for this, because it was
4354 listed in SB-EXT as a supported extension, but I found that
4355 its existing behavior was poorly specified, as well as incorrectly
4356 specified, and it looked like too much of a mess to straighten it
4357 out. I have enough on my hands trying to get ANSI stuff to work..
4358 * many patches ported from CMU CL by Martin Atzmueller, with
4359 half a dozen bug fixes in pretty-printing and the debugger, and
4360 half a dozen others elsewhere
4361 * fixed bug 13: Floating point infinities are now supported again.
4362 They might still be a little bit flaky, but thanks to bug reports
4363 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4364 as flaky as they were.
4365 * The --noprogrammer command line option is now supported. (Its
4366 behavior is slightly different in detail from what the old man
4367 page claimed it would do, but it's still appropriate under the
4368 same circumstances that the man page talks about.)
4369 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4370 are now supported, and enabled by default. Thus, the compiler can
4371 handle many floating point and complex operations much less
4372 inefficiently. (Thus e.g. you can implement a complex FFT
4374 * The compiler now detects type mismatches between DECLAIM FTYPE
4375 and DEFUN better, and implements CHECK-TYPE more correctly, and
4376 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4377 thanks to patches from Martin Atzmueller.
4378 * various fixes to make the cross-compiler more portable to
4379 ANSI-conforming-but-different cross-compilation hosts (notably
4380 Lispworks for Windows, following bug reports from Arthur Lemmens)
4381 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4382 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4383 fixed thanks to Pierre Mai's CMU CL patches.
4384 * fixes to keep the system from overflowing internal counters when
4385 it tries to use i/o buffers larger than 16M bytes
4386 * fixed bug 45a: Various internal functions required to support
4387 complex special functions have been merged from CMU CL sources.
4388 (When I was first setting up SBCL, I misunderstood a compile-time
4389 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4390 * improved support for type intersection and union, fixing bug 12
4391 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4392 more obscure bugs as well
4393 * some steps toward byte-compiling non-performance-critical
4394 parts of the system, courtesy of patches from Martin Atzmueller
4395 * Christophe Rhodes has made some debian packages of sbcl at
4396 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4397 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4398 stable, but are nonetheless usable. When he's ready, I'd be happy
4399 to add them to the SourceForge "File Releases" section. (And if
4400 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4401 * new fasl file format version number (because of changes in
4402 internal representation of (OR ..) types to accommodate the new
4403 support for (AND ..) types, among other things)
4405 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4406 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4408 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4410 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4411 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4412 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4413 * DESCRIBE now works on structure objects again.
4414 * Most function call argument type mismatches are now handled as
4415 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4416 know whether the function will be redefined before the call is
4417 executed. (The compiler could flag local calls with full WARNINGs,
4418 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4419 it doesn't keep track of enough information to know whether calls
4420 are local in this sense.)
4421 * Compiler output is now more verbose, with messages truncated
4422 later than before. (There should be some supported way for users
4423 to override the default verbosity, but I haven't decided how to
4424 provide it yet, so this behavior is still controlled by the internal
4425 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4426 src/compiler/ir1util.lisp.)
4427 * Fasl file format version numbers have increased again, because
4428 support for the Gray streams extension changes the layout of the
4429 system's STREAM objects.
4430 * The Gray subclassable streams extension now works, thanks to a
4431 patch from Martin Atzmueller.
4432 * The full LOAD-FOREIGN extension (not just the primitive
4433 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4434 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4435 but like most other programs, it defaults to copying the Unix
4436 environment from the original process instead of starting the
4437 new process in an empty environment.
4438 * Extensions which manipulate the Unix environment now support
4439 an :ENVIRONMENT keyword option which doesn't smash case or
4440 do other bad things. The CMU-CL-style :ENV option is retained
4441 for porting convenience.
4442 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4443 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4445 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4447 * A patch from Martin Atzmueller seems to have solved the SIGINT
4448 problem, and as far as we know, signal-handling now works cleanly.
4449 (If you find any new bugs, please report them!)
4450 * The system no longer defaults Lisp source file names to types
4451 ".l", ".cl", or ".lsp", but only to ".lisp".
4452 * The compiler no longer uses special default file extensions for
4453 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4454 seems to expect a single default extension for all compiled code,
4455 and there's no compelling reason to try to stretch the standard
4456 to allow two different extensions.) Instead, byte-compiled files
4457 default to the same extension as native-compiled files.
4458 * Fasl file format version numbers have increased again, because
4459 a rearrangement of internal implementation packages made some
4460 dumped symbols in old fasl files unreadable in new cores.
4461 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4462 many fewer weird special cases.
4463 * Bug #17 (differing COMPILE-FILE behavior between logical and
4464 physical pathnames) has been fixed, and some related misbehavior too,
4465 thanks to a patch from Martin Atzmueller.
4466 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4467 by Tim Moore, ported to SBCL by Martin Atzmueller.
4468 * Martin Atzmueller fixed several filesystem-related problems,
4469 including bug #36, in part by porting CMU CL patches, which were
4470 written in part by Paul Werkowski.
4471 * More compiler warnings in src/runtime/ are gone, thanks to
4472 more patches from Martin Atzmueller.
4473 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4476 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4478 * DESCRIBE now works on CONDITION objects.
4479 * The debugger now handles errors which arise when trying to print
4480 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4482 * The build system now uses an additional file, customize-target-features.lisp,
4483 to allow local modifications to the target *FEATURES* list. (The point of
4484 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4485 debugging features enabled, without having to worry about propagating it
4486 into everyone's system when I do a "cvs update".) When no
4487 customize-target-features.lisp file exists, the target *FEATURES* list
4488 should be constructed the same way as before.
4489 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4490 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4491 warn about behavior which is undefined under the ANSI spec. Note: This
4492 is specified by ANSI, but it's not very popular with programmers.
4493 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4494 macro in the SBCL sources for an example of a workaround which you
4495 might use to make portable ANSI-standard code which does what you want.
4496 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4497 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4498 when it's used as a non-toplevel form. (This is required in order
4499 to implement the DEFCONSTANT-EQX macro.)
4500 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4501 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4502 funny things, due to bugs in the implementation of EVAL-WHEN
4503 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4504 non-toplevel DEFCONSTANTs are uncommon.)
4505 * The core file version number and fasl file version number have been
4506 incremented, because the old noncompliant DEFCONSTANT behavior involved
4507 calling functions which no longer exist, and because I also took the
4508 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4509 * fixed bug 1 (error handling before read-eval-print loop starts), and
4510 redid debugger restarts and related debugger commands somewhat while
4512 ** The QUIT debugger command is gone, since it did something
4513 rather different than the SB-EXT:QUIT command, and since it never
4514 worked properly outside the main toplevel read/eval/print loop.
4515 Invoking the new TOPLEVEL restart provides the same functionality.
4516 ** The GO debugger command is also gone, since you can just invoke
4517 the CONTINUE restart directly instead.
4518 ** The TOP debugger command is also gone, since it's redundant with the
4519 FRAME 0 command, and since it interfered with abbreviations for the
4521 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4523 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4524 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4525 some porting work by Martin Atzmueller)
4526 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4527 with a boatload of patches which he ported to SBCL. Now that those
4529 ** The system tries to make sure that its low-priority messages
4530 are prefixed by semicolons, to help people who like to use
4531 syntax highlighting in their ILISP buffer. (This patch
4532 was originally due to Raymond Toy.)
4533 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4534 patches originally written by Raymond Toy.
4535 ** The compiler understands coercion between single-value and
4536 multiple-VALUES type expressions better, getting rid of some very
4537 weird behavior, thanks to patches originally by Robert MacLachlan
4538 and Douglas Crosher.
4539 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4540 lambda lists, thanks to a patch originally by Pierre Mai.
4541 ** The system no longer bogusly warns about "abbreviated type
4543 ** The compiler gets less confused by inlining and RETURN-FROM,
4544 thanks to some patches originally by Tim Moore.
4545 ** The system no longer hangs when dumping circular lists to fasl
4546 files, thanks to a patch originally from Douglas Crosher.
4547 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4548 undefined function error.
4549 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4550 in structure slot initforms) for now, documented workaround instead:-|
4551 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4552 * fixed bug 5: added stubs for various Gray stream functions called
4553 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4554 installed, at least appropriate type errors are generated
4555 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4556 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4557 consistently in DEFMETHOD forms.
4558 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4559 it doesn't seem to affect SBCL after all
4560 * The C runtime system now builds with better optimization and many
4561 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4563 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4565 * The system is now under CVS at SourceForge (instead of the
4566 CVS repository on my home machine).
4567 * The new signal handling code has been tweaked to treat register
4568 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4569 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4570 downstream, e.g. in debug-int.lisp, has implicit dependencies
4571 on the unsignedness of integer representation of machine words,
4572 and that caused the system to bomb out with infinite regress
4573 when trying to recover from type errors involving signed values,
4574 e.g. (BUTLAST '(1 2 3) -1).)
4575 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4576 (This was one of the bugs Peter Van Eynde reported back in July.)
4577 * The system now uses code inspired by Colin Walters' O(N)
4578 implementation of MAP (from the cmucl-imp@cons.org mailing
4579 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4580 inline the MAP operation, and there is more than one
4581 sequence argument to the MAP call (so that it can't just
4582 do ETYPECASE once and for all based on the type of the
4583 single sequence argument). (The old non-inline implementation
4584 of the general M-argument sequence-of-length-N case required
4585 O(M*N*N) time when any of the sequence arguments were LISTs.)
4586 * The QUIT :UNIX-CODE keyword argument has been renamed to
4587 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4588 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4590 * Raymond Wiker's patches to port dynamic loading from Linux to
4591 FreeBSD have been added.
4592 * The BUGS file is now more nearly up to date, thanks in large part
4593 to Martin Atzmueller's review of it.
4594 * The debugger now flushes standard output streams before it begins
4595 its output ("debugger invoked" and so forth).
4596 * The core version number and fasl file version number have both
4597 been incremented, because of incompatible changes in the layout
4599 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4600 used to not be called for a saved Lisp image.)
4601 * Martin Atzmueller's version of a patch to fix a compiler crash,
4602 as posted on sbcl-devel 13 September 2000, has been installed.
4603 * Instead of installing Martin Atzmueller's patch for the
4604 compiler transform for SUBSEQ, I deleted the compiler transform,
4605 and transforms for some similar consing operations.
4606 * A bug in signal handling which kept TRACE from working on OpenBSD
4608 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4609 away by constant folding
4610 * The system now defines its address space constants in one place
4611 (in the Lisp sources), and propagates them automatically elsewhere
4612 (through GENESIS and the sbcl.h file). Therefore, patching the
4613 address map is less unnecessarily tedious and error-prone. The
4614 Lisp names of address space constants have also been systematized.
4615 * CVS tags like dollar-Header-dollar have been removed from
4616 the sources, because they have never saved me trouble and
4617 they've been source of trouble working with patches and other
4618 diff-related operations.
4619 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4620 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4622 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4624 * The system has been ported to OpenBSD.
4625 * The system now compiles with a simple "sh make.sh" on the systems
4626 that it's supported on. I.e., now you no longer need to tweak
4627 text in the target-features.lisp-expr and symlinks in src/runtime/
4628 by hand, the make.sh takes care of it for you.
4629 * The system is no longer so grossly inefficient when compiling code
4630 involving vectors implemented as general (not simple) vectors (VECTOR T),
4631 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4632 which dares to use the various sequence functions on non-simple
4633 vectors, takes less of a performance hit.
4634 * There is now a primitive type predicate VECTOR-T-P
4635 to test for the (VECTOR T) type, so that e.g.
4636 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4637 can now be compiled with some semblance of efficiency. (The old code
4638 turned the type declaration into a full call to %TYPEP at runtime!)
4639 * AREF on (VECTOR T) is still not fast, since it's still compiled
4640 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4641 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4642 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4643 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4644 without making full calls to SUBTYPEP at runtime.
4645 (Some analogous efficiency issues for non-simple vectors specialized to
4646 element types other than T, or for non-simple multidimensional arrays,
4647 have not been addressed. They could almost certainly be handled the
4648 same way if anyone is motivated to do so.)
4649 * The changes in array handling break binary compatibility, so
4650 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4651 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4652 instead of (VALUES T T).
4653 * By following the instructions that Dan Barlow posted to sbcl-devel
4654 on 2 July 2000, I was able to enable primitive dynamic object
4655 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4656 functionality is not implemented (since it calls ld to resolve
4657 library references automatically, requiring RUN-PROGRAM for its
4658 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4659 try to resolve library references) is now supported.
4660 * The system now flushes the standard output streams when it terminates,
4661 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4662 them at several other probably-convenient times, e.g. in each pass of
4663 the toplevel read-eval-print loop, and after evaluating a form given
4664 as an "--eval" command-line option. (These changes were motivated by a
4665 discussion of stream flushing issues on cmucl-imp in August 2000.)
4666 * The source transform for TYPEP of array types no longer assumes
4667 that an array whose element type is a not-yet-defined type
4668 is implemented as an array of T, but instead punts, so that the
4669 type will be interpreted at runtime.
4670 * There is now some support for cross-compiling in make.sh: each of
4671 the phases of make.sh has its own script. (This should be transparent
4672 to people doing ordinary, non-cross-compile builds.)
4673 * Since my laptop doesn't have hundreds of megabytes of memory like
4674 my desktop machine, I became more motivated to do some items on
4675 my to-do list in order to reduce the size of the system a little:
4676 ** Arrange for various needed-only-at-cold-init things to be
4677 uninterned after cold init. To support this, those things have
4678 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4679 symbols with such names are now uninterned after cold init).
4680 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4681 fasl files for cold load.
4682 ** Remove the old compiler structure pooling code (which used to
4683 be conditional on the target feature :SB-ALLOC) completely.
4684 ** Redo the representation of some data in cold init to be more compact.
4685 (I also looked into supporting byte compiled code at bootstrap time,
4686 which would probably reduce the size of the system a lot, but that
4687 looked too complicated, so I punted for now.)
4688 * The maximum signal nesting depth in the src/runtime/ support code has
4689 been reduced from 4096 to 256. (I don't know any reason for the very
4690 large old value. If the new smaller value turns out to break something,
4691 I'll probably just bump it back up.)
4692 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4694 * Many, many bugs reported by Peter Van Eynde have been added to
4695 the BUGS list; some have even been fixed.
4696 * While enabling dynamic object file loading, I tried to make the
4697 code easier to understand, renaming various functions and variables
4698 with less ambiguous names, and changing some function calling
4699 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4700 * While trying to figure out how to do the OpenBSD port, I tried to
4701 clean up some of the code in src/runtime/. In particular, I dropped
4702 support for non-POSIX signal handling, added various comments,
4703 tweaked the code to reduce the number of compilation warnings, and
4704 renamed some files to increase consistency.
4705 * To support the new automatic configuration functionality in make.sh,
4706 the source file target-features.lisp-expr has been replaced with the
4707 source file base-target-features.lisp-expr and the machine-generated
4708 file local-target-features.lisp-expr.
4709 * fixed a stupid quoting error in make.sh so that using CMU CL
4710 "lisp -batch" as cross-compilation host works again
4712 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4714 * DESCRIBE no longer tries to call itself recursively to describe
4715 bound/fbound values, so that it no longer fails on symbols which are
4716 bound to themselves (like keywords, T, and NIL).
4717 * DESCRIBE now works on generic functions.
4718 * The printer now prints less-screwed-up representations of closures
4719 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4720 * A private symbol is used instead of the :EMPTY keyword previously
4721 used to mark empty slots in hash tables. Thus
4722 (DEFVAR *HT* (MAKE-HASH-TABLE))
4723 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4724 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4725 now does what ANSI says that it should. (You can still get
4726 similar noncompliant behavior if bang on the hash table
4727 implementation with all the symbols you get back from
4728 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4729 This breaks binary compatibility, since tests for equality to
4730 :EMPTY are wired into things like the macroexpansion of
4731 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4733 * There's now a minimal placeholder implementation for CL:STEP,
4734 as required by ANSI.
4735 * An obscure bug in the interaction of the normal compiler, the byte
4736 compiler, inlining, and structure predicates has been patched
4737 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4738 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4740 * Missing ordinary arguments in a macro call are now detected even
4741 when the macro lambda list contains &KEY or &REST.
4742 * The debugger no longer complains about encountering the top of the
4743 stack when you type "FRAME 0" to explicitly instruct it to go to
4744 the top of the stack. And it now prints the frame you request even
4745 if it's the current frame (instead of saying "You are here.").
4746 * As specified by ANSI, the system now always prints keywords
4747 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4749 * The default initial SIZE of HASH-TABLEs is now smaller.
4750 * Type information from CLOS class dispatch is now propagated
4751 into DEFMETHOD bodies, so that e.g.
4752 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4754 is now basically equivalent to
4755 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4756 (DECLARE (TYPE SINGLE-FLOAT X))
4758 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4759 operation, without having to do run-time type dispatch.
4760 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4761 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4762 * Since I don't seem to be making big file reorganizations very often
4763 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4764 is overflowing my ability to conveniently back them up), I've finally
4765 checked the system into CVS. (The CVS repository is on my home system,
4766 not at SourceForge -- putting it on SourceForge might come later.)
4767 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4768 high-level GC-NOTIFY-FOO functions send their output. (There's
4769 still very little control of where low-level verbose GC functions
4770 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4771 less than it used to -- the GC-NOTIFY-FOO functions are now under
4772 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4773 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4774 in only one place in the source code, and propagates it automatically
4775 everywhere that it's needed. Thus e.g. when I bump the version from
4776 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4777 * The C source files now include boilerplate legalese and documentation
4778 at the head of each file (just as the Lisp source files already did).
4779 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4780 to his page will be replaced with a link to his new CLiki service.
4782 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4784 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4785 * The build process now looks for GNU make under the default name "gmake",
4786 instead of "make" as it used to. If GNU make is not available as "gmake"
4787 on your system, you can change this default behavior by setting the
4788 GNUMAKE environment variable.
4789 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4790 can build without error under CMU CL.
4792 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4794 * There is now a partial SBCL user manual (with some new text and some
4795 text cribbed from the CMU CL manual).
4796 * The beginnings of a profiler have been added (starting with the
4797 CMU CL profiler and simplifying and cleaning up). Eventually the
4798 main interface should be through the TRACE macro, but for now,
4799 it's still accessed through vaguely CMU-CL-style functions and macros
4800 exported from the package SB-PROFILE.
4801 * Some problems left over from porting CMU CL to the new
4802 cross-compilation bootstrap process have been cleaned up:
4803 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4804 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4805 part when getting rid of the compiler *BACKEND* stuff.)
4806 ** The value of *NULL-TYPE* was screwed up, because it was
4807 being initialized before the type system knew the final
4808 definition of the 'NULL type. This screwed up several key
4809 optimizations in the compiler, causing inefficiency in all sorts
4810 of places. (I found it because I wanted to understand why
4811 GET-INTERNAL-RUN-TIME was consing.)
4812 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4813 PROCLAIM FTYPE information. Unfortunately this broke binary
4814 compatibility again, since now the forms output by DEFGENERIC
4815 to refer to functions which didn't exist in 0.6.3.
4816 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4817 can use the new (as of 0.6.3) transform for SOME into MAP into
4819 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4820 values to (INTEGER 0 1000000), so that the time code will no longer
4821 occasionally get blown up by Linux returning 1000000 microseconds
4822 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4823 its output conform to the ANSI spec. (Alas, this makes its output
4824 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4825 * A full call to MAP NIL with a single sequence argument no longer conses.
4826 * fixes to problems pointed out by Martin Atzmueller:
4827 * The manual page no longer talks about multiprocessing as though
4828 it were currently supported.
4829 * The ILISP support patches have been removed from the distribution,
4830 because as of version 5.10.1, ILISP now supports SBCL without us
4831 having to maintain patches.
4832 * added a modified version of Raymond Toy's recent CMU CL patch for
4833 EQUALP comparison of HASH-TABLE
4835 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4837 * The system still can't cross-compile itself with
4838 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4839 entails), but at least it can compile more of itself that way
4840 than it used to be able to, and various buglets which were uncovered
4841 by trying to cross-compile itself that way have now been fixed.
4842 * This release breaks binary compatibility again. This time
4843 at least I've incremented the FASL file format version to 2, so that the
4844 problem can be detected reliably instead of just causing weird errors.
4845 * various new style warnings:
4846 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4847 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4848 ** using the lexical binding of a variable named in the *FOO* style
4849 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4850 as specified by ANSI.
4851 * *RANDOM-STATE* is no longer automatically initialized from
4852 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4853 default behavior of the system is to repeat its behavior every time
4854 it's run. If you'd like to change this behavior, you can always
4855 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4856 old convention there was no comparably easy way to get the system to
4857 repeat its behavior every time it was run.
4858 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4859 FTYPE declarations has been removed, in favor of their ANSI
4860 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4861 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4862 the types of their sequence arguments can be determined at compile time.
4863 This is done through a new open code expansion for MAP which eliminates
4864 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4865 argument types can be determined at compile time.
4866 * The optimizer now transforms COERCE into an identity operation when it
4867 can prove that the coerced object is already of the correct type. (This
4868 can be a win for machine generated code, including the output of other
4869 optimization transforms, such as the MAP transform above.)
4870 * Credit information has been moved from source file headers into CREDITS.
4871 * Source file headers have been made more standard.
4872 * The CASE macro now compiles without complaining even when it has
4875 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4877 * (Note that the way that the PCL macroexpansions were rewritten
4878 to accommodate the change in DEFGENERIC below breaks binary
4879 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4880 not run under sbcl-0.6.2. Once we get out of alpha releases,
4881 i.e. hit release 1.0.0, we'll probably try to maintain binary
4882 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4883 and sbcl-1.4.4. Until then, however, it might be fairly common
4884 for maintenance releases to break binary compatibility.)
4885 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4887 * The reporting of unhandled signals has been changed to print some
4888 explanatory text as well as the report form. (Previously only
4889 the report form was printed.)
4890 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4891 it defines, so that the compiler no longer issues undefined function
4892 warnings for compiled-but-not-yet-loaded generic functions.
4893 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4894 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4895 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4896 section of the ANSI spec are supported.
4897 * The "" nickname for the "KEYWORD" package has been removed.
4898 The reader still handles symbol tokens which begin with a package marker
4899 as keywords, but it doesn't expose its mechanism for doing so in the
4900 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4901 * The system now issues STYLE-WARNINGs for contradictory TYPE
4902 proclamations. (Warnings for contradictory FTYPE proclamations would
4903 be nice too, but those can't be done usefully unless the type system
4904 is made smarter about FUNCTION types.)
4905 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4906 systematized, so that "*target-*.lisp is supposed to exist only on the
4907 target and imply that there's a related file which exists on the
4908 host, and *host-*.lisp is supposed to exist only on the host and imply
4909 that there's a related file which exists on the target. This involves a
4910 lot of renaming. Hopefully the acute confusion caused by the renaming
4911 will be justified by the reduction in chronic confusion..
4912 ** runtime-type.lisp -> early-target-type.lisp
4913 ** target-type.lisp -> late-target-type.lisp
4914 ** early-host-format.lisp -> early-format.lisp
4915 ** late-host-format.lisp -> late-format.lisp
4916 ** host-error.lisp -> misc-error.lisp
4917 ** early-error.lisp -> early-target-error.lisp
4918 ** late-error.lisp -> late-target-error.lisp
4919 ** host-defboot.lisp -> early-defboot.lisp
4920 ** code/misc.lisp -> code/target-misc.lisp
4921 ** code/host-misc.lisp -> code/misc.lisp
4922 ** code/numbers.lisp -> code/target-numbers.lisp
4923 ** code/early-numbers.lisp -> numbers.lisp
4924 ** early-host-type.lisp -> early-type.lisp
4925 ** late-host-type.lisp -> late-type.lisp
4926 ** host-typep.lisp -> typep.lisp
4927 ** load.lisp -> target-load.lisp
4928 ** host-load.lisp -> load.lisp
4929 ** host-disassem.lisp -> disassem.lisp
4930 ** host-insts.lisp -> insts.lisp
4931 ** byte-comp.lisp -> target-byte-comp.lisp
4932 ** host-byte-comp.lisp -> byte-comp.lisp
4933 ** host-signal.lisp -> signal.lisp
4934 ** host-defstruct.lisp -> defstruct.lisp
4935 ** late-target-type.lisp -> deftypes-for-target.lisp
4936 Furthermore, several other previously target-only files foo.lisp (e.g.
4937 hash-table.lisp and random.lisp) have been split into a target-and-host
4938 foo.lisp file and a target-only target-foo.lisp file, with their key type
4939 definitions in the target-and-host part, so that the cross-compiler will
4940 know more about target types.
4941 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4942 gone away. In their place are various *BACKEND-FOO* variables
4943 corresponding to the slots of the old structure.
4944 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4945 into the target SBCL has been fixed.
4946 * The chill.lisp system for loading cold code into a running SBCL
4948 * Support for the CMU CL "scavenger hook" extension has been removed.
4949 (It was undocumented and unused in the CMU CL sources that SBCL was
4950 derived from, and stale in sbcl-0.6.1.)
4951 * Various errors in the cross-compiler type system were detected
4952 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4953 (enabling various consistency checks). Many of them were fixed,
4954 but some hard problems remain, so the compiler is back to
4955 running without *TYPE-SYSTEM-INITIALIZED* for now.
4956 * As part of the cross-compiler type system cleanup, I implemented
4957 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4958 * I have started adding UNCROSS calls throughout the type system
4959 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4960 on cross-compiler input files will be able to go away, and various
4962 * CONSTANTP now returns true for quoted forms (as explicitly required
4965 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4967 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4968 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4969 (MAKE-LIST -1). In the longer term, it ought to become true
4970 that declarations are assertions even at SAFETY 1. For now, it's not
4971 quite true even at SAFETY 3, but it's at least more nearly true..
4972 (Note that this change seems to increases the size of the system by
4973 O(5%) and to decrease the speed of the compiler by 20% or more.)
4974 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4975 for the problem of printing dozens of lines of distracting information
4976 about low-level system machinery as part of the top stack frame
4977 on entry to the debugger when an undefined function was called.
4978 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4979 is not reset to COMMON-LISP-USER.
4980 * Compilation of stuff related to dyncount.lisp has been made conditional
4981 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4982 smaller. The various dyncount-related symbols have been moved into
4983 a new "SB-DYNCOUNT" package.
4984 * tty-inspect.lisp has been renamed to inspect.lisp.
4985 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4986 feature has gone away. (When we eventually port to other flavors of
4987 libc and/or Unix, we'll try to make the differences between flavors
4988 invisible at the user level.)
4989 * Various other *FEATURES* tags, and/or their associated conditionals,
4990 have been removed if obsolescent, or given better documentation, or
4991 sometimes given more-mnemonic names.
4993 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4995 * tidied up "make.sh" script
4996 * tidied up system directory structure
4997 * better "clean.sh" behavior
4998 * added doc/FOR-CMUCL-DEVELOPERS
4999 * many many small tweaks to output format, e.g. removing possibly-confusing
5000 trailing #\. character in DESCRIBE-INSTANCE
5001 * (EQUALP #\A 'A) no longer signals an error.
5002 * new hashing code, including EQUALP hashing
5003 * tidied up Lisp initialization and toplevel
5004 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5005 * command line argument processing
5006 * added POSIX-GETENV function to deal with Unix-ish environment variables
5007 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5008 terminating SBCL on EOF
5009 * non-verbose GC by default
5010 * There is no more "sbcl" shell script; the sbcl file is now the C
5011 runtime executable (just like CMU CL).
5012 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5014 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5015 symbol and package information as Lisp native symbol and package objects
5016 instead of strings naming symbols and strings naming packages. This way,
5017 whenever packages are renamed (as in warm init), debug information is
5018 transformed along with everything else.
5019 * tweaked the optimization policy declarations which control the building
5020 of SBCL itself. Now, among other things, the system no longer saves
5021 source location debugging information. (This helps two problems at once
5022 by reducing SBCL size and by keeping SBCL from trying to look for its
5023 sources -- which may not exist -- when reporting errors.)
5024 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5025 debugging and testing purposes
5026 * cleaned up printing, making the printer call PRINT-OBJECT for
5027 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5028 methods, giving nearly-ANSI behavior
5029 * converted almost all special variables to use *FOO* naming convention
5030 * deleted PARSE-TIME functionality, since it can be done portably
5031 * moved some files out of cold init into warm init
5032 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5034 * regularized formatting of source files
5035 * added an install.sh script
5036 * fixed ridiculous memory usage of cross-compiler by making
5037 compiler/alloc.lisp not try to do pooling unless it can hook
5038 itself into the GC of the cross-compilation host. Now the system
5039 builds nicely on my old laptop.
5040 * added :SB-ALLOC in target-features.lisp-expr
5041 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5042 was not implemented)
5043 * re-did condition handling and note reporting in the compiler. Notes
5044 are no longer handled by signalling conditions. Style warnings
5045 and warnings are handled more correctly and reported in such a way
5046 that it's easy to find one or the other in your output (so that you
5047 can e.g. figure out which of many problems caused COMPILE-FILE to
5049 * changed the severity of several compiler warnings from full WARNING
5050 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5051 compiler note reporting so that it doesn't use the condition system
5052 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5053 COMPILE-FILE command)
5054 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5055 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5056 no longer supported, since their implementation was deeply intertwingled
5057 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5058 * removed bogus "support" for compiler macros named (SETF FOO), and
5059 removed the compiler macro for SETF INFO (but only after making a fool
5060 of myself on the cmucl-imp mailing list by posting a bogus patch for
5061 DEFINE-COMPILER-MACRO..)
5062 * Compiled files containing forms which have side effects on the Lisp
5063 reader (such as DEFPACKAGE forms) are now handled more correctly.
5064 (Compiler queuing of top level lambdas has been suppressed by setting
5065 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5066 * deleted various currently-unused source files, e.g. gengc.lisp. They
5067 may be added back at some point e.g. when porting to other architectures,
5068 but until they are it's distracting to distribute them and to try to
5070 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5071 there were so many of them they're just distractions, and UNCROSS is
5072 known to be able to handle the current sources
5073 * moved PROFILE functionality into TRACE, so that it will be clear
5074 how the wrapping and unwrapping of functions when you profile them
5075 interacts with the wrapping and unwrapping of functions when you
5076 trace them. (Actually, the functionality isn't there yet, but at least
5077 the interface specification is there. Hopefully, the functionality will
5078 arrive with some future maintenance release.)
5079 * removed host-oops.lisp
5080 * changed signature of QUIT function to allow UNIX-CODE argument
5081 * fixed READ-SEQUENCE bug
5082 * tweaked verbose GC output so that it looks more like the progress
5083 output that ANSI specifies for functions like LOAD
5084 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5085 * added <http://sbcl.sourceforge.com> to the banner information printed by