1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes in sbcl-1.0.28 relative to 1.0.27:
3 * bug fix: timers could go off in the wrong order, be delayed indefinitely
4 (thanks to Ole Arndt for the patch)
6 changes in sbcl-1.0.27 relative to 1.0.26:
7 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
8 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
9 * improvement: the system either recovers from stack exhaustion or dies
10 properly as opposed to leaving the user uncertain of whether the handler
11 trampled on some random memory next to the stack or having to rely on
12 --lose-on-corruption (which is still a good idea to use in production
13 because stack exhaustion can happen in signal handlers which will likely
15 * bug fix: fix gc related interrupt handling bug on ppc (regression from
16 1.0.25.37, reported by Harald Hanche-Olsen)
17 * bug fix: work around signal delivery bug in darwin (regression from
18 1.0.25.44, reported by Sidney Markowitz)
19 * bug fix: fix ERROR leaking memory (reported by David Thompson)
21 changes in sbcl-1.0.26 relative to 1.0.25:
22 * incompatible change: an interruption (be it a function passed to
23 INTERRUPT-THREAD or a timer function) runs in an environment where
24 interrupts can be enabled. The interruption can use
25 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
26 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
27 potentially running out of stack. Keep in mind that in the absence
28 of WITHOUT-INTERRUPTS some potentially blocking operation such as
29 acquiring a lock can enable interrupts.
30 * incompatible change: GC-OFF and GC-ON are removed, as they were
31 always unsafe. Use WITHOUT-GCING instead.
32 * new feature: runtime option --disable-ldb
33 * new feature: runtime option --lose-on-corruption to die at the
34 slightest hint of possibly non-recoverable errors: running out of
35 memory, stack, alien stack, binding stack, encountering a memory
36 fault, etc. In the absence of --lose-on-corruption a warning is
38 * enhancement: detect binding stack exhaustion
39 * enhancement: detect alien stack exhaustion on x86/x86-64
40 * improvement: generally more stable and reliable interrupt handling
41 * improvement: there is a per thread interruption queue,
42 interruptions are executed in order of arrival
43 * improvement: a repeating timer reschedules itself when the it has
44 finished, but expiration times are spaced equally. If an
45 expiration time is in the past it will trigger after a short grace
46 period that may give a chance to other things to run.
47 * optimization: slightly faster gc on multithreaded builds
48 * optimization: faster WITHOUT-GCING
49 * bug fix: when JOIN-THREAD signals an error, do it when not holding
50 important locks so that the debugger/handler doesn't produce
51 recursive errors or deadlock.
52 * bug fix: real-time signals are not used anymore, so no more
53 hanging when the system wide real-time signal queue gets full.
54 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
56 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
57 * bug fix: fix random memory faults related to interrupts on alpha
58 * bug fix: fix random memory faults related to interrupts on sparc
59 * bug fix: fix deadlocks related to starting threads
60 * bug fix: fix deadlines on locks on futex platforms
61 * bug fix: restore errno in signal handlers
62 * bug fix: fix deadlocks related to hash tables
63 * bug fix: fix deadlocks in pcl
65 changes in sbcl-1.0.25 relative to 1.0.24:
66 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
67 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
68 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
69 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
70 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
71 better name and does not return values so stale on multiprocessor systems.
72 Also, HOLDING-MUTEX-P was added for about the only sane usage of
74 * improvement: unithread builds keep track of MUTEX-VALUE.
75 * improvement: reading from a TWO-WAY-STREAM does not touch the output
76 stream anymore making it thread safe to have a concurrent reader and
77 a writer, for instance, in a pipe.
78 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
79 arguments, making compiler-macros for SETF-functions able to inspect
80 their constant arguments.
81 * improvement: COMPILE-FILE reports times with millisecond accuracy
82 (thanks to Luis Oliveira)
83 * optimization: CHAR-CODE type derivation has been improved, making
84 TYPEP elimination on subtypes of CHARACTER work better. (reported
85 by Tobias Rittweiler, patch by Paul Khuong)
86 * bug fix: setting alien structure fields of type struct by value now
87 computes the right offset for the memory copy.
88 * bug fix: compilation problem involving inlined calls to aliens with
89 result type VOID. (reported by Ken Olum)
90 * bug fix: #235a; sequential inline expasion in different policies no
91 longer reuses the functional from the previous expansion site.
92 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
93 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
94 * bug fix: named ENUMs in multiply-referenced alien record types no longer
95 cause an implied type redefinition. Regression from 1.0.21.29.
96 * bug fix: sign-extension of small signed return values now works with C
97 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
98 * improvements to the Windows port:
99 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
100 Westfold, thanks to Rudi Schlatte)
102 changes in sbcl-1.0.24 relative to 1.0.23:
103 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
104 vector of a multidimensional SIMPLE-ARRAY.
105 * new feature: the system now signals a continuable error if standard
106 readtable modification is attempted. (thanks to Tobias Rittweiler)
107 * new feature: DIRECTORY has been extended with a non-standard keyword
108 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
109 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
110 has been robustified and documented better. (thanks to Robert Goldman)
111 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
112 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
113 * tradeoff: constant FORMAT control strings are now compiled unless
114 SPACE > SPEED (previously only when SPEED > SPACE.)
115 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
117 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
119 * bug fix: lexical type declarations are now correctly reported by
120 SB-CLTL2. (reported by Larry D'Anna)
121 * bug fix: STRING-TO-OCTETS did not handle :START properly when
122 using UTF-8 as external format. (thanks to Luis Oliveira)
123 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
124 are signalled correctly. (thanks to Stas Boukarev)
125 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
126 designator. (thanks to Tobias Rittweiler)
127 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
128 designator, and returns T instead of the function. (thanks to
130 * bug fix: direct superclasses of STANDARD-CLASS and
131 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
132 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
133 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
134 when no class with the named slot yet exists no longer causes a
135 compile-time style-warning.
136 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
137 in safe code. (reported by Didier Verna)
138 * bug fix: #430; nested structure constructors can stack allocate.
139 * bug fix: on some 64-bit platforms dynamic space size was truncated
140 to #xffff0000 bytes. (reported by Benjamin Lambert)
141 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
142 REPL unusable. (reported by Daniel Herring)
143 * bug fix: return values of READ-SEQUENCE did not take :START into
144 account on file streams, regressions since 1.0.12.22. (reported by
145 Thomas Russ, patch by Paul Khuong)
146 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
147 method specializer used to confuse permuation vector optimization.
148 * bug fix: system inserted bogus implicit type declarations for local
149 special variables in DEFMETHOD bodies.
150 * bug fix: #354; duplicated frames in backtraces due to
151 non-tail-call-optimized XEPs to functions with return type NIL
153 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
154 initializes structure object slots according to DEFSTRUCT initforms,
155 and DEFSTRUCT forms :INCLUDEind structure classes defined using
156 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
157 (reported by Bruno Haible and Stephen Wilson)
158 * bug fix: #395; fill-pointer output streams used now support
159 element-type BASE-CHAR as well.
160 * bug fix: compiler error when attempting to derive return value of
161 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
163 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
165 * bug fix: Mac OS X binaries should now be portable between Leopard
168 changes in sbcl-1.0.23 relative to 1.0.22:
169 * enhancement: when disassembling method functions, disassembly
170 for the associated fast function is also produced.
171 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
173 * optimization: printing with *PRINT-PRETTY* true is now more
174 efficient as long as the object being printed doesn't require
175 special handling by the pretty printer.
176 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
177 now interact correctly with type declarations.
178 * partial bug fix: PCL detects infinite recursion during wrapper
179 validation. (thanks to Attila Lendvai)
180 * bug fix: #426; nested function calls are inlined properly.
181 Previously if FOO was an inline function, in calls of the form
182 (FOO (FOO ...)) the outer call was not inlined.
183 * bug fix: long long arguments passed in registers to alien
184 functions on PowerPC Linux and NetBSD are now handled correctly
185 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
187 changes in sbcl-1.0.22 relative to 1.0.21:
188 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
189 for the shared object in the current directory, but passes the native
190 namestring of the designated pathname to the operation system's shared
191 object loading function as-it.
192 * minor incompatible change: --disable-debugger toplevel option now takes
193 effect before processing of initialization files and --eval or --load
195 * new feature: new commandline argument: --script, which supports
196 shebang lines. See documentation for details. (based on work by
198 * new feature: SAVE-LISP-AND-DIE can save current values of
199 --dynamic-space-size and --control-stack-size in the executable core,
200 causing it to skip normal runtime option processing. See documentation
201 for details. (thanks to Zach Beane)
202 * enhancement: inoccous calls to EVAL or generic functions dispatching
203 on subclasses of eg. STREAM no longer cause compiler notes to appear.
204 * enhancement: the system no longer resignals errors from --load and
205 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
206 associated with the original error to be lost. (thanks to Ariel
208 * enhancement: :DONT-SAVE keyword argument has been added to
209 LOAD-SHARED-OBJECT for controlling interaction with
210 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
211 the effects of an earlier LOAD-SHARED-OBJECT call.
212 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
213 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
214 provided. (reported by Cedric St-Jean)
215 * bug fix: circularity handling in the reader did not treat raw
216 structure slots correctly. (reported by Cedric St-Jean)
217 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
218 file descriptors when there were none.
219 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
220 was broken. (reporteed by Willem Broekema)
221 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
222 pathnames without a directory.
223 * bug fix: COMPILE-FILE did not warn about undefined variable
224 references at toplevel, and LOAD of the resulting fasl did
226 * bug fix: functions with non-required arguments used to end up with
227 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
228 * bug fix: redefining a function with non-required arguments didn't
229 update the system's knowledge about its call signature properly.
230 * bug fix: fixed #431; incompatible alien record type redefinitions
231 are detected and handled. (reported by Neil Haven)
232 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
233 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
235 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
236 after alien stack frames.
237 * bug fix: strings passed to mk{s,d}temp lacked null termination
239 changes in sbcl-1.0.21 relative to 1.0.20:
240 * new feature: the compiler is able to track the effective type of a
241 generic function across method addition and removal.
242 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
243 appropriately typed structure slots without locking.
244 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
245 information like those gathered by TIME using a programming-friendly
247 * new feature: TIME reports time taken even if the form performs a
248 non-local transfer of control.
249 * enhancement: reduced conservativism on GENCGC platforms: on
250 average 45% less pages pinned (measured from SBCL self build).
251 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
252 mutate constant symbols or violate declaimed type of the symbol.
253 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
254 sleeping on the semaphore in heavily contested semaphores.
255 * bug fix: semaphores and condition variables were not interrupt
257 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
258 owned by other threads anymore.
259 * bug fix: FIND on lists called KEY outside the specified
260 subsequence. (reported by budden)
261 * bug fix: LOG doesn't use single-float intermediate results when
262 given mixed integer and double-float arguments, leading to better
263 precision. (reported by Bob Felts)
264 * bug fix: LOG with base zero returned values of inconsistent type.
266 changes in sbcl-1.0.20 relative to 1.0.19:
267 * minor incompatible change: OPTIMIZE qualities
268 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
269 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
270 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
271 * documentation: some slot access efficiency guidelines have been
272 added to the user manual.
273 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
274 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
275 as ASSOC and MEMEBER.
276 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
277 to more efficient EQ-comparison versions more often.
278 * optimization: enhanced derivation of DOLIST iteration variable type
280 * optimization: constant folding of simple (LIST ...) forms as DOLIST
282 * optimization: runtime lookup of function definitions can be
283 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
284 * optimization: compiler is able to derive the return type of
285 (AREF (THE STRING X) Y) as being CHARACTER.
286 * optimization: CLRHASH on empty hash-tables no longer does pointless
287 work. (thanks to Alec Berryman)
288 * optimization: REPLACE deftransforms don't punt when :START1 or
290 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
291 no longer emits a code deletion note. (reported by Andrew
293 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
294 functions with an odd number of &OPTIONAL arguments, a &REST
295 argument, and one or more &KEY arguments at the call site.
296 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
297 arguments appeared at call sites as well.
298 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
299 READ-LINE &co can no longer read from them afterwards. (reported
301 * bug fix: fixed #427: unused local aliens no longer cause compiler
302 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
304 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
305 alien-stack leakage. (reported by Andy Hefner)
306 * bug fix: PROGV signals an error when an attempt to violate declared
307 type of a variable or bind a constant is made.
308 * bug fix: SET signals an error when an attempt to violate declared
309 type of a variable is made.
310 * bug fix: restart computation during the execution of a restart
311 test function no longer causes infinite recursion. (reported by
313 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
314 signals an error. (thanks to Pierre Mai)
315 * bug fix: EXPORT left symbol unexported in conflict situations.
316 (thanks to Michael Weber)
317 * bug fix: correctly handle name conflicts involving CL:NIL.
318 (thanks to Michael Weber)
319 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
320 conflicts arising from USEing package with conflicting symbols
321 correctly. (thanks to Michael Weber)
323 changes in sbcl-1.0.19 relative to 1.0.18:
324 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
325 warnings that go otherwise unhandled will be muffled if they are
326 of the type that's the value of this variable.
327 * optimization: stack allocation is slightly more efficient on x86
329 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
330 where a raw slot always is initialized using the initform whose
331 type is not know sufficiently well a compile-time are now compiled
332 correctly. (reported by John Morrison)
333 * bug fix: compiler no longer makes erronous assumptions in the
334 presence of non-foldable SATISFIES types.
335 * bug fix: stack analysis missed cleanups of dynamic-extent
336 arguments in non-let-converted calls to local functions.
337 * improvements to the Windows port:
338 ** adjusted address spaces for building on both Win32 and
339 Win64. (thanks for John Connors)
340 * fixed some bugs revealed by Paul Dietz' test suite:
341 ** interval arithmetic during type derivation used inexact integer
342 to single-float coercions.
343 ** arithmetic operations involving large integers and single
344 floats give the same results in compiled and interpreted code.
345 ** deriving the result type of COERCE no longer signals an error
346 if the derived type of the second argument is a MEMBER type
347 containing invalid type specifiers.
348 ** ADJOIN with constant NIL as second argument works correctly.
349 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
351 changes in sbcl-1.0.18 relative to 1.0.17:
352 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
353 profiles only the current thread.
354 * minor incompatible change: changes to SYMBOL-VALUE of constants
355 defined with DEFCONSTANT now signal an error.
356 * enhancement: SB-SPROF now has support for wallclock profiling,
357 and is also able to profile specific threads. REPORT output
358 has also additional sorting options.
359 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
361 * optimization: structure allocation has been improved
362 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
363 ** out of line constructors are ~10% faster.
364 ** inline constructors are ~15% faster.
365 ** inline constructors are capable of dynamic extent allocation
366 (generally on x86 and x86-64, in some cases on other platforms
368 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
370 * optimization: file compiler is now able to coalesce non-circular
371 lists, non-base strings, and bit-vectors. Additionally, constants
372 are never referenced through SYMBOL-VALUE at runtime.
373 * optimization: code defining methods on PRINT-OBJECT (and other
374 generic functions in the COMMON-LISP package) now loads faster.
375 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
377 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
378 fasl is now deleted. (reported by Attila Lendvai)
379 * bug fix: READ-LINE always returned NIL for the last line in files.
380 (reported by Yoshinori Tahara)
381 * bug fix: more accurate disassembly annotations of foreign function
382 calls. (thanks to Andy Hefner)
383 * bug fix: trimming non-simple strings and non-string string
384 designators when the there is nothing to trim works properly.
385 (thanks to James Knight)
386 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
387 (thanks to Travis Cross)
388 * fixed some bugs revealed by Paul Dietz' test suite:
389 ** NIL is a valid function name (regression at 1.0.13.38)
390 ** FILL on lists was missing its return value (regression at 1.0.12.27)
391 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
392 fill pointers properly (regression at 1.0.12.23)
393 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
394 displacement indices properly (regression at 1.0.12.23)
396 changes in sbcl-1.0.17 relative to 1.0.16:
397 * temporary regression: user code can no longer allocate closure
398 variable storage on stack, due to bug 419 without explicitly
399 requesting it. Please consult sbcl-devel for advice if you need to
400 use this feature in the meanwhile.
401 * new feature: runtime argument --control-stack-size can be used to
402 adjust thread default control stack size.
403 * enhancement: improved TIME output
404 ** all times are reported using the measured accuracy (milliseconds
405 for real and GC times, microseconds for everything else.)
406 ** processor cycle counts on x86 and x86-64.
407 ** interpreted forms are counted for both evaluator modes.
408 ** number of lambdas converted by the compiler is reported.
409 ** CPU percentage report (computed from real and total run time.)
410 ** more comprehensive run time reporting, using a condenced format
411 ** interperted form, lambda, and page fault counts are omitted
413 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
415 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
416 * optimization: two argument forms of LAST are upto ~10% faster
417 in normal SPEED policies.
418 * optimization: NCONC no longer needs to heap cons its &REST list
419 in normal SPEED policies.
420 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
422 * bug fix: UNION and NUNION work with :TEST-NOT once more,
423 regression since 1.0.9.1. (thanks to Eric Marsden)
424 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
425 since 1.0.15.36. (thanks to Paul Khuong)
426 * bug fix: LAST when always returned the whole list when given a bignum
427 as the second argument.
428 * bug fix: dynamic extent allocation of nested lists and vectors
429 could leak to otherwise accessible parts.
430 * bug fix: invalid optimization of heap-allocated alien variable
432 * bug fix: fasl header checking is less vulnerable to different
433 platform word lengths.
434 * bug fix: more correct assembler syntax for GNU binutils
435 2.18.50.0.4 support. (thanks to Marijn Schouten)
436 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
438 * bug fix: the fopcompiler can handle LOCALLY forms (with no
439 declarations) successfully. (reported by Attila Lendvai)
441 changes in sbcl-1.0.16 relative to 1.0.15:
442 * minor incompatible change: revert the changes to sb-posix's error
443 signaling added in 1.0.14.
444 * minor incompatible change: change PROBE-FILE back to returning
445 NIL whenever we can't get a truename, as was the case before 1.0.14.
446 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
447 inherits from ERROR instead of just CONDITION.
448 * new feature: SB-INTROSPECT can provide source locations for instances
449 as well. (thanks to Tobias Rittweiler)
450 * optimization: binding special variables now generates smaller code
451 on threaded platforms.
452 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
453 and cases where no :TEST is given but the compiler can infer that
454 the element to search is of type (OR FIXNUM (NOT NUMBER)).
455 * optimization: better LOGNOT on fixnums.
456 * optimization: modular arithmetic for a particular requested width
457 is implemented using a tagged representation unless a better
458 representation is available.
459 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
460 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
461 for wellformedness and returns a specialized vector. (reported by
462 Francois-Rene Rideau)
463 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
464 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
465 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
466 unscheduling timer at the same time as another timer fires.
467 * bug fix: don't reschedule timers for dead threads.
468 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
469 * bug fix: copying output from RUN-PROGRAM to a stream signalled
470 bogus errors if select() was interrupted.
471 * enhancement: add support for fcntl's struct flock to SB-POSIX.
473 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
474 * enhancement: cleaner backtraces for interactive interrupts, as
475 well as other cases where the interesting frames used to be
476 obscured by interrupt handling frames.
477 * enhancement: untracing a whole package using (UNTRACE "FOO") is
478 now supported, and tracing a whole package using (TRACE "FOO") now
479 traces SETF-functions as well.
480 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
481 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
482 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
483 * unreadably printed representation of hash-tables now includes
485 * bug fix: partially fixed #188: type propagation from assignments
486 is now more efficient.
487 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
488 'DOUBLE-FLOAT) are not flushable.
489 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
490 full address of the object, and none of the tag bits.
491 * bug fix: readably printing hash-tables now respects other printer
492 control variables. (reported by Cedric St-Jean)
493 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
494 keyword with MAKE-HASH-TABLE.
495 * bug fix: export SB-POSIX:MKSTEMP.
496 * bug fix: SORT was not interrupt safe.
497 * bug fix: XREF accounts for the last node of each basic-block as
499 * bug fix: MAKE-INSTANCE optimizations interacted badly with
500 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
501 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
502 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
503 create obsolete instances in the case of redefinition or
504 obsoletion of a superclass. (thanks to Andy Hefner)
505 * bug fix: Support for the Alpha architecture has been revived; it had
506 suffered somewhat from lack of maintenance since sbcl-1.0.
507 * improvements to the Windows port:
508 ** The system detects the codepage to use at startup, instead of
509 using the value from the saved core. (thanks to Kei Suzuki)
511 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
512 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
513 (see documentation for details.)
514 * revived support for OpenBSD (contributed by Josh Elsasser)
515 * partially fixed bug #108: ROOM no longer suffers from occasional
516 (AVER (SAP= CURRENT END)) failures.
517 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
518 bodies no longer cause a WARNING to be signalled. (reported by
520 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
521 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
522 the bug in TRUNCATE.) (reported by Michael Weber)
523 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
524 no samples. (reported by Andy Hefner)
525 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
526 no longer appear as (NIL ...) frames in backtraces.
527 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
528 is now actually exported from SB-EXT as documented. (reported by
530 * bug fix: sb-aclrepl now correctly understands how to inspect
531 single-floats on 64-bit platforms where single-floats are not boxed.
532 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
533 yet been finalized. (reported by Levente Meszaros)
534 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
535 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
536 SBCL-specific optimize qualities.
538 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
539 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
540 an executable in the search path, and does so in the child
541 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
542 been removed; it can be found in the CVS history, for somebody who
543 needs that search behavior (see the manual).
544 * minor incompatible change: compiler policy re. weakening type
545 checks has changed: now type checks are weakened only if SAFETY < 2
547 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
548 unparsing of directory pathnames as files. Analogously,
549 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
550 filename to parse into a directory pathname.
551 * enhancement: implicit generic function creation now signals a
552 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
553 users can bind handlers for and muffle around calls to LOAD.
554 * enhancement: RUN-PROGRAM allows unicode arguments and environments
555 to be used (using the default stream external format), and allows
556 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
557 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
558 strings and vectors whose element-type or simplicity is not fully
559 known at compile-time.
560 * optimization: STRING-TRIM and related functions no longer allocate
561 a new string when no trimming needs to be performed. These functions
562 are also faster than before when the input string has been declared
564 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
565 * optimization: READ-LINE is significantly faster for files containing
567 * optimization: non-open coded uses of character comparison operators
568 (e.g. char=) no longer cons when called with more than one parameter
569 on platforms supporting dynamic-extent allocation.
570 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
571 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
572 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
574 * bug fix: some sequence functions elided bounds checking when
576 * bug fix: too liberal weakening of union-type checks when SPEED >
578 * bug fix: more bogus fixnum declarations in ROOM implementation
581 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
582 * new feature: MAKE-HASH-TABLE now experimentally accepts a
583 :SYNCHRONIZED argument, which makes the hash-table safe for
584 concurrent accesses (but not iteration.) See also:
585 SB-EXT:WITH-LOCKED-HASH-TABLE, and
586 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
587 * optimization: CONCATENATE on strings is an order of magnitude faster
588 in code compiled with (> SPEED SPACE).
589 * optimization: SUBSEQ is ~50% faster on lists.
590 * bug fix: bug 417 fixed -- source location reporting is now more robust.
591 * bug fix: SUBSEQ on a list will now correctly signal an error if if
592 END is smaller then START.
593 * bug fix: SB-PROFILE will no longer report extra consing for nested
594 calls to profiled functions.
595 * bug fix: ROOM implementation had bogus fixnum declarations which
596 could cause type-errors when calling ROOM on large images.
597 * bug fix: if file compilation is aborted, the partial fasl is now
598 deleted, and COMPILE-FILE returns NIL as the primary value.
599 * bug fix: number of thread safety issues relating to SBCL's internal
600 hash-table usage have been fixed.
601 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
602 be returned from its body when the values were being returned
603 using unknown-values return convection and the W-P-O was wrapped
604 inside an UNWIND-PROTECT.
605 * bug fix: sb-posix should now compile again under Windows, enabling
608 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
609 * incompatible change: hash-table accessor functions are no longer
610 automatically protected by locks. Concurrent accesses on the same hash-table
611 from multiple threads can give inconsistent results or even corrupt the
612 hash-table completely. Multi-threaded applications should do their own
613 locking at the correct granularity. In the current implementation it is
614 still safe to have multiple readers access the same table, but it's not
615 guaranteed that this property will be maintained in future releases.
616 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
617 and will signal an error at runtime.
618 * enhancement: SB-THREAD package now exports a semaphore interface.
619 * enhancement: CONS can now stack-allocate on x86 and
620 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
622 * enhancement: nested lists can now be stack allocated on
623 platforms providing stack allocation support.
624 * enhancement: dynamic-extent support has been extended to support
625 cases where there are multiple possible sources for the stack
627 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
628 if the mutex is uncontested on Linux.
629 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
630 as specified by the CLHS. (thanks to Tobias Rittweiler)
631 * bug fix: NaN comparison now works on x86-64.
632 * bug fix: CLOSE :ABORT T on a stream with pending output now
634 * bug fix: instances of non-standard metaclasses using standard
635 instance structure protocol sometimes missed the slot type checks
637 * bug fix: known functions can be inlined in high-debug code.
639 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
640 * minor incompatible change: the MSI installer on Windows no longer
641 associates .lisp and .fasl files with the installed SBCL.
642 * minor incompatible change: :UNIX is no longer present in *FEATURES*
643 on Windows. (thanks to Luis Oliviera)
644 * new feature: SB-CLTL2 contrib module now implements
645 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
646 * optimization: scavenging weak pointers is now more efficient,
647 requiring O(1) instead of O(N) per weak pointer to identify
648 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
649 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
650 method bodies using constant slot-names when the first argument is
651 a specializer parameter for the method.
652 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
653 names, when the class of the instance is a direct instance
654 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
655 as slow as the constant slot-name case.)
656 * optimization: member type construction is now O(N) instead
658 * optimization: UNION and NUNION are now O(N+M) for large
659 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
661 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
662 now more readable in environments like Slime which display it.
663 (thanks to Tobias C. Rittweiler)
664 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
665 was able to infer the correct argument type for the object on which
666 the CAS operation was being performed.
667 * bug fix: copy propagation interfered with parallel assignment
668 semantics in local calls. (reported by Paul Khuong)
669 * bug fix: the signed modular fixnum shift compiled to wrong code on
670 x86 and x86-64. (spotted by a slight modification to some of
673 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
674 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
675 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
676 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
678 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
679 and generic function now print the object identity for anonymous
680 instances. (thanks to Attila Lendvai)
681 * enhancement: as an extension to MOP, SBCL now supports SETF of
682 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
683 (thanks to Attila Lendvai)
684 * enhancement: sb-cover has an alternate annotation mode which puts
685 the form annotations at the car of the form, instead of annotating
686 it completely. The benefit of this mode is that it shows better
687 which forms were instrumented by the compiler.
688 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
689 by printer control variables. (reported by Dan Corkill)
690 * bug fix: system leaked memory when delayed output was performed by
691 the OS in smaller chunks then expected. (thanks to David Smith)
692 * bug fix: system leaked memory when file streams were not closed
694 * bug fix: large objects written to slow streams that were modified
695 after the write could end up with the modified state written to
696 the underlying file descriptor.
697 * bug fix: multiple threads operating in parallel on the same stream
698 could cause buffer-overflows.
699 * bug fix: source location information is stored correctly
700 (broken since 1.0.6). This bug would generally show up as the
701 Slime debugger highlighting the wrong form when the sldb-show-source
703 * bug fix: Branch forms are again annotated as branches in the sb-cover
705 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
706 on x86-64. (reported by Gregory Vanuxem)
707 * bug fix: EQUALP could return wrong results for structures with raw
708 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
709 word). (reported by Vjacheslav Fyodorov)
710 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
713 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
714 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
715 atomic compare-and-swap operations on threaded platforms.
716 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
717 allows assigning a global minimum value to optimization qualities
718 (overriding proclamations and declarations).
719 * enhancement: closed over variables can be stack-allocated on x86
721 * performance bug fix: GETHASH and (SETF GETHASH) are once again
723 * optimization: slot definition lookup is now O(1). This speeds up
724 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
725 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
726 * optimization: ASSOC and MEMBER can now be open-coded for all
727 combinations of keyword arguments when second argument is constant
728 and SPEED >= SPACE. In other cases a specialized version is
730 * bug fix: using obsoleted structure instances with TYPEP and
731 generic functions now signals a sensible error.
732 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
733 (reported by Kristoffer Kvello)
734 * bug fix: backtrace construction is now more careful when making
735 lisp-objects from pointers on the stack, to avoid creating bogus
736 objects that can be seen by the GC.
737 * bug fix: defaulting of values in contexts expecting more than 7
738 variables now works on x86-64. (reported by Christopher Laux)
739 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
741 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
742 as the property-list of a symbol.
743 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
744 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
745 situations requested, are once again file-compileable. (reported
748 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
749 * MOP improvement: support for user-defined subclasses of
750 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
751 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
752 the expansion of DEFMETHOD, is responsible for generating a form
753 which creates a list of specializers when evaluated. Additional
754 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
755 debugging and introspective support.
756 * minor incompatible change: the (unsupported) spinlock interface
757 has changed: free spinlock now has the value NIL, and a held spinlock
758 has the owning thread as its value.
759 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
760 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
762 * enhancement: name of a socket-stream is now "a socket" instead of
764 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
765 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
766 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
767 for generic functions. (Thanks to Tobias C. Rittweiler)
768 * optimization: bignum printing speed has been improved by 20-40%
769 (depending on the bignum size.)
770 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
772 * bug fix: the cache used by the CLOS to store precomputed effective
773 methods, slot offsets, and constant return values is now thread and
775 * bug fix: generic function dispatch function updating is now thread
776 and interrupt safe (in the sense that the known issues have been
778 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
779 * bug fix: interning EQL-specializers is now thread and interrupt safe.
780 * bug fix: asdf systems with dependencies to the SB-POSIX or
781 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
782 * bug fix: interrupt safety of applicable method computation has been
785 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
786 * new contrib: sb-cover, an experimental code coverage tool, is included
788 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
789 significantly faster.
790 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
791 produce more efficient code on x86-64 if the compiler can determine
792 that (+ INDEX <constant>) does not require a bounds check and FOO
793 has an element type at least 8 bits wide.
794 * enhancement: a new, experimental synchronous timeout facility is
795 provided. Refer to SB-SYS:WITH-DEADLINE for details.
796 * enhancement: when a symbol name conflict error arises, the
797 conflicting symbols are always printed with a package prefix.
798 (thanks to Kevin Reid)
799 * enhancement: stepping is now once again supported on the SPARC and
800 MIPS platforms. (It is also now more likely to work on CheneyGC
802 * enhancement: sb-sprof can now also track and report accurate call
804 * bug fixes: the treatment of non-standard subclasses of
805 SB-MOP:SPECIALIZER is more correct.
806 * incompatible change: PURIFY no longer copies the data from the
807 dynamic space into the static and read-only spaces on platforms
808 that use the generational garbage collector
809 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
811 * bug fix: GC race condition occasionally resulting in crashes with
812 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
814 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
816 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
817 a STYLE-WARNING to be signalled (regression from 1.0.4.)
818 * bug fix: an asynchronous interrupt could previously leave the
819 system running with GC inhibited.
820 * bug fix: a DECLARE form evaluated at top-level now causes an error
821 rather than silently (or verbosely) returning NIL.
822 * bug fix: trying to dissassemble functions compiled at high DEBUG could
823 result in a error being signalled due to source form lookup errors.
824 (reported by Peter Graves)
826 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
827 * incompatible change: removed writer methods for host-ent-name,
828 host-ent-addresses -- changing the values did not update the DNS
830 * minor incompatible change: changed experimental JOIN-THREAD interface
831 * documentation: the manual now lists reader and writer methods
832 in class slot documentation sections. (thanks to Richard M Kreuter)
833 * documentation: unwinding from asyncronous events has been
834 documented as unsafe.
835 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
836 in multithreaded application code.
837 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
838 platforms. (thanks to James Anderson for the optimization hint)
839 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
841 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
842 variants no longer cons.
843 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
844 their NOT- variants no longer cons.
845 * optimization: EQUAL hash tables no longer use SXHASH for objects
846 of all data types, but instead use an EQL hash for types for which
847 EQUAL is the same as EQL.
848 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
849 are significantly faster.
850 * optimization: new STRING-OUTPUT-STREAM implementation is much
851 faster for large outputs and conses 30% less on average.
852 * enhancement: XREF information is now collected for references made
853 to global variables using SYMBOL-VALUE and a constant argument.
854 * enhancement: SIGINT now causes a specific condition
855 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
856 * bug fix: FILE-POSITION used to signal an error in some cases where
857 ANSI requires it to return NIL.
858 * bug fix: ADJUST-ARRAY is now interrupt-safe.
859 * bug fix: adding and removing fd-handlers is now interrupt-safe.
860 * bug fix: inlined calls to C now ensure 16byte stack alignment on
862 * bug fix: bad type declaration in the CLOS implementation has
863 been fixed. (reported by James Anderson)
864 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
865 (thanks to Lutz Euler)
866 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
867 required by the CLHS. (reported by Samium Gromoff)
868 * bug fix: dead unbound variable references now signal an error.
869 (reported by Marco Monteiro)
870 * bug fix: / with an unused value was being deleted in safe code.
871 (thanks to Marco Monteiro and Kevin Reid)
872 * bug fix: number of characters that can be written onto a single
873 line in a file is unlimited.
874 * bug fix: some GC deadlocks caused by asynchronous interrupts have
875 been fixed by inhibiting interrupts for when GC is disabled.
876 * bug fix: some interrupt safety issues with GETHASH, (SETF
877 GETHASH), CLRHASH and REMHASH have been fixed.
878 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
879 specifier no longer causes infinite recursion.
880 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
881 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
882 * bug fix: an &environment argument with the correct variable information
883 is passed to macros that are expanded during byte compilation.
884 (reported by Samium Gromoff)
885 * bug fix: a logic error was causing the hash function for bignums to
886 have excessive amounts of collisions. (reported by Faré Rideau)
887 * bug fix: modifying the contents of an array could change the return
888 value of SXHASH on that array, which is only allowed for strings
889 and bit vectors. (bug introduced in 0.9.16)
890 * improvement: the x86-64/darwin port now passes all tests (except
891 for the debugger tests) but should still be considered
892 experimental until this is fixed.
893 * improvement: a style-warning is signaled for CASE (etc) clauses with
894 duplicate keys. (thanks to Kevin Reid)
895 * improvement: macos/x86 and macos/x86-64 now use mach exception
896 handlers for dealing with illegal instructions (for trapping and
897 error handling) and memory protection violations (for GC).
899 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
900 * new platform: experimental support for x86-64/darwin (MacOS).
901 * incompatible change: the thread-safe (on most platforms) getaddrinfo
902 and getnameinfo sockets functions are used instead of gethostbyaddr
903 and gethostbyname, on platforms where the newer functions are available.
904 As a result, the ALIASES field of HOST-ENT will always be NIL on these
906 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
907 don't print the startup banner, but behave as if --noinform was passed
908 as a command line argument. (thanks to Kevin Reid)
909 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
910 * optimization: code using alien values with undeclared types is much faster.
911 * optimization: the compiler is now able to open code SEARCH in more cases.
912 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
913 * bug fix: using standardized COMMON-LISP special variables as loop
914 variables no longer signals bogus package lock violations. (reported
916 * bug fix: declaring local loop variables to be of a range-limited type
917 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
918 (reported by Andras Simon)
919 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
920 bugs remain on x86-64.)
921 * bug fix: the #= and ## reader macros now interact reasonably with
922 funcallable instances.
923 * bug fix: type-checks for function arguments were compiled using the
924 compiler policy settings of the wrong lexical environment (in 1.0.2
926 * bug fix: SHADOW accepts characters as string designators, as required
927 by the spec (thanks to Eric Marsden)
928 * bug fix: fixed GC safety issues when foreign functions are called with
929 non-base strings as arguments
930 * bug fix: more consistent error output for fatal-compiler-warnings, like
932 * bug fix: functions with proper names were showing up as NIL in
933 backtraces (reported by Edi Weitz, regression in 1.0.2)
935 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
936 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
937 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
938 produce more efficient code on the x86 if the compiler can determine
939 that (+ INDEX <constant>) does not require a bounds check and FOO
940 has an element type at least 8 bits wide.
941 * bug fix: references to undefined variables are handled the same way
942 in toplevel forms as in the normal compiler
943 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
945 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
946 (thanks to Jon Buller)
947 * improvement: faster compilation times for complex functions
948 * improvement: added readlink support to SB-POSIX (thanks to Richard
951 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
952 * improvement: experimental support for mach exception handling on
953 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
955 * improvement: support for GBK external format.
956 (thanks to Chun Tian (binghe))
957 * improvement: the debugger now displays variables that have been closed
958 over, in code compiled with (DEBUG 2) or higher
959 * improvement: support for executable cores on NetBSD (thanks to
961 * new feature: added a RESTART-FRAME debugger command
962 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
963 be used to provide an implementation for FILE-POSITION on Gray streams
964 (thanks to Eric Marsden)
965 * new feature: add syslog, openlog and closelog support to SB-POSIX
966 (thanks to Richard Kreuter)
967 * optimization: the function call overhead in code compiled with
968 a high DEBUG optimization setting is significantly
969 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
970 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
971 (thanks to Tony Martinez)
972 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
973 evaluated once (reported by Kevin Reid)
974 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
975 works again (thanks to Kilian Sprotte)
976 * bug fix: an error is signaled for tagbodies with duplicate tags
977 (thanks to Stephen Wilson)
978 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
979 * bug fix: Win32 port can now handle foreign code unwinding Lisp
980 stack frames from alien callbacks.
981 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
982 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
983 65k pages (thanks to David Woodhouse)
984 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
986 changes in sbcl-1.0.1 relative to sbcl-1.0:
987 * new platform: FreeBSD/x86-64, including support for threading.
988 * new feature: the compiler stores cross-referencing information
989 about function calls (who-calls), macroexpansion (who-macroexpands)
990 and special variables (who-binds, who-sets, who-references) for code
991 compiled with (< SPACE 3). This information is available through the
992 sb-introspect contrib.
993 * new feature: users may subclass SEQUENCE, and have instances of
994 these classes interoperate with standard Common Lisp functions if
995 a number of methods are defined. (This feature is experimental
996 and the interface subject to change based on feedback from SBCL
997 users and the general community)
998 * improvement: sb-sprof traces call stacks to an arbitrary depth on
999 x86/x86-64, rather than the previous fixed depth of 8
1000 * improvement: another pthread back-end of mutex "pthread-futex".
1001 * bug fix: non-ascii command-line arguments are processed correctly
1002 (thanks to Yaroslav Kavenchuk)
1003 * bug fix: non-required arguments were not passed correctly when a method
1004 defined using DEFMETHOD was called from a mop-generated method using
1005 CALL-NEXT-METHOD (reported by Pascal Costanza)
1006 * bug fix: recursion is now permitted in accessors through
1007 SLOT-UNBOUND. (reported by Pascal Costanza)
1008 * bug fix: an error was signaled at startup if the HOME environment
1009 variable was defined, but had an empty value (reported by Peter Van Eynde)
1010 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1011 signal a PROGRAM-ERROR, not a BUG.
1012 * bug fix: SB-POSIX:READDIR works when built with large file support.
1013 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1015 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1016 for better automatic indentation support. (thanks to Matt Pillsbury)
1017 * optimization: loading generic functions no longer takes O(n^2) time,
1018 proportional to the amount of methods in the generic function
1019 (reported by Todd Sabin and Jeremy Brown)
1020 * optimization: the FIND and POSITION family of sequence functions
1021 are significantly faster on arrays whose element types have been
1023 * improvements to the Windows port:
1024 ** Intermittent heap corruption problems have been fixed. (thanks
1025 to Alastair Bridgewater)
1026 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1028 ** Lisp is able to unwind foreign exception frames from alien
1029 callbacks. (thanks to Alastair Bridgewater)
1031 changes in sbcl-1.0 relative to sbcl-0.9.18:
1032 * improvement: experimental support for threading on FreeBSD/x86.
1033 (thanks to NIIMI Satoshi)
1034 * improvement: runtime option --dynamic-space-size can be used
1035 to set the size of the dynamic space reserved on startup.
1036 * improvement: floating point modes in effect are now saved in
1037 core, and restored on startup.
1038 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1039 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1040 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1041 threads are not running after *SAVE-HOOKS* have run.
1042 * improvement: writes to CLOS instance slots are type-checked in code
1043 compiled with (SAFETY 3)
1044 * improvement: floating-point exception handling on FreeBSD (thanks to
1046 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1047 (thanks to Zach Beane)
1048 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1050 * improvement: added support for the Shift-JIS external format.
1051 (contributed by NIIMI Satoshi)
1052 * improvement: callbacks are supported on Linux/PPC. (thanks to
1054 * bug fix: compiler bug triggered by a (non-standard) VALUES
1055 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1056 * bug fix: file compiler no longer confuses validated and already
1057 dumped structurres. (reported by Kaersten Poeck)
1058 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1059 fill-pointer signals a type-error as required. (thanks to
1061 * bug fix: disassemly of funcallable instances works.
1062 * bug fix: single stepping on PPC.
1063 * bug fix: fix thread-safety problems in the type system (generally
1064 manifesting as nonsensical errors like "STRING is a bad type specifier
1065 for sequences" or "The value 1 is not of type FIXNUM")
1066 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1067 (reported by Josip Gracin)
1068 * bug fix: an error is signaled for attempts to displace arrays with
1069 incompatible element types (thanks to Mario Mommer)
1070 * bug fix: more correct handling of wide characters in debug info
1071 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1072 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1073 and don't cause extra consing
1074 * optimization: MAP and MAP-INTO are significantly faster on vectors
1075 whose elements types have been declared.
1076 * Improvements to SB-SPROF:
1077 ** Support for allocation profiling
1078 ** Reduced profiling overhead, especially for long profiling runs
1079 * Improvements to the Windows port:
1080 ** floating point exceptions are now reported correctly.
1081 ** stack exhaustion detection works partially.
1082 ** more accurate GET-INTERNAL-REAL-TIME.
1083 ** better RUN-PROGRAM behaviour in case of errors in the child
1085 ** PROBE-FILE now simplifies pathnames correctly.
1086 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1087 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1088 1st 1970 (although time timezone data will be unreliable).
1089 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1091 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1092 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1093 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1095 * bug fix: two potential GC deadlocks affecting threaded builds.
1096 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1097 returns T (reported by Anton Kazennikov)
1098 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1099 with non-variable places
1100 * bug fix: remove a race condition in the setting of
1101 funcallable-instance functions, this should make threaded CLOS
1102 code more stable against memory faults.
1103 * bug fix: corruption of specials when unbinding is interrupted by an
1104 asynchronous unwind (reported by Hannu Koivisto)
1105 * improvement: the debugger will now also display local variables that
1106 are only used once, for code compiled with a DEBUG optimization quality
1109 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1110 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1111 * incompatible change: External-format support for FFI calls. The
1112 SB-ALIEN:C-STRING no longer implies an ASCII
1113 external-format. Instead the string is subject to external-format
1114 conversion. Additionally return values of type SB-ALIEN:C-STRING
1115 are by default of type (SIMPLE-ARRAY CHARACTER), not
1116 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1117 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1118 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1119 * incompatible change: SB-EXT package no longer contains the
1120 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1121 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1123 * incompatible change: the single-stepper is no longer available
1124 on Alpha, Mips and Sparc platforms.
1125 * minor incompatible change: the direct superclasses of
1126 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1127 not (STANDARD-OBJECT FUNCTION). This makes the
1128 class-precedence-lists of GENERIC-FUNCTION and
1129 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1131 * minor incompatible change: the default stream external format on
1132 non-unicode SBCL is no longer detected from the locale, but is always
1134 * new feature: Add a version of evaluator that uses an interpreter instead
1135 of the compiler. EVAL still uses the compiler by default, to switch it
1136 to use the interpreter, set the value of the variable
1137 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1138 * minor incompatible change: the single-stepper REPL has been merged
1139 with the normal debugger (see the "Stepping" heading of the debugger help
1140 for more details). The debugger command STEP will no longer switch
1141 to the single-stepper REPL.
1142 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1143 for a type now works.
1144 * bug fix: loading of default sysinit file works. (thanks to Leonid
1146 * bug fix: better detection of circularities in the file-compiler.
1147 (reported by Marco Monteiro)
1148 * bug fix: the CL pathname functions now work with files that have
1149 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1150 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1151 systems with Unix98 pty semantics.
1152 * bug fix: ASDF-INSTALL will now work with bsd tar.
1153 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1155 * bug fix: timers expiring in dead threads no longer cause a
1156 type-error (reported by Paul "Nonny Mouse").
1157 * bug fix: thanks to more lightweight single-stepper instrumentation,
1158 code compiled with (DEBUG 3) will compile and execute significantly faster,
1159 and will have more accurate type-inferencing than before
1160 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1161 whose bindings are modified
1162 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1163 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1164 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1165 CL:SOFTWARE-VERSION with proper version
1167 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1168 * feature: implemented the READER-METHOD-CLASS and
1169 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1170 as specified by AMOP.
1171 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1173 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1174 (thanks to Lutz Euler)
1175 * optimization: hashing of general arrays and vectors has been
1176 improved. (reported by Any Fingerhut)
1177 * enhancement: SB-INTROSPECT is now able to find definitions of
1178 profiled functions. (thanks to Troels Henriksen)
1179 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1180 as well, allowing compiler-macros for SETF-functions to expand.
1181 * enhancement: step-instrumentation no longer wraps known
1182 single-value functions with multiple-value context, allowing
1183 better type inference.
1184 * fixed bug #337: use of MAKE-METHOD in method combination now works
1185 even in the presence of user-defined method classes. (reported by
1186 Bruno Haible and Pascal Costanza)
1187 * fixed bug #339(c): if there are applicable methods not part of any
1188 long-form method-combination group, call INVALID-METHOD-ERROR.
1189 (reported by Bruno Haible)
1190 * fixed bug #361: the :FUNCTION initarg in the protocol for
1191 initialization of methods can now be used to override
1192 internally-produced optimized functions. (reported by Bruno
1194 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1195 system-provided lambda expression no longer cause warnings about
1196 unbound #:|pv-table| symbols.
1197 * bug fix: improved the handling of type declarations and the
1198 detection of violations for keyword arguments with non-constant
1200 * bug fix: potentially erronous calls to PATHNAME and
1201 MERGE-PATHNAMES were being flushed in some cases.
1202 (reported by Richard Kreuter)
1203 * bug fix: compiled calls to TYPEP were mishandling obsolete
1204 instances. (reported by James Bielman and Attila Lendvai)
1205 * bug fix: format strings with ~> without matching ~< no longer
1206 trigger an AVER, but signal an understandable error instead.
1207 (reported by Antonio Martinez)
1208 * bug fix: specifying an output-file without a pathname-type for
1209 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1210 of input-file instead of "fasl". (reported by Robert Dodier)
1211 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1212 (reported by James Y Knight).
1213 * bug fix: compiler-macros-function did not consider the environment
1214 argument for shadowing by local functions.
1215 * bug fix: compiler-macros expansion was inhibited by local INLINE
1217 * bug fix: inline expansions of known functions were subject to
1218 step-instrumentation in high debug policies, leading to problems
1219 with type-inference.
1220 * bug fix: compiler failed to differentiate between different CONS
1221 types in some cases.
1222 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1223 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1224 * bug fix: #368: incorrect use of expressed vs. upgraded array
1226 * bug fix: #306a: more precise unions of array types.
1227 * thread-safety improvements:
1228 ** CONDITION-WAIT could return early on Linux, if the thread was
1229 interrupted and subsequently continued with SIGCONT.
1230 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1232 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1233 * added support for the ucs-2 external format. (contributed by Ivan
1235 * minor incompatible change: pretty printing of objects of type
1236 (cons symbol) is, in the default pprint-dispatch-table, now
1237 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1239 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1240 called later in a class's lifetime, possibly as late as when the
1241 first instance of the class is created. Previously,
1242 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1243 class became finalizeable.
1244 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1245 for multibyte external-format streams. (thanks to "vbzoli")
1246 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1247 with :ALLOCATION :CLASS inherited from superclasses of the
1249 * fixed bug: anonymous classes can now be created using the :NAME
1250 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1251 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1252 * fixed bug: core-files saved with :EXECUTABLE T can again be
1253 executed when SBCL_HOME isn't set. (reported by James Knight)
1254 * fixed bug: toplevel LOCALLY forms with declarations could
1255 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1256 * fixed bug: printing from several different threads using different
1257 values of *print-case* could cause invalid output, due to
1258 some internal special variables of the printer not being bound
1259 thread-locally (reported by Max Mikhanosha)
1260 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1261 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1262 (reported by Pascal Costanza)
1263 * fixed bug: FINALIZE-INHERITANCE is called from
1264 REINITIALIZE-INSTANCE on classes when the class has previously
1265 been finalized, as required by AMOP.
1266 * minor code generation optimizations:
1267 ** better register allocation in CLOS dispatching functions
1268 ** overflow detection when coercing signed bytes to fixnums on x86-64
1269 ** is now implemented with one IMUL instruction instead of three shifts
1270 ** more efficient bit-vector access on x86 and x86-64
1271 ** more efficient access to raw structure slots on x86 and x86-64
1272 * fixed some bugs revealed by Paul Dietz' test suite:
1273 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1274 return its argument.
1276 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1277 * feature: thread support on Solaris/x86, and experimental thread support
1279 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1281 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1282 default method for SLOT-UNBOUND.
1283 * minor incompatible change: prevent the user from specializing the
1284 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1285 somewhat counter to the intent of the protocol, I (CSR) think, and
1286 additionally it just doesn't work in SBCL as currently
1287 implemented, thanks to optimizations (that are always valid for
1288 the other three SLOT-VALUEish functions, but not for the setter).
1289 * minor incompatible change: the :SB-LDB feature is now enabled by
1290 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1291 the low-level debugger.
1292 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1293 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1294 down to an SB-SYS:FD-STREAM, allowing direct communication in
1295 more cases. Temporary files and pipes are still used as a fallback
1297 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1298 on the non-reentrant C functions localtime(3) and gmtime(3).
1299 * bug fix: global optimization policy was not visible in LOCALLY and
1301 * bug fix: class objects can be used as specializers in methods.
1302 (reported by Pascal Costanza)
1303 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1304 with a type error. (reported by blitz_ on #lisp)
1305 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1306 failed with a type error. (reported by Pascal Bourguignon)
1307 * bug fix: merging pathnames against defaults with :DIRECTORY
1308 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1309 (reported by James Y Knight)
1310 * bug fix: saving large (>2GB) cores on x86-64 now works
1311 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1312 constant index and a (simple-array (signed-byte 32)) array
1313 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1314 error (patch by Robert J. Macomber)
1315 * feature: TIME now displays the amount of run-time spent in GC
1316 * bug fix: The debugger now does a better job of respecting
1317 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1318 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1319 workaround for bug 403.)
1320 * bug fix: floating point exception handling now works on FreeBSD
1321 (thanks to NIIMI Satoshi)
1322 * fixed some bugs revealed by Paul Dietz' test suite:
1323 ** MISC.641: LET-conversion were not supposed to work in late
1325 * improvements to the Win32/x86 port:
1326 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1327 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1329 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1330 * new feature: source path information is generated for macro-expansion
1331 errors for use in IDE's like Slime (thanks to Helmut Eller)
1332 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1333 * bug fix: compiler does not loop forever on an invalid type in
1335 * improvement: compilation of most CLOS applications is significantly
1337 * optimization: added a limited bytecode compiler for simple toplevel
1338 forms, speeding up compilation and FASL loading
1339 * bug fix: the statistical profiler now properly distinguishes anonymous
1342 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1343 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1344 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1345 into subclasses of the finalized class. Now user calls to
1346 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1347 system return before any subclasses are finalized.
1348 * minor incompatible change: The reader no longer ignores errors
1349 regarding non-existent packages in #+ and #- feature tests.
1350 * new feature: command line options --no-sysinit, --no-userinit to
1351 inhibit loading the corresponding init files
1352 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1353 for sending data through UDP sockets (thanks to François-René Rideau)
1354 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1355 error is signalled instead (thanks to François-René Rideau)
1356 * minor incompatible change: Error signalling behaviour of lexical
1357 operations violating package locks has changed slightly. Refer to
1358 documentation on package locks for details.
1359 * enhancement: EVAL can process IF-expressions without resorting to the
1361 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1362 constant-folded properly due to missing out-of-line functions.
1363 (reported by Mika Pihlajamäki)
1364 * bug fix: LISTEN sometimes returned T even in cases where no data was
1365 immediately available from the stream
1366 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1367 were reversed. (reported by Levente Mészáros)
1368 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1369 when given a vector displaced to an adjustable array. (reported
1371 * bug fix: derivation of float boundaries from numbers outside the
1372 appropriate float range (reported by John Wiseman)
1373 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1375 * bug fix: precomputing cache entries for generic functions with
1376 some subclasses of specializers as yet invalid does not attempt to
1377 fill a cache line with a negative offset. (reported by Levente
1379 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1380 allow condition class objects as arguments to DOCUMENTATION and
1381 (SETF DOCUMENTATION); only find and set documentation for
1382 structure names for the STRUCTURE doc-type. (suggested by Gary
1384 * improvements to the Win32/x86 port:
1385 ** (user-homedir-pathname) and default initialization file
1386 locations now know about the user's "Documents and Settings"
1387 directory (thanks to Yaroslav Kavenchuk)
1388 ** run-program is implemented (thanks to Mike Thomas)
1389 ** sockets support (thanks to Timothy Ritchey)
1390 ** better backtrace support (thanks to Alastair Bridgewater)
1391 ** sb-grovel supported
1392 ** asdf-install and sb-posix work somewhat
1393 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1394 * improvements to the Solaris/x86 port:
1395 ** works on Solaris 11/Solaris Express
1396 ** floating-point exception handling support
1397 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1398 * fixed some bugs revealed by Paul Dietz' test suite:
1399 ** REMOVE-METHOD returns its generic function argument even when
1400 the method is not one of the generic functions' methods.
1401 ** objects declared as MEMBER types can be admissible arguments to
1402 structure accessors.
1403 ** printing characters should simply be printed by the FORMAT ~:C
1405 ** compiler failure when compiling functions with hairy constant
1406 defaults for optional parameters.
1407 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1408 function, which is already optimized.
1410 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1411 * new platform: experimental support for SBCL x86/Darwin, including
1412 MacOS X 10.4.5 on Intel.
1413 * new feature: Unicode character names are now known to the system
1414 (through CHAR-NAME and NAME-CHAR).
1415 * new feature: the filesystem location of SBCL's core file is
1416 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1417 * minor incompatible change: the contrib modules SB-POSIX and
1418 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1419 this change is to make it easier to distribute
1420 location-independent binaries.
1421 * bug fix: as implied by AMOP, standardized classes no longer have
1422 slots named by external symbols of public packages. (reported by
1424 * optimization: calling functions via constant symbols -- as in
1425 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1426 via the function object as in (FUNCALL #'FOO).
1427 * optimization: on x86 and x86-64, the calling convention for the
1428 case of dealing with an unknown number of values has been altered
1429 to be friendlier to the prediction heuristics implemented,
1430 particularly with reference to CALL and RET pairing. (thanks to
1431 Alastair Bridgewater)
1432 * optimization: on x86, the code for access to thread-local symbol
1433 values and binding/unbinding of thread-local symbols is smaller.
1434 (thanks to Alastair Bridgewater)
1435 * enhancement: CONSTANTP is now able to determine constantness of
1436 more complex forms, including calls to constant-foldable standardized
1437 functions and some special forms beyond QUOTE.
1438 * fixed bug: occasional GC crashes on Solaris/x86
1439 * optimization: x86-64 supports stack allocation of results of simple
1440 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1441 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1442 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1443 (thanks to James Knight)
1444 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1445 by the standard (thanks to Stephen Compall)
1447 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1448 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1449 be used for bundling the runtime and the core file into one
1450 executable binary. This feature is not currently supported on all SBCL
1451 platforms. (thanks to James Bielman and NIIMI Satoshi)
1452 * new feature: a generational or ephemeral garbage collector is now
1453 the default on the PowerPC platform (both Linux and Darwin). The
1454 old Cheney (stop and copy) collector is a build-time option.
1455 (thanks to Cyrus Harmon, after Raymond Toy)
1456 * minor incompatible change: the method by which SBCL finds its
1457 contributed modules has changed; it no longer relies on symbolic
1458 links from an $SBCL_HOME/systems directory, but searches directly
1459 in the subdirectories of $SBCL_HOME.
1460 * enhancement: the dynamic heap size on the Linux/PPC platform is
1461 markedly larger, even using the older Cheney garbage collector.
1462 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1463 certain complicated string types. (reported by Gary King)
1464 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1465 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1466 * fixed bug: DOCUMENTATION on structures no longer signals an
1467 error if no structure type of the right name exists. (reported by
1469 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1470 applicable non-standard primary or :AROUND method on
1471 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1472 no longer cause unbound variable errors. (reported by Kevin Reid)
1473 * fixed bug: metacircle resolution in cases where methods have slots
1474 added before the slots from STANDARD-METHOD. (reported by Jean
1476 * fixed bug: the Power PC assembly code for calling into lisp
1477 sometimes computed its return address wrongly (depending on the
1478 15th bit of the address, fixed at link time). This used to
1479 manifest itself in a segmentation violation while building PCL.
1480 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1481 many others over the years)
1482 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1483 the stack pointer (treated as a signed integer) is negative.
1484 (thanks to Peter van Eynde)
1486 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1487 * new platform: experimental support for the Windows operating
1488 system has been added. (thanks to Alastair Bridgewater)
1489 * fixed several bugs in and robustified the PPC FFI (including
1490 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1491 * bug fix: allow non-simple string symbol names (reported by Paul
1493 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1495 * bug fix: saving a core corrupted callbacks on x86/x86-64
1496 * bug fix: closed a loophole in metacircularity detection and
1497 grounding in the PCL implementation of CLOS.
1498 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1499 (reported by tomppa on #lisp)
1500 * bug fix: clear the direction flag on Lisp -> C transitions, as
1501 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1502 (reported by Andrej Grozin and Hendrik Maryns)
1503 * optimization: major improvements to GC efficiency on GENCGC platforms
1504 * optimization: faster implementation of EQUAL
1505 * optimization: emit more efficient opcodes for some common
1506 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1507 * optimization: several other minor code-generation improvements on x86-64
1508 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1510 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1511 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1512 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1513 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1514 by ANSI as a generic function, and for consistency (SETF
1515 GENERIC-FUNCTION-NAME) is treated likewise.
1516 * fixed bug #233.b: make constraint propagation notice when a variable
1517 value is changed after it is referenced but before it is used
1518 * fixed bug #296: no more arbitrary behaviour from filenames with
1519 odd characters as --load arguments. (reported by Adam Warner)
1520 * fixed bug #390: :CHARACTER-SET pathname components now work as
1521 expected. (reported by Tim Daly Jr)
1522 * fixed bug #391: complicated :TYPE intersections in slot
1523 definitions no longer cause an error in PCL internals.
1524 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1525 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1526 returning the number of octets which would be written to the
1527 file-stream. (thanks to Robert J. Macomber)
1528 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1529 arguments to the internal stream functions. (thanks to David
1531 * bug fix: stack exhaustion is now handled on the x86/SunOS
1532 platform. (thanks to Daisuke Homma)
1533 * bug fix: more accurate ROOM results on GENCGC platforms
1534 * optimization: improved type inference for arithmetic-for
1535 index variables in LOOP
1536 * optimization: faster floating-point SQRT on x86-64
1537 * fixed some bugs revealed by Paul Dietz' test suite:
1538 ** DOCUMENTATION returns NIL instead of "" for method combinations
1539 that don't have a docstring
1541 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1542 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1543 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1544 therefore are not customizeable by user code (as seems to be at
1545 least permitted and maybe required by AMOP). As a consolation,
1546 however, the SBCL implementation of these functions now calls
1547 REINITIALIZE-INSTANCE as specified by AMOP.
1548 * bug fix: slot-definition documentation is propagated as per ANSI
1549 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1550 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1551 argument, as do eslotd initialization methods. (from Pascal
1552 Costanza's "Closer" project)
1553 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1554 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1556 * bug fix: it is now possible to have more than one subclass of
1557 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1558 (reported by Bruno Haible, Pascal Costanza and others)
1559 * bug fix: the dependent update protocol now works for generic
1560 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1561 and Pascal Costanza)
1562 * bug fix: condition-class instances corresponding to
1563 DEFINE-CONDITION forms are now created eagerly. (reported by
1564 Kalle Olavi Niemitalo on comp.lang.lisp)
1565 * bug fix: floating point printing is more accurate in some
1566 circumstances. (thanks to Simon Alexander)
1567 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1568 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1569 * bug fix: callbacks on OS X now preserve stack-alignment.
1570 (thanks to Cyrus Harmon)
1571 * enhancement: the x86-64 disassembler is much better at
1572 disassembling SSE instructions. (thanks to Lutz Euler)
1573 * enhancement: CHECK-TYPE on a variable now allows type-inference
1574 (thanks to Vincent Arkesteijn)
1575 * optimization: improved performance of EUC-JP external format.
1576 (thanks to NIIMI Satoshi)
1577 * optimization: performance improvements to IO on file streams of
1578 :ELEMENT-TYPE CHARACTER
1579 * optimization: much faster memory allocation on x86-64
1580 * optimization: faster garbage collections (latency and throughput)
1582 * optimization: faster bignum subtraction and division on x86 and x86-64
1583 (thanks to Lutz Euler)
1584 * optimization: more accurate type-inference for loops that use a
1585 floating point index variable or a negative step.
1587 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1588 * bug fix: add a workaround to SBCL looping infinitely at startup on
1589 Linux kernels with apparently buggy implementations of personality().
1590 (thanks to Svein Ove Aas)
1591 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1592 (thanks to David Lichteblau)
1593 * bug fix: local bindings shadow global symbol macros for
1594 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1595 * optimization: non-open coded uses of numeric comparison operators
1596 (e.g. >) no longer cons when called with more than one parameter
1597 on platforms supporting dynamic-extent allocation.
1598 * enhancement: saving cores with foreign code loaded is now supported
1599 on MIPS/Linux in addition to the previously supported platforms.
1600 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1601 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1602 * bug fix: division by zero in sb-sprof when no samples were collected
1603 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1604 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1605 (thanks to Alexander Kjeldaas)
1606 * new feature: ignoring the return values of destructive functions
1607 like NREVERSE signals a compile-time style-warning.
1608 (thanks to Kevin Reid)
1609 * new feature: the alignment of alien structure fields can be
1610 explicitly specified. (thanks to Cyrus Harmon)
1611 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1612 and dump core on SIGQUIT
1614 ** incompatible change: threads do not inherit values of specials
1615 from their parents (see manual)
1616 ** bug fix: threads stacks belonging to dead threads are freed by the
1617 next exiting thread, no need to gc to collect thread stacks anymore
1618 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1619 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1620 ** bug fix: dynamic variable and thread start related gc lossage
1621 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1623 * fixed some bugs revealed by Paul Dietz' test suite:
1624 ** SUBTYPEP is slightly more accurate on heinously complicated
1625 CONS types where some of the members have uncertain (in the
1626 NIL, NIL sense) type relationships to each other.
1628 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1629 no longer purified unless :PURIFY T is explicitly specified.
1630 ** Non-purified cores are significantly smaller than before
1632 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1633 * new feature: timers based on Zach Beane's excellent timer package
1634 * added support for the following external formats: koi8-u,
1635 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1636 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1637 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1638 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1639 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1640 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1641 * incompatible change: a threaded SBCL will no longer revert to
1642 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1643 * bug fix: interrupts are disabled until startup is complete; no
1644 more sigsegvs when receiving a signal to soon
1645 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1647 * bug fix: add a workaround for the memory randomization features in
1648 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1649 workaround will only be in effect on systems with the proc filesystem
1651 * bug fix: printing objects of type HASH-TABLE signals a
1652 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1654 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1655 have been running for over 50 days. (reported by Gilbert Baumann)
1656 * bug fix: the logic for getting names of functions gets less
1657 confused when confronded with alternate-metaclass
1658 funcallable-instances. (reported by Cyrus Harmon)
1659 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1660 consistent, even on internal alternate-metaclass objects.
1661 * bug fix: flush closure information collected by physical
1662 environment analysis prepass before the main pass. (bug reported
1664 * bug fix: compiler pack phase does not modify a hashtable, which is
1665 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1666 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1667 may be expanded. (reported by Alexander <malishev> on c.l.l)
1668 * bug fix: no more style-warnings when compiling code using LAST
1669 under high-SPEED low-SPACE settings. (reported by David Wallin
1671 * fixed some incompatibilities between SBCL's MOP and the MOP
1673 ** the METAOBJECT class is now implemented;
1674 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1675 STANDARD-OBJECT, as required;
1676 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1677 now compatible (as required by VALIDATE-SUPERCLASS); there
1678 remains a consistency requirement over the presence or absence
1679 of the FUNCTION class in the superclass list of finalized
1680 classes; see the manual for more details;
1681 ** the :AROUND method for COMPUTE-SLOTS on
1682 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1683 requested slot ordering.
1685 ** bug fix: parent thread now can be gc'ed even with a live
1687 ** bug fix: binding a special with PROGV to no value is not
1688 the same as that symbol not having been bound (thanks to
1690 ** bug fix: binding specials is thread safe (thanks to
1692 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1693 and SB-SPROF do not die with 'no handler for signal XX in
1694 interrupt_handle_now(..)' anymore
1695 ** bug fix: WITH-TIMEOUT works with multiple running threads
1696 * fixed some bugs revealed by Paul Dietz' test suite:
1697 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1698 the :method-class keyword argument.
1700 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1701 * new port: the Solaris operating system on x86 processors is now
1702 mostly supported, though some rough edges in the environment
1703 remain. (thanks to Daisuke Homma)
1704 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1705 space for its heap. (thanks to Thiemo Seufer)
1706 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1707 closures. (thanks to Thiemo Seufer)
1708 * minor incompatible change: eof selects abort in the debugger.
1709 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1710 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1712 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1713 deprecated, and will go away in a future revision of SBCL.
1714 * minor incompatible change: GC-ON and GC-OFF are no longer
1715 implemented with a counter, it does not matter how many times gc
1716 is switched on or off
1717 * bug fix: discriminating functions for generic function classes
1718 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1719 make invalid assumptions about method precedence order. (reported
1721 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1722 platforms. (thanks to Thiemo Seufer)
1723 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1724 * bug fix: correct pathname printing with printer escaping is on.
1725 (thanks to Kevin Reid)
1726 * bug fix: complex VOP definitions in "user-space" no longer trigger
1727 package locks. (reported by Zach Beane)
1728 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1729 for funcallable-instance functions, and is no different from
1731 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1732 characters in the current readtable. (reported by Nicholas Neuss)
1733 * bug fix: SUBTYPEP on various CONS types returns more a more
1734 accurate acknowledgment of its certainty.
1735 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1736 lists in some cases. This partially fixes bug 384.
1737 * flush all standard streams before prompting in the REPL and the
1739 * bug fix: signal handling and triggering gc do not conflict
1740 directly anymore, in particular a high frequency sb-sprof does
1741 not prevent gc from running
1742 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1743 approximation for timezone and DST information between the
1744 universal time epoch and the smallest negative 32-bit time_t.
1745 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1746 year 1899 so long as the encoded time is non-negative.
1747 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1748 the correct answer for non-constant rotate amounts when performing
1749 an inline 32-bit rotation.
1751 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1752 there is only one thread in the session
1753 ** bug fix: memory leak for streams created in one thread and
1754 written to in another
1755 ** bug fix: lockup when compiled with gcc4
1756 ** bug fix: race that allows the gc to be triggered when gc is
1758 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1759 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1761 ** bug fix: finished threads are gc'ed properly
1762 * fixed some bugs revealed by Paul Dietz' test suite:
1763 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1764 has arguments with a different set of applicable methods from
1765 the orignal arguments.
1766 ** The type error thrown by MAP now has a correct expected-type
1768 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1769 name a compiled function.
1770 ** Three MISC tests where a large bignum was improperly coerced to
1771 a float that couldn't represent that bignum during type
1772 derivation were fixed.
1773 ** SUBTYPEP can now handle types involving the negation of a
1774 list-form FUNCTION type.
1775 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1776 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1777 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1779 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1780 * New feature: Experimental support for bivalent streams: streams
1781 opened with :element-type :default now allow character and binary
1782 (unsigned-byte 8) I/O
1783 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1784 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1786 * bug fix: no more highly sporadic "couldn't check whether ~S is
1787 readable" when reading a stream and an interrupt hits in the middle
1788 of a select system call
1789 * compiler better recognizes complex arrays (reported by Eduardo
1791 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1792 for variables on SBCL built with linkage-tables. (reported by Luis
1794 * various error reporting improvements.
1795 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1796 (Thanks to Thiemo Seufer)
1797 * bug fix: nested reader invokations work correctly
1798 * bug fix: it is possible to have simultaneous references to foreign
1799 code and foreign data with the same name.
1801 ** added x86-64 support
1802 ** incompatible change: the threading api now works with thread
1803 objects instead of thread ids
1804 ** bug fix: threads are protected from signals and interruption when
1805 starting up or going down
1806 ** bug fix: a race where an exiting thread could lose its stack to gc
1807 ** bug fix: don't halt on infinite error in threads if possible
1808 ** fixed numerous gc deadlocks introduced in the pthread merge
1809 ** bug fix: fixed thread safety issues in read and print
1810 ** bug fix: debugger doesn't hang on session lock if interrupted at
1811 an inappropriate moment
1812 ** bug fix: run-program is now thread safe(r)
1813 ** bug fix: inner with-recursive-lock no longer releases the mutex
1814 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1815 * fixed some bugs revealed by Paul Dietz' test suite:
1816 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1817 EXPECTED-TYPE slots filled.
1818 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1820 ** MAKE-CONDITION accepts classes as type-designators.
1821 ** COMPILE may never return NIL.
1822 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1823 range before calling Unix time functions
1825 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1826 * numerous signal handling fixes to increase stability
1827 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1828 * minor incompatible change: we now correctly canonize default
1829 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1830 as per the MOP, rather than the historical (INITARG INITFUNCTION
1831 INITFORM). (reported by Bruno Haible)
1832 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1833 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1834 for more information.
1835 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1836 pathname is a directory pathname.
1837 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1838 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1840 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1841 :METACLASS argument in addition to a class name. (reported by
1842 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1843 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1844 (reported by Bruno Haible; patch from Gerd Moellmann)
1845 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1847 * bug fix: more cleanups to the floating point exception handling on
1848 x86-64 (thanks to James Knight)
1849 * bug fix: the compiler does not try to constant fold calls of
1850 COERCE and BIT-* functions when they return freshly constructed
1851 objects. (reported by Nikodemus Siivola and Paul Dietz)
1852 * optimization: improved the MIPS versions of generic (in the
1853 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1854 * optimization: direct conversion of (unsigned-byte 32) to floats on
1855 the PowerPC platform.
1856 * optimization: structure instances with raw slots now use less
1857 memory, and probably show better memory locality. (thanks to
1859 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1860 are treated as requests for stack allocation on the x86-64,
1861 Alpha-32, PPC and SPARC platforms.
1862 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1863 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1864 (thanks to Tony Martinez)
1866 ** gcing a dead thread can no longer lead to lockups
1867 ** threads block signals until they are set up properly
1868 ** errno is no longer shared by threads
1869 ** interrupt-thread restores the eflags register on x86
1870 ** fixed some lockups due to gc/thread interaction
1871 * fixed some bugs revealed by Paul Dietz' test suite:
1872 ** invalid dotted lists no longer raise a read error when
1873 *READ-SUPPRESS* is T
1874 ** signal an error if a symbol that names a declaration is used
1875 as the name of a type, or vice versa
1876 ** allow using the (declare (typespec var*)) abbreviation for
1877 (declare (type typespec var*)) with all type specifiers
1878 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1879 ** always return NIL from PPRINT-POP when OBJECT is NIL
1880 ** don't signal errors when pretty-printing malformed LABELS,
1881 FLET or MACROLET forms
1882 ** declarations in a DOLIST body are also in scope for the
1884 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1886 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1889 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1890 * fixed cross-compiler leakages that prevented building a 32-bit
1891 target with a 64-bit host compiler.
1892 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1893 opened with :IF-EXISTS :OVERWRITE.
1894 * fixed bug 281: error for an invalid qualifier in a short-form method
1895 combination method is not signalled until the faulty method is called.
1896 * bug fix: iteration variable type inferrer failed to deal with open
1897 intervals. (reported by Alan Shields)
1898 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1899 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1900 or more runtime options were provided to the sbcl binary.
1901 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1902 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1903 * optimization: there's now a fast-path for fixnum arguments in the
1904 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1905 * optimization: the code generated on x86-64 is more compact thanks
1906 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1907 * bug fix: floating-point exception handling now partly works on
1908 x86-64. (Thanks to James Knight)
1909 * improvement to the MIPS backend from Thiemo Seufer: C-style
1910 64-bit long long arguments and return values to alien functions
1912 * fixed some bugs revealed by Paul Dietz' test suite:
1913 ** the type-error signalled from WARN has a filled-in DATUM slot.
1914 ** the type-error required when a stream is not associated with
1915 a file has the stream as its datum.
1916 ** type-errors on single-floats on x86-64 no longer have
1917 :INVALID-OBJECT as the datum
1918 ** the type-errors signalled for invalid function names now have
1919 a correct expected type
1920 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1921 for structures defined with a :TYPE. Documentation strings for
1922 typed structures are no longer immediately discarded
1923 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1924 broadcast streams and synonym streams. FILE-LENGTH now also works
1925 on broadcast streams.
1927 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1928 * incompatible change: the --noprogrammer option, deprecated since
1929 version 0.7.5, has been removed. Please use the equivalent
1930 --disable-debugger option instead.
1931 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1933 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1934 inoperational for a while now) has been completely removed.
1935 * null lexical environments are now printed as #<NULL-LEXENV>,
1936 significantly reducing the amount of clutter in typical backtraces.
1937 * documentation on weak pointers, finalization, and after GC hooks
1938 has been added to the manual.
1939 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1940 as other specialized array types, is much faster. SUBSEQ and
1941 COPY-SEQ on such arrays have also been sped up.
1942 * optimization: EQL is now more efficient when at least other argument
1943 is known to be of type (OR FIXNUM (NOT NUMBER)).
1944 * fixed bug: the runtime is now less vulnerable to changes in the
1945 size of the SBCL object on OS X, and virtual memory is reserved for
1946 all spaces that need to be at a fixed address.
1947 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1948 * fixed bug: finalizers and after GC hooks that cause consing are now
1950 * fixed bug: compiler error messages and summaries are now printed to
1951 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1952 * fixed inference of the upper bound of an iteration variable.
1953 (reported by Rajat Datta).
1954 * fixed bug 211e: calling local functions with duplicated constant
1955 keyword argument no longer causes a bogus style warning about an
1957 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1958 ftype declaration to be disregarded. (reported by Dave Roberts)
1959 * fixed bug 373: caused by erronous compilation of references to alien
1960 variables in the runtime on ppc/darwin.
1961 * fixed bug 376: CONJUGATE type deriver.
1962 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1963 (reported by Baughn on #lisp)
1964 * a cross-compiler bug on non-x86 platforms has been identified and
1965 fixed. (thanks to Bruno Haible)
1966 * improvements to the MIPS runtime code for increased stability.
1967 (thanks to Thiemo Seufer)
1968 * increased the maximimum compact environment size to allow
1969 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1970 * partial workaround for bug 135: don't name the function we're
1971 calling for hairy cases of EVAL, so as not to accumulate one environment
1972 entry per call to EVAL. (reported by Kevin Reid)
1973 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1974 * optimization: type testing for non-vector arrays should be faster.
1975 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1977 * bug fix: setting 31st element of a bit vector to zero did not work
1979 * bug fix: redefining a class definition which failed due to a
1980 previous accessor / function clash now works (but see BUGS entry
1981 #380 for more problems in this area). (thanks to Zach Beane)
1982 * the long form of DEFINE-METHOD-COMBINATION disables method group
1983 checking when given a single method group with pattern *. (thanks
1985 * on x86 compiler supports stack allocation of results of simple
1986 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1987 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1988 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1989 * fixed some bugs related to Unicode integration:
1990 ** the restarts for recovering from input and output encoding
1991 errors only appear when there is in fact such an error to
1993 * fixed some bugs revealed by Paul Dietz' test suite:
1994 ** MISC.549 and similar: late transformation of unsafe type
1995 assertions into derived types caused unexpected code
1997 ** SCALE-FLOAT type deriver is less wrong.
1998 ** type derivers for EXP, LOG and similar functions compute result
1999 types for complex arguments better.
2000 ** (MISC.563) CONJUGATE type deriver works for very restricted
2002 ** out-of-line type testers for character strings are available.
2003 ** EQUAL compiler transform understands specialness of objects
2005 ** accessing double-floats stored on the stack now works on x86-64.
2006 ** debugger internals could sometimes create invalid lispobjs,
2007 resulting in GC crashes.
2008 ** MISC.548: type check weakening can convert required type into
2010 ** initialization forms for bindings are not in scope of free special
2013 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2014 * incompatible change: thread support for non-NPTL systems has
2015 been removed - locking is buggy and unreliable. A threaded
2016 SBCL build will now warn at startup and refuse to create
2017 new threads, unless futex support is detected in the kernel
2018 * incompatible change: the top level REPL now has only an ABORT
2019 restart associated with it, not TOPLEVEL and ABORT as it used to.
2020 TOP and TOPLEVEL are now available as debugger commands for
2021 returning to the top level.
2022 * incompatible change: forms evaluated in the REPL now use the
2023 global optimization policy.
2024 * incompatible change: user- and system-initialization files are
2025 no longer processed with LOAD, but by READ and EVAL; hence the
2026 global optimization policy, startup package, readtable, etc,
2028 * The .fasl file format number has been incremented because of
2029 various incompatible changes.
2030 * internal entry point details and argument counts no longer appear
2031 in backtraces unless explicitly requested by setting
2032 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2033 * built-in and standard functions no longer have names like "top
2034 level local call to FOO".
2035 * fixed bug 32: functions defined in non-null lexical environments
2036 now have more legible printed representation
2037 * fixed bug 33: functions defined in non-null lexical environemnts
2038 are now more amenable to inspection by INSPECT.
2039 * workaround for bug 354: XEPs no longer appear in backtraces unless
2040 explicitly requested.
2041 * fixed bug: receiving the signal which results from attempting to
2042 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2043 SIGBUS on other *BSDs) on architectures where the C stack is also
2044 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2045 to debugger confusion or wild instability. (reported by Cheuksan
2047 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2048 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2049 (reported by Lutz Euler)
2050 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2051 compiled code. (reported by Paul Dietz)
2052 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2053 specialized parameter to a method to something that is not TYPEP
2054 the specializer is now possible.
2055 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2056 face of package deletion.
2057 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2058 pathname. (reported by David Tolpin on c.l.l)
2059 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2060 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2061 STRING-STREAM and FILE-STREAM, respectively.
2062 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2063 than previously; in addition, its test suite is now run on build.
2064 (thanks to Luke Gorrie)
2065 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2066 * optimization: added a immediate representation for single-floats
2068 * fixed some bugs revealed by Paul Dietz' test suite:
2069 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2070 ** attempting to create a package with a colliding nickname causes
2071 correctable errors to be signalled.
2072 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2073 ** several x86-64 backend bugs related to sign-extension of immediate
2076 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2077 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2078 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2080 * fixed loading of multiply forward-referenced layouts.
2081 (thanks to Cheuksan Wang)
2082 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2083 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2084 (see COMPILE-FILE documentation for details of :PRINT option)
2085 * fixed bugs 19 and 317: fixed-format floating point printing is
2086 more accurate. This also fixes a bug reported by Adam Warner
2087 related to the ~@F format directive.
2088 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2090 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2091 dispatch table if the from-char is a dispatch macro character.
2092 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2093 by the wordsize no longer ignore the last word. (reported by Lutz
2095 * fixed bug in type checking of dynamic-extent variables. (reported
2097 * optimization: sequence traversal functions use their freedom to
2098 coerce function designators to functions.
2099 * optimization: code with many calls to CLOS methods specialized on
2100 CLOS classes has had redundant type checks removed.
2101 * optimization: added declarations to speed up operations that access
2102 the internal character database (for example STRING-UPCASE)
2103 * optimization: comparison operations between floats and sufficiently small
2104 fixnums no longer create extra rationals
2105 * fixed some bugs related to Unicode integration:
2106 ** portions of multibyte characters at the end of buffers for
2107 character-based file input are correctly transferred to the
2108 start of the buffer at the next read.
2109 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2110 passing it through to OPEN.
2111 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2112 argument given to internal calls to OPEN.
2113 ** the built-in comment readers (introduced by character sequences
2114 ";" and "#|") are more forgiving to encoding errors; they will
2115 STYLE-WARN and then attempt to resync the stream at a character
2116 boundary. (thanks to Teemu Kalvas)
2117 * fixed some bugs in the x86-64 port:
2118 ** Negative short int return values from c-calls are sign-extended
2120 ** The stack is aligned to 16-bytes for c-calls, as required by
2121 the ABI. (thanks to Cheuksan Wang)
2122 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2123 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2124 for enums. (thanks to Vincent Arkesteijn)
2125 ** Multiple small optimizations and bugfixes for floating point
2127 * fixed some bugs revealed by Paul Dietz' test suite:
2128 ** Space, Tab, Linefeed, Return and Page have the invalid
2129 secondary constituent character trait.
2130 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2132 ** WITH-INPUT-FROM-STRING should only update the index place on
2134 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2135 works more reliably.
2136 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2137 with one-dimensional arrays with fill pointers.
2138 ** TYPE-OF failed on a complex with an integer realpart and a
2140 ** compiler failure during type inference for the code of form
2141 (IF (EQL X (THE ...)) ...) (MISC.535).
2143 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2144 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2145 platform. The port supports 61-bit fixnums, large memory spaces
2146 and reloading shared object files.
2147 * enhancement: saving cores with foreign code loaded is now
2148 supported on ppc/Darwin in addition to the previously supported
2150 * enhancement: the statistical profiler now walks deeper into the
2151 call stack for more meaningful call-graphs and accrued time
2152 reports (x86/x86-64 only). It also now reports time spent in
2154 * enhancement: it is now possible to trace most individual methods
2155 of a generic function in addition to tracing the generic function
2157 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2159 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2160 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2161 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2162 produces an error. (thanks to Vincent Arkesteijn)
2163 * bug fix: NAMESTRING on pathnames with :WILD components in their
2164 directories works correctly. (thanks to Artem V. Andreev)
2165 * fixed bug 125: compiler preserves identity of closures. (reported
2167 * bug fixed: functions with &REST arguments sometimes failed with
2168 "Undefined function" when compiled with (DEBUG 3). (reported by
2170 * bug fix: overflow during compiling of setting element of a bit
2171 vector with constant index and value. (reported by Timmy Douglas)
2172 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2173 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2174 * on x86 compiler partially supports stack allocation of dynamic-extent
2176 * GO and RETURN-FROM do not check the extent of their exit points
2177 when compiled with SAFETY 0.
2178 * fixed some bugs related to Unicode integration:
2179 ** encoding and decoding errors are now much more robustly
2180 handled; it should now be possible to recover even from invalid
2181 input or output to the terminal. (thanks to Teemu Kalvas)
2182 ** provided a first cut at implementing STRING-TO-OCTETS and
2183 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2184 ** altered the SB-MD5 contributed module slightly, changing the
2185 interface just enough for it to be supportable for builds where
2186 lisp characters are not eight bits.
2187 * fixed some bugs revealed by Paul Dietz' test suite:
2188 ** the FORMATTER-generated functions for ~V[ conditionals require
2189 the correct number of arguments.
2190 ** READ-FROM-STRING returns the mandated second value when applied
2191 to displaced strings.
2192 ** the #\Rubout and #\Backspace characters are treated as invalid
2193 constituent characters by the tokenizer.
2195 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2196 * new feature: reloading changed shared object files with
2197 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2198 * new feature: references to foreign variables and functions
2199 can now be compiled and loaded before the corresponding shared
2200 object file is loaded, as long as the foreign definitions are
2201 available at runtime.
2202 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2203 supported. (thanks to Dan Debertin)
2204 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2205 just keywords. (thanks to Vincent Arkesteijn)
2206 * fixed bug #331: structure-class instances corresponding to
2207 DEFSTRUCT forms are now created eagerly.
2208 * fixed bug #345: backtraces from calls to undefined functions work
2209 on x86 as well. Related bug #61 is now also partially fixed on x86
2210 (backtraces from throws to unknown catch tags.)
2211 * bug fix: lambda-list parsing is now stricter vrt. order and number
2212 of lambda-list keywords.
2213 * bug fix: as specified by AMOP, an error is signalled if a
2214 class-option appears multiple times in a DEFCLASS form. (reported
2216 * bug fix: RANDOM can be compiled when the compiler derives the type
2217 of its numeric argument as a disjoint set of small integers.
2218 (reported by Paul Dietz)
2219 * bug fix: starting a core saved with shared objects loaded when
2220 those objects are not available no longer causes threaded SBCL to
2221 hang. (reported by Sean Ross)
2222 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2224 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2225 argument, which must always be NIL. (reported by Kalle Niemitalo)
2226 * bug fix: printing 1.0d+23 no longer results in an error.
2227 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2228 * bug fix: structure slot setters preserve evaluation order. (thanks
2230 * fixed some bugs related to Unicode integration:
2231 ** RUN-PROGRAM can allow its child to take input from a Lisp
2232 stream. (reported by Stefan Scholl)
2233 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2234 locales. (reported by Ken Causey)
2235 * fixed some bugs revealed by Paul Dietz' test suite:
2236 ** INCF, DECF and REMF evaluate their place form as specified in
2238 ** FORMATTER expands ~{ ~} iteration directives with V or #
2239 parameters correctly.
2240 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2241 consequent uses no arguments correctly.
2242 ** the system has a partial understanding of the (COMPLEX RATIO)
2244 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2247 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2248 * new feature: a build-time option (controlled by the :SB-UNICODE
2249 keyword feature, enabled by default) for building the system with
2250 support for the entire 21-bit character space defined by the
2252 * new feature: the system now has rudimentary external-format
2253 support; the primary user-visible change at this time is that
2254 characters with the high bit set (such as lower-case-e-acute) will
2255 print correctly to a terminal in a UTF-8 environment.
2256 * minor incompatible change: BASE-CHAR no longer names a class;
2257 however, CHARACTER continues to do so, as required by ANSI.
2258 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2259 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2260 has been moved to the SB-EXT package (temporarily re-exported from
2262 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2263 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2265 * fixed bug #351: better error-handlind and reporting for malformed
2267 * fixed bug #350: bignum-printing is now more memory-efficient,
2268 allowing printing of very large bignums, eg. (expt 2 10000000).
2269 (reported by Bruno Haible)
2270 * fixed bug #302: better primitive-type selection for intersection
2272 * fixed bug #308: non-graphic characters now all have names, as
2273 required. (reported by Bruno Haible)
2274 * bug fix: redefining a class with different superclasses now correctly
2275 removes it from the direct-subclasses of its previous superclasses.
2276 (reported by David Morse)
2277 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2278 new value now works. (reported by Bruno Haible)
2279 * bug fix: correct canonicalization of multiple non-standard slot
2280 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2281 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2282 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2283 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2284 now exists, an signals an error.
2285 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2286 during macroexpansion for non-iteration variables as well. (reported
2287 by Bruno Haible for CMUCL)
2288 * bug fix: Cyclic structures and unprintable objects in compiler
2289 messages no longer cause errors. (reported by Bruno Haible)
2290 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2291 and READ-FROM-STRING all now return a primary value of NIL if
2292 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2293 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2294 (reported by Bruno Haible for CMUCL)
2295 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2296 do not designate a condition. (reported by Bruno Haible for
2298 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2299 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2300 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2301 permitting resolution in favour of any of the conflicting symbols.
2302 (reported by Bruno Haible for CMUCL)
2303 * bug fix: EQUAL compiler optimizations is less aggressive on
2304 strings which can potentially compare true despite having distinct
2305 specialized array element types.
2306 * bug fix: unit enumerations can be defined without dividing by
2307 zero. (thanks to Vincent Arkesteijn)
2308 * FORMAT compile-time argument count checking has been enhanced.
2309 (report from Bruno Haible for CMUCL)
2310 * a partial workaround for the bug 262: the compiler does not try to
2311 inline-expand a local function doing RETURN-FROM from a deleted
2312 BLOCK. (thanks to Peter Denno for the bug report and to David
2313 Wragg for the simple test case)
2314 * fixed some bugs revealed by Paul Dietz' test suite:
2315 ** RENAME-PACKAGE allows all package designators as new package
2317 ** constraint propagation and lambda variable substitution are
2318 more cautious in dealing with partially deleted code.
2319 ** compiler sometimes lost reoptimization passes.
2320 ** CERROR, given a condition as condition designator, passes its
2321 remaining arguments to the continue format control without
2323 ** Case-altering FORMAT directives work correctly on non-ASCII
2325 ** The REAL type specifier handles bounds outside the floating
2326 point ranges without signalling FLOATING-POINT-OVERFLOW.
2327 ** Functions with IR1-transformations can create intercomponent
2328 references to global functions.
2329 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2331 ** FORMAT ~R treats a nil value for its first parameter correctly.
2333 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2334 * enhancement: saving cores with foreign code loaded is now
2335 supported on x86/NetBSD and sparc/Linux in addition to the previously
2336 supported platforms.
2337 * bug fix: on some platforms repeated installations caused multiple
2338 copies of HTML documentation to be installed -- should not happen
2339 any more. (reported by Stefan Scholl)
2340 * bug fix: parsing self-recursive alien record types multiple times
2341 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2342 original patch by Helmut Eller for CMUCL)
2343 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2344 (thanks to Richard Kreuter)
2345 * bug fix: defining classes whose accessors are methods on existing
2346 generic functions in other (locked) packages no longer signals
2347 bogus package lock violations. (reported by François-René Rideau)
2348 * bug fix: special variables as DEFMETHOD parameters no longer have
2349 associated bogus type declarations. (reported by David Wragg and
2351 * bug fix: read-write consistency on streams of element-type
2352 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2353 * bug fix: redefiniton of the only method of a generic function with
2354 no DEFGENERIC no longer emits a full WARNING. In addition,
2355 redefinition of generic functions with no DEFGENERIC to an
2356 incompatible lambda list now signals an error. (thanks to Zach
2358 * bug fix: DEFGENERIC now works even when there's a function of the
2359 same name in an enclosing lexical environment. (thanks to Zach
2361 * fixed compiler failure, caused by instrumenting code during
2362 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2363 * optimization: added loop analysis and improved register allocation
2364 * fixed some bugs revealed by Paul Dietz' test suite:
2365 ** POSITION on displaced vectors with non-zero displacement
2366 returns the right answer.
2367 ** (SIMPLE-STRING) is a valid type specifier for sequence
2369 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2371 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2372 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2374 ** The justification version of the FORMAT ~< directive treats
2375 non-zero minpad parameter correctly.
2377 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2378 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2379 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2380 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2381 the supported interface.
2382 * new feature: Single-stepping of code compiled with DEBUG 2 or
2383 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2384 * new feature: saving cores with foreign code loaded is now
2385 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2386 Timothy Moore's work for CMUCL)
2387 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2388 parameters to * instead of NIL if no initform is supplied.
2389 (reported by Johan Bockgaard)
2390 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2391 correctly when FUNCALL appears as the car of the form. Note:
2392 despite this FUNCALL forms are not currently subject to
2393 compiler-macro expansion. (port of Raymond Toy's fix for the
2394 same from CMUCL, reported by Johan Bockgaard)
2395 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2396 (thanks for Teemu Kalvas)
2397 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2398 case instead of terminating the iteration (thanks for Julian
2399 Squires, Sean Champ and Raymond Toy)
2400 * bug fix: incorrect expansion of defgeneric that caused a style
2401 warning. (thanks for Zach Beane)
2402 * bug fix: slot accessor effective method computation works properly
2403 for classes with multiple non-standard applicable methods on
2404 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2405 * on x86 compiler supports stack allocation of results of LIST and
2406 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2407 CMUCL implementation by Gerd Moellmann)
2408 * fixed some bugs revealed by Paul Dietz' test suite:
2409 ** FORMAT strings with both the ~<~:;~> form of the justification
2410 directive and pretty-printing directives cause an error.
2412 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2413 * incompatible change: the internal functions
2414 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2415 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2416 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2417 instead of the old functions.
2418 * new feature: on platforms where "dladdr" is available foreign
2419 function names now appear in backtraces. (based on Helmut Eller's
2421 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2422 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2424 * documentation: ASDF manual is now distributed with SBCL.
2425 * bug fix: SBCL can now load its contributed modules with REQUIRE
2426 even if the system-provided entries have been removed from
2427 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2429 * bug fix: backtraces involving undefined functions or assembly
2430 routines are more informative. (thanks to Brian Downing)
2431 * bug fix: mutually referent alien structures now work correctly.
2432 (reported by Rick Taube)
2433 * bug fix: structures defined by WITH-ALIEN can be referred to
2434 within other definitions in the same WITH-ALIEN.
2435 * bug fix: division operators (MOD, TRUNCATE and the like) with
2436 constant zero divisors and integer dividends no longer generate
2438 * bug fix: provide default methods for INPUT-STREAM-P and
2439 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2440 * bug fix: improve the detection and resolution of MOP metacycles
2441 (where effective method computation depends on the generic function
2442 having its effective method computed). (reported by Bruno Haible)
2443 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2444 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2445 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2446 * bug fix: pathnames with relative directory components can be
2447 represented relative to default pathnames.
2448 * optimization: in taking the GCD of bignums, reduce the two bignums
2449 to approximately the same size (using Euclid's algorithm) before
2450 applying the more sophisticated binary GCD. (thanks to Juho
2452 * optimization: COUNT on bitvectors now operates word-at-a-time.
2453 * optimization: ASH with a positive, but not necessarily constant,
2454 (leftwards) shift, when in a modular context, is compiled to a
2456 * fixed some bugs revealed by Paul Dietz' test suite:
2457 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2458 if the corresponding argument is NIL.
2459 ** FORMAT directives accepting a minpad parameter treat negative
2461 ** literal commas as character parameters to FORMAT directives are
2463 ** literal spaces directly after ~<Newline> directives within a
2464 format-logical-block (~:< ~@:>) do not induce :FILL-style
2465 conditional newlines.
2466 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2467 as specified: it no longer includes conditional newlines).
2468 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2470 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2471 :PER-LINE-PREFIX argument does not evaluate to a string.
2473 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2474 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2475 the manual for details; for now, package locks can be disabled by
2476 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2477 if no major problems are found then it is likely that they will be
2478 compiled in unconditionally.
2479 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2480 unsupported operators on all platforms. To load a shared library
2481 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2482 object file, link it into a shared library outside of SBCL and
2483 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2485 * bug workaround: ROOM T and the GC were not working together
2486 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2487 reporting facility aren't preserved. That reporting has been
2488 disabled, so now until and unless someone figures out how to make it
2489 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2490 * minor incompatible change: as threatened around sbcl-0.8.0, the
2491 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2492 The symbols are also exported from SB-PCL for backwards
2493 compatibility, but more so than before SB-PCL should be treated as
2494 an implementation-internal package.
2495 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2497 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2498 It is known as of this release not to work on the Alpha, however.
2499 * fixed bug #167: errors signalled due to illegal syntax in method
2500 bodies are now more legible.
2501 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2502 designators and can hence be used with TYPEP.
2503 * fixed bug #333: CHECK-TYPE now ensures that the type error
2504 signalled, if any, has the right object to be accessed by
2505 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2506 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2507 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2509 * fixed bug #334: programmatic addition of slots using specialized
2510 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2511 and :ALLOCATION :CLASS slots.
2512 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2513 float exponents. (rereported by Peter Seibel)
2514 * fixed a bug: #\Space (and other whitespace characters) are no
2515 longer considered to be macro characters in standard syntax by
2516 GET-MACRO-CHARACTER.
2517 * fixed bug: initialization of condition class metaobjects no longer
2518 causes an instance of the condition to be created. (reported by
2520 * fixed bug: it is now possible to have slots such that
2521 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2522 description is neither :INSTANCE nor :CLASS.
2523 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2524 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2525 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2526 non-standard SLOT-DEFINITION-ALLOCATION values.
2527 * fixed bugs: various race conditions handling exiting threads.
2528 CL-PPCRE's thread torture test now passes.
2529 * fixed arguably-a-bug: GC time proportional to number of threads in
2530 system even when most of them are idle
2531 * optimization: improved performance of BIT and SBIT on bit-vectors.
2532 * .fasl file incompatibility: The fasl file version number has
2533 been incremented because of changes associated with package locks.
2535 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2536 * minor incompatible change: the system no longer provides
2537 optimization hints (conditions which are TYPEP
2538 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2539 mode; these hints are still emitted when higher SPEED optimization
2541 * new contrib module: a sampling profiler (profiling by statistical
2542 sampling, rather than by instrumenting functions) is available as
2543 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2544 * the behaviour of the standard function ED is now customizeable by
2545 third parties through a hook variable: see ED's documentation
2546 string for information on the protocol.
2547 * the compiler no longer emits efficiency notes for (FUNCALL X)
2548 when the type of X is uncertain under default optimization
2550 * fixed bug 276: mutating a binding of a specialized parameter to a
2551 method to something that is not TYPEP the specializer is now
2553 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2554 exhibit the required behaviour on the x86 platform. (thanks to
2555 Peter van Eynde, Eric Marsden and Bruno Haible)
2556 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2557 even for difficult arguments. (reported by Peter Graves)
2558 * fixed bug 141a: the backquote printer now descends quoted
2560 * fixed another bug in backquote printing: no more destructive
2561 modification of the form's list structure. (reported by Brian
2563 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2564 move between its address being taken and the call to
2565 interrupt_thread, fixing a crashing race condition.
2566 * the SB-POSIX contrib implementation has been adjusted so that it
2567 no longer exhibits ridiculously poor performance when constructing
2568 instances corresponding to C structs.
2570 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2571 * minor incompatible change: the sb-grovel contrib now treats C
2572 structures as alien (in the sense of SB-ALIEN) objects rather than
2573 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2574 has implications for memory management of client code
2575 (sb-grovel-returned objects must now be manually managed) and for
2576 type safety (alien objects now have full types).
2577 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2578 used to control emission of compiler diagnostics, rather than the
2579 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2580 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2581 quality should be considered deprecated.
2582 * install.sh now installs the user manual as well
2583 * (not quite a new documentable feature, but worth considering in
2584 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2585 beginnings of a semantically meaningful condition hierarchy is
2586 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2588 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2589 designator as the defaults argument.
2590 * fixed bug: Displaced arrays whose displaced-to array has become
2591 too small now cause ARRAY-DIMENSION to signal an error, providing
2592 for safer bounds-checking. (reported by Bruno Haible)
2593 * fixed bug: DEFCLASS slot definitions with identical :READER and
2594 :WRITER names now signal a reasonable error. (reported by Thomas
2596 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2598 * fixed bug: Invalid slot specification errors now print correctly.
2599 (thanks to Zach Beane)
2600 * fixed bug 320: Shared to local slot value transfers in class
2601 redefinitions now happen corrently from superclasses as
2602 well. (reported by Bruno Haible)
2603 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2605 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2606 improper lists correctly. (reported by Bruno Haible)
2607 * fixed bug 313: source-transform for <fun-name> was erroneously
2608 applied to a call of a value of a variable with name <fun-name>.
2609 (reported by Antonio Menezes Leitao)
2610 * fixed bug 307: The obsolete instance protocol ensures that
2611 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2612 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2613 remove dead unknown-values globs from the middle of the stack.
2614 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2615 easier to resume long computations after using *BREAK-ON-SIGNALS*
2616 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2617 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2618 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2619 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2620 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2621 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2623 * fixed a (fairly theoretical) bug in string printing: if
2624 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2625 does not have array-element-type equal to the most general string
2627 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2628 function name. (thanks to Nikodemus Siivola)
2629 * fixed bug: compile-time format string checker failed on
2630 non-closed ~{. (reported by Thomas F Burdick)
2631 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2632 don't warn on legal loop code involving more than one aggregate
2633 boolean. (thanks to Nikodemus Siivola)
2634 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2635 in keyword position to MAKE-PACKAGE was not being reported as
2637 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2638 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2640 * fixed bugs 280 and 312: the checking for multiple definitions in a
2641 file is less likely to become confused by uses of inline
2643 * fixed bug: the #S reader macro performs the keyword coercion
2644 specified for slot names. (reported by Kalle Niemitalo)
2645 * fixed bug: lambda lists may contain symbols whose names start with
2646 & but are not lambda-list-keywords; their occurrence triggers a
2648 * fixed bug 321: define-method-combination argument lambda lists do
2649 not cause invalid code to be generated when &optional or &aux
2650 variables have default values. (reported by Bruno Haible)
2651 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2652 have CLOS classes; this appears also to have fixed reported
2653 instability in stack exhaustion detection.
2654 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2655 MUFFLE-WARNING when no associated restart is present is now
2657 * optimization: rearranged the expansion of various defining macros
2658 so that each expands into only one top-level form in a
2659 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2661 * optimization: used a previously little-used slot in symbols to
2662 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2664 * fixed some bugs revealed by Paul Dietz' test suite:
2665 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2667 ** Fixed an optimization bug related to inheritance of initforms
2668 from local to shared slots.
2669 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2670 ** CAST optimizer forgot to flush argument derived type.
2671 ** print/read consistency on floats is now orders of magnitude
2672 more likely. (thanks also to Bruno Haible for a similar report
2674 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2675 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2676 for objects of type REAL. Make it so.
2677 ** SXHASH is specified (infelicitously) to respect similarity,
2678 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2679 it so. (thanks to Markus Ziegler)
2680 ** on the Alpha, the compiler succeeds in compiling functions
2681 returning a known number of arguments greater than 63.
2682 ** fixed handling of invalid NIL arguments in keyword position
2684 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2686 ** provided a readably-printable representation for RANDOM-STATE
2688 ** ensured that pathnames, where they have a namestring, always
2689 print using #P"..." syntax.
2691 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2692 * Support for the forthcoming 2.0 version of the NetBSD kernel
2693 running on x86 hardware has been added. (thanks to Perry
2694 E. Metzger most immediately, and others for their past work)
2695 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2696 current release version; SBCL's OpenBSD support had been broken
2697 since about the time of OpenBSD's switch to ELF binary format.)
2698 * [placeholder for DX summary]
2699 ** user code with &REST lists declared dynamic-extent, under high
2700 speed or space and low safety and debug optimization policy.
2701 * The manual has been converted to Texinfo format and the debugger
2702 chapter from the cmucl manual has been added.
2703 * A facility has been added to extract documentation strings from
2704 sbcl and store them as Texinfo-formatted snippets for inclusion in
2705 the manual (via Texinfo's @include directive)
2706 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2707 the test case to Dave Roberts)
2708 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2709 pointers. (reported by Sean Ross)
2710 * bug fix: PROFILE output is printed nicely even for large numerical
2711 values. (thanks to Zach Beane)
2712 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2713 greater than 32 handle EOF correctly.
2714 * bug fix: on X86 an immediate argument of the IMUL instruction is
2715 correctly printed in disassembly. (thanks to Lutz Euler)
2716 * bug fix: class slots in redefined classes preserve their old
2717 values. (thanks to Bruno Haible and Nikodemus Siivola)
2718 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2719 succeeds. (reported by Marco Baringer)
2720 * fixed some bugs revealed by Paul Dietz' test suite:
2721 ** READ-SEQUENCE now works on ECHO-STREAMs.
2722 ** RATIONALIZE works more according to its specification. (thanks
2725 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2726 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2727 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2728 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2729 should matter to you only if you rebind the printer control
2730 variables and then find you want different bindings in the
2731 debugger than in the ordinary execution of your program.)
2732 * The runtime build system has been tweaked to support building
2733 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2734 assembler and linker. (thanks to Nikodemus Siivola)
2735 * Unbound, undefined, undeclared variables now trigger full
2736 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2737 more often programmer error than deliberate exploitation of undefined
2739 * optimization: the hash algorithm for strings has changed to one
2740 that is less vulnerable to spurious collisions. (thanks to Juho
2742 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2743 needless bounds checking. (thanks to Juho Snellman)
2744 * optimization: implemented multiplication as a modular
2745 (UNSIGNED-BYTE 32) operation on the PPC backend.
2746 * fixed some bugs revealed by Paul Dietz' test suite:
2747 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2749 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2750 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2751 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2753 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2755 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2757 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2758 their output stream on EOF from read.
2759 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2760 have been read to end-of-file.
2761 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2763 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2764 description of determination of which consecutive characters
2766 ** Printing the "Space" character with escaping on now yields "#\\ ",
2767 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2768 ** Reading floating-point numbers with *READ-BASE* set to a number
2769 less than 10 works correctly.
2770 ** Reading floating-point numbers with *READ-BASE* set to a number
2771 more than 10 works correctly.
2772 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2773 the readtable currently in effect.
2775 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2776 * minor incompatible change: parsing of namestrings on a physical
2777 (Unix) host has changed; numbers after the final #\. in a
2778 namestring are no longer interpreted as a version field. This is
2779 intented to be largely invisible to the user, except that the
2780 meaning of the namestring "*.*.*" has changed: it now refers to a
2781 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2782 should usually be replaced by
2783 (make-pathname :name :wild :type :wild :version :wild)
2784 with the added benefit that this is more likely to be portable.
2785 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2786 option to OPEN now signals an error if the file being opened
2787 exists; this may have an impact on existing code.
2788 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2789 signals received, on the PowerPC platforms (both Linux and
2790 Darwin). (thanks to Pierre Mai for pointing out the location of
2792 * several fixes on OS X: The system now builds and runs cleanly on
2793 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2794 (thanks to Brian Mastenbrook)
2795 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2796 second-resolution: integer multiples of 1/3600 between -24 and 24.
2797 (thanks to Vincent Arkesteijn)
2798 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2799 type when called with 1 argument; PEEK-CHAR checked type of
2800 PEEK-TYPE only after having read first character from a
2801 stream. (reported by Peter Graves)
2802 * bug fix: the garbage collector now has much better locality
2803 behaviour, and in particular no longer treats all memory as being
2804 exhausted when many small objects point to each other in a deeply
2806 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2807 recognized as being TYPEP their class.
2808 * bug fix: the PUSHNEW documentation string has been corrected.
2809 (thanks to Vincent Arkesteijn)
2810 * bug fix: defaulting of the value for the last of an atypically
2811 large number of multiple values being bound was not being
2812 performed correctly on the Alpha or PPC platforms
2813 * optimization: implemented multiplication as a modular
2814 (UNSIGNED-BYTE 32) operation on the x86 backend.
2815 * optimization: SEARCH on simple-base-strings can now be open-coded.
2816 (see also contrib/compiler-extras.lisp for inspiration for
2817 teaching the compiler about the Boyer-Moore algorithm).
2818 * value, returned by MAX (and MIN) called with several EQUALP, but
2819 not EQL, arguments now does not depend on compiler settings.
2820 * fixed some bugs revealed by Paul Dietz' test suite:
2821 ** in stack analysis liveness information is propagated from
2822 non-local entry points.
2823 ** pathwise CAST removing failed when the CAST node did not start
2825 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2826 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2828 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2829 host is already defined.
2830 ** RENAME-FILE works on streams instead of signalling an internal
2832 ** PEEK-CHAR uses the current readtable when determining whether
2833 or not a character is whitespace.
2834 ** MERGE-PATHNAMES handles the case when the pathname does not
2835 specify a name while the default-pathname specifies a version.
2836 ** Pathnames now stand a better chance of respecting print/read
2838 ** Attempting to use standardized file system operators with a
2839 pathname with invalid :DIRECTORY components signals a
2841 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2843 ** DIRECTORY on logical pathnames is more correct.
2844 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2845 signal a TYPE-ERROR if their argument is not a stream.
2846 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2847 designator argument does not designate a stream.
2848 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2849 examining the synonym.
2850 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2852 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2853 element-type larger than ([UN]SIGNED-BYTE 32).
2855 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2856 * When built with the :SB-FUTEX feature, threaded builds now take
2857 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2858 for faster/more reliable mutex and condition variable support.
2859 * Incompatible change (but one you probably shouldn't have been using
2860 anyway): the interface and code for arbitrating between multiple
2861 threads in the same user session has been redesigned.
2862 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2863 error when called without an explicit environment argument.
2864 (thanks to Nikodemus Siivola)
2865 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2866 confused about their position. (thanks to Adam Warner and Gerd
2868 * bug fix: Pretty printing backquoted forms with unquotations in the
2869 argument list position of various code constructs such as LAMBDA
2870 now works correctly. (reported by Paul Dietz)
2871 * bug fix: Pretty printing unquotations no longer loses all
2872 stream position information.
2873 * optimization: performance of string output streams is now less
2874 poor for multiple small sequence writes.
2875 * optimization: performance of CSUBTYPEP in the presence of complex
2876 expressions involving CONS and NOT many times has been improved.
2877 (reported by Paul Dietz)
2878 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2880 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2881 streams for the REPL, for the debugger, and so on. (thanks to
2883 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2884 * fixed some bugs revealed by Paul Dietz' test suite:
2885 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2886 is ignored if it is too small, rather than propagating through
2887 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2888 ** extremely complex negations of CONS types were not being
2889 sufficiently canonicalized, leading to inconsistencies in
2891 ** VALUES tranformer lost derived type.
2893 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2894 * fixed a bootstrapping bug: the build process no longer assumes
2895 that the various BOOLE-related constants have the same value in
2896 host and target lisps. (noted by Paul Dietz' test suite on an
2897 SBCL binary built from CLISP)
2898 * The system can now be dynamically linked on the MIPS platform,
2899 which enables dynamic loading of foreign code from Lisp. (thanks
2900 to Ralf Baechle for discussions on the MIPS ABI)
2901 * The system now records debugging information for its own source
2902 files in a filesystem-position-independent manner, relative to
2903 the "SYS" logical host.
2904 * fixed a compiler bug: MV-LET convertion did not check references
2905 to the "max args" entry point. (reported by Brian Downing)
2906 * tweaked disassembly notes to be less confident about proclaiming
2907 some instruction as an LRA. (thanks to Brian Downing)
2908 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2909 now each have their own history, command character, and other
2910 characteristics. (thanks to David Lichteblau)
2911 * fixed some bugs revealed by Paul Dietz' test suite:
2912 ** compiler failure in compiling LOGAND expressions including a
2914 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2915 shift greater than 32.
2916 ** FLUSH-DEST did not mark blocks for type check regeneration.
2917 ** HANDLER-CASE failed to accept declarations in handler clauses
2918 in some circumstances.
2920 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2921 * New code in contrib/sb-introspect (still probably not entirely
2922 stable yet) provides some support for smart Lisp development
2923 environments like SLIME.
2924 * The conditions signalled for errors occurring when loading .fasl
2925 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2926 in a way which should help ASDF recover gracefully.
2927 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2928 stuff has been cleaned up. If you code contrib/ stuff, this might
2929 affect you, and you can look at contrib/README, contrib/STANDARDS,
2930 and/or the 0.8.4.27 diff to check.
2931 * In full calls the compiler now does not generate checks for declared
2932 argument types for all arguments.
2933 * various threading fixes
2934 ** and some experimental patches which didn't make it into
2935 the main tree for this release, but which are shipped in
2936 contrib/experimental-thread.patch as a possible fix for some
2937 failures (deadlock, spinning...) in GC-intensive multithreaded
2939 * fixed PPC build problem (source code incompatibility of different
2940 library versions): added offsetof() hackery which attempts to divine
2941 where glibc maintainers put uc_mcontext today
2942 * fixed bug 282: compiler does not trust type assertions while passing
2943 arguments to a full call.
2944 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2945 &OPTIONAL VALUES type parameter.
2946 * fix bug 214: algorithm for noting rejected templates is now more
2947 similar to that of template seletion. (also reported by rydis on
2949 * fixed bug 141b: printing backquoted information readably and prettily
2950 inserts a space where necessary.
2951 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2952 (CHARACTER 10) are now reported as errors, rather than propagated
2953 as unknown types. (reported by piso on #lisp)
2954 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2955 with values NIL and :ERROR. (thanks to Milan Zamazal)
2956 * fixed bug 191c: CLOS now does proper keyword argument checking as
2957 described in CLHS 7.6.5 and 7.6.5.1.
2958 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2959 longer raise an error; further, using a list as a for-as-arithmetic
2960 counter now raises a meaningful error.
2961 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2962 now understood by sequence creation functions such as MAKE-SEQUENCE
2964 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2965 a STRING-INPUT-STREAM.
2966 * compiler enhancement: SIGNUM is now better able to derive the type
2968 * type declarations inside WITH-SLOTS are checked. (reported by
2970 * fixed some bugs revealed by Paul Dietz' test suite:
2971 ** incorrect optimization of TRUNCATE for a positive first
2972 argument and negative second.
2973 ** compiler failure in let-convertion during flushing dead code.
2974 ** compiler failure while deriving type of TRUNCATE on an
2975 interval, containing 0.
2976 ** ASH of a negative bignum by a negative bignum count now returns
2978 ** intersection of CONS types now canonicalizes properly, fixing
2979 inconsistencies in SUBTYPEP.
2981 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2982 * incompatible change: The --disable-debugger command line
2983 option now clobbers the debugger at a more fundamental
2984 level, by redefining #'INVOKE-DEBUGGER instead of by
2985 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2986 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2987 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2988 enter the debugger REPL and then suffer recursive errors
2989 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2990 while under the new system, BREAK in --disable-debugger mode
2991 terminates the system just as an unhandled error would.
2992 * fixed compiler performance when processing loops with a step >1;
2993 * bug fix: DOCUMENTATION now retrieves generic function
2994 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2995 support has been systematized, and now supports the methods
2996 specified by ANSI, along with a default method and a method for
2997 slot documentation. (reported by Nathan Froyd)
2998 * bug fix: effective methods associated with a generic function are
2999 no longer cached over a change of that generic function's method
3000 combination. (reported by Andreas Fuchs)
3001 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3002 names a non-existent directory. (thanks to Andreas Fuchs)
3003 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3004 the PPC platform, lead to stack corruption; this has been fixed.
3005 (reported by Rainer Joswig)
3006 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3007 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3008 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3009 arithmetic in general, is now much more reliable on the PPC
3011 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3012 platform now returns the right answer.
3013 * optimization: restored some effective method precomputation in
3014 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3015 precomputation is now tunable.
3016 * optimization: compiler-internal data structure use has been
3017 reviewed, and changes have been made that should improve the
3018 performance of the compiler by about 20%.
3019 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3020 simple-base-strings and simple-bit-vectors is improved.
3021 * optimization: the optimization of 32-bit logical and arithmetic
3022 functions introduced in version 0.8.3 on the x86 has been
3023 implemented on the mips, ppc and sparc platforms; an
3024 implementation of the same facility, but for 64-bit arithmetic,
3025 has been added for the alpha.
3026 * microoptimization: the compiler is better able to make use of the
3027 x86 LEA instruction for multiplication by constants.
3028 * bug fix: in some situations compiler did not report usage of
3029 generic arithmetic in (SPEED 3) policy.
3030 * bug 145b fix: compiler used wrong type specifier while converting
3031 MEMBER-types to numeric.
3032 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3034 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3035 index argument now works properly on the Alpha platform.
3036 * bug fix: floating point exception treatment on the Alpha platform
3038 * bug fix: FILE-POSITION works much better on string input and
3039 output streams. (thanks to Nikodemus Siivola)
3040 * bug fix: many threading/garbage collection symptoms sorted.
3041 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3042 might be pseudo-atomic.
3043 * internal change: Stopping for GC is now done with signals not
3044 ptrace. GC is now done in whichever thread wanted it, instead of
3046 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3048 * bug fix: result form in DO is not contained in the implicit
3050 * incompatible change: ICR structure is changed; the value part of
3051 CONTINUATION is now called LVAR; corresponding functions are
3052 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3053 * added type deriver for ISQRT (thanks to Robert E. Brown).
3054 * bug fix: better support for loading from the command line when an
3055 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3057 * fixed some bugs revealed by Paul Dietz' test suite:
3058 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3059 ** ROUND and FROUND now give the right answer when given very
3060 small float arguments.
3061 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3063 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3064 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3065 ** LCM should return a non-negative integer.
3066 ** PARSE-INTEGER returned the index of a terminator instead of the
3067 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3068 ** PARSE-INTEGER returned an incorrect index being applied to a
3070 ** LCM with two arguments of 0 returns 0 rather than signalling
3072 ** unsigned addition of a 32-bit constant with the high bit set no
3073 longer causes an internal compiler error on the x86.
3074 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3075 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3076 with negative last argument.
3077 ** byte specifiers with zero size and position no longer cause
3078 an error during type derivation.
3079 ** bignum multiplication on the Alpha platform now returns the
3081 * porting: The system now builds on SuSE AMD64, although it still
3082 generates a 32-bit binary.
3083 * .fasl file incompatibility: The fasl file version number has
3084 been incremented (because of the changes to internal compiler
3085 data structures referred to above).
3087 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3088 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3089 more accurately, on the Darwin kernel running on PowerPC hardware.
3090 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3091 * Compiler code deletion notes now signal a condition of type
3092 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3093 an associated MUFFLE-WARNING restart.
3094 * The compiler now performs limited argument count validation of
3095 constant format strings in FORMAT, and where appropriate in ERROR,
3096 CERROR and WARN. (thanks to Gerd Moellmann)
3097 * New ASDF-INSTALL contrib can be used for automatic download and
3098 installation of third-party Lisp code from CCLAN or other sites
3100 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3101 which forces another thread to execute a function supplied by the
3103 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3104 now accept and act upon their :ELEMENT-TYPE keyword argument.
3105 (reported by Martin Atzmueller, Edi Weitz)
3106 * bug fix: FILE-POSITION now accepts position designators up to
3107 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3108 is the greater. (thanks to Patrik Nordebo)
3109 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3110 Kalle Olavi Niemitalo)
3111 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3112 printable. (reported by Eric Marsden)
3113 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3114 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3115 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3116 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3117 (thanks to Stig Sandoe)
3118 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3119 UNREAD-CHAR now correctly includes the unread character in the
3120 target sequence. (thanks to Gerd Moellmann)
3121 * bug fix in threaded builds: the system can now be suspended and
3122 resumed by shell job control with minimal disruption.
3123 * bug fixes in times and timezones >2038AD
3124 * better handling of "where is GNU make?" problem in build scripts
3125 (thanks to Nikodemus Siivola)
3126 * new optimization: inside a named function any reference to a
3127 function with the same name is considered to be a self-reference;
3128 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3129 optimization quality.
3130 * new optimization on x86: logical functions and + now have
3131 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3132 used when the result is truncated to 32 bits.
3133 * VALUES declaration is partially enabled.
3134 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3135 * fixed some bugs revealed by Paul Dietz' test suite:
3136 ** The system now obeys the constraint imposed by
3137 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3138 types form a lattice under type intersection.
3139 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3140 ** ASSOC now ignores NIL elements in an alist.
3141 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3142 and (1+ MOST-POSITIVE-FIXNUM) answers.
3143 ** The addition of a method with invalid qualifiers to a generic
3144 function does not cause an error to be signalled immediately;
3145 a warning is signalled, and the error is generated only on
3146 calling the generic function.
3147 * changed .fasl file version number, in part to add type codes for
3148 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3149 obscure ANSI requirements
3151 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3152 * fixed bug 148: failure to inline-expand a local function left
3153 garbage, confusing the compiler.
3154 * fixed bugs 3cd: structure slot readers perform type check if the
3155 slot can have an invalid value (i.e. it is either not initialized
3156 or can be written with a less specific slot writer).
3157 * bug fix: the compiler now traps array references to elements off
3158 the end of an array; previously, the bounds checking in some
3159 circumstances could go off-by-one.
3160 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3162 * type declarations for array element types now obey the description
3163 on the CLHS page "Declaration TYPE", as per discussions on
3164 sbcl-help around 2003-05-08. This means that a declaration
3165 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3166 declaration, all references to BAR will be asserted or assumed
3167 (with THE, so dependent on compiler policy) to involve objects of
3168 type FOO. Note that no such declaration is implied in
3169 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3170 * declared types of functions from the "Conditions"
3171 chapter. (reported by Paul Dietz)
3172 * bug fix: CERROR accepts a function as its first argument.
3173 * bug fix: NTH an NTHCDR accept a bignum as index
3174 arguments. (reported by Adam Warner)
3175 * optimization: character compare routines now optimize comparing
3176 against a constant character. (reported by Gilbert Baumann)
3177 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3178 argument now works properly on the MIPS platform.
3179 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3180 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3181 declared to have a functional type, does not cause an error
3183 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3184 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3185 * fixed bug: if last continuation of a deleted block has a
3186 destination, this destination should be deleted too. (reported by
3188 * fixed a bug in the bootstrap process: the host compiler's values
3189 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3190 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3191 test case from Patrik Nordebo)
3192 * improved the ability of the disassembler on the PPC platform to
3193 provide helpful disassembly notes.
3194 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3195 the class in more cases than previously.
3196 * bug fix: FILE-POSITION now understands :START and :END for
3197 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3198 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3199 8) data. (thanks to David Lichteblau)
3200 * bug fix: it is possible to add a method to a generic function
3201 without lambda list.
3202 * bug fix: reader failed to signal END-OF-FILE inside an
3203 object representation. (reported by Nikodemus Siivola)
3204 * fixed some bugs revealed by Paul Dietz' test suite:
3205 ** LAST and [N]BUTLAST should accept a bignum.
3206 ** condition slot accessors are methods.
3207 ** (VECTOR NIL) is a subtype of STRING.
3209 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3210 * minor incompatible change: some nonsensical specialized lambda
3211 lists (used in DEFMETHOD) which were previously ignored now signal
3213 * minor incompatible change: the system is now aware of the types of
3214 variables in the COMMON-LISP package, and will signal errors for
3215 most violations of these type constraints (where previously they
3216 were silently accepted).
3217 * minor incompatible change: COMPILE-FILE now uses the freedom
3218 afforded (ANSI 3.2.2.3) to use derived function types for
3219 functions defined in the same file. This also permits the system
3220 to warn on static type mismatches and function
3221 redefinition. (Currently it does not work with high DEBUG level.)
3222 * minor incompatible change: VALUES declaration is disabled.
3223 * When issuing notes, the compiler now signals a condition of type
3224 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3225 restart for use in user handlers. It is expected that the
3226 COMPILER-NOTE condition will eventually become a condition
3227 supertype to a hierarchy of note types, which will then be
3228 handleable in a similar fashion. However, at the moment, no such
3229 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3230 but it's an implementation detail, not a classification for the
3232 * Changes in type checking closed the following bugs:
3233 ** type checking of unused values (192b, 194d, 203);
3234 ** template selection based on unsafe type assertions (192c, 236);
3235 ** type checking in branches (194bc).
3236 * A short form of VALUES type specifier has ANSI meaning (it has
3237 increased the number of situations when SBCL cannot perform type
3239 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3240 keywords or constants is permissible.
3241 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3242 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3243 classes. (thanks to Antonio Martinez)
3244 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3245 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3246 argument to a pretty-print stream. Instead, it leaves any such
3247 operations to DESCRIBE-OBJECT methods.
3248 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3250 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3251 lambda lists. (thanks to David Lichteblau)
3252 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3253 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3254 (thanks to Gerd Moellmann)
3255 * fixed bug 246: increased compilation speed of long
3256 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3258 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3260 * DEFCONSTANT now throws a condition of type
3261 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3262 constant to a non-EQL value; CONTINUE and ABORT restarts
3263 respectively change and preserve the value.
3264 * fixed bug 63: The code walker, part of the implementation of CLOS,
3265 is now better at handling symbol macros.
3266 * bug fix: There is no longer an internal implementation type named
3267 CL:LENGTH. (reported by Raymond Toy)
3268 * bug fix: In macro-like defining macros/special operators the
3269 implicit block does not enclose the lambda list.
3270 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3271 enclosed in parentheses) are not suitable as type specifiers, and
3272 their use properly signals an error now.
3273 * bug fix: An argument count mismatch for a type specifier in code
3274 being compiled no longer causes an unhandled error at compile
3275 time, but signals a compile-time warning.
3276 * fixed simple vector readable printing
3277 * bug fix: DESCRIBE takes more care over whether the class
3278 precedence list slot of a class is bound before accessing it.
3279 (reported by Markus Krummenacker)
3280 * bug fix: FORMATTER can successfully compile pretty-printer format
3281 strings which use variants of the ~* directive inside.
3282 * bug fix: SEARCH now applies its TEST predicate to the elements of
3283 the arguments in the correct order. (thanks to Wolfhard Buss)
3284 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3285 in type inference. (thanks to Robert E. Brown)
3286 * bug fix: Reading in symbols with an explicit package name of ""
3287 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3288 * fixed some bugs revealed by Paul Dietz' test suite:
3289 ** NIL is now allowed as a structure slot name.
3290 ** Arbitrary numbers, not just REALs, are allowed in certain
3291 circumstances in LOOP for-as-arithmetic clauses.
3292 ** Multiple class redefinitions before slot access no longer
3293 causes a type error.
3294 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3295 association between the name and a class.
3296 ** Generic functions with non-standard method-combination and over
3297 five methods all of which return constants no longer return NIL
3298 after the first few invocations. (thanks to Gerd Moellmann)
3299 ** CALL-NEXT-METHOD with no arguments now passes the original
3300 values of the arguments, even in the presence of assignment.
3301 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3303 ** DISASSEMBLE works with closures and funcallable instances.
3304 ** ADD-METHOD now returns the generic function, not the new method.
3305 ** FIND-METHOD signals an error if the lengths of the specializers
3306 is incompatible with the generic function, even if the ERRORP
3308 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3309 which its argument is a member.
3310 ** DEFCLASS only redefines the class named by its class-name
3311 argument if that name is the proper name of the class;
3312 otherwise, it creates a new class.
3313 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3314 of the UNBOUND-SLOT condition to the name of the slot.
3315 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3316 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3317 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3318 SLOT-MAKUNBOUND in the specified fashion.
3320 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3321 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3322 cross-compilation host. As a consequence, we can now bootstrap our
3323 way up to SBCL starting with a bare gcc toolchain and human-readable
3324 source code (first the source to CLISP, then the source to SBCL).
3325 * A contributed module containing a partial implementation of the
3326 simple-streams interface has been included. (thanks to Rudi
3328 * A contributed module implementing the RFC1321 Message Digest
3329 Algorithm, known as MD5, has been included.
3330 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3331 no longer has any effect, as the code controlled by this feature
3332 has been deleted. (As far as we know, no-one has ever built using
3333 this feature, and its semantics were confused in any case).
3334 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3335 work on conditions (as required by the ANSI specification),
3336 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3337 expected behaviour on conditions. Users should note, however,
3338 that such behaviour is not required by the ANSI specification,
3339 and so use of this behaviour may render their code unportable.
3340 * fixed some bugs revealed by Paul Dietz' test suite:
3341 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3343 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3344 ** SXHASH on condition objects no longer returns NIL.
3345 ** :ALLOCATION :CLASS slots are better treated; their values are
3346 updated on class redefinition, and initforms inherited from
3347 superclasses are applied.
3348 ** REMOVE-METHOD returns its generic function argument even when
3349 no method was removed.
3350 ** SHARED-INITIALIZE now initializes the values of the requested
3351 slots, including those with :ALLOCATION :CLASS.
3352 ** ALLOCATE-INSTANCE now works on structure classes defined via
3353 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3355 ** SLOT-EXISTS-P now works on conditions, as well as structures
3357 ** MAKE-LOAD-FORM now has the required methods on
3358 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3359 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3360 keyword, and now implements the SLOT-NAMES argument.
3361 ** methods with &OPTIONAL arguments no longer allow too many
3362 arguments to be passed in the call without error.
3363 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3364 option is consistent with the required arguments of the generic
3365 function lambda list.
3366 * bug fix: REQUIRE accepts a string designator. (Thanks to
3368 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3369 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3370 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3371 * bug fix: The compiler now checks for duplicated variables in macro
3373 * bug fix: SETQ on globals returns the correct value.
3374 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3375 not a COMPILER-ERROR (followed by some other strange error on
3376 choosing the CONTINUE restart).
3377 * bug fix: make.sh and friends are now more consistent in the way that
3378 they look for GNU "make".
3380 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3381 * experimental native threads support (on x86 Linux >=2.4 only).
3382 This is not compiled in by default: you need to add :SB-THREAD to
3383 the target features. See the "Beyond ANSI" chapter of the manual
3385 * fix for longstanding nonANSIism: The old distinction between
3386 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3387 The return value from CL:FIND-CLASS is now a CLOS class, and
3388 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3389 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3390 * An interface to the MetaObject Protocol, as described in Kiczales,
3391 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3392 MIT Press, 1991, is available from the SB-MOP package.
3393 * incompatible change: the SB-PCL package should now be considered
3394 a private implementation detail, and no longer a semi-private MOP
3396 * minor incompatible change: due to rearrangement for threads, the
3397 control stack and binding stack are now allocated at arbitrary
3398 addresses instead of being hardcoded per-port. Users affected by
3399 this probably have to be doing advanced things with shared
3400 libraries, and will know who they are.
3401 * minor incompatible change: Previously, all --eval forms used were
3402 processed with READ before any of them were processed with EVAL.
3403 Now each --eval form is processed with both READ and EVAL before
3404 the next --eval form is processed. (Thus package operations like
3405 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3406 work as the user might reasonably expect.)
3407 * minor incompatible change: *STANDARD-INPUT* is now only an
3408 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3410 * minor incompatible change: Y-OR-N-P is now character-oriented,
3411 not line oriented. Also, YES-OR-NO-P now works without errors.
3412 (thanks to Antonio Martinez)
3413 * sb-aclrepl module improvements: an integrated inspector, added
3414 repl features, and a bug fix to :trace command.
3415 * Known functions, which cannot be open coded by the backend, are
3416 considered to be able to check types of their arguments. (fixing
3417 a bug report by Nathan J. Froyd)
3418 * fixed a bug in computing method discriminating functions: It is
3419 now possible to define methods specialized on classes which have
3420 forward-referenced superclasses. (thanks to Gerd Moellmann)
3421 * fixed evaluation order in optional entries (reported by Gilbert
3423 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3424 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3425 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3426 required-by-AMOP NAME argument, as well as CLASS and
3427 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3428 * fixed bug 20: DEFMETHOD can define methods using names that are
3429 not the proper names of classes to designate class specializers.
3430 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3431 Unix file descriptors, instead of blowing up. (thanks to Antonio
3433 * Garbage collection refactoring: user-visible change is that a
3434 call to the GC function during WITHOUT-GCING will not do garbage
3435 collection until the end of the WITHOUT-GCING. If you were doing
3436 this you were probably losing anyway.
3437 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3438 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3439 TYPEP the latter but not the former.
3440 * The compiler issues a full WARNING for calls to undefined functions
3441 with names from the CL package.
3442 * MAP-INTO for a vector destination is open coded. (reported by
3443 Brian Downing on c.l.l)
3444 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3445 documentation string.
3446 * fixed some bugs revealed by Paul Dietz' test suite:
3447 ** COPY-ALIST now signals an error if its argument is a dotted
3449 ** Condition slots are now accessed more correctly in the presence
3450 of multiple initargs for a given slot.
3451 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3452 exclude restarts of the same name associated with a different
3454 ** DEFCLASS of forward-referenced classes with another
3455 forward-referenced class in the superclasses list no longer
3457 ** Condition slots are now initialized once each, not multiple
3458 times. (thanks to Gerd Moellmann)
3459 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3461 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3462 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3463 arguments contain duplicated elements.
3464 ** RESTART-CASE understands local macros.
3465 ** RESTART-CASE associates exactly its own restarts with a condition.
3466 ** ENDP in safe mode checks its argument to be of type LIST.
3467 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3468 in question is unbound.
3469 ** Optimized MAKE-INSTANCE functions no longer cause internal
3470 assertion failures in the presence of duplicate initargs.
3471 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3472 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3473 propagated even in the ctor optimized implementation of
3475 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3477 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3478 * a better implementation of SXHASH on (simple) bit vectors,
3479 measured both in execution speed and in distribution of results
3480 over the positive fixnums, has been installed. Likewise, a better
3481 implementation of EQUAL for simple bit vectors is now available.
3482 * fixed CEILING optimization for a divisor of form 2^k.
3483 * fixed bug 240 (emitting extra style warnings "using the lexical
3484 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3485 by Antonio Martinez)
3486 * fixed SXHASH, giving different results for NIL depending on type
3487 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3488 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3489 variable. (found by Rolf Wester)
3490 * SBCL does not ignore type declarations for special
3491 variables. (reported by rif on c.l.l 2003-03-05)
3492 * some bug fixes in contrib/sb-aclrepl/
3493 * fixed some bugs revealed by Paul Dietz' test suite:
3494 ** a bug in the CONS type specifier, whereby the CAR and CDR
3495 types got intertwined, has been fixed;
3496 ** the type system is now able to reason about the interaction
3497 between INTEGER and RATIO types more completely;
3498 ** APPEND, [N]REVERSE and NRECONC check that those their
3499 arguments, which must be proper lists, are really so;
3500 ** An array specialized to be unable to hold elements has been
3501 implemented, as required -- yes, really -- by ANSI;
3502 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3503 on malformed property lists;
3505 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3506 * incompatible packaging change: in line with Unix convention,
3507 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3508 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3509 * REQUIRE and PROVIDE are now optionally capable of doing something
3510 useful. See the documentation string for REQUIRE.
3511 * infrastructure for a managed SBCL contrib system: contributed
3512 modules in this release include:
3513 ** the ASDF system definition facility;
3514 ** an interface to the BSD Sockets API;
3515 ** an ACL-like convenience interface to the repl;
3516 (thanks to Kevin Rosenberg)
3517 ** an implementation of ROTATE-BYTE, with efficient implementation
3519 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3520 gives a closer approximation to the right answer than previously.
3521 (thanks to Raymond Toy)
3522 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3523 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3524 optional environment arguments, as required by ANSI.
3525 * fixed bugs in other functions taking environment objects, allowing
3526 calls with an explicit NIL environment argument to be compiled
3528 * fixed bug 228: primary return values from
3529 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3530 COMPILE or FUNCTION.
3531 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3532 :TYPE will no longer signal errors on innocuous objects.
3533 * fixed bug 231b: SETQ is better at respecting type declarations in
3534 the lexical environment.
3535 * fixed a bug in DEFCLASS: classes named by symbols with no or
3536 unprintable packages can now be defined.
3537 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3538 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3539 * fixed compiler failure related to checking types of functions.
3540 (reported by Robert E. Brown)
3541 * the compiler is now much more consistent in its error-checking
3542 treatment of bounding index arguments to sequence functions: in
3543 (SAFETY 3) code, errors will be signalled in almost all cases if
3544 invalid sequence bounding indices are passed to functions defined
3545 by ANSI to operate on sequences.
3546 * fixed a bug in the build procedure: documentation of SBCL-specific
3547 packages is now preserved and available in the final Lisp image.
3548 * lifted FDEFINITION lookup out of loops in the implementation of
3549 many list operations. (thanks to Robert E. Brown)
3550 * fixed a bug in the reader: the #n# reader macro now works for
3551 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3552 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3553 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3554 E. Brown and Tony Martinez respectively)
3555 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3556 SLOT-BOUNDP: in particular, it is now able to optimize them much
3557 better, and is now not vulnerable to having packages renamed.
3558 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3559 always called when a slot is not present in an instance. (thanks
3561 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3562 longer signal an error on unreading a character following EOF on
3563 the previous constituent stream. (thanks to Tony Martinez)
3564 * fixed some bugs revealed by Paul Dietz' test suite:
3565 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3566 not just nonnegative fixnums;
3567 ** the logical bit-array operators such as BIT-AND now accept an
3568 explicit NIL for their "opt-arg" argument (to indicate a
3569 freshly-consed result bit-array);
3570 ** ELT now signals an error on an invalid sequence index in safe
3572 ** the type system is now cleverer about negations of numeric
3573 types, and consequently understands the BIGNUM and RATIO types
3575 ** the type system is now cleverer about the interaction between
3576 INTEGER and RATIO types: while bugs still remain, many more
3577 cases are accurately computed;
3578 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3579 if it is in the last clause;
3580 ** CONSTANTLY now correctly returns a side-effect-free function in
3582 ** DECLARE is no longer treated as a special-operator; in
3583 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3584 * incremented fasl file version number due to the change in the
3585 DEFSTRUCT-SLOT-DESCRIPTION structure.
3587 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3588 * minor incompatible change: code processed by the "interpreter" or
3589 EVAL now has a compilation optimization policy of (DEBUG 2)
3590 (changed from (DEBUG 1)) to improve debuggability of interactive
3591 development, and to allow the use of the debug RETURN command in
3593 * an experimental implementation of the RETURN command for the
3594 debugger has been included. (thanks to Frederik Kuivinen)
3595 * fixed bug 62: constraints were not propagated into a loop.
3596 * fixed bug in embedded calls of SORT (reported and investigated by
3598 * fixed some bugs revealed by Paul Dietz' test suite:
3599 ** printing and reading of arrays with some dimensions having
3600 length 0 (thanks to Gerd Moellmann);
3601 ** BOA constructor with &AUX argument without a default value does
3602 not cause a type error;
3603 ** CONSTANTP now returns true for all self-evaluating objects.
3605 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3606 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3607 accessors that are related by inheritance, as specified in the
3608 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3609 Valtteri Vuorikoski)
3610 * The compiler is now able to inline functions that were defined in
3611 a complex lexical environment (e.g. inside a MACROLET).
3612 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3614 * Support for the upcoming FreeBSD-5.0 release has been included.
3615 (thanks to Dag-Erling Smorgrav)
3616 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3617 effect when it is not in a toplevel context.
3618 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3619 stand a better chance of being correct. (thanks to Gerd
3621 * fixed bug in COERCE, which now signals an error on coercing a
3622 rational to a bounded real type which excludes the expected
3624 * The compiler is now able to derive types more accurately from the
3625 COERCE and COMPILE functions.
3626 * fixed bug 223: functional binding is considered to be constant
3627 only for symbols in the CL package.
3628 * fixed bug 231: SETQ did not check the type of a variable being set
3629 (reported by Robert E. Brown)
3630 * A new optimization for MAKE-INSTANCE has been included, fixing
3631 various bugs (including relating to :ALLOCATION :CLASS slots and
3632 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3634 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3635 ** As required by ANSI, LOOP now disallows anonymous collection
3636 clauses such as COLLECT I in conjunction with aggregate boolean
3637 clauses such as THEREIS (= I 1);
3638 ** LOOP now signals an error when any variable is reused in the
3639 same loop (including the potentially useful construct analogous
3640 to WITH A = 1 WITH A = (1+ A);
3641 ** IT is only a special loop symbol within the first clause of a
3642 conditional loop clause;
3643 ** LOOP with a typed iteration variable over a hashtable now
3644 signals a type error iff it should.
3645 * fixed some other bugs revealed by Paul Dietz' test suite:
3646 ** FILE-STREAM now names the class previously known as FD-STREAM;
3647 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3648 argument) no longer signals an error;
3649 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3650 of the slot symbol, rather than using the current package
3651 ((:CONC-NAME "") continues to intern the slot's name in the
3653 * incremented fasl file version number, because of the incompatible
3654 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3655 of the new implementation of DEFINE-COMPILER-MACRO.
3657 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3658 * Support for building SBCL for MIPS platforms running in
3659 little-endian mode has now been checked in, and basic
3660 functionality on said platforms verified.
3661 * minor incompatible change: PCL now records the pathname of a file
3662 in which methods and the like are defined, rather than its
3664 * minor incompatible change: TRUENAME now considers the truename of
3665 a file naming a directory to be the pathname with :DIRECTORY
3666 component indicating that directory.
3667 * minor incompatible change: a NAMED clause in the extended form of
3668 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3669 reason for the previous behaviour is unclear.
3670 * more systematization and improvement of CLOS and MOP conformance
3671 in PCL (thanks to Gerd Moellman and Pierre Mai):
3672 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3674 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3675 primary methods with no specializers;
3676 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3678 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3679 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3680 FORWARD-REFERENCED-CLASSes; error reporting on
3681 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3683 ** SXHASH on CLOS instances now uses a slot internal to the
3684 instance to return different numbers on distinct instances,
3685 while preserving the same return value through invocations of
3687 ** DEFMETHOD signals errors when methods with longer incongruent
3688 lambda lists are added to generic functions;
3689 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3690 CLASS, as specified in AMOP;
3691 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3692 on the order returned by the primary method for classes of
3693 class STANDARD-CLASS;
3694 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3695 * fixed some bugs shown by Paul Dietz' test suite:
3696 ** DOLIST puts its body in TAGBODY;
3697 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3699 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3700 value producing form;
3701 ** if more variables are given to PROGV than values, extra
3702 variables are bound and made to have no value;
3703 ** NSUBSTITUTE on list arguments gets the right answer with
3705 ** ELT signals an error of type TYPE-ERROR when the index argument
3706 is not a valid sequence index;
3707 ** LOOP signals (at macroexpansion time) an error of type
3708 PROGRAM-ERROR when duplicate variable names are found;
3709 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3710 ** FUNCALL of special-operators now cause an error of type
3712 ** PSETQ now works as required in the presence of side-effecting
3713 symbol-macro places;
3714 ** NCONC accepts any object as its last argument;
3715 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3717 ** loop-for-as-package does not require a package to be explicitely
3719 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3721 * fixed bug 166: compiler preserves "there is a way to go"
3722 invariant when deleting code.
3723 * fixed bug 172: macro lambda lists with required arguments after
3724 &REST arguments now cause an error to be signalled. (thanks to
3726 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3727 bug. (thanks to Matthew Danish)
3728 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3730 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3731 arguments are lexically rebound. (thanks to Gerd Moellmann and
3733 * fixed bug 194: error messages are now more informative when there
3734 is no primary method applicable in a call to a generic
3735 function. (thanks to Gerd Moellmann)
3736 * fixed bug in command line argument checking (thanks to Julian
3738 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3739 (thanks to Matthew Danish)
3740 * incremented fasl file version number, because of the
3741 SXHASH-related changes in the layout of CLOS data structures
3743 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3744 * minor incompatible change: The runtime (the Unix executable named
3745 "sbcl") is now much pickier about the .core files it will load.
3746 Essentially it now requires .core files to descend from the same
3747 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3748 as the runtime does. (The intent is to prevent the crashes which
3749 can occur, and which can even be reported as mysterious failures,
3750 when people patch the sources or change the build parameters
3751 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3752 sbcl and .core files.)
3753 * fixed bug: VALUES-LIST is no longer optimized away.
3754 * fixed bug 142: The FFI conversion of C string values to Lisp
3755 string values no longer conses excessively. (thanks to Nathan
3756 Froyd porting Raymond Toy's fix to CMU CL)
3757 * began to systematize and improve MOP conformance in PCL (thanks to
3758 Nathan Froyd, Gerd Moellman and Pierre Mai):
3759 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3761 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3762 ** FINALIZE-INHERITANCE is now called on class finalization;
3763 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3764 argument precedence order.
3765 * fixed bug 202: The compiler no longer fails on functions whose
3766 derived types contradict their declared type.
3767 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3768 so it can be non-toplevel.
3769 * The fasl file version number has changed (because of the new
3770 implementation of DEFMACRO).
3771 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3772 safe code, that any length requirement by their type-specifier
3773 argument is valid. The exceptions to this are described in bug
3775 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3776 functions now check the types of their inputs as required by ANSI.
3777 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3778 introduced symbol is DECLAREd to be SPECIAL.
3779 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3780 * fixed bug: :COUNT argument to sequence functions may be negative
3781 * fixed bug: body of DO-SYMBOLS may contain declarations
3782 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3783 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3784 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3785 symbol macro only once
3786 * fixed printing of call frame when argument list is unavailable
3787 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3788 * compiler no longer signals WARNING on unknown keyword
3791 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3792 * A beta-quality port to the mips architecture running Linux,
3793 based on the old CMUCL backend, has been made. It has been tested
3794 on a big-endian kernel, and works sufficiently well to be able to
3795 rebuild itself; it has not been tested in little-endian mode.
3796 * fixed an inconsistency between gencgc.c and purify.c which made
3797 dumping/loading .core files unreliable
3798 * fixed bug 120a: The compiler now deals correctly with IFs where
3799 the consequent is the same as the alternative, instead of
3800 misderiving the return type. (thanks to Alexey Dejneka)
3801 * fixed bug 113: Logical pathnames are now dumpable (the logical
3802 host is resolved at load-time, throwing an error if it is not
3804 * fixed bug 174: FORMAT's error message is slightly clearer when a
3805 non-printing character is used in a format directive.
3806 * fixed several bugs in compiler checking of type declarations, i.e.
3807 violations of the Python "declarations are assertions" principle
3808 (thanks to Alexey Dejneka)
3809 * fixed several bugs in PCL's error checking (thanks to Gerd
3811 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3813 * fixed bug in compilation of functions as first class values
3814 (thanks to Antonio Martinez-Shotton)
3815 * The compiler's handling TYPE-ERRORs which it can prove will
3816 inevitably happen at runtime has been cleaned up and corrected
3817 in several ways. (thanks to Alexey Dejneka)
3818 * improved argument type checking for various basic arithmetic
3819 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3820 lavished on them in the past that they can be compiled in many
3821 ways in different special cases
3822 * fixed bug 181: compiler checks validity of user supplied type
3824 * cleaned up code flushing in optimization: Function calls which
3825 should signal errors for safety purposes (e.g. which ANSI says
3826 should signal errors when their arguments are of incorrect type)
3827 are no longer optimized away.
3828 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3829 * incremented fasl file version number, because changes in the
3830 implementation of sequence functions like COERCE caused
3831 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3832 in old inline expansions) to become undefined. (Actually these
3833 changes were later undone, so we might very well be binary
3834 compatible with 0.7.7 after all, but leaving the version number
3835 incremented seemed like the simplest and most conservative
3838 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3839 * An alpha-quality port to the parisc architecture running Linux,
3840 based on the old CMUCL backend, has been made. This, even more so
3841 than the other backends, should be considered still a work in
3842 progress; known problems include that the Linux kernel in 64-bit
3843 mode does not propagate the correct sigcontext structure to
3844 userspace, and consequently SBCL on a parisc64 kernel will not
3846 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3847 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3848 suppressed.) Also now that the compiler is looking at declarations
3849 in the environment, it checks optimization declarations as well,
3850 and suppresses inlining when (> DEBUG SPEED).
3851 * More fixes have been made to treatment of floating point exception
3852 treatment and other Unix signals. In particular, floating point
3853 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3854 * The detection and handling of control stack exhaustion (infinite
3855 or very deeply nested recursion) has changed. Stack exhaustion
3856 detection is now done by write-protecting pages at the OS level
3857 and applies at all optimization settings; when found, a
3858 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3859 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3860 be caught using IGNORE-ERRORS.
3861 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3862 correctly when transforming recursive LABELS functions to LETs.
3863 (thanks to Alexey Dejneka)
3864 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3865 that are names of constants or global variables.
3866 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3867 alien routines with docstrings.
3868 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3869 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3871 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3872 (thanks to Eric Marsden)
3873 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3874 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3875 * Bug fix: The SPARC backend can now compile functions involving
3876 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3877 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3878 to a functions expecting a PATHNAME-DESIGNATOR.
3879 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3880 lambda lists. (thanks to Alexey Dejneka)
3881 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3882 functions so that EXPORTing the name of the class doesn't cause
3883 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3884 to Antonio Martinez for reporting this)
3885 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3886 constant-folded without failing an assertion. (thanks to Einar
3887 Floystad Dorum for reporting this)
3888 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3889 types are now dealt with more correctly by the compiler.
3890 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3891 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3892 behaviour analogous to RENAME-FILE. This puts its behaviour more
3893 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3895 * The fasl file version number has changed again. (because of the
3896 bug fix involving the names of PCL MAKE-INSTANCE functions)
3898 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3899 * bug fix: Floating point exceptions are treated much more
3900 consistently on the x86/Linux and PPC/Linux platforms.
3901 * Array initialization with :INITIAL-ELEMENT is now much faster for
3902 cases when the compiler cannot open code the array creation, but
3903 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3904 array accesses have also seen a speed increase.
3905 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3906 specified. (This was at the root of some bad interactions between
3907 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3909 * bug fix: Internal error arguments for undefined functions are now
3910 computed correctly on the PPC/Linux platform.
3911 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3912 Raymond Toy's patch for CMU CL)
3913 * Support for the Solaris 9 operating environment has been included
3914 (thanks to Daniel Merritt)
3915 * A very ugly but hopefully complete draft of the missing FFI chapter
3916 of the manual has been created by reformatting the corresponding
3917 CMU CL manual chapter into (currently very ugly and incoherent)
3918 DocBook and bringing it up to date for SBCL behavior. Thus, the
3919 manual is now essentially complete, at least by my extreme
3920 once-and-only-once standards, whereby it's acceptable to refer to
3921 the doc strings of SB-EXT functions as the primary documentation.
3922 * The fasl file version number has changed again, due to cleanup of
3923 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3924 is no longer a static symbol.)
3926 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3927 * SBCL now builds with OpenMCL (version 0.12) as the
3928 cross-compilation host; also, more progress has been made toward
3929 bootstrapping under CLISP.
3930 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3932 * bug 158 fixed: The compiler can now deal with integer loop
3933 increments different from 1; fixing this turned out also to fix
3935 * bug 169 fixed: no more bogus warnings about using lexical bindings
3936 despite the presence of perfectly good SPECIAL declarations (thanks
3937 to David Lichteblau)
3938 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3939 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3940 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3942 * bug 180 fixed: Method combination specifications no longer ignore
3943 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3944 * bug fix: Structure type predicate functions now check their argument
3945 count as they should.
3946 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3947 correctly. (thanks to Pierre Mai)
3948 * minor incompatible change: The --noprogrammer option is deprecated
3949 in favor of the new --disable-debugger option, which is very similar.
3950 (The major difference is that it takes effect at a slightly different
3951 time at startup, causing handling of errors in --sysinit and
3952 --userinit files will be affected differently.) The
3953 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3954 been added to allow this functionality to be controlled from ordinary
3955 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3956 maintainers, who might want to run non-interactive scripts to
3957 build SBCL cores which will later be used interactively.)
3958 * minor incompatible change: The LOAD function no longer, when given
3959 a wild pathname to load, loads all files matching that pathname.
3960 Instead, an error of type FILE-ERROR is signalled.
3962 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3963 * bug 147 fixed: The compiler preserves its block link/count
3964 invariants more correctly now so that it doesn't crash. (thanks
3966 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3968 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3969 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3970 no-value-for-this-character as NIL (as specified by ANSI).
3971 * HOST-NAMESTRING on physical pathnames now returns a string that is
3972 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3973 (thanks to Christophe Rhodes)
3974 * The Alpha port handles icache flushing more correctly. (thanks to
3976 * More progress has been made toward bootstrapping under CLISP. (thanks
3977 to Christophe Rhodes)
3978 * The fasl file format has changed again, because dynamic loading
3979 on OpenBSD (which has non-ELF object files) motivated some cleanups
3980 in the way that foreign symbols are transformed and passed around.
3981 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3982 is no longer treated as whitespace by the reader, but instead as
3983 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3984 |A
\7fB|, instead of A as it used to.
3986 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3987 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3988 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3989 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3990 this as of 0.7.1.45, but was left out of the previous news section
3991 (thanks to Dan Barlow)
3992 * SBCL now runs on the Solaris operating system on SPARC architectures
3993 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3994 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3995 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3996 * SPARC backend cleanups, allowing builds of cores optimized for V8
3997 and V9 SPARCS, and also emission of code targeted to a particular
3998 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4000 * SBCL is closer to bootstrapping under CLISP, thanks to various
4001 fixes by Christophe Rhodes.
4002 * The fasl file format has changed again, to allow the compiler's
4003 INFO database to support symbol macros.
4004 * The user manual (in doc/) is formatted into HTML more nicely.
4005 (thanks to coreythomas)
4006 * The system is smarter about SUBTYPEP relationships, especially
4007 those involving NOT types (including types such as ATOM which are
4008 represented internally using NOT types). Thus SUBTYPEP is less
4009 likely to return (VALUES NIL NIL) in general, and in particular
4010 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4012 * The fasl file format has changed again, because the internal
4013 representation of types now includes a new slot to support the new
4014 SUBTYPEP-of-NOT-types logic.
4015 * (not a change in the main branch of SBCL, but a related prototype
4016 which can hopefully be merged into the main branch of SBCL in the
4017 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4018 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4019 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4020 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4021 dispatch macro character combinations. (thanks to Alexey Dejneka)
4022 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4023 that superficially look like logical namestrings correctly.
4024 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4026 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4027 default supplied-p) for &optional and &key arguments. (thanks to
4030 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4031 * incompatible change: The compiler is now less aggressive about
4032 tail call optimization, doing it only when (> SPACE DEBUG) or
4033 (> SPEED DEBUG). (This is an incompatible change because there are
4034 programs which relied on the old CMU-CL-style behavior to optimize
4035 away their unbounded recursion which will now die of stack overflow.)
4036 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4037 for non-GENCGC systems has been increased to 20M (since that
4038 seems much closer to the likely performance optimum for modern
4039 systems than the old 4M value was)
4040 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4041 and *DEBUG-PRINT-LEVEL*
4042 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4043 of CMU CL's support for SPARC, and various endianness and other
4044 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4045 * new syntactic sugar for the Unix command line: --load foo.bar is now
4046 an alternate notation for --eval '(load "foo.bar")'.
4048 ** The system now detects stack overflow and handles it gracefully,
4049 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4050 optimization settings. (This is a good thing in general, and
4051 its introduction in this version should be particularly timely
4052 for anyone whose code fails because of suppression of tail
4054 ** The system now hunts for the C variable "environ" in a more
4055 devious way, to avoid segfaults when the C library version differs
4056 between compile time and run time. (thanks to Christophe Rhodes)
4057 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4058 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4059 ** The compiler no longer issues bogus style warnings for undefined
4060 classes in the same source file as the DEFCLASSes which defined
4061 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4063 ** fixes in CONDITION class precedence list for undefined function
4064 errors (thanks to Alexei Dejneka)
4065 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4066 correctly. (thanks to Dan Barlow)
4067 ** portability fixes aiming at bootstrapping under CLISP (thanks
4068 to Dave McDonald and Christophe Rhodes)
4069 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4070 ** fixes in type translation and and type inference (thanks to
4072 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4073 ** various fixes in the new ports (thanks to Dan Barlow)
4074 * several changes related to debugging:
4075 ** suppression of tail recursion, as noted above
4076 ** stack overflow detection, as noted above
4077 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4078 is now the default. (For some time encapsulation has been more
4079 reliable than the breakpoint-based :ENCAPSULATE NIL
4080 implementation, at least on X86 systems; and I just noticed that
4081 encapsulation also seems closer to the spirit of the ANSI
4084 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4086 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4087 up properly again. (There was a packaging bug in 0.7.0 which
4088 left their definitions in SB-SYS::LOAD-FOREIGN and
4089 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4090 vital for most things which interface to C-level interfaces,
4091 like extensions working with sockets or databases or
4092 Perl-compatible regexes or whatever, and the need to fix
4093 this bug is the main reason that 0.7.1 was released so
4095 ** DEFGENERIC is now choosier about the methods it redefines, so that
4096 reLOADing a previously-LOADed file containing DEFGENERICs does
4097 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4098 cycle now works as it should. (thanks to Alexey Dejneka)
4099 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4100 fixed by Christophe Rhodes. (He actually submitted this patch
4101 months ago, and I delayed until after 0.7.0.)
4102 ** Bug 111 (internal compiler confusion about runtime checks on
4103 FUNCTION types) was fixed by Alexey Dejneka.
4104 * Some internal cleanups (getting rid of variables which aren't
4105 needed now that the byte interpreter is gone) caused the fasl
4106 file format number to change again.
4108 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4109 * major incompatible change: The default fasl file extension, i.e. the
4110 default extension for files produced by COMPILE-FILE, has changed
4111 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4113 ** There are many changes in the implementation of the compiler.
4114 SBCL is now essentially a compiler-only implementation of ANSI
4115 Common Lisp. EVAL still "interprets" a few special cases, but
4116 almost all the interesting cases are handled by creating
4117 a LAMBDA expression, calling COMPILE on it, then calling
4118 FUNCALL on the result.
4119 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4120 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4121 Since the code is newer, there might still be some new bugs
4122 (though not as many as before Martin Atzmueller's fixes:-). But
4123 the new code is substantially simpler and clearer, and hopefully
4124 any remaining bugs will be simpler, less fundamental, and more
4125 fixable then the bugs in the old code.
4126 ** The revised compiler is still a little unsteady on its feet.
4128 *** The debugging information it produces (particularly the names
4129 of FUNCTION objects) is sometimes much less useful than what
4130 the old compiler produced.
4131 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4132 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4133 within a MACROLET) has been temporarily weakened.
4134 ** There are new compiler optimizations for various functions:
4135 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4136 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4137 *** the math functions TRUNCATE, FLOOR, and CEILING
4138 *** the function-of-all-trades COERCE
4139 Mostly these should be transparent, but there's one
4140 potentially-annoying problem (bug 117): when the compiler
4141 inline-expands a function and does type analysis on the result,
4142 it can create control paths which have type mismatches, and
4143 when it can't prove that those control paths aren't taken,
4144 it will issue WARNINGs about the type mismatches. This is
4145 a particular problem in practice for the new sequence functions.
4146 It's not clear how this should be fixed, and for now, a
4147 workaround is given in the entry for 117 in the BUGS file.
4148 ** (Because of the interaction between the two previous items --
4149 occasional inlining problems and new inline expansions -- some
4150 of the new sequence function optimizations won't really kick in
4151 completely until debugging information, and then inlining, are
4152 straightened out in some future version.)
4153 * minor incompatible changes:
4154 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4155 DIRECTORY no longer implicitly promotes NIL slots of its
4156 pathname argument to :WILD. In particular, when you ask for the
4157 contents of a directory (which you used to be able to do without
4158 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4159 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4160 ** changes in behavior that ANSI explicitly defines to be
4161 implementation dependent:
4162 *** The new compiler-only implementation still conforms with ANSI,
4163 but acts a little different than before. Besides the obvious
4164 changes in performance tradeoffs (that the cost per form passed
4165 to EVAL has gone up, and the cost per form executed by EVAL
4166 has gone down), the behavior of the system changes a little
4167 because there are no longer any interpreted function objects.
4168 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4169 e.g. doing COMPILE on the output of interactive DEFUN is
4171 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4173 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4174 has changed from (:CL) to NIL.
4175 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4176 e.g. #\Newline and #\Tab, don't have names specified in the
4177 ANSI Common Lisp standard, is now based on their ASCII symbolic
4178 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4179 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4180 are no longer used for output.
4181 ** changes in internal implementation constants:
4182 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4183 4 million. (If your application spends a lot of time GCing and
4184 you have a lot of RAM, you might want to experiment with
4185 increasing it even more.)
4186 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4187 However, almost all old code should still continue to work without
4188 immediate update, as SB-C-CALL is now a (deprecated) nickname
4190 ** Old operator names in the style DEF-FOO are now deprecated in
4191 favor of new corresponding names DEFINE-FOO, for consistency with
4192 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4193 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4194 internal symbols, but a few supported extensions like
4195 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4196 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4197 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4198 etc. as you get deeper into recursive calls to the debugger
4199 command loop, instead of the old "5]", "5]]", "5]]]"
4200 sequence. (I was motivated to do this when squabbles between
4201 ILISP and SBCL left me very deeply nested in the debugger. In the
4202 short term, this change will probably provoke more ILISP/SBCL
4203 squabbles, but hopefully it will be an improvement in the long run.)
4204 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4205 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4206 function, since declaiming return types involving aliens is
4207 (1) annoyingly messy to do by hand and (2) vital to efficient
4208 compilation of code which calls such functions.
4209 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4210 longer reexported by the SB-EXT package. They're solely useful
4211 for alien code, so it seems more logical that you should get
4212 them from the SB-ALIEN package, not in SB-EXT.
4213 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4214 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4215 features. Instead, the code that they used to control is always
4216 built into the system.
4217 * many other bug fixes
4218 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4219 advantage of the new EVAL-WHEN stuff and to clean them up in
4220 general, and they are now more ANSI-compliant in a number of
4221 ways. Martin Atzmueller is responsible for a lot of this.
4222 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4224 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4225 *** correct ERROR type for various file operations
4226 *** some fixes for Lisp streams
4227 *** DEFMETHOD syntax checking
4228 *** changing old weird representation of debug information as
4229 strings (which, among their other deficiencies, don't transform
4230 correctly when you rename packages, and don't change their
4231 print representation when you change things like *PACKAGE*
4232 and *PRINT-LENGTH*) to symbols and lists of symbols
4233 He also made several improvements and fixed several bugs in DESCRIBE.
4234 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4236 *** misbehavior of WRITE-STRING/WRITE-LINE
4237 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4238 and several other LOOP problems as well
4239 *** DIRECTORY when similar filenames are present
4240 *** DEFGENERIC with :METHOD options
4241 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4242 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4243 *** argument ordering in FIND with :TEST option
4244 *** mishandled package designator argument in APROPOS-LIST
4245 *** various problems in the backquote readmacro
4246 *** a bug in APROPOS
4247 *** probably some others that I'm not describing very well here,
4248 since the CVS log documents them by reference to sbcl-devel
4249 messages, and the SourceForge archives aren't working well.:-(
4250 ** Dan Barlow improved the Alpha port (and is making progress on the
4251 PPC port, for those of you who think different).
4252 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4253 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4254 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4255 continues to work on the SPARC port (for those of you in a position
4256 to look down upon our little PC-compatible boxes from a great height).
4257 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4258 and uses that copy, rather than the current dynamic value, when
4259 it's trying to decide whether to truncate output. Thus e.g.
4260 (let ((*print-lines* 50))
4261 (pprint-logical-block (stream nil)
4263 (let ((*print-lines* 8))
4264 (print (aref possiblybigthings i) stream)))))
4265 should now truncate the logical block only at 50 lines, instead of
4266 often truncating it at 8 lines, as it did before.
4267 * The doc/cmucl/ directory, containing old CMU CL documentation
4268 from the time of the fork, is no longer part of the base system.
4269 SourceForge has shut down its anonymous FTP service, and with it
4270 my original plan for distributing the old CMU CL documentation
4271 there. For now, if you need these files you can download an old
4272 SBCL source release and extract them from it.
4273 * The fasl file version number changed again, for dozens of reasons,
4274 some of which are apparent above.
4276 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4277 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4278 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4279 patch, so that DISASSEMBLE output is much nicer.
4280 * The code in the SB-PROFILE package now seems reasonably stable.
4281 I still haven't decided what the final interface should look like
4282 (I'd like PROFILE to interact cleanly with TRACE, since both
4283 facilities use function encapsulation) but if you have a need
4284 for profiling now, you can probably use it successfully with
4285 the current CMU-CL-style interface.
4286 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4287 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4288 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4289 link now returns the dangling link itself, and for similar
4290 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4291 link itself. (In these cases the old code signalled an error and
4292 looped endlessly, respectively.) Thus, DIRECTORY now works even
4293 in the presence of dangling and cyclic symbolic links.
4294 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4295 is now a supported extension again, since the consensus on
4296 sbcl-devel was that it can be useful for ordinary development
4297 work, not just for debugging SBCL itself.
4298 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4299 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4300 that currently-defined functions might be redefined later with
4301 different return types.
4302 * Hash tables can be printed readably, as inspired by CMU CL code
4303 of Eric Marsden and SBCL code of Martin Atzmueller.
4304 * better error handling in CLOS method combination, thanks to
4305 Martin Atzmueller porting Pierre Mai's CMU CL patches
4306 * more overflow fixes for >16Mbyte I/O buffers
4307 * A bug in READ has been fixed, so that now a single Ctrl-D
4308 character suffices to cause end-of-file on character streams.
4309 In particular, now you only need one Ctrl-D at the command
4310 line (not two) to exit SBCL.
4311 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4313 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4314 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4315 does the right thing.
4316 * The implementation of some type tests, especially for CONDITION
4317 types, is now tidier and maybe faster, due to CMU CL code
4318 originally by Douglas Crosher, ported by Martin Atzmueller.
4319 * Some math functions have been fixed, and there are new
4320 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4321 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4322 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4323 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4324 likely to be built into sbcl-0.7.0.)
4325 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4326 CL, ported to SBCL by Martin Atzmueller)
4327 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4328 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4329 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4330 * There's a new slam.sh hack to shorten the edit/compile/debug
4331 cycle for low-level changes to SBCL itself, and a new
4332 :SB-AFTER-XC-CORE target feature to control the generation of
4333 the after-xc.core file needed by slam.sh.
4334 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4335 COMPILE-FILE is no longer supported, so that now every function
4336 gets an entry point, so that block compilation looks a little
4337 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4338 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4339 returns the number of bytes consed since the system started,
4340 rather than the number consed since the first time the function
4341 was called. (The new definition parallels ANSI functions like
4342 CL:GET-INTERNAL-RUN-TIME.)
4343 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4344 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4345 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4346 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4347 * Fasl file version numbers are now independent of the target CPU,
4348 since historically most system changes which required version
4349 number changes have affected all CPUs equally. Similarly,
4350 the byte fasl file version is now equal to the ordinary
4353 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4354 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4355 is no longer recognized. I apologize for this, because it was
4356 listed in SB-EXT as a supported extension, but I found that
4357 its existing behavior was poorly specified, as well as incorrectly
4358 specified, and it looked like too much of a mess to straighten it
4359 out. I have enough on my hands trying to get ANSI stuff to work..
4360 * many patches ported from CMU CL by Martin Atzmueller, with
4361 half a dozen bug fixes in pretty-printing and the debugger, and
4362 half a dozen others elsewhere
4363 * fixed bug 13: Floating point infinities are now supported again.
4364 They might still be a little bit flaky, but thanks to bug reports
4365 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4366 as flaky as they were.
4367 * The --noprogrammer command line option is now supported. (Its
4368 behavior is slightly different in detail from what the old man
4369 page claimed it would do, but it's still appropriate under the
4370 same circumstances that the man page talks about.)
4371 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4372 are now supported, and enabled by default. Thus, the compiler can
4373 handle many floating point and complex operations much less
4374 inefficiently. (Thus e.g. you can implement a complex FFT
4376 * The compiler now detects type mismatches between DECLAIM FTYPE
4377 and DEFUN better, and implements CHECK-TYPE more correctly, and
4378 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4379 thanks to patches from Martin Atzmueller.
4380 * various fixes to make the cross-compiler more portable to
4381 ANSI-conforming-but-different cross-compilation hosts (notably
4382 Lispworks for Windows, following bug reports from Arthur Lemmens)
4383 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4384 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4385 fixed thanks to Pierre Mai's CMU CL patches.
4386 * fixes to keep the system from overflowing internal counters when
4387 it tries to use i/o buffers larger than 16M bytes
4388 * fixed bug 45a: Various internal functions required to support
4389 complex special functions have been merged from CMU CL sources.
4390 (When I was first setting up SBCL, I misunderstood a compile-time
4391 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4392 * improved support for type intersection and union, fixing bug 12
4393 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4394 more obscure bugs as well
4395 * some steps toward byte-compiling non-performance-critical
4396 parts of the system, courtesy of patches from Martin Atzmueller
4397 * Christophe Rhodes has made some debian packages of sbcl at
4398 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4399 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4400 stable, but are nonetheless usable. When he's ready, I'd be happy
4401 to add them to the SourceForge "File Releases" section. (And if
4402 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4403 * new fasl file format version number (because of changes in
4404 internal representation of (OR ..) types to accommodate the new
4405 support for (AND ..) types, among other things)
4407 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4408 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4410 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4412 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4413 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4414 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4415 * DESCRIBE now works on structure objects again.
4416 * Most function call argument type mismatches are now handled as
4417 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4418 know whether the function will be redefined before the call is
4419 executed. (The compiler could flag local calls with full WARNINGs,
4420 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4421 it doesn't keep track of enough information to know whether calls
4422 are local in this sense.)
4423 * Compiler output is now more verbose, with messages truncated
4424 later than before. (There should be some supported way for users
4425 to override the default verbosity, but I haven't decided how to
4426 provide it yet, so this behavior is still controlled by the internal
4427 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4428 src/compiler/ir1util.lisp.)
4429 * Fasl file format version numbers have increased again, because
4430 support for the Gray streams extension changes the layout of the
4431 system's STREAM objects.
4432 * The Gray subclassable streams extension now works, thanks to a
4433 patch from Martin Atzmueller.
4434 * The full LOAD-FOREIGN extension (not just the primitive
4435 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4436 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4437 but like most other programs, it defaults to copying the Unix
4438 environment from the original process instead of starting the
4439 new process in an empty environment.
4440 * Extensions which manipulate the Unix environment now support
4441 an :ENVIRONMENT keyword option which doesn't smash case or
4442 do other bad things. The CMU-CL-style :ENV option is retained
4443 for porting convenience.
4444 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4445 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4447 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4449 * A patch from Martin Atzmueller seems to have solved the SIGINT
4450 problem, and as far as we know, signal-handling now works cleanly.
4451 (If you find any new bugs, please report them!)
4452 * The system no longer defaults Lisp source file names to types
4453 ".l", ".cl", or ".lsp", but only to ".lisp".
4454 * The compiler no longer uses special default file extensions for
4455 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4456 seems to expect a single default extension for all compiled code,
4457 and there's no compelling reason to try to stretch the standard
4458 to allow two different extensions.) Instead, byte-compiled files
4459 default to the same extension as native-compiled files.
4460 * Fasl file format version numbers have increased again, because
4461 a rearrangement of internal implementation packages made some
4462 dumped symbols in old fasl files unreadable in new cores.
4463 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4464 many fewer weird special cases.
4465 * Bug #17 (differing COMPILE-FILE behavior between logical and
4466 physical pathnames) has been fixed, and some related misbehavior too,
4467 thanks to a patch from Martin Atzmueller.
4468 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4469 by Tim Moore, ported to SBCL by Martin Atzmueller.
4470 * Martin Atzmueller fixed several filesystem-related problems,
4471 including bug #36, in part by porting CMU CL patches, which were
4472 written in part by Paul Werkowski.
4473 * More compiler warnings in src/runtime/ are gone, thanks to
4474 more patches from Martin Atzmueller.
4475 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4478 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4480 * DESCRIBE now works on CONDITION objects.
4481 * The debugger now handles errors which arise when trying to print
4482 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4484 * The build system now uses an additional file, customize-target-features.lisp,
4485 to allow local modifications to the target *FEATURES* list. (The point of
4486 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4487 debugging features enabled, without having to worry about propagating it
4488 into everyone's system when I do a "cvs update".) When no
4489 customize-target-features.lisp file exists, the target *FEATURES* list
4490 should be constructed the same way as before.
4491 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4492 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4493 warn about behavior which is undefined under the ANSI spec. Note: This
4494 is specified by ANSI, but it's not very popular with programmers.
4495 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4496 macro in the SBCL sources for an example of a workaround which you
4497 might use to make portable ANSI-standard code which does what you want.
4498 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4499 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4500 when it's used as a non-toplevel form. (This is required in order
4501 to implement the DEFCONSTANT-EQX macro.)
4502 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4503 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4504 funny things, due to bugs in the implementation of EVAL-WHEN
4505 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4506 non-toplevel DEFCONSTANTs are uncommon.)
4507 * The core file version number and fasl file version number have been
4508 incremented, because the old noncompliant DEFCONSTANT behavior involved
4509 calling functions which no longer exist, and because I also took the
4510 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4511 * fixed bug 1 (error handling before read-eval-print loop starts), and
4512 redid debugger restarts and related debugger commands somewhat while
4514 ** The QUIT debugger command is gone, since it did something
4515 rather different than the SB-EXT:QUIT command, and since it never
4516 worked properly outside the main toplevel read/eval/print loop.
4517 Invoking the new TOPLEVEL restart provides the same functionality.
4518 ** The GO debugger command is also gone, since you can just invoke
4519 the CONTINUE restart directly instead.
4520 ** The TOP debugger command is also gone, since it's redundant with the
4521 FRAME 0 command, and since it interfered with abbreviations for the
4523 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4525 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4526 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4527 some porting work by Martin Atzmueller)
4528 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4529 with a boatload of patches which he ported to SBCL. Now that those
4531 ** The system tries to make sure that its low-priority messages
4532 are prefixed by semicolons, to help people who like to use
4533 syntax highlighting in their ILISP buffer. (This patch
4534 was originally due to Raymond Toy.)
4535 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4536 patches originally written by Raymond Toy.
4537 ** The compiler understands coercion between single-value and
4538 multiple-VALUES type expressions better, getting rid of some very
4539 weird behavior, thanks to patches originally by Robert MacLachlan
4540 and Douglas Crosher.
4541 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4542 lambda lists, thanks to a patch originally by Pierre Mai.
4543 ** The system no longer bogusly warns about "abbreviated type
4545 ** The compiler gets less confused by inlining and RETURN-FROM,
4546 thanks to some patches originally by Tim Moore.
4547 ** The system no longer hangs when dumping circular lists to fasl
4548 files, thanks to a patch originally from Douglas Crosher.
4549 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4550 undefined function error.
4551 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4552 in structure slot initforms) for now, documented workaround instead:-|
4553 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4554 * fixed bug 5: added stubs for various Gray stream functions called
4555 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4556 installed, at least appropriate type errors are generated
4557 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4558 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4559 consistently in DEFMETHOD forms.
4560 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4561 it doesn't seem to affect SBCL after all
4562 * The C runtime system now builds with better optimization and many
4563 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4565 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4567 * The system is now under CVS at SourceForge (instead of the
4568 CVS repository on my home machine).
4569 * The new signal handling code has been tweaked to treat register
4570 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4571 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4572 downstream, e.g. in debug-int.lisp, has implicit dependencies
4573 on the unsignedness of integer representation of machine words,
4574 and that caused the system to bomb out with infinite regress
4575 when trying to recover from type errors involving signed values,
4576 e.g. (BUTLAST '(1 2 3) -1).)
4577 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4578 (This was one of the bugs Peter Van Eynde reported back in July.)
4579 * The system now uses code inspired by Colin Walters' O(N)
4580 implementation of MAP (from the cmucl-imp@cons.org mailing
4581 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4582 inline the MAP operation, and there is more than one
4583 sequence argument to the MAP call (so that it can't just
4584 do ETYPECASE once and for all based on the type of the
4585 single sequence argument). (The old non-inline implementation
4586 of the general M-argument sequence-of-length-N case required
4587 O(M*N*N) time when any of the sequence arguments were LISTs.)
4588 * The QUIT :UNIX-CODE keyword argument has been renamed to
4589 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4590 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4592 * Raymond Wiker's patches to port dynamic loading from Linux to
4593 FreeBSD have been added.
4594 * The BUGS file is now more nearly up to date, thanks in large part
4595 to Martin Atzmueller's review of it.
4596 * The debugger now flushes standard output streams before it begins
4597 its output ("debugger invoked" and so forth).
4598 * The core version number and fasl file version number have both
4599 been incremented, because of incompatible changes in the layout
4601 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4602 used to not be called for a saved Lisp image.)
4603 * Martin Atzmueller's version of a patch to fix a compiler crash,
4604 as posted on sbcl-devel 13 September 2000, has been installed.
4605 * Instead of installing Martin Atzmueller's patch for the
4606 compiler transform for SUBSEQ, I deleted the compiler transform,
4607 and transforms for some similar consing operations.
4608 * A bug in signal handling which kept TRACE from working on OpenBSD
4610 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4611 away by constant folding
4612 * The system now defines its address space constants in one place
4613 (in the Lisp sources), and propagates them automatically elsewhere
4614 (through GENESIS and the sbcl.h file). Therefore, patching the
4615 address map is less unnecessarily tedious and error-prone. The
4616 Lisp names of address space constants have also been systematized.
4617 * CVS tags like dollar-Header-dollar have been removed from
4618 the sources, because they have never saved me trouble and
4619 they've been source of trouble working with patches and other
4620 diff-related operations.
4621 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4622 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4624 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4626 * The system has been ported to OpenBSD.
4627 * The system now compiles with a simple "sh make.sh" on the systems
4628 that it's supported on. I.e., now you no longer need to tweak
4629 text in the target-features.lisp-expr and symlinks in src/runtime/
4630 by hand, the make.sh takes care of it for you.
4631 * The system is no longer so grossly inefficient when compiling code
4632 involving vectors implemented as general (not simple) vectors (VECTOR T),
4633 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4634 which dares to use the various sequence functions on non-simple
4635 vectors, takes less of a performance hit.
4636 * There is now a primitive type predicate VECTOR-T-P
4637 to test for the (VECTOR T) type, so that e.g.
4638 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4639 can now be compiled with some semblance of efficiency. (The old code
4640 turned the type declaration into a full call to %TYPEP at runtime!)
4641 * AREF on (VECTOR T) is still not fast, since it's still compiled
4642 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4643 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4644 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4645 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4646 without making full calls to SUBTYPEP at runtime.
4647 (Some analogous efficiency issues for non-simple vectors specialized to
4648 element types other than T, or for non-simple multidimensional arrays,
4649 have not been addressed. They could almost certainly be handled the
4650 same way if anyone is motivated to do so.)
4651 * The changes in array handling break binary compatibility, so
4652 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4653 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4654 instead of (VALUES T T).
4655 * By following the instructions that Dan Barlow posted to sbcl-devel
4656 on 2 July 2000, I was able to enable primitive dynamic object
4657 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4658 functionality is not implemented (since it calls ld to resolve
4659 library references automatically, requiring RUN-PROGRAM for its
4660 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4661 try to resolve library references) is now supported.
4662 * The system now flushes the standard output streams when it terminates,
4663 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4664 them at several other probably-convenient times, e.g. in each pass of
4665 the toplevel read-eval-print loop, and after evaluating a form given
4666 as an "--eval" command-line option. (These changes were motivated by a
4667 discussion of stream flushing issues on cmucl-imp in August 2000.)
4668 * The source transform for TYPEP of array types no longer assumes
4669 that an array whose element type is a not-yet-defined type
4670 is implemented as an array of T, but instead punts, so that the
4671 type will be interpreted at runtime.
4672 * There is now some support for cross-compiling in make.sh: each of
4673 the phases of make.sh has its own script. (This should be transparent
4674 to people doing ordinary, non-cross-compile builds.)
4675 * Since my laptop doesn't have hundreds of megabytes of memory like
4676 my desktop machine, I became more motivated to do some items on
4677 my to-do list in order to reduce the size of the system a little:
4678 ** Arrange for various needed-only-at-cold-init things to be
4679 uninterned after cold init. To support this, those things have
4680 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4681 symbols with such names are now uninterned after cold init).
4682 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4683 fasl files for cold load.
4684 ** Remove the old compiler structure pooling code (which used to
4685 be conditional on the target feature :SB-ALLOC) completely.
4686 ** Redo the representation of some data in cold init to be more compact.
4687 (I also looked into supporting byte compiled code at bootstrap time,
4688 which would probably reduce the size of the system a lot, but that
4689 looked too complicated, so I punted for now.)
4690 * The maximum signal nesting depth in the src/runtime/ support code has
4691 been reduced from 4096 to 256. (I don't know any reason for the very
4692 large old value. If the new smaller value turns out to break something,
4693 I'll probably just bump it back up.)
4694 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4696 * Many, many bugs reported by Peter Van Eynde have been added to
4697 the BUGS list; some have even been fixed.
4698 * While enabling dynamic object file loading, I tried to make the
4699 code easier to understand, renaming various functions and variables
4700 with less ambiguous names, and changing some function calling
4701 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4702 * While trying to figure out how to do the OpenBSD port, I tried to
4703 clean up some of the code in src/runtime/. In particular, I dropped
4704 support for non-POSIX signal handling, added various comments,
4705 tweaked the code to reduce the number of compilation warnings, and
4706 renamed some files to increase consistency.
4707 * To support the new automatic configuration functionality in make.sh,
4708 the source file target-features.lisp-expr has been replaced with the
4709 source file base-target-features.lisp-expr and the machine-generated
4710 file local-target-features.lisp-expr.
4711 * fixed a stupid quoting error in make.sh so that using CMU CL
4712 "lisp -batch" as cross-compilation host works again
4714 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4716 * DESCRIBE no longer tries to call itself recursively to describe
4717 bound/fbound values, so that it no longer fails on symbols which are
4718 bound to themselves (like keywords, T, and NIL).
4719 * DESCRIBE now works on generic functions.
4720 * The printer now prints less-screwed-up representations of closures
4721 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4722 * A private symbol is used instead of the :EMPTY keyword previously
4723 used to mark empty slots in hash tables. Thus
4724 (DEFVAR *HT* (MAKE-HASH-TABLE))
4725 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4726 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4727 now does what ANSI says that it should. (You can still get
4728 similar noncompliant behavior if bang on the hash table
4729 implementation with all the symbols you get back from
4730 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4731 This breaks binary compatibility, since tests for equality to
4732 :EMPTY are wired into things like the macroexpansion of
4733 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4735 * There's now a minimal placeholder implementation for CL:STEP,
4736 as required by ANSI.
4737 * An obscure bug in the interaction of the normal compiler, the byte
4738 compiler, inlining, and structure predicates has been patched
4739 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4740 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4742 * Missing ordinary arguments in a macro call are now detected even
4743 when the macro lambda list contains &KEY or &REST.
4744 * The debugger no longer complains about encountering the top of the
4745 stack when you type "FRAME 0" to explicitly instruct it to go to
4746 the top of the stack. And it now prints the frame you request even
4747 if it's the current frame (instead of saying "You are here.").
4748 * As specified by ANSI, the system now always prints keywords
4749 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4751 * The default initial SIZE of HASH-TABLEs is now smaller.
4752 * Type information from CLOS class dispatch is now propagated
4753 into DEFMETHOD bodies, so that e.g.
4754 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4756 is now basically equivalent to
4757 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4758 (DECLARE (TYPE SINGLE-FLOAT X))
4760 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4761 operation, without having to do run-time type dispatch.
4762 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4763 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4764 * Since I don't seem to be making big file reorganizations very often
4765 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4766 is overflowing my ability to conveniently back them up), I've finally
4767 checked the system into CVS. (The CVS repository is on my home system,
4768 not at SourceForge -- putting it on SourceForge might come later.)
4769 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4770 high-level GC-NOTIFY-FOO functions send their output. (There's
4771 still very little control of where low-level verbose GC functions
4772 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4773 less than it used to -- the GC-NOTIFY-FOO functions are now under
4774 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4775 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4776 in only one place in the source code, and propagates it automatically
4777 everywhere that it's needed. Thus e.g. when I bump the version from
4778 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4779 * The C source files now include boilerplate legalese and documentation
4780 at the head of each file (just as the Lisp source files already did).
4781 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4782 to his page will be replaced with a link to his new CLiki service.
4784 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4786 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4787 * The build process now looks for GNU make under the default name "gmake",
4788 instead of "make" as it used to. If GNU make is not available as "gmake"
4789 on your system, you can change this default behavior by setting the
4790 GNUMAKE environment variable.
4791 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4792 can build without error under CMU CL.
4794 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4796 * There is now a partial SBCL user manual (with some new text and some
4797 text cribbed from the CMU CL manual).
4798 * The beginnings of a profiler have been added (starting with the
4799 CMU CL profiler and simplifying and cleaning up). Eventually the
4800 main interface should be through the TRACE macro, but for now,
4801 it's still accessed through vaguely CMU-CL-style functions and macros
4802 exported from the package SB-PROFILE.
4803 * Some problems left over from porting CMU CL to the new
4804 cross-compilation bootstrap process have been cleaned up:
4805 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4806 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4807 part when getting rid of the compiler *BACKEND* stuff.)
4808 ** The value of *NULL-TYPE* was screwed up, because it was
4809 being initialized before the type system knew the final
4810 definition of the 'NULL type. This screwed up several key
4811 optimizations in the compiler, causing inefficiency in all sorts
4812 of places. (I found it because I wanted to understand why
4813 GET-INTERNAL-RUN-TIME was consing.)
4814 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4815 PROCLAIM FTYPE information. Unfortunately this broke binary
4816 compatibility again, since now the forms output by DEFGENERIC
4817 to refer to functions which didn't exist in 0.6.3.
4818 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4819 can use the new (as of 0.6.3) transform for SOME into MAP into
4821 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4822 values to (INTEGER 0 1000000), so that the time code will no longer
4823 occasionally get blown up by Linux returning 1000000 microseconds
4824 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4825 its output conform to the ANSI spec. (Alas, this makes its output
4826 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4827 * A full call to MAP NIL with a single sequence argument no longer conses.
4828 * fixes to problems pointed out by Martin Atzmueller:
4829 * The manual page no longer talks about multiprocessing as though
4830 it were currently supported.
4831 * The ILISP support patches have been removed from the distribution,
4832 because as of version 5.10.1, ILISP now supports SBCL without us
4833 having to maintain patches.
4834 * added a modified version of Raymond Toy's recent CMU CL patch for
4835 EQUALP comparison of HASH-TABLE
4837 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4839 * The system still can't cross-compile itself with
4840 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4841 entails), but at least it can compile more of itself that way
4842 than it used to be able to, and various buglets which were uncovered
4843 by trying to cross-compile itself that way have now been fixed.
4844 * This release breaks binary compatibility again. This time
4845 at least I've incremented the FASL file format version to 2, so that the
4846 problem can be detected reliably instead of just causing weird errors.
4847 * various new style warnings:
4848 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4849 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4850 ** using the lexical binding of a variable named in the *FOO* style
4851 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4852 as specified by ANSI.
4853 * *RANDOM-STATE* is no longer automatically initialized from
4854 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4855 default behavior of the system is to repeat its behavior every time
4856 it's run. If you'd like to change this behavior, you can always
4857 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4858 old convention there was no comparably easy way to get the system to
4859 repeat its behavior every time it was run.
4860 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4861 FTYPE declarations has been removed, in favor of their ANSI
4862 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4863 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4864 the types of their sequence arguments can be determined at compile time.
4865 This is done through a new open code expansion for MAP which eliminates
4866 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4867 argument types can be determined at compile time.
4868 * The optimizer now transforms COERCE into an identity operation when it
4869 can prove that the coerced object is already of the correct type. (This
4870 can be a win for machine generated code, including the output of other
4871 optimization transforms, such as the MAP transform above.)
4872 * Credit information has been moved from source file headers into CREDITS.
4873 * Source file headers have been made more standard.
4874 * The CASE macro now compiles without complaining even when it has
4877 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4879 * (Note that the way that the PCL macroexpansions were rewritten
4880 to accommodate the change in DEFGENERIC below breaks binary
4881 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4882 not run under sbcl-0.6.2. Once we get out of alpha releases,
4883 i.e. hit release 1.0.0, we'll probably try to maintain binary
4884 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4885 and sbcl-1.4.4. Until then, however, it might be fairly common
4886 for maintenance releases to break binary compatibility.)
4887 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4889 * The reporting of unhandled signals has been changed to print some
4890 explanatory text as well as the report form. (Previously only
4891 the report form was printed.)
4892 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4893 it defines, so that the compiler no longer issues undefined function
4894 warnings for compiled-but-not-yet-loaded generic functions.
4895 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4896 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4897 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4898 section of the ANSI spec are supported.
4899 * The "" nickname for the "KEYWORD" package has been removed.
4900 The reader still handles symbol tokens which begin with a package marker
4901 as keywords, but it doesn't expose its mechanism for doing so in the
4902 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4903 * The system now issues STYLE-WARNINGs for contradictory TYPE
4904 proclamations. (Warnings for contradictory FTYPE proclamations would
4905 be nice too, but those can't be done usefully unless the type system
4906 is made smarter about FUNCTION types.)
4907 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4908 systematized, so that "*target-*.lisp is supposed to exist only on the
4909 target and imply that there's a related file which exists on the
4910 host, and *host-*.lisp is supposed to exist only on the host and imply
4911 that there's a related file which exists on the target. This involves a
4912 lot of renaming. Hopefully the acute confusion caused by the renaming
4913 will be justified by the reduction in chronic confusion..
4914 ** runtime-type.lisp -> early-target-type.lisp
4915 ** target-type.lisp -> late-target-type.lisp
4916 ** early-host-format.lisp -> early-format.lisp
4917 ** late-host-format.lisp -> late-format.lisp
4918 ** host-error.lisp -> misc-error.lisp
4919 ** early-error.lisp -> early-target-error.lisp
4920 ** late-error.lisp -> late-target-error.lisp
4921 ** host-defboot.lisp -> early-defboot.lisp
4922 ** code/misc.lisp -> code/target-misc.lisp
4923 ** code/host-misc.lisp -> code/misc.lisp
4924 ** code/numbers.lisp -> code/target-numbers.lisp
4925 ** code/early-numbers.lisp -> numbers.lisp
4926 ** early-host-type.lisp -> early-type.lisp
4927 ** late-host-type.lisp -> late-type.lisp
4928 ** host-typep.lisp -> typep.lisp
4929 ** load.lisp -> target-load.lisp
4930 ** host-load.lisp -> load.lisp
4931 ** host-disassem.lisp -> disassem.lisp
4932 ** host-insts.lisp -> insts.lisp
4933 ** byte-comp.lisp -> target-byte-comp.lisp
4934 ** host-byte-comp.lisp -> byte-comp.lisp
4935 ** host-signal.lisp -> signal.lisp
4936 ** host-defstruct.lisp -> defstruct.lisp
4937 ** late-target-type.lisp -> deftypes-for-target.lisp
4938 Furthermore, several other previously target-only files foo.lisp (e.g.
4939 hash-table.lisp and random.lisp) have been split into a target-and-host
4940 foo.lisp file and a target-only target-foo.lisp file, with their key type
4941 definitions in the target-and-host part, so that the cross-compiler will
4942 know more about target types.
4943 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4944 gone away. In their place are various *BACKEND-FOO* variables
4945 corresponding to the slots of the old structure.
4946 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4947 into the target SBCL has been fixed.
4948 * The chill.lisp system for loading cold code into a running SBCL
4950 * Support for the CMU CL "scavenger hook" extension has been removed.
4951 (It was undocumented and unused in the CMU CL sources that SBCL was
4952 derived from, and stale in sbcl-0.6.1.)
4953 * Various errors in the cross-compiler type system were detected
4954 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4955 (enabling various consistency checks). Many of them were fixed,
4956 but some hard problems remain, so the compiler is back to
4957 running without *TYPE-SYSTEM-INITIALIZED* for now.
4958 * As part of the cross-compiler type system cleanup, I implemented
4959 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4960 * I have started adding UNCROSS calls throughout the type system
4961 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4962 on cross-compiler input files will be able to go away, and various
4964 * CONSTANTP now returns true for quoted forms (as explicitly required
4967 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4969 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4970 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4971 (MAKE-LIST -1). In the longer term, it ought to become true
4972 that declarations are assertions even at SAFETY 1. For now, it's not
4973 quite true even at SAFETY 3, but it's at least more nearly true..
4974 (Note that this change seems to increases the size of the system by
4975 O(5%) and to decrease the speed of the compiler by 20% or more.)
4976 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4977 for the problem of printing dozens of lines of distracting information
4978 about low-level system machinery as part of the top stack frame
4979 on entry to the debugger when an undefined function was called.
4980 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4981 is not reset to COMMON-LISP-USER.
4982 * Compilation of stuff related to dyncount.lisp has been made conditional
4983 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4984 smaller. The various dyncount-related symbols have been moved into
4985 a new "SB-DYNCOUNT" package.
4986 * tty-inspect.lisp has been renamed to inspect.lisp.
4987 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4988 feature has gone away. (When we eventually port to other flavors of
4989 libc and/or Unix, we'll try to make the differences between flavors
4990 invisible at the user level.)
4991 * Various other *FEATURES* tags, and/or their associated conditionals,
4992 have been removed if obsolescent, or given better documentation, or
4993 sometimes given more-mnemonic names.
4995 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4997 * tidied up "make.sh" script
4998 * tidied up system directory structure
4999 * better "clean.sh" behavior
5000 * added doc/FOR-CMUCL-DEVELOPERS
5001 * many many small tweaks to output format, e.g. removing possibly-confusing
5002 trailing #\. character in DESCRIBE-INSTANCE
5003 * (EQUALP #\A 'A) no longer signals an error.
5004 * new hashing code, including EQUALP hashing
5005 * tidied up Lisp initialization and toplevel
5006 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5007 * command line argument processing
5008 * added POSIX-GETENV function to deal with Unix-ish environment variables
5009 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5010 terminating SBCL on EOF
5011 * non-verbose GC by default
5012 * There is no more "sbcl" shell script; the sbcl file is now the C
5013 runtime executable (just like CMU CL).
5014 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5016 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5017 symbol and package information as Lisp native symbol and package objects
5018 instead of strings naming symbols and strings naming packages. This way,
5019 whenever packages are renamed (as in warm init), debug information is
5020 transformed along with everything else.
5021 * tweaked the optimization policy declarations which control the building
5022 of SBCL itself. Now, among other things, the system no longer saves
5023 source location debugging information. (This helps two problems at once
5024 by reducing SBCL size and by keeping SBCL from trying to look for its
5025 sources -- which may not exist -- when reporting errors.)
5026 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5027 debugging and testing purposes
5028 * cleaned up printing, making the printer call PRINT-OBJECT for
5029 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5030 methods, giving nearly-ANSI behavior
5031 * converted almost all special variables to use *FOO* naming convention
5032 * deleted PARSE-TIME functionality, since it can be done portably
5033 * moved some files out of cold init into warm init
5034 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5036 * regularized formatting of source files
5037 * added an install.sh script
5038 * fixed ridiculous memory usage of cross-compiler by making
5039 compiler/alloc.lisp not try to do pooling unless it can hook
5040 itself into the GC of the cross-compilation host. Now the system
5041 builds nicely on my old laptop.
5042 * added :SB-ALLOC in target-features.lisp-expr
5043 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5044 was not implemented)
5045 * re-did condition handling and note reporting in the compiler. Notes
5046 are no longer handled by signalling conditions. Style warnings
5047 and warnings are handled more correctly and reported in such a way
5048 that it's easy to find one or the other in your output (so that you
5049 can e.g. figure out which of many problems caused COMPILE-FILE to
5051 * changed the severity of several compiler warnings from full WARNING
5052 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5053 compiler note reporting so that it doesn't use the condition system
5054 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5055 COMPILE-FILE command)
5056 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5057 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5058 no longer supported, since their implementation was deeply intertwingled
5059 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5060 * removed bogus "support" for compiler macros named (SETF FOO), and
5061 removed the compiler macro for SETF INFO (but only after making a fool
5062 of myself on the cmucl-imp mailing list by posting a bogus patch for
5063 DEFINE-COMPILER-MACRO..)
5064 * Compiled files containing forms which have side effects on the Lisp
5065 reader (such as DEFPACKAGE forms) are now handled more correctly.
5066 (Compiler queuing of top level lambdas has been suppressed by setting
5067 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5068 * deleted various currently-unused source files, e.g. gengc.lisp. They
5069 may be added back at some point e.g. when porting to other architectures,
5070 but until they are it's distracting to distribute them and to try to
5072 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5073 there were so many of them they're just distractions, and UNCROSS is
5074 known to be able to handle the current sources
5075 * moved PROFILE functionality into TRACE, so that it will be clear
5076 how the wrapping and unwrapping of functions when you profile them
5077 interacts with the wrapping and unwrapping of functions when you
5078 trace them. (Actually, the functionality isn't there yet, but at least
5079 the interface specification is there. Hopefully, the functionality will
5080 arrive with some future maintenance release.)
5081 * removed host-oops.lisp
5082 * changed signature of QUIT function to allow UNIX-CODE argument
5083 * fixed READ-SEQUENCE bug
5084 * tweaked verbose GC output so that it looks more like the progress
5085 output that ANSI specifies for functions like LOAD
5086 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5087 * added <http://sbcl.sourceforge.com> to the banner information printed by