1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
4 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
5 allows the compiler to safely elide boundedness checks for special
7 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
8 the symbol, prohibits both lexical and dynamic binding. This is mainly an
9 efficiency measure for threaded platforms, but also valueable in
11 * improvement: SBCL now emits a compiler note where stack allocation was
12 requested but could not be provided.
13 * optimization: compiler now generates faster array typechecking code.
14 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
16 * optimization: multidimensional array accesses in the absence of type
17 information regarding array rank are approximately 10% faster due to
18 open coding of ARRAY-RANK.
19 * documentation: CLOS slot typechecing policy has been documented.
20 * bug fix: potential miscompilation of array stack allocation on x86 and
21 x86-64. (reported by Time Tossavainen)
22 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
23 result in their subforms being treated as top level forms. (reported by
25 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
26 than a machine register could leave garbage in the high bits of the
27 result register (bug 316325).
28 * bug fix: disable address space randomization Linux/x86-64 as well,
29 not just x86-64. (reported by Ken Olum)
30 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
31 generate incorrect code.
32 * bug fix: #201; type inference for CONS and ARRAY types could derive
33 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
34 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
35 could not be assigned to or bound with PROGV. (reported by Lorenz
37 * bug fix: the value of CL:- in the inspector was the previous expression
38 evaluated rather than the expression being evaluated.
39 * bug fix: constants can no longer be locally declared special.
41 changes in sbcl-1.0.28 relative to 1.0.27:
42 * a number of bugs in cross-compilation have been fixed, with the ultimate
43 result that building under (at least) clisp should be much more reliable.
44 * minor incompatible changes: echo-streams now propagate unread-char to the
45 underlying input stream, and no longer permit unreading more than one
47 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
48 frames, allowing for instance more reliable backtraces.
49 * improvement: the debugger REPL can now reference lexical variables
50 by name directly for code compiled with (DEBUG 3).
51 * improvement: errors from malformed declarations now have better source
52 paths associated with them. (thanks to Tobias Rittweiler)
53 * optimization: faster local calls on x86/x86-64
54 * bug fix: some error messages for out-of-bound array indexes confused the
55 index and the bound. (thanks to Stas Boukarev)
56 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
58 * bug fix: running regressions tests in shells without OSTYPE set now works.
59 (reported by Harald Hanche-Olsen)
60 * bug fix: more robust static space exhaustion signalling from
61 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
62 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
63 docstring away instead of storing it under names such as (LAMBDA (X)).
64 (reported by Leslie Polzer)
65 * bug fix: timers could go off in the wrong order, be delayed indefinitely
66 (thanks to Ole Arndt for the patch)
67 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
68 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
69 potentially-invalid effective methods in its cache.
70 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
71 instances as well (thanks to Paul Khuong)
72 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
73 SB-POSIX:WAITPID and their C equivalents.
74 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
76 changes in sbcl-1.0.27 relative to 1.0.26:
77 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
78 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
79 * improvement: the system either recovers from stack exhaustion or dies
80 properly as opposed to leaving the user uncertain of whether the handler
81 trampled on some random memory next to the stack or having to rely on
82 --lose-on-corruption (which is still a good idea to use in production
83 because stack exhaustion can happen in signal handlers which will likely
85 * bug fix: fix gc related interrupt handling bug on ppc (regression from
86 1.0.25.37, reported by Harald Hanche-Olsen)
87 * bug fix: work around signal delivery bug in darwin (regression from
88 1.0.25.44, reported by Sidney Markowitz)
89 * bug fix: fix ERROR leaking memory (reported by David Thompson)
91 changes in sbcl-1.0.26 relative to 1.0.25:
92 * incompatible change: an interruption (be it a function passed to
93 INTERRUPT-THREAD or a timer function) runs in an environment where
94 interrupts can be enabled. The interruption can use
95 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
96 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
97 potentially running out of stack. Keep in mind that in the absence
98 of WITHOUT-INTERRUPTS some potentially blocking operation such as
99 acquiring a lock can enable interrupts.
100 * incompatible change: GC-OFF and GC-ON are removed, as they were
101 always unsafe. Use WITHOUT-GCING instead.
102 * new feature: runtime option --disable-ldb
103 * new feature: runtime option --lose-on-corruption to die at the
104 slightest hint of possibly non-recoverable errors: running out of
105 memory, stack, alien stack, binding stack, encountering a memory
106 fault, etc. In the absence of --lose-on-corruption a warning is
108 * enhancement: detect binding stack exhaustion
109 * enhancement: detect alien stack exhaustion on x86/x86-64
110 * improvement: generally more stable and reliable interrupt handling
111 * improvement: there is a per thread interruption queue,
112 interruptions are executed in order of arrival
113 * improvement: a repeating timer reschedules itself when the it has
114 finished, but expiration times are spaced equally. If an
115 expiration time is in the past it will trigger after a short grace
116 period that may give a chance to other things to run.
117 * optimization: slightly faster gc on multithreaded builds
118 * optimization: faster WITHOUT-GCING
119 * bug fix: when JOIN-THREAD signals an error, do it when not holding
120 important locks so that the debugger/handler doesn't produce
121 recursive errors or deadlock.
122 * bug fix: real-time signals are not used anymore, so no more
123 hanging when the system wide real-time signal queue gets full.
124 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
126 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
127 * bug fix: fix random memory faults related to interrupts on alpha
128 * bug fix: fix random memory faults related to interrupts on sparc
129 * bug fix: fix deadlocks related to starting threads
130 * bug fix: fix deadlines on locks on futex platforms
131 * bug fix: restore errno in signal handlers
132 * bug fix: fix deadlocks related to hash tables
133 * bug fix: fix deadlocks in pcl
135 changes in sbcl-1.0.25 relative to 1.0.24:
136 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
137 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
138 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
139 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
140 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
141 better name and does not return values so stale on multiprocessor systems.
142 Also, HOLDING-MUTEX-P was added for about the only sane usage of
144 * improvement: unithread builds keep track of MUTEX-VALUE.
145 * improvement: reading from a TWO-WAY-STREAM does not touch the output
146 stream anymore making it thread safe to have a concurrent reader and
147 a writer, for instance, in a pipe.
148 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
149 arguments, making compiler-macros for SETF-functions able to inspect
150 their constant arguments.
151 * improvement: COMPILE-FILE reports times with millisecond accuracy
152 (thanks to Luis Oliveira)
153 * optimization: CHAR-CODE type derivation has been improved, making
154 TYPEP elimination on subtypes of CHARACTER work better. (reported
155 by Tobias Rittweiler, patch by Paul Khuong)
156 * bug fix: setting alien structure fields of type struct by value now
157 computes the right offset for the memory copy.
158 * bug fix: compilation problem involving inlined calls to aliens with
159 result type VOID. (reported by Ken Olum)
160 * bug fix: #235a; sequential inline expasion in different policies no
161 longer reuses the functional from the previous expansion site.
162 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
163 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
164 * bug fix: named ENUMs in multiply-referenced alien record types no longer
165 cause an implied type redefinition. Regression from 1.0.21.29.
166 * bug fix: sign-extension of small signed return values now works with C
167 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
168 * improvements to the Windows port:
169 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
170 Westfold, thanks to Rudi Schlatte)
172 changes in sbcl-1.0.24 relative to 1.0.23:
173 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
174 vector of a multidimensional SIMPLE-ARRAY.
175 * new feature: the system now signals a continuable error if standard
176 readtable modification is attempted. (thanks to Tobias Rittweiler)
177 * new feature: DIRECTORY has been extended with a non-standard keyword
178 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
179 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
180 has been robustified and documented better. (thanks to Robert Goldman)
181 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
182 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
183 * tradeoff: constant FORMAT control strings are now compiled unless
184 SPACE > SPEED (previously only when SPEED > SPACE.)
185 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
187 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
189 * bug fix: lexical type declarations are now correctly reported by
190 SB-CLTL2. (reported by Larry D'Anna)
191 * bug fix: STRING-TO-OCTETS did not handle :START properly when
192 using UTF-8 as external format. (thanks to Luis Oliveira)
193 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
194 are signalled correctly. (thanks to Stas Boukarev)
195 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
196 designator. (thanks to Tobias Rittweiler)
197 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
198 designator, and returns T instead of the function. (thanks to
200 * bug fix: direct superclasses of STANDARD-CLASS and
201 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
202 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
203 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
204 when no class with the named slot yet exists no longer causes a
205 compile-time style-warning.
206 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
207 in safe code. (reported by Didier Verna)
208 * bug fix: #430; nested structure constructors can stack allocate.
209 * bug fix: on some 64-bit platforms dynamic space size was truncated
210 to #xffff0000 bytes. (reported by Benjamin Lambert)
211 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
212 REPL unusable. (reported by Daniel Herring)
213 * bug fix: return values of READ-SEQUENCE did not take :START into
214 account on file streams, regressions since 1.0.12.22. (reported by
215 Thomas Russ, patch by Paul Khuong)
216 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
217 method specializer used to confuse permuation vector optimization.
218 * bug fix: system inserted bogus implicit type declarations for local
219 special variables in DEFMETHOD bodies.
220 * bug fix: #354; duplicated frames in backtraces due to
221 non-tail-call-optimized XEPs to functions with return type NIL
223 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
224 initializes structure object slots according to DEFSTRUCT initforms,
225 and DEFSTRUCT forms :INCLUDEind structure classes defined using
226 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
227 (reported by Bruno Haible and Stephen Wilson)
228 * bug fix: #395; fill-pointer output streams used now support
229 element-type BASE-CHAR as well.
230 * bug fix: compiler error when attempting to derive return value of
231 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
233 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
235 * bug fix: Mac OS X binaries should now be portable between Leopard
238 changes in sbcl-1.0.23 relative to 1.0.22:
239 * enhancement: when disassembling method functions, disassembly
240 for the associated fast function is also produced.
241 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
243 * optimization: printing with *PRINT-PRETTY* true is now more
244 efficient as long as the object being printed doesn't require
245 special handling by the pretty printer.
246 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
247 now interact correctly with type declarations.
248 * partial bug fix: PCL detects infinite recursion during wrapper
249 validation. (thanks to Attila Lendvai)
250 * bug fix: #426; nested function calls are inlined properly.
251 Previously if FOO was an inline function, in calls of the form
252 (FOO (FOO ...)) the outer call was not inlined.
253 * bug fix: long long arguments passed in registers to alien
254 functions on PowerPC Linux and NetBSD are now handled correctly
255 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
257 changes in sbcl-1.0.22 relative to 1.0.21:
258 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
259 for the shared object in the current directory, but passes the native
260 namestring of the designated pathname to the operation system's shared
261 object loading function as-it.
262 * minor incompatible change: --disable-debugger toplevel option now takes
263 effect before processing of initialization files and --eval or --load
265 * new feature: new commandline argument: --script, which supports
266 shebang lines. See documentation for details. (based on work by
268 * new feature: SAVE-LISP-AND-DIE can save current values of
269 --dynamic-space-size and --control-stack-size in the executable core,
270 causing it to skip normal runtime option processing. See documentation
271 for details. (thanks to Zach Beane)
272 * enhancement: inoccous calls to EVAL or generic functions dispatching
273 on subclasses of eg. STREAM no longer cause compiler notes to appear.
274 * enhancement: the system no longer resignals errors from --load and
275 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
276 associated with the original error to be lost. (thanks to Ariel
278 * enhancement: :DONT-SAVE keyword argument has been added to
279 LOAD-SHARED-OBJECT for controlling interaction with
280 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
281 the effects of an earlier LOAD-SHARED-OBJECT call.
282 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
283 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
284 provided. (reported by Cedric St-Jean)
285 * bug fix: circularity handling in the reader did not treat raw
286 structure slots correctly. (reported by Cedric St-Jean)
287 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
288 file descriptors when there were none.
289 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
290 was broken. (reporteed by Willem Broekema)
291 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
292 pathnames without a directory.
293 * bug fix: COMPILE-FILE did not warn about undefined variable
294 references at toplevel, and LOAD of the resulting fasl did
296 * bug fix: functions with non-required arguments used to end up with
297 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
298 * bug fix: redefining a function with non-required arguments didn't
299 update the system's knowledge about its call signature properly.
300 * bug fix: fixed #431; incompatible alien record type redefinitions
301 are detected and handled. (reported by Neil Haven)
302 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
303 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
305 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
306 after alien stack frames.
307 * bug fix: strings passed to mk{s,d}temp lacked null termination
309 changes in sbcl-1.0.21 relative to 1.0.20:
310 * new feature: the compiler is able to track the effective type of a
311 generic function across method addition and removal.
312 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
313 appropriately typed structure slots without locking.
314 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
315 information like those gathered by TIME using a programming-friendly
317 * new feature: TIME reports time taken even if the form performs a
318 non-local transfer of control.
319 * enhancement: reduced conservativism on GENCGC platforms: on
320 average 45% less pages pinned (measured from SBCL self build).
321 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
322 mutate constant symbols or violate declaimed type of the symbol.
323 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
324 sleeping on the semaphore in heavily contested semaphores.
325 * bug fix: semaphores and condition variables were not interrupt
327 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
328 owned by other threads anymore.
329 * bug fix: FIND on lists called KEY outside the specified
330 subsequence. (reported by budden)
331 * bug fix: LOG doesn't use single-float intermediate results when
332 given mixed integer and double-float arguments, leading to better
333 precision. (reported by Bob Felts)
334 * bug fix: LOG with base zero returned values of inconsistent type.
336 changes in sbcl-1.0.20 relative to 1.0.19:
337 * minor incompatible change: OPTIMIZE qualities
338 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
339 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
340 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
341 * documentation: some slot access efficiency guidelines have been
342 added to the user manual.
343 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
344 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
345 as ASSOC and MEMEBER.
346 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
347 to more efficient EQ-comparison versions more often.
348 * optimization: enhanced derivation of DOLIST iteration variable type
350 * optimization: constant folding of simple (LIST ...) forms as DOLIST
352 * optimization: runtime lookup of function definitions can be
353 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
354 * optimization: compiler is able to derive the return type of
355 (AREF (THE STRING X) Y) as being CHARACTER.
356 * optimization: CLRHASH on empty hash-tables no longer does pointless
357 work. (thanks to Alec Berryman)
358 * optimization: REPLACE deftransforms don't punt when :START1 or
360 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
361 no longer emits a code deletion note. (reported by Andrew
363 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
364 functions with an odd number of &OPTIONAL arguments, a &REST
365 argument, and one or more &KEY arguments at the call site.
366 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
367 arguments appeared at call sites as well.
368 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
369 READ-LINE &co can no longer read from them afterwards. (reported
371 * bug fix: fixed #427: unused local aliens no longer cause compiler
372 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
374 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
375 alien-stack leakage. (reported by Andy Hefner)
376 * bug fix: PROGV signals an error when an attempt to violate declared
377 type of a variable or bind a constant is made.
378 * bug fix: SET signals an error when an attempt to violate declared
379 type of a variable is made.
380 * bug fix: restart computation during the execution of a restart
381 test function no longer causes infinite recursion. (reported by
383 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
384 signals an error. (thanks to Pierre Mai)
385 * bug fix: EXPORT left symbol unexported in conflict situations.
386 (thanks to Michael Weber)
387 * bug fix: correctly handle name conflicts involving CL:NIL.
388 (thanks to Michael Weber)
389 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
390 conflicts arising from USEing package with conflicting symbols
391 correctly. (thanks to Michael Weber)
393 changes in sbcl-1.0.19 relative to 1.0.18:
394 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
395 warnings that go otherwise unhandled will be muffled if they are
396 of the type that's the value of this variable.
397 * optimization: stack allocation is slightly more efficient on x86
399 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
400 where a raw slot always is initialized using the initform whose
401 type is not know sufficiently well a compile-time are now compiled
402 correctly. (reported by John Morrison)
403 * bug fix: compiler no longer makes erronous assumptions in the
404 presence of non-foldable SATISFIES types.
405 * bug fix: stack analysis missed cleanups of dynamic-extent
406 arguments in non-let-converted calls to local functions.
407 * improvements to the Windows port:
408 ** adjusted address spaces for building on both Win32 and
409 Win64. (thanks for John Connors)
410 * fixed some bugs revealed by Paul Dietz' test suite:
411 ** interval arithmetic during type derivation used inexact integer
412 to single-float coercions.
413 ** arithmetic operations involving large integers and single
414 floats give the same results in compiled and interpreted code.
415 ** deriving the result type of COERCE no longer signals an error
416 if the derived type of the second argument is a MEMBER type
417 containing invalid type specifiers.
418 ** ADJOIN with constant NIL as second argument works correctly.
419 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
421 changes in sbcl-1.0.18 relative to 1.0.17:
422 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
423 profiles only the current thread.
424 * minor incompatible change: changes to SYMBOL-VALUE of constants
425 defined with DEFCONSTANT now signal an error.
426 * enhancement: SB-SPROF now has support for wallclock profiling,
427 and is also able to profile specific threads. REPORT output
428 has also additional sorting options.
429 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
431 * optimization: structure allocation has been improved
432 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
433 ** out of line constructors are ~10% faster.
434 ** inline constructors are ~15% faster.
435 ** inline constructors are capable of dynamic extent allocation
436 (generally on x86 and x86-64, in some cases on other platforms
438 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
440 * optimization: file compiler is now able to coalesce non-circular
441 lists, non-base strings, and bit-vectors. Additionally, constants
442 are never referenced through SYMBOL-VALUE at runtime.
443 * optimization: code defining methods on PRINT-OBJECT (and other
444 generic functions in the COMMON-LISP package) now loads faster.
445 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
447 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
448 fasl is now deleted. (reported by Attila Lendvai)
449 * bug fix: READ-LINE always returned NIL for the last line in files.
450 (reported by Yoshinori Tahara)
451 * bug fix: more accurate disassembly annotations of foreign function
452 calls. (thanks to Andy Hefner)
453 * bug fix: trimming non-simple strings and non-string string
454 designators when the there is nothing to trim works properly.
455 (thanks to James Knight)
456 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
457 (thanks to Travis Cross)
458 * fixed some bugs revealed by Paul Dietz' test suite:
459 ** NIL is a valid function name (regression at 1.0.13.38)
460 ** FILL on lists was missing its return value (regression at 1.0.12.27)
461 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
462 fill pointers properly (regression at 1.0.12.23)
463 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
464 displacement indices properly (regression at 1.0.12.23)
466 changes in sbcl-1.0.17 relative to 1.0.16:
467 * temporary regression: user code can no longer allocate closure
468 variable storage on stack, due to bug 419 without explicitly
469 requesting it. Please consult sbcl-devel for advice if you need to
470 use this feature in the meanwhile.
471 * new feature: runtime argument --control-stack-size can be used to
472 adjust thread default control stack size.
473 * enhancement: improved TIME output
474 ** all times are reported using the measured accuracy (milliseconds
475 for real and GC times, microseconds for everything else.)
476 ** processor cycle counts on x86 and x86-64.
477 ** interpreted forms are counted for both evaluator modes.
478 ** number of lambdas converted by the compiler is reported.
479 ** CPU percentage report (computed from real and total run time.)
480 ** more comprehensive run time reporting, using a condenced format
481 ** interperted form, lambda, and page fault counts are omitted
483 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
485 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
486 * optimization: two argument forms of LAST are upto ~10% faster
487 in normal SPEED policies.
488 * optimization: NCONC no longer needs to heap cons its &REST list
489 in normal SPEED policies.
490 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
492 * bug fix: UNION and NUNION work with :TEST-NOT once more,
493 regression since 1.0.9.1. (thanks to Eric Marsden)
494 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
495 since 1.0.15.36. (thanks to Paul Khuong)
496 * bug fix: LAST when always returned the whole list when given a bignum
497 as the second argument.
498 * bug fix: dynamic extent allocation of nested lists and vectors
499 could leak to otherwise accessible parts.
500 * bug fix: invalid optimization of heap-allocated alien variable
502 * bug fix: fasl header checking is less vulnerable to different
503 platform word lengths.
504 * bug fix: more correct assembler syntax for GNU binutils
505 2.18.50.0.4 support. (thanks to Marijn Schouten)
506 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
508 * bug fix: the fopcompiler can handle LOCALLY forms (with no
509 declarations) successfully. (reported by Attila Lendvai)
511 changes in sbcl-1.0.16 relative to 1.0.15:
512 * minor incompatible change: revert the changes to sb-posix's error
513 signaling added in 1.0.14.
514 * minor incompatible change: change PROBE-FILE back to returning
515 NIL whenever we can't get a truename, as was the case before 1.0.14.
516 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
517 inherits from ERROR instead of just CONDITION.
518 * new feature: SB-INTROSPECT can provide source locations for instances
519 as well. (thanks to Tobias Rittweiler)
520 * optimization: binding special variables now generates smaller code
521 on threaded platforms.
522 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
523 and cases where no :TEST is given but the compiler can infer that
524 the element to search is of type (OR FIXNUM (NOT NUMBER)).
525 * optimization: better LOGNOT on fixnums.
526 * optimization: modular arithmetic for a particular requested width
527 is implemented using a tagged representation unless a better
528 representation is available.
529 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
530 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
531 for wellformedness and returns a specialized vector. (reported by
532 Francois-Rene Rideau)
533 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
534 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
535 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
536 unscheduling timer at the same time as another timer fires.
537 * bug fix: don't reschedule timers for dead threads.
538 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
539 * bug fix: copying output from RUN-PROGRAM to a stream signalled
540 bogus errors if select() was interrupted.
541 * enhancement: add support for fcntl's struct flock to SB-POSIX.
543 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
544 * enhancement: cleaner backtraces for interactive interrupts, as
545 well as other cases where the interesting frames used to be
546 obscured by interrupt handling frames.
547 * enhancement: untracing a whole package using (UNTRACE "FOO") is
548 now supported, and tracing a whole package using (TRACE "FOO") now
549 traces SETF-functions as well.
550 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
551 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
552 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
553 * unreadably printed representation of hash-tables now includes
555 * bug fix: partially fixed #188: type propagation from assignments
556 is now more efficient.
557 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
558 'DOUBLE-FLOAT) are not flushable.
559 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
560 full address of the object, and none of the tag bits.
561 * bug fix: readably printing hash-tables now respects other printer
562 control variables. (reported by Cedric St-Jean)
563 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
564 keyword with MAKE-HASH-TABLE.
565 * bug fix: export SB-POSIX:MKSTEMP.
566 * bug fix: SORT was not interrupt safe.
567 * bug fix: XREF accounts for the last node of each basic-block as
569 * bug fix: MAKE-INSTANCE optimizations interacted badly with
570 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
571 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
572 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
573 create obsolete instances in the case of redefinition or
574 obsoletion of a superclass. (thanks to Andy Hefner)
575 * bug fix: Support for the Alpha architecture has been revived; it had
576 suffered somewhat from lack of maintenance since sbcl-1.0.
577 * improvements to the Windows port:
578 ** The system detects the codepage to use at startup, instead of
579 using the value from the saved core. (thanks to Kei Suzuki)
581 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
582 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
583 (see documentation for details.)
584 * revived support for OpenBSD (contributed by Josh Elsasser)
585 * partially fixed bug #108: ROOM no longer suffers from occasional
586 (AVER (SAP= CURRENT END)) failures.
587 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
588 bodies no longer cause a WARNING to be signalled. (reported by
590 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
591 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
592 the bug in TRUNCATE.) (reported by Michael Weber)
593 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
594 no samples. (reported by Andy Hefner)
595 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
596 no longer appear as (NIL ...) frames in backtraces.
597 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
598 is now actually exported from SB-EXT as documented. (reported by
600 * bug fix: sb-aclrepl now correctly understands how to inspect
601 single-floats on 64-bit platforms where single-floats are not boxed.
602 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
603 yet been finalized. (reported by Levente Meszaros)
604 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
605 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
606 SBCL-specific optimize qualities.
608 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
609 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
610 an executable in the search path, and does so in the child
611 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
612 been removed; it can be found in the CVS history, for somebody who
613 needs that search behavior (see the manual).
614 * minor incompatible change: compiler policy re. weakening type
615 checks has changed: now type checks are weakened only if SAFETY < 2
617 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
618 unparsing of directory pathnames as files. Analogously,
619 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
620 filename to parse into a directory pathname.
621 * enhancement: implicit generic function creation now signals a
622 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
623 users can bind handlers for and muffle around calls to LOAD.
624 * enhancement: RUN-PROGRAM allows unicode arguments and environments
625 to be used (using the default stream external format), and allows
626 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
627 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
628 strings and vectors whose element-type or simplicity is not fully
629 known at compile-time.
630 * optimization: STRING-TRIM and related functions no longer allocate
631 a new string when no trimming needs to be performed. These functions
632 are also faster than before when the input string has been declared
634 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
635 * optimization: READ-LINE is significantly faster for files containing
637 * optimization: non-open coded uses of character comparison operators
638 (e.g. char=) no longer cons when called with more than one parameter
639 on platforms supporting dynamic-extent allocation.
640 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
641 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
642 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
644 * bug fix: some sequence functions elided bounds checking when
646 * bug fix: too liberal weakening of union-type checks when SPEED >
648 * bug fix: more bogus fixnum declarations in ROOM implementation
651 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
652 * new feature: MAKE-HASH-TABLE now experimentally accepts a
653 :SYNCHRONIZED argument, which makes the hash-table safe for
654 concurrent accesses (but not iteration.) See also:
655 SB-EXT:WITH-LOCKED-HASH-TABLE, and
656 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
657 * optimization: CONCATENATE on strings is an order of magnitude faster
658 in code compiled with (> SPEED SPACE).
659 * optimization: SUBSEQ is ~50% faster on lists.
660 * bug fix: bug 417 fixed -- source location reporting is now more robust.
661 * bug fix: SUBSEQ on a list will now correctly signal an error if if
662 END is smaller then START.
663 * bug fix: SB-PROFILE will no longer report extra consing for nested
664 calls to profiled functions.
665 * bug fix: ROOM implementation had bogus fixnum declarations which
666 could cause type-errors when calling ROOM on large images.
667 * bug fix: if file compilation is aborted, the partial fasl is now
668 deleted, and COMPILE-FILE returns NIL as the primary value.
669 * bug fix: number of thread safety issues relating to SBCL's internal
670 hash-table usage have been fixed.
671 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
672 be returned from its body when the values were being returned
673 using unknown-values return convection and the W-P-O was wrapped
674 inside an UNWIND-PROTECT.
675 * bug fix: sb-posix should now compile again under Windows, enabling
678 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
679 * incompatible change: hash-table accessor functions are no longer
680 automatically protected by locks. Concurrent accesses on the same hash-table
681 from multiple threads can give inconsistent results or even corrupt the
682 hash-table completely. Multi-threaded applications should do their own
683 locking at the correct granularity. In the current implementation it is
684 still safe to have multiple readers access the same table, but it's not
685 guaranteed that this property will be maintained in future releases.
686 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
687 and will signal an error at runtime.
688 * enhancement: SB-THREAD package now exports a semaphore interface.
689 * enhancement: CONS can now stack-allocate on x86 and
690 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
692 * enhancement: nested lists can now be stack allocated on
693 platforms providing stack allocation support.
694 * enhancement: dynamic-extent support has been extended to support
695 cases where there are multiple possible sources for the stack
697 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
698 if the mutex is uncontested on Linux.
699 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
700 as specified by the CLHS. (thanks to Tobias Rittweiler)
701 * bug fix: NaN comparison now works on x86-64.
702 * bug fix: CLOSE :ABORT T on a stream with pending output now
704 * bug fix: instances of non-standard metaclasses using standard
705 instance structure protocol sometimes missed the slot type checks
707 * bug fix: known functions can be inlined in high-debug code.
709 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
710 * minor incompatible change: the MSI installer on Windows no longer
711 associates .lisp and .fasl files with the installed SBCL.
712 * minor incompatible change: :UNIX is no longer present in *FEATURES*
713 on Windows. (thanks to Luis Oliviera)
714 * new feature: SB-CLTL2 contrib module now implements
715 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
716 * optimization: scavenging weak pointers is now more efficient,
717 requiring O(1) instead of O(N) per weak pointer to identify
718 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
719 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
720 method bodies using constant slot-names when the first argument is
721 a specializer parameter for the method.
722 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
723 names, when the class of the instance is a direct instance
724 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
725 as slow as the constant slot-name case.)
726 * optimization: member type construction is now O(N) instead
728 * optimization: UNION and NUNION are now O(N+M) for large
729 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
731 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
732 now more readable in environments like Slime which display it.
733 (thanks to Tobias C. Rittweiler)
734 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
735 was able to infer the correct argument type for the object on which
736 the CAS operation was being performed.
737 * bug fix: copy propagation interfered with parallel assignment
738 semantics in local calls. (reported by Paul Khuong)
739 * bug fix: the signed modular fixnum shift compiled to wrong code on
740 x86 and x86-64. (spotted by a slight modification to some of
743 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
744 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
745 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
746 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
748 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
749 and generic function now print the object identity for anonymous
750 instances. (thanks to Attila Lendvai)
751 * enhancement: as an extension to MOP, SBCL now supports SETF of
752 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
753 (thanks to Attila Lendvai)
754 * enhancement: sb-cover has an alternate annotation mode which puts
755 the form annotations at the car of the form, instead of annotating
756 it completely. The benefit of this mode is that it shows better
757 which forms were instrumented by the compiler.
758 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
759 by printer control variables. (reported by Dan Corkill)
760 * bug fix: system leaked memory when delayed output was performed by
761 the OS in smaller chunks then expected. (thanks to David Smith)
762 * bug fix: system leaked memory when file streams were not closed
764 * bug fix: large objects written to slow streams that were modified
765 after the write could end up with the modified state written to
766 the underlying file descriptor.
767 * bug fix: multiple threads operating in parallel on the same stream
768 could cause buffer-overflows.
769 * bug fix: source location information is stored correctly
770 (broken since 1.0.6). This bug would generally show up as the
771 Slime debugger highlighting the wrong form when the sldb-show-source
773 * bug fix: Branch forms are again annotated as branches in the sb-cover
775 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
776 on x86-64. (reported by Gregory Vanuxem)
777 * bug fix: EQUALP could return wrong results for structures with raw
778 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
779 word). (reported by Vjacheslav Fyodorov)
780 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
783 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
784 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
785 atomic compare-and-swap operations on threaded platforms.
786 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
787 allows assigning a global minimum value to optimization qualities
788 (overriding proclamations and declarations).
789 * enhancement: closed over variables can be stack-allocated on x86
791 * performance bug fix: GETHASH and (SETF GETHASH) are once again
793 * optimization: slot definition lookup is now O(1). This speeds up
794 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
795 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
796 * optimization: ASSOC and MEMBER can now be open-coded for all
797 combinations of keyword arguments when second argument is constant
798 and SPEED >= SPACE. In other cases a specialized version is
800 * bug fix: using obsoleted structure instances with TYPEP and
801 generic functions now signals a sensible error.
802 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
803 (reported by Kristoffer Kvello)
804 * bug fix: backtrace construction is now more careful when making
805 lisp-objects from pointers on the stack, to avoid creating bogus
806 objects that can be seen by the GC.
807 * bug fix: defaulting of values in contexts expecting more than 7
808 variables now works on x86-64. (reported by Christopher Laux)
809 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
811 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
812 as the property-list of a symbol.
813 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
814 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
815 situations requested, are once again file-compileable. (reported
818 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
819 * MOP improvement: support for user-defined subclasses of
820 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
821 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
822 the expansion of DEFMETHOD, is responsible for generating a form
823 which creates a list of specializers when evaluated. Additional
824 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
825 debugging and introspective support.
826 * minor incompatible change: the (unsupported) spinlock interface
827 has changed: free spinlock now has the value NIL, and a held spinlock
828 has the owning thread as its value.
829 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
830 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
832 * enhancement: name of a socket-stream is now "a socket" instead of
834 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
835 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
836 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
837 for generic functions. (Thanks to Tobias C. Rittweiler)
838 * optimization: bignum printing speed has been improved by 20-40%
839 (depending on the bignum size.)
840 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
842 * bug fix: the cache used by the CLOS to store precomputed effective
843 methods, slot offsets, and constant return values is now thread and
845 * bug fix: generic function dispatch function updating is now thread
846 and interrupt safe (in the sense that the known issues have been
848 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
849 * bug fix: interning EQL-specializers is now thread and interrupt safe.
850 * bug fix: asdf systems with dependencies to the SB-POSIX or
851 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
852 * bug fix: interrupt safety of applicable method computation has been
855 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
856 * new contrib: sb-cover, an experimental code coverage tool, is included
858 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
859 significantly faster.
860 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
861 produce more efficient code on x86-64 if the compiler can determine
862 that (+ INDEX <constant>) does not require a bounds check and FOO
863 has an element type at least 8 bits wide.
864 * enhancement: a new, experimental synchronous timeout facility is
865 provided. Refer to SB-SYS:WITH-DEADLINE for details.
866 * enhancement: when a symbol name conflict error arises, the
867 conflicting symbols are always printed with a package prefix.
868 (thanks to Kevin Reid)
869 * enhancement: stepping is now once again supported on the SPARC and
870 MIPS platforms. (It is also now more likely to work on CheneyGC
872 * enhancement: sb-sprof can now also track and report accurate call
874 * bug fixes: the treatment of non-standard subclasses of
875 SB-MOP:SPECIALIZER is more correct.
876 * incompatible change: PURIFY no longer copies the data from the
877 dynamic space into the static and read-only spaces on platforms
878 that use the generational garbage collector
879 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
881 * bug fix: GC race condition occasionally resulting in crashes with
882 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
884 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
886 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
887 a STYLE-WARNING to be signalled (regression from 1.0.4.)
888 * bug fix: an asynchronous interrupt could previously leave the
889 system running with GC inhibited.
890 * bug fix: a DECLARE form evaluated at top-level now causes an error
891 rather than silently (or verbosely) returning NIL.
892 * bug fix: trying to dissassemble functions compiled at high DEBUG could
893 result in a error being signalled due to source form lookup errors.
894 (reported by Peter Graves)
896 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
897 * incompatible change: removed writer methods for host-ent-name,
898 host-ent-addresses -- changing the values did not update the DNS
900 * minor incompatible change: changed experimental JOIN-THREAD interface
901 * documentation: the manual now lists reader and writer methods
902 in class slot documentation sections. (thanks to Richard M Kreuter)
903 * documentation: unwinding from asyncronous events has been
904 documented as unsafe.
905 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
906 in multithreaded application code.
907 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
908 platforms. (thanks to James Anderson for the optimization hint)
909 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
911 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
912 variants no longer cons.
913 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
914 their NOT- variants no longer cons.
915 * optimization: EQUAL hash tables no longer use SXHASH for objects
916 of all data types, but instead use an EQL hash for types for which
917 EQUAL is the same as EQL.
918 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
919 are significantly faster.
920 * optimization: new STRING-OUTPUT-STREAM implementation is much
921 faster for large outputs and conses 30% less on average.
922 * enhancement: XREF information is now collected for references made
923 to global variables using SYMBOL-VALUE and a constant argument.
924 * enhancement: SIGINT now causes a specific condition
925 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
926 * bug fix: FILE-POSITION used to signal an error in some cases where
927 ANSI requires it to return NIL.
928 * bug fix: ADJUST-ARRAY is now interrupt-safe.
929 * bug fix: adding and removing fd-handlers is now interrupt-safe.
930 * bug fix: inlined calls to C now ensure 16byte stack alignment on
932 * bug fix: bad type declaration in the CLOS implementation has
933 been fixed. (reported by James Anderson)
934 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
935 (thanks to Lutz Euler)
936 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
937 required by the CLHS. (reported by Samium Gromoff)
938 * bug fix: dead unbound variable references now signal an error.
939 (reported by Marco Monteiro)
940 * bug fix: / with an unused value was being deleted in safe code.
941 (thanks to Marco Monteiro and Kevin Reid)
942 * bug fix: number of characters that can be written onto a single
943 line in a file is unlimited.
944 * bug fix: some GC deadlocks caused by asynchronous interrupts have
945 been fixed by inhibiting interrupts for when GC is disabled.
946 * bug fix: some interrupt safety issues with GETHASH, (SETF
947 GETHASH), CLRHASH and REMHASH have been fixed.
948 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
949 specifier no longer causes infinite recursion.
950 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
951 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
952 * bug fix: an &environment argument with the correct variable information
953 is passed to macros that are expanded during byte compilation.
954 (reported by Samium Gromoff)
955 * bug fix: a logic error was causing the hash function for bignums to
956 have excessive amounts of collisions. (reported by Faré Rideau)
957 * bug fix: modifying the contents of an array could change the return
958 value of SXHASH on that array, which is only allowed for strings
959 and bit vectors. (bug introduced in 0.9.16)
960 * improvement: the x86-64/darwin port now passes all tests (except
961 for the debugger tests) but should still be considered
962 experimental until this is fixed.
963 * improvement: a style-warning is signaled for CASE (etc) clauses with
964 duplicate keys. (thanks to Kevin Reid)
965 * improvement: macos/x86 and macos/x86-64 now use mach exception
966 handlers for dealing with illegal instructions (for trapping and
967 error handling) and memory protection violations (for GC).
969 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
970 * new platform: experimental support for x86-64/darwin (MacOS).
971 * incompatible change: the thread-safe (on most platforms) getaddrinfo
972 and getnameinfo sockets functions are used instead of gethostbyaddr
973 and gethostbyname, on platforms where the newer functions are available.
974 As a result, the ALIASES field of HOST-ENT will always be NIL on these
976 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
977 don't print the startup banner, but behave as if --noinform was passed
978 as a command line argument. (thanks to Kevin Reid)
979 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
980 * optimization: code using alien values with undeclared types is much faster.
981 * optimization: the compiler is now able to open code SEARCH in more cases.
982 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
983 * bug fix: using standardized COMMON-LISP special variables as loop
984 variables no longer signals bogus package lock violations. (reported
986 * bug fix: declaring local loop variables to be of a range-limited type
987 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
988 (reported by Andras Simon)
989 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
990 bugs remain on x86-64.)
991 * bug fix: the #= and ## reader macros now interact reasonably with
992 funcallable instances.
993 * bug fix: type-checks for function arguments were compiled using the
994 compiler policy settings of the wrong lexical environment (in 1.0.2
996 * bug fix: SHADOW accepts characters as string designators, as required
997 by the spec (thanks to Eric Marsden)
998 * bug fix: fixed GC safety issues when foreign functions are called with
999 non-base strings as arguments
1000 * bug fix: more consistent error output for fatal-compiler-warnings, like
1002 * bug fix: functions with proper names were showing up as NIL in
1003 backtraces (reported by Edi Weitz, regression in 1.0.2)
1005 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1006 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1007 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1008 produce more efficient code on the x86 if the compiler can determine
1009 that (+ INDEX <constant>) does not require a bounds check and FOO
1010 has an element type at least 8 bits wide.
1011 * bug fix: references to undefined variables are handled the same way
1012 in toplevel forms as in the normal compiler
1013 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1015 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1016 (thanks to Jon Buller)
1017 * improvement: faster compilation times for complex functions
1018 * improvement: added readlink support to SB-POSIX (thanks to Richard
1021 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1022 * improvement: experimental support for mach exception handling on
1023 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1025 * improvement: support for GBK external format.
1026 (thanks to Chun Tian (binghe))
1027 * improvement: the debugger now displays variables that have been closed
1028 over, in code compiled with (DEBUG 2) or higher
1029 * improvement: support for executable cores on NetBSD (thanks to
1031 * new feature: added a RESTART-FRAME debugger command
1032 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1033 be used to provide an implementation for FILE-POSITION on Gray streams
1034 (thanks to Eric Marsden)
1035 * new feature: add syslog, openlog and closelog support to SB-POSIX
1036 (thanks to Richard Kreuter)
1037 * optimization: the function call overhead in code compiled with
1038 a high DEBUG optimization setting is significantly
1039 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1040 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1041 (thanks to Tony Martinez)
1042 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1043 evaluated once (reported by Kevin Reid)
1044 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1045 works again (thanks to Kilian Sprotte)
1046 * bug fix: an error is signaled for tagbodies with duplicate tags
1047 (thanks to Stephen Wilson)
1048 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1049 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1050 stack frames from alien callbacks.
1051 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1052 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1053 65k pages (thanks to David Woodhouse)
1054 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1056 changes in sbcl-1.0.1 relative to sbcl-1.0:
1057 * new platform: FreeBSD/x86-64, including support for threading.
1058 * new feature: the compiler stores cross-referencing information
1059 about function calls (who-calls), macroexpansion (who-macroexpands)
1060 and special variables (who-binds, who-sets, who-references) for code
1061 compiled with (< SPACE 3). This information is available through the
1062 sb-introspect contrib.
1063 * new feature: users may subclass SEQUENCE, and have instances of
1064 these classes interoperate with standard Common Lisp functions if
1065 a number of methods are defined. (This feature is experimental
1066 and the interface subject to change based on feedback from SBCL
1067 users and the general community)
1068 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1069 x86/x86-64, rather than the previous fixed depth of 8
1070 * improvement: another pthread back-end of mutex "pthread-futex".
1071 * bug fix: non-ascii command-line arguments are processed correctly
1072 (thanks to Yaroslav Kavenchuk)
1073 * bug fix: non-required arguments were not passed correctly when a method
1074 defined using DEFMETHOD was called from a mop-generated method using
1075 CALL-NEXT-METHOD (reported by Pascal Costanza)
1076 * bug fix: recursion is now permitted in accessors through
1077 SLOT-UNBOUND. (reported by Pascal Costanza)
1078 * bug fix: an error was signaled at startup if the HOME environment
1079 variable was defined, but had an empty value (reported by Peter Van Eynde)
1080 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1081 signal a PROGRAM-ERROR, not a BUG.
1082 * bug fix: SB-POSIX:READDIR works when built with large file support.
1083 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1085 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1086 for better automatic indentation support. (thanks to Matt Pillsbury)
1087 * optimization: loading generic functions no longer takes O(n^2) time,
1088 proportional to the amount of methods in the generic function
1089 (reported by Todd Sabin and Jeremy Brown)
1090 * optimization: the FIND and POSITION family of sequence functions
1091 are significantly faster on arrays whose element types have been
1093 * improvements to the Windows port:
1094 ** Intermittent heap corruption problems have been fixed. (thanks
1095 to Alastair Bridgewater)
1096 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1098 ** Lisp is able to unwind foreign exception frames from alien
1099 callbacks. (thanks to Alastair Bridgewater)
1101 changes in sbcl-1.0 relative to sbcl-0.9.18:
1102 * improvement: experimental support for threading on FreeBSD/x86.
1103 (thanks to NIIMI Satoshi)
1104 * improvement: runtime option --dynamic-space-size can be used
1105 to set the size of the dynamic space reserved on startup.
1106 * improvement: floating point modes in effect are now saved in
1107 core, and restored on startup.
1108 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1109 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1110 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1111 threads are not running after *SAVE-HOOKS* have run.
1112 * improvement: writes to CLOS instance slots are type-checked in code
1113 compiled with (SAFETY 3)
1114 * improvement: floating-point exception handling on FreeBSD (thanks to
1116 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1117 (thanks to Zach Beane)
1118 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1120 * improvement: added support for the Shift-JIS external format.
1121 (contributed by NIIMI Satoshi)
1122 * improvement: callbacks are supported on Linux/PPC. (thanks to
1124 * bug fix: compiler bug triggered by a (non-standard) VALUES
1125 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1126 * bug fix: file compiler no longer confuses validated and already
1127 dumped structurres. (reported by Kaersten Poeck)
1128 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1129 fill-pointer signals a type-error as required. (thanks to
1131 * bug fix: disassemly of funcallable instances works.
1132 * bug fix: single stepping on PPC.
1133 * bug fix: fix thread-safety problems in the type system (generally
1134 manifesting as nonsensical errors like "STRING is a bad type specifier
1135 for sequences" or "The value 1 is not of type FIXNUM")
1136 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1137 (reported by Josip Gracin)
1138 * bug fix: an error is signaled for attempts to displace arrays with
1139 incompatible element types (thanks to Mario Mommer)
1140 * bug fix: more correct handling of wide characters in debug info
1141 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1142 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1143 and don't cause extra consing
1144 * optimization: MAP and MAP-INTO are significantly faster on vectors
1145 whose elements types have been declared.
1146 * Improvements to SB-SPROF:
1147 ** Support for allocation profiling
1148 ** Reduced profiling overhead, especially for long profiling runs
1149 * Improvements to the Windows port:
1150 ** floating point exceptions are now reported correctly.
1151 ** stack exhaustion detection works partially.
1152 ** more accurate GET-INTERNAL-REAL-TIME.
1153 ** better RUN-PROGRAM behaviour in case of errors in the child
1155 ** PROBE-FILE now simplifies pathnames correctly.
1156 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1157 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1158 1st 1970 (although time timezone data will be unreliable).
1159 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1161 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1162 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1163 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1165 * bug fix: two potential GC deadlocks affecting threaded builds.
1166 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1167 returns T (reported by Anton Kazennikov)
1168 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1169 with non-variable places
1170 * bug fix: remove a race condition in the setting of
1171 funcallable-instance functions, this should make threaded CLOS
1172 code more stable against memory faults.
1173 * bug fix: corruption of specials when unbinding is interrupted by an
1174 asynchronous unwind (reported by Hannu Koivisto)
1175 * improvement: the debugger will now also display local variables that
1176 are only used once, for code compiled with a DEBUG optimization quality
1179 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1180 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1181 * incompatible change: External-format support for FFI calls. The
1182 SB-ALIEN:C-STRING no longer implies an ASCII
1183 external-format. Instead the string is subject to external-format
1184 conversion. Additionally return values of type SB-ALIEN:C-STRING
1185 are by default of type (SIMPLE-ARRAY CHARACTER), not
1186 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1187 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1188 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1189 * incompatible change: SB-EXT package no longer contains the
1190 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1191 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1193 * incompatible change: the single-stepper is no longer available
1194 on Alpha, Mips and Sparc platforms.
1195 * minor incompatible change: the direct superclasses of
1196 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1197 not (STANDARD-OBJECT FUNCTION). This makes the
1198 class-precedence-lists of GENERIC-FUNCTION and
1199 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1201 * minor incompatible change: the default stream external format on
1202 non-unicode SBCL is no longer detected from the locale, but is always
1204 * new feature: Add a version of evaluator that uses an interpreter instead
1205 of the compiler. EVAL still uses the compiler by default, to switch it
1206 to use the interpreter, set the value of the variable
1207 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1208 * minor incompatible change: the single-stepper REPL has been merged
1209 with the normal debugger (see the "Stepping" heading of the debugger help
1210 for more details). The debugger command STEP will no longer switch
1211 to the single-stepper REPL.
1212 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1213 for a type now works.
1214 * bug fix: loading of default sysinit file works. (thanks to Leonid
1216 * bug fix: better detection of circularities in the file-compiler.
1217 (reported by Marco Monteiro)
1218 * bug fix: the CL pathname functions now work with files that have
1219 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1220 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1221 systems with Unix98 pty semantics.
1222 * bug fix: ASDF-INSTALL will now work with bsd tar.
1223 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1225 * bug fix: timers expiring in dead threads no longer cause a
1226 type-error (reported by Paul "Nonny Mouse").
1227 * bug fix: thanks to more lightweight single-stepper instrumentation,
1228 code compiled with (DEBUG 3) will compile and execute significantly faster,
1229 and will have more accurate type-inferencing than before
1230 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1231 whose bindings are modified
1232 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1233 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1234 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1235 CL:SOFTWARE-VERSION with proper version
1237 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1238 * feature: implemented the READER-METHOD-CLASS and
1239 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1240 as specified by AMOP.
1241 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1243 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1244 (thanks to Lutz Euler)
1245 * optimization: hashing of general arrays and vectors has been
1246 improved. (reported by Any Fingerhut)
1247 * enhancement: SB-INTROSPECT is now able to find definitions of
1248 profiled functions. (thanks to Troels Henriksen)
1249 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1250 as well, allowing compiler-macros for SETF-functions to expand.
1251 * enhancement: step-instrumentation no longer wraps known
1252 single-value functions with multiple-value context, allowing
1253 better type inference.
1254 * fixed bug #337: use of MAKE-METHOD in method combination now works
1255 even in the presence of user-defined method classes. (reported by
1256 Bruno Haible and Pascal Costanza)
1257 * fixed bug #339(c): if there are applicable methods not part of any
1258 long-form method-combination group, call INVALID-METHOD-ERROR.
1259 (reported by Bruno Haible)
1260 * fixed bug #361: the :FUNCTION initarg in the protocol for
1261 initialization of methods can now be used to override
1262 internally-produced optimized functions. (reported by Bruno
1264 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1265 system-provided lambda expression no longer cause warnings about
1266 unbound #:|pv-table| symbols.
1267 * bug fix: improved the handling of type declarations and the
1268 detection of violations for keyword arguments with non-constant
1270 * bug fix: potentially erronous calls to PATHNAME and
1271 MERGE-PATHNAMES were being flushed in some cases.
1272 (reported by Richard Kreuter)
1273 * bug fix: compiled calls to TYPEP were mishandling obsolete
1274 instances. (reported by James Bielman and Attila Lendvai)
1275 * bug fix: format strings with ~> without matching ~< no longer
1276 trigger an AVER, but signal an understandable error instead.
1277 (reported by Antonio Martinez)
1278 * bug fix: specifying an output-file without a pathname-type for
1279 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1280 of input-file instead of "fasl". (reported by Robert Dodier)
1281 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1282 (reported by James Y Knight).
1283 * bug fix: compiler-macros-function did not consider the environment
1284 argument for shadowing by local functions.
1285 * bug fix: compiler-macros expansion was inhibited by local INLINE
1287 * bug fix: inline expansions of known functions were subject to
1288 step-instrumentation in high debug policies, leading to problems
1289 with type-inference.
1290 * bug fix: compiler failed to differentiate between different CONS
1291 types in some cases.
1292 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1293 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1294 * bug fix: #368: incorrect use of expressed vs. upgraded array
1296 * bug fix: #306a: more precise unions of array types.
1297 * thread-safety improvements:
1298 ** CONDITION-WAIT could return early on Linux, if the thread was
1299 interrupted and subsequently continued with SIGCONT.
1300 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1302 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1303 * added support for the ucs-2 external format. (contributed by Ivan
1305 * minor incompatible change: pretty printing of objects of type
1306 (cons symbol) is, in the default pprint-dispatch-table, now
1307 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1309 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1310 called later in a class's lifetime, possibly as late as when the
1311 first instance of the class is created. Previously,
1312 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1313 class became finalizeable.
1314 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1315 for multibyte external-format streams. (thanks to "vbzoli")
1316 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1317 with :ALLOCATION :CLASS inherited from superclasses of the
1319 * fixed bug: anonymous classes can now be created using the :NAME
1320 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1321 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1322 * fixed bug: core-files saved with :EXECUTABLE T can again be
1323 executed when SBCL_HOME isn't set. (reported by James Knight)
1324 * fixed bug: toplevel LOCALLY forms with declarations could
1325 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1326 * fixed bug: printing from several different threads using different
1327 values of *print-case* could cause invalid output, due to
1328 some internal special variables of the printer not being bound
1329 thread-locally (reported by Max Mikhanosha)
1330 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1331 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1332 (reported by Pascal Costanza)
1333 * fixed bug: FINALIZE-INHERITANCE is called from
1334 REINITIALIZE-INSTANCE on classes when the class has previously
1335 been finalized, as required by AMOP.
1336 * minor code generation optimizations:
1337 ** better register allocation in CLOS dispatching functions
1338 ** overflow detection when coercing signed bytes to fixnums on x86-64
1339 ** is now implemented with one IMUL instruction instead of three shifts
1340 ** more efficient bit-vector access on x86 and x86-64
1341 ** more efficient access to raw structure slots on x86 and x86-64
1342 * fixed some bugs revealed by Paul Dietz' test suite:
1343 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1344 return its argument.
1346 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1347 * feature: thread support on Solaris/x86, and experimental thread support
1349 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1351 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1352 default method for SLOT-UNBOUND.
1353 * minor incompatible change: prevent the user from specializing the
1354 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1355 somewhat counter to the intent of the protocol, I (CSR) think, and
1356 additionally it just doesn't work in SBCL as currently
1357 implemented, thanks to optimizations (that are always valid for
1358 the other three SLOT-VALUEish functions, but not for the setter).
1359 * minor incompatible change: the :SB-LDB feature is now enabled by
1360 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1361 the low-level debugger.
1362 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1363 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1364 down to an SB-SYS:FD-STREAM, allowing direct communication in
1365 more cases. Temporary files and pipes are still used as a fallback
1367 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1368 on the non-reentrant C functions localtime(3) and gmtime(3).
1369 * bug fix: global optimization policy was not visible in LOCALLY and
1371 * bug fix: class objects can be used as specializers in methods.
1372 (reported by Pascal Costanza)
1373 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1374 with a type error. (reported by blitz_ on #lisp)
1375 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1376 failed with a type error. (reported by Pascal Bourguignon)
1377 * bug fix: merging pathnames against defaults with :DIRECTORY
1378 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1379 (reported by James Y Knight)
1380 * bug fix: saving large (>2GB) cores on x86-64 now works
1381 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1382 constant index and a (simple-array (signed-byte 32)) array
1383 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1384 error (patch by Robert J. Macomber)
1385 * feature: TIME now displays the amount of run-time spent in GC
1386 * bug fix: The debugger now does a better job of respecting
1387 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1388 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1389 workaround for bug 403.)
1390 * bug fix: floating point exception handling now works on FreeBSD
1391 (thanks to NIIMI Satoshi)
1392 * fixed some bugs revealed by Paul Dietz' test suite:
1393 ** MISC.641: LET-conversion were not supposed to work in late
1395 * improvements to the Win32/x86 port:
1396 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1397 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1399 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1400 * new feature: source path information is generated for macro-expansion
1401 errors for use in IDE's like Slime (thanks to Helmut Eller)
1402 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1403 * bug fix: compiler does not loop forever on an invalid type in
1405 * improvement: compilation of most CLOS applications is significantly
1407 * optimization: added a limited bytecode compiler for simple toplevel
1408 forms, speeding up compilation and FASL loading
1409 * bug fix: the statistical profiler now properly distinguishes anonymous
1412 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1413 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1414 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1415 into subclasses of the finalized class. Now user calls to
1416 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1417 system return before any subclasses are finalized.
1418 * minor incompatible change: The reader no longer ignores errors
1419 regarding non-existent packages in #+ and #- feature tests.
1420 * new feature: command line options --no-sysinit, --no-userinit to
1421 inhibit loading the corresponding init files
1422 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1423 for sending data through UDP sockets (thanks to François-René Rideau)
1424 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1425 error is signalled instead (thanks to François-René Rideau)
1426 * minor incompatible change: Error signalling behaviour of lexical
1427 operations violating package locks has changed slightly. Refer to
1428 documentation on package locks for details.
1429 * enhancement: EVAL can process IF-expressions without resorting to the
1431 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1432 constant-folded properly due to missing out-of-line functions.
1433 (reported by Mika Pihlajamäki)
1434 * bug fix: LISTEN sometimes returned T even in cases where no data was
1435 immediately available from the stream
1436 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1437 were reversed. (reported by Levente Mészáros)
1438 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1439 when given a vector displaced to an adjustable array. (reported
1441 * bug fix: derivation of float boundaries from numbers outside the
1442 appropriate float range (reported by John Wiseman)
1443 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1445 * bug fix: precomputing cache entries for generic functions with
1446 some subclasses of specializers as yet invalid does not attempt to
1447 fill a cache line with a negative offset. (reported by Levente
1449 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1450 allow condition class objects as arguments to DOCUMENTATION and
1451 (SETF DOCUMENTATION); only find and set documentation for
1452 structure names for the STRUCTURE doc-type. (suggested by Gary
1454 * improvements to the Win32/x86 port:
1455 ** (user-homedir-pathname) and default initialization file
1456 locations now know about the user's "Documents and Settings"
1457 directory (thanks to Yaroslav Kavenchuk)
1458 ** run-program is implemented (thanks to Mike Thomas)
1459 ** sockets support (thanks to Timothy Ritchey)
1460 ** better backtrace support (thanks to Alastair Bridgewater)
1461 ** sb-grovel supported
1462 ** asdf-install and sb-posix work somewhat
1463 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1464 * improvements to the Solaris/x86 port:
1465 ** works on Solaris 11/Solaris Express
1466 ** floating-point exception handling support
1467 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1468 * fixed some bugs revealed by Paul Dietz' test suite:
1469 ** REMOVE-METHOD returns its generic function argument even when
1470 the method is not one of the generic functions' methods.
1471 ** objects declared as MEMBER types can be admissible arguments to
1472 structure accessors.
1473 ** printing characters should simply be printed by the FORMAT ~:C
1475 ** compiler failure when compiling functions with hairy constant
1476 defaults for optional parameters.
1477 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1478 function, which is already optimized.
1480 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1481 * new platform: experimental support for SBCL x86/Darwin, including
1482 MacOS X 10.4.5 on Intel.
1483 * new feature: Unicode character names are now known to the system
1484 (through CHAR-NAME and NAME-CHAR).
1485 * new feature: the filesystem location of SBCL's core file is
1486 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1487 * minor incompatible change: the contrib modules SB-POSIX and
1488 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1489 this change is to make it easier to distribute
1490 location-independent binaries.
1491 * bug fix: as implied by AMOP, standardized classes no longer have
1492 slots named by external symbols of public packages. (reported by
1494 * optimization: calling functions via constant symbols -- as in
1495 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1496 via the function object as in (FUNCALL #'FOO).
1497 * optimization: on x86 and x86-64, the calling convention for the
1498 case of dealing with an unknown number of values has been altered
1499 to be friendlier to the prediction heuristics implemented,
1500 particularly with reference to CALL and RET pairing. (thanks to
1501 Alastair Bridgewater)
1502 * optimization: on x86, the code for access to thread-local symbol
1503 values and binding/unbinding of thread-local symbols is smaller.
1504 (thanks to Alastair Bridgewater)
1505 * enhancement: CONSTANTP is now able to determine constantness of
1506 more complex forms, including calls to constant-foldable standardized
1507 functions and some special forms beyond QUOTE.
1508 * fixed bug: occasional GC crashes on Solaris/x86
1509 * optimization: x86-64 supports stack allocation of results of simple
1510 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1511 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1512 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1513 (thanks to James Knight)
1514 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1515 by the standard (thanks to Stephen Compall)
1517 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1518 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1519 be used for bundling the runtime and the core file into one
1520 executable binary. This feature is not currently supported on all SBCL
1521 platforms. (thanks to James Bielman and NIIMI Satoshi)
1522 * new feature: a generational or ephemeral garbage collector is now
1523 the default on the PowerPC platform (both Linux and Darwin). The
1524 old Cheney (stop and copy) collector is a build-time option.
1525 (thanks to Cyrus Harmon, after Raymond Toy)
1526 * minor incompatible change: the method by which SBCL finds its
1527 contributed modules has changed; it no longer relies on symbolic
1528 links from an $SBCL_HOME/systems directory, but searches directly
1529 in the subdirectories of $SBCL_HOME.
1530 * enhancement: the dynamic heap size on the Linux/PPC platform is
1531 markedly larger, even using the older Cheney garbage collector.
1532 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1533 certain complicated string types. (reported by Gary King)
1534 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1535 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1536 * fixed bug: DOCUMENTATION on structures no longer signals an
1537 error if no structure type of the right name exists. (reported by
1539 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1540 applicable non-standard primary or :AROUND method on
1541 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1542 no longer cause unbound variable errors. (reported by Kevin Reid)
1543 * fixed bug: metacircle resolution in cases where methods have slots
1544 added before the slots from STANDARD-METHOD. (reported by Jean
1546 * fixed bug: the Power PC assembly code for calling into lisp
1547 sometimes computed its return address wrongly (depending on the
1548 15th bit of the address, fixed at link time). This used to
1549 manifest itself in a segmentation violation while building PCL.
1550 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1551 many others over the years)
1552 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1553 the stack pointer (treated as a signed integer) is negative.
1554 (thanks to Peter van Eynde)
1556 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1557 * new platform: experimental support for the Windows operating
1558 system has been added. (thanks to Alastair Bridgewater)
1559 * fixed several bugs in and robustified the PPC FFI (including
1560 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1561 * bug fix: allow non-simple string symbol names (reported by Paul
1563 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1565 * bug fix: saving a core corrupted callbacks on x86/x86-64
1566 * bug fix: closed a loophole in metacircularity detection and
1567 grounding in the PCL implementation of CLOS.
1568 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1569 (reported by tomppa on #lisp)
1570 * bug fix: clear the direction flag on Lisp -> C transitions, as
1571 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1572 (reported by Andrej Grozin and Hendrik Maryns)
1573 * optimization: major improvements to GC efficiency on GENCGC platforms
1574 * optimization: faster implementation of EQUAL
1575 * optimization: emit more efficient opcodes for some common
1576 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1577 * optimization: several other minor code-generation improvements on x86-64
1578 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1580 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1581 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1582 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1583 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1584 by ANSI as a generic function, and for consistency (SETF
1585 GENERIC-FUNCTION-NAME) is treated likewise.
1586 * fixed bug #233.b: make constraint propagation notice when a variable
1587 value is changed after it is referenced but before it is used
1588 * fixed bug #296: no more arbitrary behaviour from filenames with
1589 odd characters as --load arguments. (reported by Adam Warner)
1590 * fixed bug #390: :CHARACTER-SET pathname components now work as
1591 expected. (reported by Tim Daly Jr)
1592 * fixed bug #391: complicated :TYPE intersections in slot
1593 definitions no longer cause an error in PCL internals.
1594 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1595 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1596 returning the number of octets which would be written to the
1597 file-stream. (thanks to Robert J. Macomber)
1598 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1599 arguments to the internal stream functions. (thanks to David
1601 * bug fix: stack exhaustion is now handled on the x86/SunOS
1602 platform. (thanks to Daisuke Homma)
1603 * bug fix: more accurate ROOM results on GENCGC platforms
1604 * optimization: improved type inference for arithmetic-for
1605 index variables in LOOP
1606 * optimization: faster floating-point SQRT on x86-64
1607 * fixed some bugs revealed by Paul Dietz' test suite:
1608 ** DOCUMENTATION returns NIL instead of "" for method combinations
1609 that don't have a docstring
1611 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1612 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1613 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1614 therefore are not customizeable by user code (as seems to be at
1615 least permitted and maybe required by AMOP). As a consolation,
1616 however, the SBCL implementation of these functions now calls
1617 REINITIALIZE-INSTANCE as specified by AMOP.
1618 * bug fix: slot-definition documentation is propagated as per ANSI
1619 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1620 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1621 argument, as do eslotd initialization methods. (from Pascal
1622 Costanza's "Closer" project)
1623 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1624 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1626 * bug fix: it is now possible to have more than one subclass of
1627 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1628 (reported by Bruno Haible, Pascal Costanza and others)
1629 * bug fix: the dependent update protocol now works for generic
1630 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1631 and Pascal Costanza)
1632 * bug fix: condition-class instances corresponding to
1633 DEFINE-CONDITION forms are now created eagerly. (reported by
1634 Kalle Olavi Niemitalo on comp.lang.lisp)
1635 * bug fix: floating point printing is more accurate in some
1636 circumstances. (thanks to Simon Alexander)
1637 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1638 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1639 * bug fix: callbacks on OS X now preserve stack-alignment.
1640 (thanks to Cyrus Harmon)
1641 * enhancement: the x86-64 disassembler is much better at
1642 disassembling SSE instructions. (thanks to Lutz Euler)
1643 * enhancement: CHECK-TYPE on a variable now allows type-inference
1644 (thanks to Vincent Arkesteijn)
1645 * optimization: improved performance of EUC-JP external format.
1646 (thanks to NIIMI Satoshi)
1647 * optimization: performance improvements to IO on file streams of
1648 :ELEMENT-TYPE CHARACTER
1649 * optimization: much faster memory allocation on x86-64
1650 * optimization: faster garbage collections (latency and throughput)
1652 * optimization: faster bignum subtraction and division on x86 and x86-64
1653 (thanks to Lutz Euler)
1654 * optimization: more accurate type-inference for loops that use a
1655 floating point index variable or a negative step.
1657 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1658 * bug fix: add a workaround to SBCL looping infinitely at startup on
1659 Linux kernels with apparently buggy implementations of personality().
1660 (thanks to Svein Ove Aas)
1661 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1662 (thanks to David Lichteblau)
1663 * bug fix: local bindings shadow global symbol macros for
1664 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1665 * optimization: non-open coded uses of numeric comparison operators
1666 (e.g. >) no longer cons when called with more than one parameter
1667 on platforms supporting dynamic-extent allocation.
1668 * enhancement: saving cores with foreign code loaded is now supported
1669 on MIPS/Linux in addition to the previously supported platforms.
1670 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1671 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1672 * bug fix: division by zero in sb-sprof when no samples were collected
1673 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1674 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1675 (thanks to Alexander Kjeldaas)
1676 * new feature: ignoring the return values of destructive functions
1677 like NREVERSE signals a compile-time style-warning.
1678 (thanks to Kevin Reid)
1679 * new feature: the alignment of alien structure fields can be
1680 explicitly specified. (thanks to Cyrus Harmon)
1681 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1682 and dump core on SIGQUIT
1684 ** incompatible change: threads do not inherit values of specials
1685 from their parents (see manual)
1686 ** bug fix: threads stacks belonging to dead threads are freed by the
1687 next exiting thread, no need to gc to collect thread stacks anymore
1688 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1689 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1690 ** bug fix: dynamic variable and thread start related gc lossage
1691 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1693 * fixed some bugs revealed by Paul Dietz' test suite:
1694 ** SUBTYPEP is slightly more accurate on heinously complicated
1695 CONS types where some of the members have uncertain (in the
1696 NIL, NIL sense) type relationships to each other.
1698 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1699 no longer purified unless :PURIFY T is explicitly specified.
1700 ** Non-purified cores are significantly smaller than before
1702 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1703 * new feature: timers based on Zach Beane's excellent timer package
1704 * added support for the following external formats: koi8-u,
1705 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1706 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1707 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1708 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1709 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1710 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1711 * incompatible change: a threaded SBCL will no longer revert to
1712 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1713 * bug fix: interrupts are disabled until startup is complete; no
1714 more sigsegvs when receiving a signal to soon
1715 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1717 * bug fix: add a workaround for the memory randomization features in
1718 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1719 workaround will only be in effect on systems with the proc filesystem
1721 * bug fix: printing objects of type HASH-TABLE signals a
1722 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1724 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1725 have been running for over 50 days. (reported by Gilbert Baumann)
1726 * bug fix: the logic for getting names of functions gets less
1727 confused when confronded with alternate-metaclass
1728 funcallable-instances. (reported by Cyrus Harmon)
1729 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1730 consistent, even on internal alternate-metaclass objects.
1731 * bug fix: flush closure information collected by physical
1732 environment analysis prepass before the main pass. (bug reported
1734 * bug fix: compiler pack phase does not modify a hashtable, which is
1735 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1736 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1737 may be expanded. (reported by Alexander <malishev> on c.l.l)
1738 * bug fix: no more style-warnings when compiling code using LAST
1739 under high-SPEED low-SPACE settings. (reported by David Wallin
1741 * fixed some incompatibilities between SBCL's MOP and the MOP
1743 ** the METAOBJECT class is now implemented;
1744 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1745 STANDARD-OBJECT, as required;
1746 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1747 now compatible (as required by VALIDATE-SUPERCLASS); there
1748 remains a consistency requirement over the presence or absence
1749 of the FUNCTION class in the superclass list of finalized
1750 classes; see the manual for more details;
1751 ** the :AROUND method for COMPUTE-SLOTS on
1752 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1753 requested slot ordering.
1755 ** bug fix: parent thread now can be gc'ed even with a live
1757 ** bug fix: binding a special with PROGV to no value is not
1758 the same as that symbol not having been bound (thanks to
1760 ** bug fix: binding specials is thread safe (thanks to
1762 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1763 and SB-SPROF do not die with 'no handler for signal XX in
1764 interrupt_handle_now(..)' anymore
1765 ** bug fix: WITH-TIMEOUT works with multiple running threads
1766 * fixed some bugs revealed by Paul Dietz' test suite:
1767 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1768 the :method-class keyword argument.
1770 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1771 * new port: the Solaris operating system on x86 processors is now
1772 mostly supported, though some rough edges in the environment
1773 remain. (thanks to Daisuke Homma)
1774 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1775 space for its heap. (thanks to Thiemo Seufer)
1776 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1777 closures. (thanks to Thiemo Seufer)
1778 * minor incompatible change: eof selects abort in the debugger.
1779 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1780 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1782 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1783 deprecated, and will go away in a future revision of SBCL.
1784 * minor incompatible change: GC-ON and GC-OFF are no longer
1785 implemented with a counter, it does not matter how many times gc
1786 is switched on or off
1787 * bug fix: discriminating functions for generic function classes
1788 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1789 make invalid assumptions about method precedence order. (reported
1791 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1792 platforms. (thanks to Thiemo Seufer)
1793 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1794 * bug fix: correct pathname printing with printer escaping is on.
1795 (thanks to Kevin Reid)
1796 * bug fix: complex VOP definitions in "user-space" no longer trigger
1797 package locks. (reported by Zach Beane)
1798 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1799 for funcallable-instance functions, and is no different from
1801 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1802 characters in the current readtable. (reported by Nicholas Neuss)
1803 * bug fix: SUBTYPEP on various CONS types returns more a more
1804 accurate acknowledgment of its certainty.
1805 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1806 lists in some cases. This partially fixes bug 384.
1807 * flush all standard streams before prompting in the REPL and the
1809 * bug fix: signal handling and triggering gc do not conflict
1810 directly anymore, in particular a high frequency sb-sprof does
1811 not prevent gc from running
1812 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1813 approximation for timezone and DST information between the
1814 universal time epoch and the smallest negative 32-bit time_t.
1815 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1816 year 1899 so long as the encoded time is non-negative.
1817 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1818 the correct answer for non-constant rotate amounts when performing
1819 an inline 32-bit rotation.
1821 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1822 there is only one thread in the session
1823 ** bug fix: memory leak for streams created in one thread and
1824 written to in another
1825 ** bug fix: lockup when compiled with gcc4
1826 ** bug fix: race that allows the gc to be triggered when gc is
1828 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1829 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1831 ** bug fix: finished threads are gc'ed properly
1832 * fixed some bugs revealed by Paul Dietz' test suite:
1833 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1834 has arguments with a different set of applicable methods from
1835 the orignal arguments.
1836 ** The type error thrown by MAP now has a correct expected-type
1838 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1839 name a compiled function.
1840 ** Three MISC tests where a large bignum was improperly coerced to
1841 a float that couldn't represent that bignum during type
1842 derivation were fixed.
1843 ** SUBTYPEP can now handle types involving the negation of a
1844 list-form FUNCTION type.
1845 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1846 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1847 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1849 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1850 * New feature: Experimental support for bivalent streams: streams
1851 opened with :element-type :default now allow character and binary
1852 (unsigned-byte 8) I/O
1853 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1854 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1856 * bug fix: no more highly sporadic "couldn't check whether ~S is
1857 readable" when reading a stream and an interrupt hits in the middle
1858 of a select system call
1859 * compiler better recognizes complex arrays (reported by Eduardo
1861 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1862 for variables on SBCL built with linkage-tables. (reported by Luis
1864 * various error reporting improvements.
1865 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1866 (Thanks to Thiemo Seufer)
1867 * bug fix: nested reader invokations work correctly
1868 * bug fix: it is possible to have simultaneous references to foreign
1869 code and foreign data with the same name.
1871 ** added x86-64 support
1872 ** incompatible change: the threading api now works with thread
1873 objects instead of thread ids
1874 ** bug fix: threads are protected from signals and interruption when
1875 starting up or going down
1876 ** bug fix: a race where an exiting thread could lose its stack to gc
1877 ** bug fix: don't halt on infinite error in threads if possible
1878 ** fixed numerous gc deadlocks introduced in the pthread merge
1879 ** bug fix: fixed thread safety issues in read and print
1880 ** bug fix: debugger doesn't hang on session lock if interrupted at
1881 an inappropriate moment
1882 ** bug fix: run-program is now thread safe(r)
1883 ** bug fix: inner with-recursive-lock no longer releases the mutex
1884 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1885 * fixed some bugs revealed by Paul Dietz' test suite:
1886 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1887 EXPECTED-TYPE slots filled.
1888 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1890 ** MAKE-CONDITION accepts classes as type-designators.
1891 ** COMPILE may never return NIL.
1892 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1893 range before calling Unix time functions
1895 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1896 * numerous signal handling fixes to increase stability
1897 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1898 * minor incompatible change: we now correctly canonize default
1899 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1900 as per the MOP, rather than the historical (INITARG INITFUNCTION
1901 INITFORM). (reported by Bruno Haible)
1902 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1903 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1904 for more information.
1905 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1906 pathname is a directory pathname.
1907 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1908 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1910 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1911 :METACLASS argument in addition to a class name. (reported by
1912 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1913 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1914 (reported by Bruno Haible; patch from Gerd Moellmann)
1915 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1917 * bug fix: more cleanups to the floating point exception handling on
1918 x86-64 (thanks to James Knight)
1919 * bug fix: the compiler does not try to constant fold calls of
1920 COERCE and BIT-* functions when they return freshly constructed
1921 objects. (reported by Nikodemus Siivola and Paul Dietz)
1922 * optimization: improved the MIPS versions of generic (in the
1923 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1924 * optimization: direct conversion of (unsigned-byte 32) to floats on
1925 the PowerPC platform.
1926 * optimization: structure instances with raw slots now use less
1927 memory, and probably show better memory locality. (thanks to
1929 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1930 are treated as requests for stack allocation on the x86-64,
1931 Alpha-32, PPC and SPARC platforms.
1932 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1933 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1934 (thanks to Tony Martinez)
1936 ** gcing a dead thread can no longer lead to lockups
1937 ** threads block signals until they are set up properly
1938 ** errno is no longer shared by threads
1939 ** interrupt-thread restores the eflags register on x86
1940 ** fixed some lockups due to gc/thread interaction
1941 * fixed some bugs revealed by Paul Dietz' test suite:
1942 ** invalid dotted lists no longer raise a read error when
1943 *READ-SUPPRESS* is T
1944 ** signal an error if a symbol that names a declaration is used
1945 as the name of a type, or vice versa
1946 ** allow using the (declare (typespec var*)) abbreviation for
1947 (declare (type typespec var*)) with all type specifiers
1948 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1949 ** always return NIL from PPRINT-POP when OBJECT is NIL
1950 ** don't signal errors when pretty-printing malformed LABELS,
1951 FLET or MACROLET forms
1952 ** declarations in a DOLIST body are also in scope for the
1954 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1956 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1959 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1960 * fixed cross-compiler leakages that prevented building a 32-bit
1961 target with a 64-bit host compiler.
1962 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1963 opened with :IF-EXISTS :OVERWRITE.
1964 * fixed bug 281: error for an invalid qualifier in a short-form method
1965 combination method is not signalled until the faulty method is called.
1966 * bug fix: iteration variable type inferrer failed to deal with open
1967 intervals. (reported by Alan Shields)
1968 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1969 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1970 or more runtime options were provided to the sbcl binary.
1971 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1972 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1973 * optimization: there's now a fast-path for fixnum arguments in the
1974 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1975 * optimization: the code generated on x86-64 is more compact thanks
1976 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1977 * bug fix: floating-point exception handling now partly works on
1978 x86-64. (Thanks to James Knight)
1979 * improvement to the MIPS backend from Thiemo Seufer: C-style
1980 64-bit long long arguments and return values to alien functions
1982 * fixed some bugs revealed by Paul Dietz' test suite:
1983 ** the type-error signalled from WARN has a filled-in DATUM slot.
1984 ** the type-error required when a stream is not associated with
1985 a file has the stream as its datum.
1986 ** type-errors on single-floats on x86-64 no longer have
1987 :INVALID-OBJECT as the datum
1988 ** the type-errors signalled for invalid function names now have
1989 a correct expected type
1990 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1991 for structures defined with a :TYPE. Documentation strings for
1992 typed structures are no longer immediately discarded
1993 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1994 broadcast streams and synonym streams. FILE-LENGTH now also works
1995 on broadcast streams.
1997 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1998 * incompatible change: the --noprogrammer option, deprecated since
1999 version 0.7.5, has been removed. Please use the equivalent
2000 --disable-debugger option instead.
2001 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2003 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2004 inoperational for a while now) has been completely removed.
2005 * null lexical environments are now printed as #<NULL-LEXENV>,
2006 significantly reducing the amount of clutter in typical backtraces.
2007 * documentation on weak pointers, finalization, and after GC hooks
2008 has been added to the manual.
2009 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2010 as other specialized array types, is much faster. SUBSEQ and
2011 COPY-SEQ on such arrays have also been sped up.
2012 * optimization: EQL is now more efficient when at least other argument
2013 is known to be of type (OR FIXNUM (NOT NUMBER)).
2014 * fixed bug: the runtime is now less vulnerable to changes in the
2015 size of the SBCL object on OS X, and virtual memory is reserved for
2016 all spaces that need to be at a fixed address.
2017 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2018 * fixed bug: finalizers and after GC hooks that cause consing are now
2020 * fixed bug: compiler error messages and summaries are now printed to
2021 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2022 * fixed inference of the upper bound of an iteration variable.
2023 (reported by Rajat Datta).
2024 * fixed bug 211e: calling local functions with duplicated constant
2025 keyword argument no longer causes a bogus style warning about an
2027 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2028 ftype declaration to be disregarded. (reported by Dave Roberts)
2029 * fixed bug 373: caused by erronous compilation of references to alien
2030 variables in the runtime on ppc/darwin.
2031 * fixed bug 376: CONJUGATE type deriver.
2032 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2033 (reported by Baughn on #lisp)
2034 * a cross-compiler bug on non-x86 platforms has been identified and
2035 fixed. (thanks to Bruno Haible)
2036 * improvements to the MIPS runtime code for increased stability.
2037 (thanks to Thiemo Seufer)
2038 * increased the maximimum compact environment size to allow
2039 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2040 * partial workaround for bug 135: don't name the function we're
2041 calling for hairy cases of EVAL, so as not to accumulate one environment
2042 entry per call to EVAL. (reported by Kevin Reid)
2043 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2044 * optimization: type testing for non-vector arrays should be faster.
2045 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2047 * bug fix: setting 31st element of a bit vector to zero did not work
2049 * bug fix: redefining a class definition which failed due to a
2050 previous accessor / function clash now works (but see BUGS entry
2051 #380 for more problems in this area). (thanks to Zach Beane)
2052 * the long form of DEFINE-METHOD-COMBINATION disables method group
2053 checking when given a single method group with pattern *. (thanks
2055 * on x86 compiler supports stack allocation of results of simple
2056 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2057 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2058 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2059 * fixed some bugs related to Unicode integration:
2060 ** the restarts for recovering from input and output encoding
2061 errors only appear when there is in fact such an error to
2063 * fixed some bugs revealed by Paul Dietz' test suite:
2064 ** MISC.549 and similar: late transformation of unsafe type
2065 assertions into derived types caused unexpected code
2067 ** SCALE-FLOAT type deriver is less wrong.
2068 ** type derivers for EXP, LOG and similar functions compute result
2069 types for complex arguments better.
2070 ** (MISC.563) CONJUGATE type deriver works for very restricted
2072 ** out-of-line type testers for character strings are available.
2073 ** EQUAL compiler transform understands specialness of objects
2075 ** accessing double-floats stored on the stack now works on x86-64.
2076 ** debugger internals could sometimes create invalid lispobjs,
2077 resulting in GC crashes.
2078 ** MISC.548: type check weakening can convert required type into
2080 ** initialization forms for bindings are not in scope of free special
2083 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2084 * incompatible change: thread support for non-NPTL systems has
2085 been removed - locking is buggy and unreliable. A threaded
2086 SBCL build will now warn at startup and refuse to create
2087 new threads, unless futex support is detected in the kernel
2088 * incompatible change: the top level REPL now has only an ABORT
2089 restart associated with it, not TOPLEVEL and ABORT as it used to.
2090 TOP and TOPLEVEL are now available as debugger commands for
2091 returning to the top level.
2092 * incompatible change: forms evaluated in the REPL now use the
2093 global optimization policy.
2094 * incompatible change: user- and system-initialization files are
2095 no longer processed with LOAD, but by READ and EVAL; hence the
2096 global optimization policy, startup package, readtable, etc,
2098 * The .fasl file format number has been incremented because of
2099 various incompatible changes.
2100 * internal entry point details and argument counts no longer appear
2101 in backtraces unless explicitly requested by setting
2102 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2103 * built-in and standard functions no longer have names like "top
2104 level local call to FOO".
2105 * fixed bug 32: functions defined in non-null lexical environments
2106 now have more legible printed representation
2107 * fixed bug 33: functions defined in non-null lexical environemnts
2108 are now more amenable to inspection by INSPECT.
2109 * workaround for bug 354: XEPs no longer appear in backtraces unless
2110 explicitly requested.
2111 * fixed bug: receiving the signal which results from attempting to
2112 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2113 SIGBUS on other *BSDs) on architectures where the C stack is also
2114 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2115 to debugger confusion or wild instability. (reported by Cheuksan
2117 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2118 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2119 (reported by Lutz Euler)
2120 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2121 compiled code. (reported by Paul Dietz)
2122 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2123 specialized parameter to a method to something that is not TYPEP
2124 the specializer is now possible.
2125 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2126 face of package deletion.
2127 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2128 pathname. (reported by David Tolpin on c.l.l)
2129 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2130 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2131 STRING-STREAM and FILE-STREAM, respectively.
2132 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2133 than previously; in addition, its test suite is now run on build.
2134 (thanks to Luke Gorrie)
2135 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2136 * optimization: added a immediate representation for single-floats
2138 * fixed some bugs revealed by Paul Dietz' test suite:
2139 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2140 ** attempting to create a package with a colliding nickname causes
2141 correctable errors to be signalled.
2142 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2143 ** several x86-64 backend bugs related to sign-extension of immediate
2146 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2147 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2148 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2150 * fixed loading of multiply forward-referenced layouts.
2151 (thanks to Cheuksan Wang)
2152 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2153 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2154 (see COMPILE-FILE documentation for details of :PRINT option)
2155 * fixed bugs 19 and 317: fixed-format floating point printing is
2156 more accurate. This also fixes a bug reported by Adam Warner
2157 related to the ~@F format directive.
2158 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2160 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2161 dispatch table if the from-char is a dispatch macro character.
2162 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2163 by the wordsize no longer ignore the last word. (reported by Lutz
2165 * fixed bug in type checking of dynamic-extent variables. (reported
2167 * optimization: sequence traversal functions use their freedom to
2168 coerce function designators to functions.
2169 * optimization: code with many calls to CLOS methods specialized on
2170 CLOS classes has had redundant type checks removed.
2171 * optimization: added declarations to speed up operations that access
2172 the internal character database (for example STRING-UPCASE)
2173 * optimization: comparison operations between floats and sufficiently small
2174 fixnums no longer create extra rationals
2175 * fixed some bugs related to Unicode integration:
2176 ** portions of multibyte characters at the end of buffers for
2177 character-based file input are correctly transferred to the
2178 start of the buffer at the next read.
2179 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2180 passing it through to OPEN.
2181 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2182 argument given to internal calls to OPEN.
2183 ** the built-in comment readers (introduced by character sequences
2184 ";" and "#|") are more forgiving to encoding errors; they will
2185 STYLE-WARN and then attempt to resync the stream at a character
2186 boundary. (thanks to Teemu Kalvas)
2187 * fixed some bugs in the x86-64 port:
2188 ** Negative short int return values from c-calls are sign-extended
2190 ** The stack is aligned to 16-bytes for c-calls, as required by
2191 the ABI. (thanks to Cheuksan Wang)
2192 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2193 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2194 for enums. (thanks to Vincent Arkesteijn)
2195 ** Multiple small optimizations and bugfixes for floating point
2197 * fixed some bugs revealed by Paul Dietz' test suite:
2198 ** Space, Tab, Linefeed, Return and Page have the invalid
2199 secondary constituent character trait.
2200 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2202 ** WITH-INPUT-FROM-STRING should only update the index place on
2204 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2205 works more reliably.
2206 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2207 with one-dimensional arrays with fill pointers.
2208 ** TYPE-OF failed on a complex with an integer realpart and a
2210 ** compiler failure during type inference for the code of form
2211 (IF (EQL X (THE ...)) ...) (MISC.535).
2213 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2214 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2215 platform. The port supports 61-bit fixnums, large memory spaces
2216 and reloading shared object files.
2217 * enhancement: saving cores with foreign code loaded is now
2218 supported on ppc/Darwin in addition to the previously supported
2220 * enhancement: the statistical profiler now walks deeper into the
2221 call stack for more meaningful call-graphs and accrued time
2222 reports (x86/x86-64 only). It also now reports time spent in
2224 * enhancement: it is now possible to trace most individual methods
2225 of a generic function in addition to tracing the generic function
2227 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2229 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2230 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2231 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2232 produces an error. (thanks to Vincent Arkesteijn)
2233 * bug fix: NAMESTRING on pathnames with :WILD components in their
2234 directories works correctly. (thanks to Artem V. Andreev)
2235 * fixed bug 125: compiler preserves identity of closures. (reported
2237 * bug fixed: functions with &REST arguments sometimes failed with
2238 "Undefined function" when compiled with (DEBUG 3). (reported by
2240 * bug fix: overflow during compiling of setting element of a bit
2241 vector with constant index and value. (reported by Timmy Douglas)
2242 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2243 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2244 * on x86 compiler partially supports stack allocation of dynamic-extent
2246 * GO and RETURN-FROM do not check the extent of their exit points
2247 when compiled with SAFETY 0.
2248 * fixed some bugs related to Unicode integration:
2249 ** encoding and decoding errors are now much more robustly
2250 handled; it should now be possible to recover even from invalid
2251 input or output to the terminal. (thanks to Teemu Kalvas)
2252 ** provided a first cut at implementing STRING-TO-OCTETS and
2253 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2254 ** altered the SB-MD5 contributed module slightly, changing the
2255 interface just enough for it to be supportable for builds where
2256 lisp characters are not eight bits.
2257 * fixed some bugs revealed by Paul Dietz' test suite:
2258 ** the FORMATTER-generated functions for ~V[ conditionals require
2259 the correct number of arguments.
2260 ** READ-FROM-STRING returns the mandated second value when applied
2261 to displaced strings.
2262 ** the #\Rubout and #\Backspace characters are treated as invalid
2263 constituent characters by the tokenizer.
2265 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2266 * new feature: reloading changed shared object files with
2267 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2268 * new feature: references to foreign variables and functions
2269 can now be compiled and loaded before the corresponding shared
2270 object file is loaded, as long as the foreign definitions are
2271 available at runtime.
2272 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2273 supported. (thanks to Dan Debertin)
2274 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2275 just keywords. (thanks to Vincent Arkesteijn)
2276 * fixed bug #331: structure-class instances corresponding to
2277 DEFSTRUCT forms are now created eagerly.
2278 * fixed bug #345: backtraces from calls to undefined functions work
2279 on x86 as well. Related bug #61 is now also partially fixed on x86
2280 (backtraces from throws to unknown catch tags.)
2281 * bug fix: lambda-list parsing is now stricter vrt. order and number
2282 of lambda-list keywords.
2283 * bug fix: as specified by AMOP, an error is signalled if a
2284 class-option appears multiple times in a DEFCLASS form. (reported
2286 * bug fix: RANDOM can be compiled when the compiler derives the type
2287 of its numeric argument as a disjoint set of small integers.
2288 (reported by Paul Dietz)
2289 * bug fix: starting a core saved with shared objects loaded when
2290 those objects are not available no longer causes threaded SBCL to
2291 hang. (reported by Sean Ross)
2292 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2294 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2295 argument, which must always be NIL. (reported by Kalle Niemitalo)
2296 * bug fix: printing 1.0d+23 no longer results in an error.
2297 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2298 * bug fix: structure slot setters preserve evaluation order. (thanks
2300 * fixed some bugs related to Unicode integration:
2301 ** RUN-PROGRAM can allow its child to take input from a Lisp
2302 stream. (reported by Stefan Scholl)
2303 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2304 locales. (reported by Ken Causey)
2305 * fixed some bugs revealed by Paul Dietz' test suite:
2306 ** INCF, DECF and REMF evaluate their place form as specified in
2308 ** FORMATTER expands ~{ ~} iteration directives with V or #
2309 parameters correctly.
2310 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2311 consequent uses no arguments correctly.
2312 ** the system has a partial understanding of the (COMPLEX RATIO)
2314 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2317 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2318 * new feature: a build-time option (controlled by the :SB-UNICODE
2319 keyword feature, enabled by default) for building the system with
2320 support for the entire 21-bit character space defined by the
2322 * new feature: the system now has rudimentary external-format
2323 support; the primary user-visible change at this time is that
2324 characters with the high bit set (such as lower-case-e-acute) will
2325 print correctly to a terminal in a UTF-8 environment.
2326 * minor incompatible change: BASE-CHAR no longer names a class;
2327 however, CHARACTER continues to do so, as required by ANSI.
2328 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2329 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2330 has been moved to the SB-EXT package (temporarily re-exported from
2332 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2333 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2335 * fixed bug #351: better error-handlind and reporting for malformed
2337 * fixed bug #350: bignum-printing is now more memory-efficient,
2338 allowing printing of very large bignums, eg. (expt 2 10000000).
2339 (reported by Bruno Haible)
2340 * fixed bug #302: better primitive-type selection for intersection
2342 * fixed bug #308: non-graphic characters now all have names, as
2343 required. (reported by Bruno Haible)
2344 * bug fix: redefining a class with different superclasses now correctly
2345 removes it from the direct-subclasses of its previous superclasses.
2346 (reported by David Morse)
2347 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2348 new value now works. (reported by Bruno Haible)
2349 * bug fix: correct canonicalization of multiple non-standard slot
2350 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2351 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2352 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2353 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2354 now exists, an signals an error.
2355 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2356 during macroexpansion for non-iteration variables as well. (reported
2357 by Bruno Haible for CMUCL)
2358 * bug fix: Cyclic structures and unprintable objects in compiler
2359 messages no longer cause errors. (reported by Bruno Haible)
2360 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2361 and READ-FROM-STRING all now return a primary value of NIL if
2362 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2363 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2364 (reported by Bruno Haible for CMUCL)
2365 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2366 do not designate a condition. (reported by Bruno Haible for
2368 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2369 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2370 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2371 permitting resolution in favour of any of the conflicting symbols.
2372 (reported by Bruno Haible for CMUCL)
2373 * bug fix: EQUAL compiler optimizations is less aggressive on
2374 strings which can potentially compare true despite having distinct
2375 specialized array element types.
2376 * bug fix: unit enumerations can be defined without dividing by
2377 zero. (thanks to Vincent Arkesteijn)
2378 * FORMAT compile-time argument count checking has been enhanced.
2379 (report from Bruno Haible for CMUCL)
2380 * a partial workaround for the bug 262: the compiler does not try to
2381 inline-expand a local function doing RETURN-FROM from a deleted
2382 BLOCK. (thanks to Peter Denno for the bug report and to David
2383 Wragg for the simple test case)
2384 * fixed some bugs revealed by Paul Dietz' test suite:
2385 ** RENAME-PACKAGE allows all package designators as new package
2387 ** constraint propagation and lambda variable substitution are
2388 more cautious in dealing with partially deleted code.
2389 ** compiler sometimes lost reoptimization passes.
2390 ** CERROR, given a condition as condition designator, passes its
2391 remaining arguments to the continue format control without
2393 ** Case-altering FORMAT directives work correctly on non-ASCII
2395 ** The REAL type specifier handles bounds outside the floating
2396 point ranges without signalling FLOATING-POINT-OVERFLOW.
2397 ** Functions with IR1-transformations can create intercomponent
2398 references to global functions.
2399 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2401 ** FORMAT ~R treats a nil value for its first parameter correctly.
2403 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2404 * enhancement: saving cores with foreign code loaded is now
2405 supported on x86/NetBSD and sparc/Linux in addition to the previously
2406 supported platforms.
2407 * bug fix: on some platforms repeated installations caused multiple
2408 copies of HTML documentation to be installed -- should not happen
2409 any more. (reported by Stefan Scholl)
2410 * bug fix: parsing self-recursive alien record types multiple times
2411 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2412 original patch by Helmut Eller for CMUCL)
2413 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2414 (thanks to Richard Kreuter)
2415 * bug fix: defining classes whose accessors are methods on existing
2416 generic functions in other (locked) packages no longer signals
2417 bogus package lock violations. (reported by François-René Rideau)
2418 * bug fix: special variables as DEFMETHOD parameters no longer have
2419 associated bogus type declarations. (reported by David Wragg and
2421 * bug fix: read-write consistency on streams of element-type
2422 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2423 * bug fix: redefiniton of the only method of a generic function with
2424 no DEFGENERIC no longer emits a full WARNING. In addition,
2425 redefinition of generic functions with no DEFGENERIC to an
2426 incompatible lambda list now signals an error. (thanks to Zach
2428 * bug fix: DEFGENERIC now works even when there's a function of the
2429 same name in an enclosing lexical environment. (thanks to Zach
2431 * fixed compiler failure, caused by instrumenting code during
2432 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2433 * optimization: added loop analysis and improved register allocation
2434 * fixed some bugs revealed by Paul Dietz' test suite:
2435 ** POSITION on displaced vectors with non-zero displacement
2436 returns the right answer.
2437 ** (SIMPLE-STRING) is a valid type specifier for sequence
2439 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2441 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2442 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2444 ** The justification version of the FORMAT ~< directive treats
2445 non-zero minpad parameter correctly.
2447 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2448 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2449 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2450 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2451 the supported interface.
2452 * new feature: Single-stepping of code compiled with DEBUG 2 or
2453 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2454 * new feature: saving cores with foreign code loaded is now
2455 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2456 Timothy Moore's work for CMUCL)
2457 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2458 parameters to * instead of NIL if no initform is supplied.
2459 (reported by Johan Bockgaard)
2460 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2461 correctly when FUNCALL appears as the car of the form. Note:
2462 despite this FUNCALL forms are not currently subject to
2463 compiler-macro expansion. (port of Raymond Toy's fix for the
2464 same from CMUCL, reported by Johan Bockgaard)
2465 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2466 (thanks for Teemu Kalvas)
2467 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2468 case instead of terminating the iteration (thanks for Julian
2469 Squires, Sean Champ and Raymond Toy)
2470 * bug fix: incorrect expansion of defgeneric that caused a style
2471 warning. (thanks for Zach Beane)
2472 * bug fix: slot accessor effective method computation works properly
2473 for classes with multiple non-standard applicable methods on
2474 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2475 * on x86 compiler supports stack allocation of results of LIST and
2476 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2477 CMUCL implementation by Gerd Moellmann)
2478 * fixed some bugs revealed by Paul Dietz' test suite:
2479 ** FORMAT strings with both the ~<~:;~> form of the justification
2480 directive and pretty-printing directives cause an error.
2482 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2483 * incompatible change: the internal functions
2484 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2485 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2486 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2487 instead of the old functions.
2488 * new feature: on platforms where "dladdr" is available foreign
2489 function names now appear in backtraces. (based on Helmut Eller's
2491 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2492 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2494 * documentation: ASDF manual is now distributed with SBCL.
2495 * bug fix: SBCL can now load its contributed modules with REQUIRE
2496 even if the system-provided entries have been removed from
2497 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2499 * bug fix: backtraces involving undefined functions or assembly
2500 routines are more informative. (thanks to Brian Downing)
2501 * bug fix: mutually referent alien structures now work correctly.
2502 (reported by Rick Taube)
2503 * bug fix: structures defined by WITH-ALIEN can be referred to
2504 within other definitions in the same WITH-ALIEN.
2505 * bug fix: division operators (MOD, TRUNCATE and the like) with
2506 constant zero divisors and integer dividends no longer generate
2508 * bug fix: provide default methods for INPUT-STREAM-P and
2509 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2510 * bug fix: improve the detection and resolution of MOP metacycles
2511 (where effective method computation depends on the generic function
2512 having its effective method computed). (reported by Bruno Haible)
2513 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2514 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2515 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2516 * bug fix: pathnames with relative directory components can be
2517 represented relative to default pathnames.
2518 * optimization: in taking the GCD of bignums, reduce the two bignums
2519 to approximately the same size (using Euclid's algorithm) before
2520 applying the more sophisticated binary GCD. (thanks to Juho
2522 * optimization: COUNT on bitvectors now operates word-at-a-time.
2523 * optimization: ASH with a positive, but not necessarily constant,
2524 (leftwards) shift, when in a modular context, is compiled to a
2526 * fixed some bugs revealed by Paul Dietz' test suite:
2527 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2528 if the corresponding argument is NIL.
2529 ** FORMAT directives accepting a minpad parameter treat negative
2531 ** literal commas as character parameters to FORMAT directives are
2533 ** literal spaces directly after ~<Newline> directives within a
2534 format-logical-block (~:< ~@:>) do not induce :FILL-style
2535 conditional newlines.
2536 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2537 as specified: it no longer includes conditional newlines).
2538 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2540 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2541 :PER-LINE-PREFIX argument does not evaluate to a string.
2543 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2544 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2545 the manual for details; for now, package locks can be disabled by
2546 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2547 if no major problems are found then it is likely that they will be
2548 compiled in unconditionally.
2549 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2550 unsupported operators on all platforms. To load a shared library
2551 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2552 object file, link it into a shared library outside of SBCL and
2553 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2555 * bug workaround: ROOM T and the GC were not working together
2556 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2557 reporting facility aren't preserved. That reporting has been
2558 disabled, so now until and unless someone figures out how to make it
2559 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2560 * minor incompatible change: as threatened around sbcl-0.8.0, the
2561 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2562 The symbols are also exported from SB-PCL for backwards
2563 compatibility, but more so than before SB-PCL should be treated as
2564 an implementation-internal package.
2565 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2567 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2568 It is known as of this release not to work on the Alpha, however.
2569 * fixed bug #167: errors signalled due to illegal syntax in method
2570 bodies are now more legible.
2571 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2572 designators and can hence be used with TYPEP.
2573 * fixed bug #333: CHECK-TYPE now ensures that the type error
2574 signalled, if any, has the right object to be accessed by
2575 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2576 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2577 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2579 * fixed bug #334: programmatic addition of slots using specialized
2580 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2581 and :ALLOCATION :CLASS slots.
2582 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2583 float exponents. (rereported by Peter Seibel)
2584 * fixed a bug: #\Space (and other whitespace characters) are no
2585 longer considered to be macro characters in standard syntax by
2586 GET-MACRO-CHARACTER.
2587 * fixed bug: initialization of condition class metaobjects no longer
2588 causes an instance of the condition to be created. (reported by
2590 * fixed bug: it is now possible to have slots such that
2591 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2592 description is neither :INSTANCE nor :CLASS.
2593 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2594 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2595 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2596 non-standard SLOT-DEFINITION-ALLOCATION values.
2597 * fixed bugs: various race conditions handling exiting threads.
2598 CL-PPCRE's thread torture test now passes.
2599 * fixed arguably-a-bug: GC time proportional to number of threads in
2600 system even when most of them are idle
2601 * optimization: improved performance of BIT and SBIT on bit-vectors.
2602 * .fasl file incompatibility: The fasl file version number has
2603 been incremented because of changes associated with package locks.
2605 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2606 * minor incompatible change: the system no longer provides
2607 optimization hints (conditions which are TYPEP
2608 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2609 mode; these hints are still emitted when higher SPEED optimization
2611 * new contrib module: a sampling profiler (profiling by statistical
2612 sampling, rather than by instrumenting functions) is available as
2613 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2614 * the behaviour of the standard function ED is now customizeable by
2615 third parties through a hook variable: see ED's documentation
2616 string for information on the protocol.
2617 * the compiler no longer emits efficiency notes for (FUNCALL X)
2618 when the type of X is uncertain under default optimization
2620 * fixed bug 276: mutating a binding of a specialized parameter to a
2621 method to something that is not TYPEP the specializer is now
2623 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2624 exhibit the required behaviour on the x86 platform. (thanks to
2625 Peter van Eynde, Eric Marsden and Bruno Haible)
2626 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2627 even for difficult arguments. (reported by Peter Graves)
2628 * fixed bug 141a: the backquote printer now descends quoted
2630 * fixed another bug in backquote printing: no more destructive
2631 modification of the form's list structure. (reported by Brian
2633 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2634 move between its address being taken and the call to
2635 interrupt_thread, fixing a crashing race condition.
2636 * the SB-POSIX contrib implementation has been adjusted so that it
2637 no longer exhibits ridiculously poor performance when constructing
2638 instances corresponding to C structs.
2640 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2641 * minor incompatible change: the sb-grovel contrib now treats C
2642 structures as alien (in the sense of SB-ALIEN) objects rather than
2643 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2644 has implications for memory management of client code
2645 (sb-grovel-returned objects must now be manually managed) and for
2646 type safety (alien objects now have full types).
2647 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2648 used to control emission of compiler diagnostics, rather than the
2649 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2650 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2651 quality should be considered deprecated.
2652 * install.sh now installs the user manual as well
2653 * (not quite a new documentable feature, but worth considering in
2654 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2655 beginnings of a semantically meaningful condition hierarchy is
2656 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2658 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2659 designator as the defaults argument.
2660 * fixed bug: Displaced arrays whose displaced-to array has become
2661 too small now cause ARRAY-DIMENSION to signal an error, providing
2662 for safer bounds-checking. (reported by Bruno Haible)
2663 * fixed bug: DEFCLASS slot definitions with identical :READER and
2664 :WRITER names now signal a reasonable error. (reported by Thomas
2666 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2668 * fixed bug: Invalid slot specification errors now print correctly.
2669 (thanks to Zach Beane)
2670 * fixed bug 320: Shared to local slot value transfers in class
2671 redefinitions now happen corrently from superclasses as
2672 well. (reported by Bruno Haible)
2673 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2675 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2676 improper lists correctly. (reported by Bruno Haible)
2677 * fixed bug 313: source-transform for <fun-name> was erroneously
2678 applied to a call of a value of a variable with name <fun-name>.
2679 (reported by Antonio Menezes Leitao)
2680 * fixed bug 307: The obsolete instance protocol ensures that
2681 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2682 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2683 remove dead unknown-values globs from the middle of the stack.
2684 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2685 easier to resume long computations after using *BREAK-ON-SIGNALS*
2686 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2687 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2688 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2689 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2690 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2691 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2693 * fixed a (fairly theoretical) bug in string printing: if
2694 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2695 does not have array-element-type equal to the most general string
2697 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2698 function name. (thanks to Nikodemus Siivola)
2699 * fixed bug: compile-time format string checker failed on
2700 non-closed ~{. (reported by Thomas F Burdick)
2701 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2702 don't warn on legal loop code involving more than one aggregate
2703 boolean. (thanks to Nikodemus Siivola)
2704 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2705 in keyword position to MAKE-PACKAGE was not being reported as
2707 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2708 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2710 * fixed bugs 280 and 312: the checking for multiple definitions in a
2711 file is less likely to become confused by uses of inline
2713 * fixed bug: the #S reader macro performs the keyword coercion
2714 specified for slot names. (reported by Kalle Niemitalo)
2715 * fixed bug: lambda lists may contain symbols whose names start with
2716 & but are not lambda-list-keywords; their occurrence triggers a
2718 * fixed bug 321: define-method-combination argument lambda lists do
2719 not cause invalid code to be generated when &optional or &aux
2720 variables have default values. (reported by Bruno Haible)
2721 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2722 have CLOS classes; this appears also to have fixed reported
2723 instability in stack exhaustion detection.
2724 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2725 MUFFLE-WARNING when no associated restart is present is now
2727 * optimization: rearranged the expansion of various defining macros
2728 so that each expands into only one top-level form in a
2729 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2731 * optimization: used a previously little-used slot in symbols to
2732 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2734 * fixed some bugs revealed by Paul Dietz' test suite:
2735 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2737 ** Fixed an optimization bug related to inheritance of initforms
2738 from local to shared slots.
2739 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2740 ** CAST optimizer forgot to flush argument derived type.
2741 ** print/read consistency on floats is now orders of magnitude
2742 more likely. (thanks also to Bruno Haible for a similar report
2744 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2745 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2746 for objects of type REAL. Make it so.
2747 ** SXHASH is specified (infelicitously) to respect similarity,
2748 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2749 it so. (thanks to Markus Ziegler)
2750 ** on the Alpha, the compiler succeeds in compiling functions
2751 returning a known number of arguments greater than 63.
2752 ** fixed handling of invalid NIL arguments in keyword position
2754 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2756 ** provided a readably-printable representation for RANDOM-STATE
2758 ** ensured that pathnames, where they have a namestring, always
2759 print using #P"..." syntax.
2761 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2762 * Support for the forthcoming 2.0 version of the NetBSD kernel
2763 running on x86 hardware has been added. (thanks to Perry
2764 E. Metzger most immediately, and others for their past work)
2765 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2766 current release version; SBCL's OpenBSD support had been broken
2767 since about the time of OpenBSD's switch to ELF binary format.)
2768 * [placeholder for DX summary]
2769 ** user code with &REST lists declared dynamic-extent, under high
2770 speed or space and low safety and debug optimization policy.
2771 * The manual has been converted to Texinfo format and the debugger
2772 chapter from the cmucl manual has been added.
2773 * A facility has been added to extract documentation strings from
2774 sbcl and store them as Texinfo-formatted snippets for inclusion in
2775 the manual (via Texinfo's @include directive)
2776 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2777 the test case to Dave Roberts)
2778 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2779 pointers. (reported by Sean Ross)
2780 * bug fix: PROFILE output is printed nicely even for large numerical
2781 values. (thanks to Zach Beane)
2782 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2783 greater than 32 handle EOF correctly.
2784 * bug fix: on X86 an immediate argument of the IMUL instruction is
2785 correctly printed in disassembly. (thanks to Lutz Euler)
2786 * bug fix: class slots in redefined classes preserve their old
2787 values. (thanks to Bruno Haible and Nikodemus Siivola)
2788 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2789 succeeds. (reported by Marco Baringer)
2790 * fixed some bugs revealed by Paul Dietz' test suite:
2791 ** READ-SEQUENCE now works on ECHO-STREAMs.
2792 ** RATIONALIZE works more according to its specification. (thanks
2795 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2796 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2797 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2798 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2799 should matter to you only if you rebind the printer control
2800 variables and then find you want different bindings in the
2801 debugger than in the ordinary execution of your program.)
2802 * The runtime build system has been tweaked to support building
2803 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2804 assembler and linker. (thanks to Nikodemus Siivola)
2805 * Unbound, undefined, undeclared variables now trigger full
2806 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2807 more often programmer error than deliberate exploitation of undefined
2809 * optimization: the hash algorithm for strings has changed to one
2810 that is less vulnerable to spurious collisions. (thanks to Juho
2812 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2813 needless bounds checking. (thanks to Juho Snellman)
2814 * optimization: implemented multiplication as a modular
2815 (UNSIGNED-BYTE 32) operation on the PPC backend.
2816 * fixed some bugs revealed by Paul Dietz' test suite:
2817 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2819 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2820 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2821 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2823 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2825 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2827 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2828 their output stream on EOF from read.
2829 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2830 have been read to end-of-file.
2831 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2833 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2834 description of determination of which consecutive characters
2836 ** Printing the "Space" character with escaping on now yields "#\\ ",
2837 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2838 ** Reading floating-point numbers with *READ-BASE* set to a number
2839 less than 10 works correctly.
2840 ** Reading floating-point numbers with *READ-BASE* set to a number
2841 more than 10 works correctly.
2842 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2843 the readtable currently in effect.
2845 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2846 * minor incompatible change: parsing of namestrings on a physical
2847 (Unix) host has changed; numbers after the final #\. in a
2848 namestring are no longer interpreted as a version field. This is
2849 intented to be largely invisible to the user, except that the
2850 meaning of the namestring "*.*.*" has changed: it now refers to a
2851 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2852 should usually be replaced by
2853 (make-pathname :name :wild :type :wild :version :wild)
2854 with the added benefit that this is more likely to be portable.
2855 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2856 option to OPEN now signals an error if the file being opened
2857 exists; this may have an impact on existing code.
2858 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2859 signals received, on the PowerPC platforms (both Linux and
2860 Darwin). (thanks to Pierre Mai for pointing out the location of
2862 * several fixes on OS X: The system now builds and runs cleanly on
2863 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2864 (thanks to Brian Mastenbrook)
2865 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2866 second-resolution: integer multiples of 1/3600 between -24 and 24.
2867 (thanks to Vincent Arkesteijn)
2868 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2869 type when called with 1 argument; PEEK-CHAR checked type of
2870 PEEK-TYPE only after having read first character from a
2871 stream. (reported by Peter Graves)
2872 * bug fix: the garbage collector now has much better locality
2873 behaviour, and in particular no longer treats all memory as being
2874 exhausted when many small objects point to each other in a deeply
2876 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2877 recognized as being TYPEP their class.
2878 * bug fix: the PUSHNEW documentation string has been corrected.
2879 (thanks to Vincent Arkesteijn)
2880 * bug fix: defaulting of the value for the last of an atypically
2881 large number of multiple values being bound was not being
2882 performed correctly on the Alpha or PPC platforms
2883 * optimization: implemented multiplication as a modular
2884 (UNSIGNED-BYTE 32) operation on the x86 backend.
2885 * optimization: SEARCH on simple-base-strings can now be open-coded.
2886 (see also contrib/compiler-extras.lisp for inspiration for
2887 teaching the compiler about the Boyer-Moore algorithm).
2888 * value, returned by MAX (and MIN) called with several EQUALP, but
2889 not EQL, arguments now does not depend on compiler settings.
2890 * fixed some bugs revealed by Paul Dietz' test suite:
2891 ** in stack analysis liveness information is propagated from
2892 non-local entry points.
2893 ** pathwise CAST removing failed when the CAST node did not start
2895 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2896 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2898 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2899 host is already defined.
2900 ** RENAME-FILE works on streams instead of signalling an internal
2902 ** PEEK-CHAR uses the current readtable when determining whether
2903 or not a character is whitespace.
2904 ** MERGE-PATHNAMES handles the case when the pathname does not
2905 specify a name while the default-pathname specifies a version.
2906 ** Pathnames now stand a better chance of respecting print/read
2908 ** Attempting to use standardized file system operators with a
2909 pathname with invalid :DIRECTORY components signals a
2911 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2913 ** DIRECTORY on logical pathnames is more correct.
2914 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2915 signal a TYPE-ERROR if their argument is not a stream.
2916 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2917 designator argument does not designate a stream.
2918 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2919 examining the synonym.
2920 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2922 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2923 element-type larger than ([UN]SIGNED-BYTE 32).
2925 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2926 * When built with the :SB-FUTEX feature, threaded builds now take
2927 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2928 for faster/more reliable mutex and condition variable support.
2929 * Incompatible change (but one you probably shouldn't have been using
2930 anyway): the interface and code for arbitrating between multiple
2931 threads in the same user session has been redesigned.
2932 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2933 error when called without an explicit environment argument.
2934 (thanks to Nikodemus Siivola)
2935 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2936 confused about their position. (thanks to Adam Warner and Gerd
2938 * bug fix: Pretty printing backquoted forms with unquotations in the
2939 argument list position of various code constructs such as LAMBDA
2940 now works correctly. (reported by Paul Dietz)
2941 * bug fix: Pretty printing unquotations no longer loses all
2942 stream position information.
2943 * optimization: performance of string output streams is now less
2944 poor for multiple small sequence writes.
2945 * optimization: performance of CSUBTYPEP in the presence of complex
2946 expressions involving CONS and NOT many times has been improved.
2947 (reported by Paul Dietz)
2948 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2950 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2951 streams for the REPL, for the debugger, and so on. (thanks to
2953 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2954 * fixed some bugs revealed by Paul Dietz' test suite:
2955 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2956 is ignored if it is too small, rather than propagating through
2957 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2958 ** extremely complex negations of CONS types were not being
2959 sufficiently canonicalized, leading to inconsistencies in
2961 ** VALUES tranformer lost derived type.
2963 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2964 * fixed a bootstrapping bug: the build process no longer assumes
2965 that the various BOOLE-related constants have the same value in
2966 host and target lisps. (noted by Paul Dietz' test suite on an
2967 SBCL binary built from CLISP)
2968 * The system can now be dynamically linked on the MIPS platform,
2969 which enables dynamic loading of foreign code from Lisp. (thanks
2970 to Ralf Baechle for discussions on the MIPS ABI)
2971 * The system now records debugging information for its own source
2972 files in a filesystem-position-independent manner, relative to
2973 the "SYS" logical host.
2974 * fixed a compiler bug: MV-LET convertion did not check references
2975 to the "max args" entry point. (reported by Brian Downing)
2976 * tweaked disassembly notes to be less confident about proclaiming
2977 some instruction as an LRA. (thanks to Brian Downing)
2978 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2979 now each have their own history, command character, and other
2980 characteristics. (thanks to David Lichteblau)
2981 * fixed some bugs revealed by Paul Dietz' test suite:
2982 ** compiler failure in compiling LOGAND expressions including a
2984 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2985 shift greater than 32.
2986 ** FLUSH-DEST did not mark blocks for type check regeneration.
2987 ** HANDLER-CASE failed to accept declarations in handler clauses
2988 in some circumstances.
2990 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2991 * New code in contrib/sb-introspect (still probably not entirely
2992 stable yet) provides some support for smart Lisp development
2993 environments like SLIME.
2994 * The conditions signalled for errors occurring when loading .fasl
2995 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2996 in a way which should help ASDF recover gracefully.
2997 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2998 stuff has been cleaned up. If you code contrib/ stuff, this might
2999 affect you, and you can look at contrib/README, contrib/STANDARDS,
3000 and/or the 0.8.4.27 diff to check.
3001 * In full calls the compiler now does not generate checks for declared
3002 argument types for all arguments.
3003 * various threading fixes
3004 ** and some experimental patches which didn't make it into
3005 the main tree for this release, but which are shipped in
3006 contrib/experimental-thread.patch as a possible fix for some
3007 failures (deadlock, spinning...) in GC-intensive multithreaded
3009 * fixed PPC build problem (source code incompatibility of different
3010 library versions): added offsetof() hackery which attempts to divine
3011 where glibc maintainers put uc_mcontext today
3012 * fixed bug 282: compiler does not trust type assertions while passing
3013 arguments to a full call.
3014 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3015 &OPTIONAL VALUES type parameter.
3016 * fix bug 214: algorithm for noting rejected templates is now more
3017 similar to that of template seletion. (also reported by rydis on
3019 * fixed bug 141b: printing backquoted information readably and prettily
3020 inserts a space where necessary.
3021 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3022 (CHARACTER 10) are now reported as errors, rather than propagated
3023 as unknown types. (reported by piso on #lisp)
3024 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3025 with values NIL and :ERROR. (thanks to Milan Zamazal)
3026 * fixed bug 191c: CLOS now does proper keyword argument checking as
3027 described in CLHS 7.6.5 and 7.6.5.1.
3028 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3029 longer raise an error; further, using a list as a for-as-arithmetic
3030 counter now raises a meaningful error.
3031 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3032 now understood by sequence creation functions such as MAKE-SEQUENCE
3034 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3035 a STRING-INPUT-STREAM.
3036 * compiler enhancement: SIGNUM is now better able to derive the type
3038 * type declarations inside WITH-SLOTS are checked. (reported by
3040 * fixed some bugs revealed by Paul Dietz' test suite:
3041 ** incorrect optimization of TRUNCATE for a positive first
3042 argument and negative second.
3043 ** compiler failure in let-convertion during flushing dead code.
3044 ** compiler failure while deriving type of TRUNCATE on an
3045 interval, containing 0.
3046 ** ASH of a negative bignum by a negative bignum count now returns
3048 ** intersection of CONS types now canonicalizes properly, fixing
3049 inconsistencies in SUBTYPEP.
3051 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3052 * incompatible change: The --disable-debugger command line
3053 option now clobbers the debugger at a more fundamental
3054 level, by redefining #'INVOKE-DEBUGGER instead of by
3055 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3056 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3057 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3058 enter the debugger REPL and then suffer recursive errors
3059 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3060 while under the new system, BREAK in --disable-debugger mode
3061 terminates the system just as an unhandled error would.
3062 * fixed compiler performance when processing loops with a step >1;
3063 * bug fix: DOCUMENTATION now retrieves generic function
3064 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3065 support has been systematized, and now supports the methods
3066 specified by ANSI, along with a default method and a method for
3067 slot documentation. (reported by Nathan Froyd)
3068 * bug fix: effective methods associated with a generic function are
3069 no longer cached over a change of that generic function's method
3070 combination. (reported by Andreas Fuchs)
3071 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3072 names a non-existent directory. (thanks to Andreas Fuchs)
3073 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3074 the PPC platform, lead to stack corruption; this has been fixed.
3075 (reported by Rainer Joswig)
3076 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3077 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3078 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3079 arithmetic in general, is now much more reliable on the PPC
3081 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3082 platform now returns the right answer.
3083 * optimization: restored some effective method precomputation in
3084 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3085 precomputation is now tunable.
3086 * optimization: compiler-internal data structure use has been
3087 reviewed, and changes have been made that should improve the
3088 performance of the compiler by about 20%.
3089 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3090 simple-base-strings and simple-bit-vectors is improved.
3091 * optimization: the optimization of 32-bit logical and arithmetic
3092 functions introduced in version 0.8.3 on the x86 has been
3093 implemented on the mips, ppc and sparc platforms; an
3094 implementation of the same facility, but for 64-bit arithmetic,
3095 has been added for the alpha.
3096 * microoptimization: the compiler is better able to make use of the
3097 x86 LEA instruction for multiplication by constants.
3098 * bug fix: in some situations compiler did not report usage of
3099 generic arithmetic in (SPEED 3) policy.
3100 * bug 145b fix: compiler used wrong type specifier while converting
3101 MEMBER-types to numeric.
3102 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3104 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3105 index argument now works properly on the Alpha platform.
3106 * bug fix: floating point exception treatment on the Alpha platform
3108 * bug fix: FILE-POSITION works much better on string input and
3109 output streams. (thanks to Nikodemus Siivola)
3110 * bug fix: many threading/garbage collection symptoms sorted.
3111 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3112 might be pseudo-atomic.
3113 * internal change: Stopping for GC is now done with signals not
3114 ptrace. GC is now done in whichever thread wanted it, instead of
3116 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3118 * bug fix: result form in DO is not contained in the implicit
3120 * incompatible change: ICR structure is changed; the value part of
3121 CONTINUATION is now called LVAR; corresponding functions are
3122 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3123 * added type deriver for ISQRT (thanks to Robert E. Brown).
3124 * bug fix: better support for loading from the command line when an
3125 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3127 * fixed some bugs revealed by Paul Dietz' test suite:
3128 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3129 ** ROUND and FROUND now give the right answer when given very
3130 small float arguments.
3131 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3133 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3134 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3135 ** LCM should return a non-negative integer.
3136 ** PARSE-INTEGER returned the index of a terminator instead of the
3137 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3138 ** PARSE-INTEGER returned an incorrect index being applied to a
3140 ** LCM with two arguments of 0 returns 0 rather than signalling
3142 ** unsigned addition of a 32-bit constant with the high bit set no
3143 longer causes an internal compiler error on the x86.
3144 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3145 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3146 with negative last argument.
3147 ** byte specifiers with zero size and position no longer cause
3148 an error during type derivation.
3149 ** bignum multiplication on the Alpha platform now returns the
3151 * porting: The system now builds on SuSE AMD64, although it still
3152 generates a 32-bit binary.
3153 * .fasl file incompatibility: The fasl file version number has
3154 been incremented (because of the changes to internal compiler
3155 data structures referred to above).
3157 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3158 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3159 more accurately, on the Darwin kernel running on PowerPC hardware.
3160 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3161 * Compiler code deletion notes now signal a condition of type
3162 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3163 an associated MUFFLE-WARNING restart.
3164 * The compiler now performs limited argument count validation of
3165 constant format strings in FORMAT, and where appropriate in ERROR,
3166 CERROR and WARN. (thanks to Gerd Moellmann)
3167 * New ASDF-INSTALL contrib can be used for automatic download and
3168 installation of third-party Lisp code from CCLAN or other sites
3170 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3171 which forces another thread to execute a function supplied by the
3173 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3174 now accept and act upon their :ELEMENT-TYPE keyword argument.
3175 (reported by Martin Atzmueller, Edi Weitz)
3176 * bug fix: FILE-POSITION now accepts position designators up to
3177 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3178 is the greater. (thanks to Patrik Nordebo)
3179 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3180 Kalle Olavi Niemitalo)
3181 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3182 printable. (reported by Eric Marsden)
3183 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3184 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3185 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3186 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3187 (thanks to Stig Sandoe)
3188 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3189 UNREAD-CHAR now correctly includes the unread character in the
3190 target sequence. (thanks to Gerd Moellmann)
3191 * bug fix in threaded builds: the system can now be suspended and
3192 resumed by shell job control with minimal disruption.
3193 * bug fixes in times and timezones >2038AD
3194 * better handling of "where is GNU make?" problem in build scripts
3195 (thanks to Nikodemus Siivola)
3196 * new optimization: inside a named function any reference to a
3197 function with the same name is considered to be a self-reference;
3198 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3199 optimization quality.
3200 * new optimization on x86: logical functions and + now have
3201 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3202 used when the result is truncated to 32 bits.
3203 * VALUES declaration is partially enabled.
3204 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3205 * fixed some bugs revealed by Paul Dietz' test suite:
3206 ** The system now obeys the constraint imposed by
3207 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3208 types form a lattice under type intersection.
3209 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3210 ** ASSOC now ignores NIL elements in an alist.
3211 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3212 and (1+ MOST-POSITIVE-FIXNUM) answers.
3213 ** The addition of a method with invalid qualifiers to a generic
3214 function does not cause an error to be signalled immediately;
3215 a warning is signalled, and the error is generated only on
3216 calling the generic function.
3217 * changed .fasl file version number, in part to add type codes for
3218 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3219 obscure ANSI requirements
3221 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3222 * fixed bug 148: failure to inline-expand a local function left
3223 garbage, confusing the compiler.
3224 * fixed bugs 3cd: structure slot readers perform type check if the
3225 slot can have an invalid value (i.e. it is either not initialized
3226 or can be written with a less specific slot writer).
3227 * bug fix: the compiler now traps array references to elements off
3228 the end of an array; previously, the bounds checking in some
3229 circumstances could go off-by-one.
3230 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3232 * type declarations for array element types now obey the description
3233 on the CLHS page "Declaration TYPE", as per discussions on
3234 sbcl-help around 2003-05-08. This means that a declaration
3235 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3236 declaration, all references to BAR will be asserted or assumed
3237 (with THE, so dependent on compiler policy) to involve objects of
3238 type FOO. Note that no such declaration is implied in
3239 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3240 * declared types of functions from the "Conditions"
3241 chapter. (reported by Paul Dietz)
3242 * bug fix: CERROR accepts a function as its first argument.
3243 * bug fix: NTH an NTHCDR accept a bignum as index
3244 arguments. (reported by Adam Warner)
3245 * optimization: character compare routines now optimize comparing
3246 against a constant character. (reported by Gilbert Baumann)
3247 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3248 argument now works properly on the MIPS platform.
3249 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3250 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3251 declared to have a functional type, does not cause an error
3253 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3254 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3255 * fixed bug: if last continuation of a deleted block has a
3256 destination, this destination should be deleted too. (reported by
3258 * fixed a bug in the bootstrap process: the host compiler's values
3259 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3260 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3261 test case from Patrik Nordebo)
3262 * improved the ability of the disassembler on the PPC platform to
3263 provide helpful disassembly notes.
3264 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3265 the class in more cases than previously.
3266 * bug fix: FILE-POSITION now understands :START and :END for
3267 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3268 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3269 8) data. (thanks to David Lichteblau)
3270 * bug fix: it is possible to add a method to a generic function
3271 without lambda list.
3272 * bug fix: reader failed to signal END-OF-FILE inside an
3273 object representation. (reported by Nikodemus Siivola)
3274 * fixed some bugs revealed by Paul Dietz' test suite:
3275 ** LAST and [N]BUTLAST should accept a bignum.
3276 ** condition slot accessors are methods.
3277 ** (VECTOR NIL) is a subtype of STRING.
3279 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3280 * minor incompatible change: some nonsensical specialized lambda
3281 lists (used in DEFMETHOD) which were previously ignored now signal
3283 * minor incompatible change: the system is now aware of the types of
3284 variables in the COMMON-LISP package, and will signal errors for
3285 most violations of these type constraints (where previously they
3286 were silently accepted).
3287 * minor incompatible change: COMPILE-FILE now uses the freedom
3288 afforded (ANSI 3.2.2.3) to use derived function types for
3289 functions defined in the same file. This also permits the system
3290 to warn on static type mismatches and function
3291 redefinition. (Currently it does not work with high DEBUG level.)
3292 * minor incompatible change: VALUES declaration is disabled.
3293 * When issuing notes, the compiler now signals a condition of type
3294 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3295 restart for use in user handlers. It is expected that the
3296 COMPILER-NOTE condition will eventually become a condition
3297 supertype to a hierarchy of note types, which will then be
3298 handleable in a similar fashion. However, at the moment, no such
3299 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3300 but it's an implementation detail, not a classification for the
3302 * Changes in type checking closed the following bugs:
3303 ** type checking of unused values (192b, 194d, 203);
3304 ** template selection based on unsafe type assertions (192c, 236);
3305 ** type checking in branches (194bc).
3306 * A short form of VALUES type specifier has ANSI meaning (it has
3307 increased the number of situations when SBCL cannot perform type
3309 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3310 keywords or constants is permissible.
3311 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3312 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3313 classes. (thanks to Antonio Martinez)
3314 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3315 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3316 argument to a pretty-print stream. Instead, it leaves any such
3317 operations to DESCRIBE-OBJECT methods.
3318 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3320 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3321 lambda lists. (thanks to David Lichteblau)
3322 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3323 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3324 (thanks to Gerd Moellmann)
3325 * fixed bug 246: increased compilation speed of long
3326 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3328 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3330 * DEFCONSTANT now throws a condition of type
3331 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3332 constant to a non-EQL value; CONTINUE and ABORT restarts
3333 respectively change and preserve the value.
3334 * fixed bug 63: The code walker, part of the implementation of CLOS,
3335 is now better at handling symbol macros.
3336 * bug fix: There is no longer an internal implementation type named
3337 CL:LENGTH. (reported by Raymond Toy)
3338 * bug fix: In macro-like defining macros/special operators the
3339 implicit block does not enclose the lambda list.
3340 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3341 enclosed in parentheses) are not suitable as type specifiers, and
3342 their use properly signals an error now.
3343 * bug fix: An argument count mismatch for a type specifier in code
3344 being compiled no longer causes an unhandled error at compile
3345 time, but signals a compile-time warning.
3346 * fixed simple vector readable printing
3347 * bug fix: DESCRIBE takes more care over whether the class
3348 precedence list slot of a class is bound before accessing it.
3349 (reported by Markus Krummenacker)
3350 * bug fix: FORMATTER can successfully compile pretty-printer format
3351 strings which use variants of the ~* directive inside.
3352 * bug fix: SEARCH now applies its TEST predicate to the elements of
3353 the arguments in the correct order. (thanks to Wolfhard Buss)
3354 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3355 in type inference. (thanks to Robert E. Brown)
3356 * bug fix: Reading in symbols with an explicit package name of ""
3357 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3358 * fixed some bugs revealed by Paul Dietz' test suite:
3359 ** NIL is now allowed as a structure slot name.
3360 ** Arbitrary numbers, not just REALs, are allowed in certain
3361 circumstances in LOOP for-as-arithmetic clauses.
3362 ** Multiple class redefinitions before slot access no longer
3363 causes a type error.
3364 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3365 association between the name and a class.
3366 ** Generic functions with non-standard method-combination and over
3367 five methods all of which return constants no longer return NIL
3368 after the first few invocations. (thanks to Gerd Moellmann)
3369 ** CALL-NEXT-METHOD with no arguments now passes the original
3370 values of the arguments, even in the presence of assignment.
3371 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3373 ** DISASSEMBLE works with closures and funcallable instances.
3374 ** ADD-METHOD now returns the generic function, not the new method.
3375 ** FIND-METHOD signals an error if the lengths of the specializers
3376 is incompatible with the generic function, even if the ERRORP
3378 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3379 which its argument is a member.
3380 ** DEFCLASS only redefines the class named by its class-name
3381 argument if that name is the proper name of the class;
3382 otherwise, it creates a new class.
3383 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3384 of the UNBOUND-SLOT condition to the name of the slot.
3385 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3386 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3387 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3388 SLOT-MAKUNBOUND in the specified fashion.
3390 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3391 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3392 cross-compilation host. As a consequence, we can now bootstrap our
3393 way up to SBCL starting with a bare gcc toolchain and human-readable
3394 source code (first the source to CLISP, then the source to SBCL).
3395 * A contributed module containing a partial implementation of the
3396 simple-streams interface has been included. (thanks to Rudi
3398 * A contributed module implementing the RFC1321 Message Digest
3399 Algorithm, known as MD5, has been included.
3400 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3401 no longer has any effect, as the code controlled by this feature
3402 has been deleted. (As far as we know, no-one has ever built using
3403 this feature, and its semantics were confused in any case).
3404 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3405 work on conditions (as required by the ANSI specification),
3406 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3407 expected behaviour on conditions. Users should note, however,
3408 that such behaviour is not required by the ANSI specification,
3409 and so use of this behaviour may render their code unportable.
3410 * fixed some bugs revealed by Paul Dietz' test suite:
3411 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3413 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3414 ** SXHASH on condition objects no longer returns NIL.
3415 ** :ALLOCATION :CLASS slots are better treated; their values are
3416 updated on class redefinition, and initforms inherited from
3417 superclasses are applied.
3418 ** REMOVE-METHOD returns its generic function argument even when
3419 no method was removed.
3420 ** SHARED-INITIALIZE now initializes the values of the requested
3421 slots, including those with :ALLOCATION :CLASS.
3422 ** ALLOCATE-INSTANCE now works on structure classes defined via
3423 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3425 ** SLOT-EXISTS-P now works on conditions, as well as structures
3427 ** MAKE-LOAD-FORM now has the required methods on
3428 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3429 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3430 keyword, and now implements the SLOT-NAMES argument.
3431 ** methods with &OPTIONAL arguments no longer allow too many
3432 arguments to be passed in the call without error.
3433 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3434 option is consistent with the required arguments of the generic
3435 function lambda list.
3436 * bug fix: REQUIRE accepts a string designator. (Thanks to
3438 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3439 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3440 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3441 * bug fix: The compiler now checks for duplicated variables in macro
3443 * bug fix: SETQ on globals returns the correct value.
3444 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3445 not a COMPILER-ERROR (followed by some other strange error on
3446 choosing the CONTINUE restart).
3447 * bug fix: make.sh and friends are now more consistent in the way that
3448 they look for GNU "make".
3450 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3451 * experimental native threads support (on x86 Linux >=2.4 only).
3452 This is not compiled in by default: you need to add :SB-THREAD to
3453 the target features. See the "Beyond ANSI" chapter of the manual
3455 * fix for longstanding nonANSIism: The old distinction between
3456 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3457 The return value from CL:FIND-CLASS is now a CLOS class, and
3458 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3459 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3460 * An interface to the MetaObject Protocol, as described in Kiczales,
3461 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3462 MIT Press, 1991, is available from the SB-MOP package.
3463 * incompatible change: the SB-PCL package should now be considered
3464 a private implementation detail, and no longer a semi-private MOP
3466 * minor incompatible change: due to rearrangement for threads, the
3467 control stack and binding stack are now allocated at arbitrary
3468 addresses instead of being hardcoded per-port. Users affected by
3469 this probably have to be doing advanced things with shared
3470 libraries, and will know who they are.
3471 * minor incompatible change: Previously, all --eval forms used were
3472 processed with READ before any of them were processed with EVAL.
3473 Now each --eval form is processed with both READ and EVAL before
3474 the next --eval form is processed. (Thus package operations like
3475 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3476 work as the user might reasonably expect.)
3477 * minor incompatible change: *STANDARD-INPUT* is now only an
3478 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3480 * minor incompatible change: Y-OR-N-P is now character-oriented,
3481 not line oriented. Also, YES-OR-NO-P now works without errors.
3482 (thanks to Antonio Martinez)
3483 * sb-aclrepl module improvements: an integrated inspector, added
3484 repl features, and a bug fix to :trace command.
3485 * Known functions, which cannot be open coded by the backend, are
3486 considered to be able to check types of their arguments. (fixing
3487 a bug report by Nathan J. Froyd)
3488 * fixed a bug in computing method discriminating functions: It is
3489 now possible to define methods specialized on classes which have
3490 forward-referenced superclasses. (thanks to Gerd Moellmann)
3491 * fixed evaluation order in optional entries (reported by Gilbert
3493 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3494 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3495 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3496 required-by-AMOP NAME argument, as well as CLASS and
3497 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3498 * fixed bug 20: DEFMETHOD can define methods using names that are
3499 not the proper names of classes to designate class specializers.
3500 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3501 Unix file descriptors, instead of blowing up. (thanks to Antonio
3503 * Garbage collection refactoring: user-visible change is that a
3504 call to the GC function during WITHOUT-GCING will not do garbage
3505 collection until the end of the WITHOUT-GCING. If you were doing
3506 this you were probably losing anyway.
3507 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3508 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3509 TYPEP the latter but not the former.
3510 * The compiler issues a full WARNING for calls to undefined functions
3511 with names from the CL package.
3512 * MAP-INTO for a vector destination is open coded. (reported by
3513 Brian Downing on c.l.l)
3514 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3515 documentation string.
3516 * fixed some bugs revealed by Paul Dietz' test suite:
3517 ** COPY-ALIST now signals an error if its argument is a dotted
3519 ** Condition slots are now accessed more correctly in the presence
3520 of multiple initargs for a given slot.
3521 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3522 exclude restarts of the same name associated with a different
3524 ** DEFCLASS of forward-referenced classes with another
3525 forward-referenced class in the superclasses list no longer
3527 ** Condition slots are now initialized once each, not multiple
3528 times. (thanks to Gerd Moellmann)
3529 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3531 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3532 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3533 arguments contain duplicated elements.
3534 ** RESTART-CASE understands local macros.
3535 ** RESTART-CASE associates exactly its own restarts with a condition.
3536 ** ENDP in safe mode checks its argument to be of type LIST.
3537 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3538 in question is unbound.
3539 ** Optimized MAKE-INSTANCE functions no longer cause internal
3540 assertion failures in the presence of duplicate initargs.
3541 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3542 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3543 propagated even in the ctor optimized implementation of
3545 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3547 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3548 * a better implementation of SXHASH on (simple) bit vectors,
3549 measured both in execution speed and in distribution of results
3550 over the positive fixnums, has been installed. Likewise, a better
3551 implementation of EQUAL for simple bit vectors is now available.
3552 * fixed CEILING optimization for a divisor of form 2^k.
3553 * fixed bug 240 (emitting extra style warnings "using the lexical
3554 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3555 by Antonio Martinez)
3556 * fixed SXHASH, giving different results for NIL depending on type
3557 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3558 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3559 variable. (found by Rolf Wester)
3560 * SBCL does not ignore type declarations for special
3561 variables. (reported by rif on c.l.l 2003-03-05)
3562 * some bug fixes in contrib/sb-aclrepl/
3563 * fixed some bugs revealed by Paul Dietz' test suite:
3564 ** a bug in the CONS type specifier, whereby the CAR and CDR
3565 types got intertwined, has been fixed;
3566 ** the type system is now able to reason about the interaction
3567 between INTEGER and RATIO types more completely;
3568 ** APPEND, [N]REVERSE and NRECONC check that those their
3569 arguments, which must be proper lists, are really so;
3570 ** An array specialized to be unable to hold elements has been
3571 implemented, as required -- yes, really -- by ANSI;
3572 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3573 on malformed property lists;
3575 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3576 * incompatible packaging change: in line with Unix convention,
3577 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3578 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3579 * REQUIRE and PROVIDE are now optionally capable of doing something
3580 useful. See the documentation string for REQUIRE.
3581 * infrastructure for a managed SBCL contrib system: contributed
3582 modules in this release include:
3583 ** the ASDF system definition facility;
3584 ** an interface to the BSD Sockets API;
3585 ** an ACL-like convenience interface to the repl;
3586 (thanks to Kevin Rosenberg)
3587 ** an implementation of ROTATE-BYTE, with efficient implementation
3589 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3590 gives a closer approximation to the right answer than previously.
3591 (thanks to Raymond Toy)
3592 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3593 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3594 optional environment arguments, as required by ANSI.
3595 * fixed bugs in other functions taking environment objects, allowing
3596 calls with an explicit NIL environment argument to be compiled
3598 * fixed bug 228: primary return values from
3599 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3600 COMPILE or FUNCTION.
3601 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3602 :TYPE will no longer signal errors on innocuous objects.
3603 * fixed bug 231b: SETQ is better at respecting type declarations in
3604 the lexical environment.
3605 * fixed a bug in DEFCLASS: classes named by symbols with no or
3606 unprintable packages can now be defined.
3607 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3608 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3609 * fixed compiler failure related to checking types of functions.
3610 (reported by Robert E. Brown)
3611 * the compiler is now much more consistent in its error-checking
3612 treatment of bounding index arguments to sequence functions: in
3613 (SAFETY 3) code, errors will be signalled in almost all cases if
3614 invalid sequence bounding indices are passed to functions defined
3615 by ANSI to operate on sequences.
3616 * fixed a bug in the build procedure: documentation of SBCL-specific
3617 packages is now preserved and available in the final Lisp image.
3618 * lifted FDEFINITION lookup out of loops in the implementation of
3619 many list operations. (thanks to Robert E. Brown)
3620 * fixed a bug in the reader: the #n# reader macro now works for
3621 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3622 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3623 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3624 E. Brown and Tony Martinez respectively)
3625 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3626 SLOT-BOUNDP: in particular, it is now able to optimize them much
3627 better, and is now not vulnerable to having packages renamed.
3628 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3629 always called when a slot is not present in an instance. (thanks
3631 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3632 longer signal an error on unreading a character following EOF on
3633 the previous constituent stream. (thanks to Tony Martinez)
3634 * fixed some bugs revealed by Paul Dietz' test suite:
3635 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3636 not just nonnegative fixnums;
3637 ** the logical bit-array operators such as BIT-AND now accept an
3638 explicit NIL for their "opt-arg" argument (to indicate a
3639 freshly-consed result bit-array);
3640 ** ELT now signals an error on an invalid sequence index in safe
3642 ** the type system is now cleverer about negations of numeric
3643 types, and consequently understands the BIGNUM and RATIO types
3645 ** the type system is now cleverer about the interaction between
3646 INTEGER and RATIO types: while bugs still remain, many more
3647 cases are accurately computed;
3648 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3649 if it is in the last clause;
3650 ** CONSTANTLY now correctly returns a side-effect-free function in
3652 ** DECLARE is no longer treated as a special-operator; in
3653 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3654 * incremented fasl file version number due to the change in the
3655 DEFSTRUCT-SLOT-DESCRIPTION structure.
3657 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3658 * minor incompatible change: code processed by the "interpreter" or
3659 EVAL now has a compilation optimization policy of (DEBUG 2)
3660 (changed from (DEBUG 1)) to improve debuggability of interactive
3661 development, and to allow the use of the debug RETURN command in
3663 * an experimental implementation of the RETURN command for the
3664 debugger has been included. (thanks to Frederik Kuivinen)
3665 * fixed bug 62: constraints were not propagated into a loop.
3666 * fixed bug in embedded calls of SORT (reported and investigated by
3668 * fixed some bugs revealed by Paul Dietz' test suite:
3669 ** printing and reading of arrays with some dimensions having
3670 length 0 (thanks to Gerd Moellmann);
3671 ** BOA constructor with &AUX argument without a default value does
3672 not cause a type error;
3673 ** CONSTANTP now returns true for all self-evaluating objects.
3675 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3676 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3677 accessors that are related by inheritance, as specified in the
3678 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3679 Valtteri Vuorikoski)
3680 * The compiler is now able to inline functions that were defined in
3681 a complex lexical environment (e.g. inside a MACROLET).
3682 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3684 * Support for the upcoming FreeBSD-5.0 release has been included.
3685 (thanks to Dag-Erling Smorgrav)
3686 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3687 effect when it is not in a toplevel context.
3688 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3689 stand a better chance of being correct. (thanks to Gerd
3691 * fixed bug in COERCE, which now signals an error on coercing a
3692 rational to a bounded real type which excludes the expected
3694 * The compiler is now able to derive types more accurately from the
3695 COERCE and COMPILE functions.
3696 * fixed bug 223: functional binding is considered to be constant
3697 only for symbols in the CL package.
3698 * fixed bug 231: SETQ did not check the type of a variable being set
3699 (reported by Robert E. Brown)
3700 * A new optimization for MAKE-INSTANCE has been included, fixing
3701 various bugs (including relating to :ALLOCATION :CLASS slots and
3702 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3704 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3705 ** As required by ANSI, LOOP now disallows anonymous collection
3706 clauses such as COLLECT I in conjunction with aggregate boolean
3707 clauses such as THEREIS (= I 1);
3708 ** LOOP now signals an error when any variable is reused in the
3709 same loop (including the potentially useful construct analogous
3710 to WITH A = 1 WITH A = (1+ A);
3711 ** IT is only a special loop symbol within the first clause of a
3712 conditional loop clause;
3713 ** LOOP with a typed iteration variable over a hashtable now
3714 signals a type error iff it should.
3715 * fixed some other bugs revealed by Paul Dietz' test suite:
3716 ** FILE-STREAM now names the class previously known as FD-STREAM;
3717 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3718 argument) no longer signals an error;
3719 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3720 of the slot symbol, rather than using the current package
3721 ((:CONC-NAME "") continues to intern the slot's name in the
3723 * incremented fasl file version number, because of the incompatible
3724 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3725 of the new implementation of DEFINE-COMPILER-MACRO.
3727 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3728 * Support for building SBCL for MIPS platforms running in
3729 little-endian mode has now been checked in, and basic
3730 functionality on said platforms verified.
3731 * minor incompatible change: PCL now records the pathname of a file
3732 in which methods and the like are defined, rather than its
3734 * minor incompatible change: TRUENAME now considers the truename of
3735 a file naming a directory to be the pathname with :DIRECTORY
3736 component indicating that directory.
3737 * minor incompatible change: a NAMED clause in the extended form of
3738 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3739 reason for the previous behaviour is unclear.
3740 * more systematization and improvement of CLOS and MOP conformance
3741 in PCL (thanks to Gerd Moellman and Pierre Mai):
3742 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3744 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3745 primary methods with no specializers;
3746 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3748 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3749 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3750 FORWARD-REFERENCED-CLASSes; error reporting on
3751 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3753 ** SXHASH on CLOS instances now uses a slot internal to the
3754 instance to return different numbers on distinct instances,
3755 while preserving the same return value through invocations of
3757 ** DEFMETHOD signals errors when methods with longer incongruent
3758 lambda lists are added to generic functions;
3759 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3760 CLASS, as specified in AMOP;
3761 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3762 on the order returned by the primary method for classes of
3763 class STANDARD-CLASS;
3764 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3765 * fixed some bugs shown by Paul Dietz' test suite:
3766 ** DOLIST puts its body in TAGBODY;
3767 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3769 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3770 value producing form;
3771 ** if more variables are given to PROGV than values, extra
3772 variables are bound and made to have no value;
3773 ** NSUBSTITUTE on list arguments gets the right answer with
3775 ** ELT signals an error of type TYPE-ERROR when the index argument
3776 is not a valid sequence index;
3777 ** LOOP signals (at macroexpansion time) an error of type
3778 PROGRAM-ERROR when duplicate variable names are found;
3779 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3780 ** FUNCALL of special-operators now cause an error of type
3782 ** PSETQ now works as required in the presence of side-effecting
3783 symbol-macro places;
3784 ** NCONC accepts any object as its last argument;
3785 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3787 ** loop-for-as-package does not require a package to be explicitely
3789 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3791 * fixed bug 166: compiler preserves "there is a way to go"
3792 invariant when deleting code.
3793 * fixed bug 172: macro lambda lists with required arguments after
3794 &REST arguments now cause an error to be signalled. (thanks to
3796 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3797 bug. (thanks to Matthew Danish)
3798 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3800 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3801 arguments are lexically rebound. (thanks to Gerd Moellmann and
3803 * fixed bug 194: error messages are now more informative when there
3804 is no primary method applicable in a call to a generic
3805 function. (thanks to Gerd Moellmann)
3806 * fixed bug in command line argument checking (thanks to Julian
3808 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3809 (thanks to Matthew Danish)
3810 * incremented fasl file version number, because of the
3811 SXHASH-related changes in the layout of CLOS data structures
3813 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3814 * minor incompatible change: The runtime (the Unix executable named
3815 "sbcl") is now much pickier about the .core files it will load.
3816 Essentially it now requires .core files to descend from the same
3817 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3818 as the runtime does. (The intent is to prevent the crashes which
3819 can occur, and which can even be reported as mysterious failures,
3820 when people patch the sources or change the build parameters
3821 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3822 sbcl and .core files.)
3823 * fixed bug: VALUES-LIST is no longer optimized away.
3824 * fixed bug 142: The FFI conversion of C string values to Lisp
3825 string values no longer conses excessively. (thanks to Nathan
3826 Froyd porting Raymond Toy's fix to CMU CL)
3827 * began to systematize and improve MOP conformance in PCL (thanks to
3828 Nathan Froyd, Gerd Moellman and Pierre Mai):
3829 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3831 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3832 ** FINALIZE-INHERITANCE is now called on class finalization;
3833 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3834 argument precedence order.
3835 * fixed bug 202: The compiler no longer fails on functions whose
3836 derived types contradict their declared type.
3837 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3838 so it can be non-toplevel.
3839 * The fasl file version number has changed (because of the new
3840 implementation of DEFMACRO).
3841 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3842 safe code, that any length requirement by their type-specifier
3843 argument is valid. The exceptions to this are described in bug
3845 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3846 functions now check the types of their inputs as required by ANSI.
3847 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3848 introduced symbol is DECLAREd to be SPECIAL.
3849 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3850 * fixed bug: :COUNT argument to sequence functions may be negative
3851 * fixed bug: body of DO-SYMBOLS may contain declarations
3852 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3853 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3854 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3855 symbol macro only once
3856 * fixed printing of call frame when argument list is unavailable
3857 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3858 * compiler no longer signals WARNING on unknown keyword
3861 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3862 * A beta-quality port to the mips architecture running Linux,
3863 based on the old CMUCL backend, has been made. It has been tested
3864 on a big-endian kernel, and works sufficiently well to be able to
3865 rebuild itself; it has not been tested in little-endian mode.
3866 * fixed an inconsistency between gencgc.c and purify.c which made
3867 dumping/loading .core files unreliable
3868 * fixed bug 120a: The compiler now deals correctly with IFs where
3869 the consequent is the same as the alternative, instead of
3870 misderiving the return type. (thanks to Alexey Dejneka)
3871 * fixed bug 113: Logical pathnames are now dumpable (the logical
3872 host is resolved at load-time, throwing an error if it is not
3874 * fixed bug 174: FORMAT's error message is slightly clearer when a
3875 non-printing character is used in a format directive.
3876 * fixed several bugs in compiler checking of type declarations, i.e.
3877 violations of the Python "declarations are assertions" principle
3878 (thanks to Alexey Dejneka)
3879 * fixed several bugs in PCL's error checking (thanks to Gerd
3881 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3883 * fixed bug in compilation of functions as first class values
3884 (thanks to Antonio Martinez-Shotton)
3885 * The compiler's handling TYPE-ERRORs which it can prove will
3886 inevitably happen at runtime has been cleaned up and corrected
3887 in several ways. (thanks to Alexey Dejneka)
3888 * improved argument type checking for various basic arithmetic
3889 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3890 lavished on them in the past that they can be compiled in many
3891 ways in different special cases
3892 * fixed bug 181: compiler checks validity of user supplied type
3894 * cleaned up code flushing in optimization: Function calls which
3895 should signal errors for safety purposes (e.g. which ANSI says
3896 should signal errors when their arguments are of incorrect type)
3897 are no longer optimized away.
3898 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3899 * incremented fasl file version number, because changes in the
3900 implementation of sequence functions like COERCE caused
3901 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3902 in old inline expansions) to become undefined. (Actually these
3903 changes were later undone, so we might very well be binary
3904 compatible with 0.7.7 after all, but leaving the version number
3905 incremented seemed like the simplest and most conservative
3908 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3909 * An alpha-quality port to the parisc architecture running Linux,
3910 based on the old CMUCL backend, has been made. This, even more so
3911 than the other backends, should be considered still a work in
3912 progress; known problems include that the Linux kernel in 64-bit
3913 mode does not propagate the correct sigcontext structure to
3914 userspace, and consequently SBCL on a parisc64 kernel will not
3916 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3917 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3918 suppressed.) Also now that the compiler is looking at declarations
3919 in the environment, it checks optimization declarations as well,
3920 and suppresses inlining when (> DEBUG SPEED).
3921 * More fixes have been made to treatment of floating point exception
3922 treatment and other Unix signals. In particular, floating point
3923 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3924 * The detection and handling of control stack exhaustion (infinite
3925 or very deeply nested recursion) has changed. Stack exhaustion
3926 detection is now done by write-protecting pages at the OS level
3927 and applies at all optimization settings; when found, a
3928 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3929 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3930 be caught using IGNORE-ERRORS.
3931 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3932 correctly when transforming recursive LABELS functions to LETs.
3933 (thanks to Alexey Dejneka)
3934 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3935 that are names of constants or global variables.
3936 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3937 alien routines with docstrings.
3938 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3939 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3941 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3942 (thanks to Eric Marsden)
3943 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3944 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3945 * Bug fix: The SPARC backend can now compile functions involving
3946 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3947 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3948 to a functions expecting a PATHNAME-DESIGNATOR.
3949 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3950 lambda lists. (thanks to Alexey Dejneka)
3951 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3952 functions so that EXPORTing the name of the class doesn't cause
3953 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3954 to Antonio Martinez for reporting this)
3955 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3956 constant-folded without failing an assertion. (thanks to Einar
3957 Floystad Dorum for reporting this)
3958 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3959 types are now dealt with more correctly by the compiler.
3960 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3961 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3962 behaviour analogous to RENAME-FILE. This puts its behaviour more
3963 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3965 * The fasl file version number has changed again. (because of the
3966 bug fix involving the names of PCL MAKE-INSTANCE functions)
3968 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3969 * bug fix: Floating point exceptions are treated much more
3970 consistently on the x86/Linux and PPC/Linux platforms.
3971 * Array initialization with :INITIAL-ELEMENT is now much faster for
3972 cases when the compiler cannot open code the array creation, but
3973 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3974 array accesses have also seen a speed increase.
3975 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3976 specified. (This was at the root of some bad interactions between
3977 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3979 * bug fix: Internal error arguments for undefined functions are now
3980 computed correctly on the PPC/Linux platform.
3981 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3982 Raymond Toy's patch for CMU CL)
3983 * Support for the Solaris 9 operating environment has been included
3984 (thanks to Daniel Merritt)
3985 * A very ugly but hopefully complete draft of the missing FFI chapter
3986 of the manual has been created by reformatting the corresponding
3987 CMU CL manual chapter into (currently very ugly and incoherent)
3988 DocBook and bringing it up to date for SBCL behavior. Thus, the
3989 manual is now essentially complete, at least by my extreme
3990 once-and-only-once standards, whereby it's acceptable to refer to
3991 the doc strings of SB-EXT functions as the primary documentation.
3992 * The fasl file version number has changed again, due to cleanup of
3993 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3994 is no longer a static symbol.)
3996 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3997 * SBCL now builds with OpenMCL (version 0.12) as the
3998 cross-compilation host; also, more progress has been made toward
3999 bootstrapping under CLISP.
4000 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4002 * bug 158 fixed: The compiler can now deal with integer loop
4003 increments different from 1; fixing this turned out also to fix
4005 * bug 169 fixed: no more bogus warnings about using lexical bindings
4006 despite the presence of perfectly good SPECIAL declarations (thanks
4007 to David Lichteblau)
4008 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4009 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4010 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4012 * bug 180 fixed: Method combination specifications no longer ignore
4013 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4014 * bug fix: Structure type predicate functions now check their argument
4015 count as they should.
4016 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4017 correctly. (thanks to Pierre Mai)
4018 * minor incompatible change: The --noprogrammer option is deprecated
4019 in favor of the new --disable-debugger option, which is very similar.
4020 (The major difference is that it takes effect at a slightly different
4021 time at startup, causing handling of errors in --sysinit and
4022 --userinit files will be affected differently.) The
4023 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4024 been added to allow this functionality to be controlled from ordinary
4025 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4026 maintainers, who might want to run non-interactive scripts to
4027 build SBCL cores which will later be used interactively.)
4028 * minor incompatible change: The LOAD function no longer, when given
4029 a wild pathname to load, loads all files matching that pathname.
4030 Instead, an error of type FILE-ERROR is signalled.
4032 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4033 * bug 147 fixed: The compiler preserves its block link/count
4034 invariants more correctly now so that it doesn't crash. (thanks
4036 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4038 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4039 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4040 no-value-for-this-character as NIL (as specified by ANSI).
4041 * HOST-NAMESTRING on physical pathnames now returns a string that is
4042 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4043 (thanks to Christophe Rhodes)
4044 * The Alpha port handles icache flushing more correctly. (thanks to
4046 * More progress has been made toward bootstrapping under CLISP. (thanks
4047 to Christophe Rhodes)
4048 * The fasl file format has changed again, because dynamic loading
4049 on OpenBSD (which has non-ELF object files) motivated some cleanups
4050 in the way that foreign symbols are transformed and passed around.
4051 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4052 is no longer treated as whitespace by the reader, but instead as
4053 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4054 |A
\7fB|, instead of A as it used to.
4056 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4057 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4058 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4059 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4060 this as of 0.7.1.45, but was left out of the previous news section
4061 (thanks to Dan Barlow)
4062 * SBCL now runs on the Solaris operating system on SPARC architectures
4063 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4064 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4065 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4066 * SPARC backend cleanups, allowing builds of cores optimized for V8
4067 and V9 SPARCS, and also emission of code targeted to a particular
4068 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4070 * SBCL is closer to bootstrapping under CLISP, thanks to various
4071 fixes by Christophe Rhodes.
4072 * The fasl file format has changed again, to allow the compiler's
4073 INFO database to support symbol macros.
4074 * The user manual (in doc/) is formatted into HTML more nicely.
4075 (thanks to coreythomas)
4076 * The system is smarter about SUBTYPEP relationships, especially
4077 those involving NOT types (including types such as ATOM which are
4078 represented internally using NOT types). Thus SUBTYPEP is less
4079 likely to return (VALUES NIL NIL) in general, and in particular
4080 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4082 * The fasl file format has changed again, because the internal
4083 representation of types now includes a new slot to support the new
4084 SUBTYPEP-of-NOT-types logic.
4085 * (not a change in the main branch of SBCL, but a related prototype
4086 which can hopefully be merged into the main branch of SBCL in the
4087 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4088 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4089 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4090 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4091 dispatch macro character combinations. (thanks to Alexey Dejneka)
4092 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4093 that superficially look like logical namestrings correctly.
4094 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4096 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4097 default supplied-p) for &optional and &key arguments. (thanks to
4100 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4101 * incompatible change: The compiler is now less aggressive about
4102 tail call optimization, doing it only when (> SPACE DEBUG) or
4103 (> SPEED DEBUG). (This is an incompatible change because there are
4104 programs which relied on the old CMU-CL-style behavior to optimize
4105 away their unbounded recursion which will now die of stack overflow.)
4106 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4107 for non-GENCGC systems has been increased to 20M (since that
4108 seems much closer to the likely performance optimum for modern
4109 systems than the old 4M value was)
4110 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4111 and *DEBUG-PRINT-LEVEL*
4112 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4113 of CMU CL's support for SPARC, and various endianness and other
4114 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4115 * new syntactic sugar for the Unix command line: --load foo.bar is now
4116 an alternate notation for --eval '(load "foo.bar")'.
4118 ** The system now detects stack overflow and handles it gracefully,
4119 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4120 optimization settings. (This is a good thing in general, and
4121 its introduction in this version should be particularly timely
4122 for anyone whose code fails because of suppression of tail
4124 ** The system now hunts for the C variable "environ" in a more
4125 devious way, to avoid segfaults when the C library version differs
4126 between compile time and run time. (thanks to Christophe Rhodes)
4127 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4128 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4129 ** The compiler no longer issues bogus style warnings for undefined
4130 classes in the same source file as the DEFCLASSes which defined
4131 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4133 ** fixes in CONDITION class precedence list for undefined function
4134 errors (thanks to Alexei Dejneka)
4135 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4136 correctly. (thanks to Dan Barlow)
4137 ** portability fixes aiming at bootstrapping under CLISP (thanks
4138 to Dave McDonald and Christophe Rhodes)
4139 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4140 ** fixes in type translation and and type inference (thanks to
4142 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4143 ** various fixes in the new ports (thanks to Dan Barlow)
4144 * several changes related to debugging:
4145 ** suppression of tail recursion, as noted above
4146 ** stack overflow detection, as noted above
4147 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4148 is now the default. (For some time encapsulation has been more
4149 reliable than the breakpoint-based :ENCAPSULATE NIL
4150 implementation, at least on X86 systems; and I just noticed that
4151 encapsulation also seems closer to the spirit of the ANSI
4154 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4156 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4157 up properly again. (There was a packaging bug in 0.7.0 which
4158 left their definitions in SB-SYS::LOAD-FOREIGN and
4159 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4160 vital for most things which interface to C-level interfaces,
4161 like extensions working with sockets or databases or
4162 Perl-compatible regexes or whatever, and the need to fix
4163 this bug is the main reason that 0.7.1 was released so
4165 ** DEFGENERIC is now choosier about the methods it redefines, so that
4166 reLOADing a previously-LOADed file containing DEFGENERICs does
4167 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4168 cycle now works as it should. (thanks to Alexey Dejneka)
4169 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4170 fixed by Christophe Rhodes. (He actually submitted this patch
4171 months ago, and I delayed until after 0.7.0.)
4172 ** Bug 111 (internal compiler confusion about runtime checks on
4173 FUNCTION types) was fixed by Alexey Dejneka.
4174 * Some internal cleanups (getting rid of variables which aren't
4175 needed now that the byte interpreter is gone) caused the fasl
4176 file format number to change again.
4178 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4179 * major incompatible change: The default fasl file extension, i.e. the
4180 default extension for files produced by COMPILE-FILE, has changed
4181 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4183 ** There are many changes in the implementation of the compiler.
4184 SBCL is now essentially a compiler-only implementation of ANSI
4185 Common Lisp. EVAL still "interprets" a few special cases, but
4186 almost all the interesting cases are handled by creating
4187 a LAMBDA expression, calling COMPILE on it, then calling
4188 FUNCALL on the result.
4189 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4190 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4191 Since the code is newer, there might still be some new bugs
4192 (though not as many as before Martin Atzmueller's fixes:-). But
4193 the new code is substantially simpler and clearer, and hopefully
4194 any remaining bugs will be simpler, less fundamental, and more
4195 fixable then the bugs in the old code.
4196 ** The revised compiler is still a little unsteady on its feet.
4198 *** The debugging information it produces (particularly the names
4199 of FUNCTION objects) is sometimes much less useful than what
4200 the old compiler produced.
4201 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4202 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4203 within a MACROLET) has been temporarily weakened.
4204 ** There are new compiler optimizations for various functions:
4205 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4206 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4207 *** the math functions TRUNCATE, FLOOR, and CEILING
4208 *** the function-of-all-trades COERCE
4209 Mostly these should be transparent, but there's one
4210 potentially-annoying problem (bug 117): when the compiler
4211 inline-expands a function and does type analysis on the result,
4212 it can create control paths which have type mismatches, and
4213 when it can't prove that those control paths aren't taken,
4214 it will issue WARNINGs about the type mismatches. This is
4215 a particular problem in practice for the new sequence functions.
4216 It's not clear how this should be fixed, and for now, a
4217 workaround is given in the entry for 117 in the BUGS file.
4218 ** (Because of the interaction between the two previous items --
4219 occasional inlining problems and new inline expansions -- some
4220 of the new sequence function optimizations won't really kick in
4221 completely until debugging information, and then inlining, are
4222 straightened out in some future version.)
4223 * minor incompatible changes:
4224 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4225 DIRECTORY no longer implicitly promotes NIL slots of its
4226 pathname argument to :WILD. In particular, when you ask for the
4227 contents of a directory (which you used to be able to do without
4228 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4229 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4230 ** changes in behavior that ANSI explicitly defines to be
4231 implementation dependent:
4232 *** The new compiler-only implementation still conforms with ANSI,
4233 but acts a little different than before. Besides the obvious
4234 changes in performance tradeoffs (that the cost per form passed
4235 to EVAL has gone up, and the cost per form executed by EVAL
4236 has gone down), the behavior of the system changes a little
4237 because there are no longer any interpreted function objects.
4238 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4239 e.g. doing COMPILE on the output of interactive DEFUN is
4241 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4243 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4244 has changed from (:CL) to NIL.
4245 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4246 e.g. #\Newline and #\Tab, don't have names specified in the
4247 ANSI Common Lisp standard, is now based on their ASCII symbolic
4248 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4249 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4250 are no longer used for output.
4251 ** changes in internal implementation constants:
4252 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4253 4 million. (If your application spends a lot of time GCing and
4254 you have a lot of RAM, you might want to experiment with
4255 increasing it even more.)
4256 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4257 However, almost all old code should still continue to work without
4258 immediate update, as SB-C-CALL is now a (deprecated) nickname
4260 ** Old operator names in the style DEF-FOO are now deprecated in
4261 favor of new corresponding names DEFINE-FOO, for consistency with
4262 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4263 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4264 internal symbols, but a few supported extensions like
4265 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4266 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4267 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4268 etc. as you get deeper into recursive calls to the debugger
4269 command loop, instead of the old "5]", "5]]", "5]]]"
4270 sequence. (I was motivated to do this when squabbles between
4271 ILISP and SBCL left me very deeply nested in the debugger. In the
4272 short term, this change will probably provoke more ILISP/SBCL
4273 squabbles, but hopefully it will be an improvement in the long run.)
4274 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4275 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4276 function, since declaiming return types involving aliens is
4277 (1) annoyingly messy to do by hand and (2) vital to efficient
4278 compilation of code which calls such functions.
4279 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4280 longer reexported by the SB-EXT package. They're solely useful
4281 for alien code, so it seems more logical that you should get
4282 them from the SB-ALIEN package, not in SB-EXT.
4283 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4284 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4285 features. Instead, the code that they used to control is always
4286 built into the system.
4287 * many other bug fixes
4288 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4289 advantage of the new EVAL-WHEN stuff and to clean them up in
4290 general, and they are now more ANSI-compliant in a number of
4291 ways. Martin Atzmueller is responsible for a lot of this.
4292 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4294 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4295 *** correct ERROR type for various file operations
4296 *** some fixes for Lisp streams
4297 *** DEFMETHOD syntax checking
4298 *** changing old weird representation of debug information as
4299 strings (which, among their other deficiencies, don't transform
4300 correctly when you rename packages, and don't change their
4301 print representation when you change things like *PACKAGE*
4302 and *PRINT-LENGTH*) to symbols and lists of symbols
4303 He also made several improvements and fixed several bugs in DESCRIBE.
4304 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4306 *** misbehavior of WRITE-STRING/WRITE-LINE
4307 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4308 and several other LOOP problems as well
4309 *** DIRECTORY when similar filenames are present
4310 *** DEFGENERIC with :METHOD options
4311 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4312 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4313 *** argument ordering in FIND with :TEST option
4314 *** mishandled package designator argument in APROPOS-LIST
4315 *** various problems in the backquote readmacro
4316 *** a bug in APROPOS
4317 *** probably some others that I'm not describing very well here,
4318 since the CVS log documents them by reference to sbcl-devel
4319 messages, and the SourceForge archives aren't working well.:-(
4320 ** Dan Barlow improved the Alpha port (and is making progress on the
4321 PPC port, for those of you who think different).
4322 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4323 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4324 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4325 continues to work on the SPARC port (for those of you in a position
4326 to look down upon our little PC-compatible boxes from a great height).
4327 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4328 and uses that copy, rather than the current dynamic value, when
4329 it's trying to decide whether to truncate output. Thus e.g.
4330 (let ((*print-lines* 50))
4331 (pprint-logical-block (stream nil)
4333 (let ((*print-lines* 8))
4334 (print (aref possiblybigthings i) stream)))))
4335 should now truncate the logical block only at 50 lines, instead of
4336 often truncating it at 8 lines, as it did before.
4337 * The doc/cmucl/ directory, containing old CMU CL documentation
4338 from the time of the fork, is no longer part of the base system.
4339 SourceForge has shut down its anonymous FTP service, and with it
4340 my original plan for distributing the old CMU CL documentation
4341 there. For now, if you need these files you can download an old
4342 SBCL source release and extract them from it.
4343 * The fasl file version number changed again, for dozens of reasons,
4344 some of which are apparent above.
4346 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4347 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4348 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4349 patch, so that DISASSEMBLE output is much nicer.
4350 * The code in the SB-PROFILE package now seems reasonably stable.
4351 I still haven't decided what the final interface should look like
4352 (I'd like PROFILE to interact cleanly with TRACE, since both
4353 facilities use function encapsulation) but if you have a need
4354 for profiling now, you can probably use it successfully with
4355 the current CMU-CL-style interface.
4356 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4357 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4358 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4359 link now returns the dangling link itself, and for similar
4360 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4361 link itself. (In these cases the old code signalled an error and
4362 looped endlessly, respectively.) Thus, DIRECTORY now works even
4363 in the presence of dangling and cyclic symbolic links.
4364 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4365 is now a supported extension again, since the consensus on
4366 sbcl-devel was that it can be useful for ordinary development
4367 work, not just for debugging SBCL itself.
4368 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4369 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4370 that currently-defined functions might be redefined later with
4371 different return types.
4372 * Hash tables can be printed readably, as inspired by CMU CL code
4373 of Eric Marsden and SBCL code of Martin Atzmueller.
4374 * better error handling in CLOS method combination, thanks to
4375 Martin Atzmueller porting Pierre Mai's CMU CL patches
4376 * more overflow fixes for >16Mbyte I/O buffers
4377 * A bug in READ has been fixed, so that now a single Ctrl-D
4378 character suffices to cause end-of-file on character streams.
4379 In particular, now you only need one Ctrl-D at the command
4380 line (not two) to exit SBCL.
4381 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4383 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4384 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4385 does the right thing.
4386 * The implementation of some type tests, especially for CONDITION
4387 types, is now tidier and maybe faster, due to CMU CL code
4388 originally by Douglas Crosher, ported by Martin Atzmueller.
4389 * Some math functions have been fixed, and there are new
4390 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4391 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4392 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4393 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4394 likely to be built into sbcl-0.7.0.)
4395 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4396 CL, ported to SBCL by Martin Atzmueller)
4397 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4398 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4399 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4400 * There's a new slam.sh hack to shorten the edit/compile/debug
4401 cycle for low-level changes to SBCL itself, and a new
4402 :SB-AFTER-XC-CORE target feature to control the generation of
4403 the after-xc.core file needed by slam.sh.
4404 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4405 COMPILE-FILE is no longer supported, so that now every function
4406 gets an entry point, so that block compilation looks a little
4407 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4408 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4409 returns the number of bytes consed since the system started,
4410 rather than the number consed since the first time the function
4411 was called. (The new definition parallels ANSI functions like
4412 CL:GET-INTERNAL-RUN-TIME.)
4413 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4414 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4415 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4416 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4417 * Fasl file version numbers are now independent of the target CPU,
4418 since historically most system changes which required version
4419 number changes have affected all CPUs equally. Similarly,
4420 the byte fasl file version is now equal to the ordinary
4423 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4424 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4425 is no longer recognized. I apologize for this, because it was
4426 listed in SB-EXT as a supported extension, but I found that
4427 its existing behavior was poorly specified, as well as incorrectly
4428 specified, and it looked like too much of a mess to straighten it
4429 out. I have enough on my hands trying to get ANSI stuff to work..
4430 * many patches ported from CMU CL by Martin Atzmueller, with
4431 half a dozen bug fixes in pretty-printing and the debugger, and
4432 half a dozen others elsewhere
4433 * fixed bug 13: Floating point infinities are now supported again.
4434 They might still be a little bit flaky, but thanks to bug reports
4435 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4436 as flaky as they were.
4437 * The --noprogrammer command line option is now supported. (Its
4438 behavior is slightly different in detail from what the old man
4439 page claimed it would do, but it's still appropriate under the
4440 same circumstances that the man page talks about.)
4441 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4442 are now supported, and enabled by default. Thus, the compiler can
4443 handle many floating point and complex operations much less
4444 inefficiently. (Thus e.g. you can implement a complex FFT
4446 * The compiler now detects type mismatches between DECLAIM FTYPE
4447 and DEFUN better, and implements CHECK-TYPE more correctly, and
4448 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4449 thanks to patches from Martin Atzmueller.
4450 * various fixes to make the cross-compiler more portable to
4451 ANSI-conforming-but-different cross-compilation hosts (notably
4452 Lispworks for Windows, following bug reports from Arthur Lemmens)
4453 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4454 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4455 fixed thanks to Pierre Mai's CMU CL patches.
4456 * fixes to keep the system from overflowing internal counters when
4457 it tries to use i/o buffers larger than 16M bytes
4458 * fixed bug 45a: Various internal functions required to support
4459 complex special functions have been merged from CMU CL sources.
4460 (When I was first setting up SBCL, I misunderstood a compile-time
4461 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4462 * improved support for type intersection and union, fixing bug 12
4463 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4464 more obscure bugs as well
4465 * some steps toward byte-compiling non-performance-critical
4466 parts of the system, courtesy of patches from Martin Atzmueller
4467 * Christophe Rhodes has made some debian packages of sbcl at
4468 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4469 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4470 stable, but are nonetheless usable. When he's ready, I'd be happy
4471 to add them to the SourceForge "File Releases" section. (And if
4472 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4473 * new fasl file format version number (because of changes in
4474 internal representation of (OR ..) types to accommodate the new
4475 support for (AND ..) types, among other things)
4477 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4478 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4480 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4482 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4483 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4484 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4485 * DESCRIBE now works on structure objects again.
4486 * Most function call argument type mismatches are now handled as
4487 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4488 know whether the function will be redefined before the call is
4489 executed. (The compiler could flag local calls with full WARNINGs,
4490 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4491 it doesn't keep track of enough information to know whether calls
4492 are local in this sense.)
4493 * Compiler output is now more verbose, with messages truncated
4494 later than before. (There should be some supported way for users
4495 to override the default verbosity, but I haven't decided how to
4496 provide it yet, so this behavior is still controlled by the internal
4497 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4498 src/compiler/ir1util.lisp.)
4499 * Fasl file format version numbers have increased again, because
4500 support for the Gray streams extension changes the layout of the
4501 system's STREAM objects.
4502 * The Gray subclassable streams extension now works, thanks to a
4503 patch from Martin Atzmueller.
4504 * The full LOAD-FOREIGN extension (not just the primitive
4505 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4506 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4507 but like most other programs, it defaults to copying the Unix
4508 environment from the original process instead of starting the
4509 new process in an empty environment.
4510 * Extensions which manipulate the Unix environment now support
4511 an :ENVIRONMENT keyword option which doesn't smash case or
4512 do other bad things. The CMU-CL-style :ENV option is retained
4513 for porting convenience.
4514 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4515 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4517 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4519 * A patch from Martin Atzmueller seems to have solved the SIGINT
4520 problem, and as far as we know, signal-handling now works cleanly.
4521 (If you find any new bugs, please report them!)
4522 * The system no longer defaults Lisp source file names to types
4523 ".l", ".cl", or ".lsp", but only to ".lisp".
4524 * The compiler no longer uses special default file extensions for
4525 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4526 seems to expect a single default extension for all compiled code,
4527 and there's no compelling reason to try to stretch the standard
4528 to allow two different extensions.) Instead, byte-compiled files
4529 default to the same extension as native-compiled files.
4530 * Fasl file format version numbers have increased again, because
4531 a rearrangement of internal implementation packages made some
4532 dumped symbols in old fasl files unreadable in new cores.
4533 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4534 many fewer weird special cases.
4535 * Bug #17 (differing COMPILE-FILE behavior between logical and
4536 physical pathnames) has been fixed, and some related misbehavior too,
4537 thanks to a patch from Martin Atzmueller.
4538 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4539 by Tim Moore, ported to SBCL by Martin Atzmueller.
4540 * Martin Atzmueller fixed several filesystem-related problems,
4541 including bug #36, in part by porting CMU CL patches, which were
4542 written in part by Paul Werkowski.
4543 * More compiler warnings in src/runtime/ are gone, thanks to
4544 more patches from Martin Atzmueller.
4545 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4548 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4550 * DESCRIBE now works on CONDITION objects.
4551 * The debugger now handles errors which arise when trying to print
4552 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4554 * The build system now uses an additional file, customize-target-features.lisp,
4555 to allow local modifications to the target *FEATURES* list. (The point of
4556 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4557 debugging features enabled, without having to worry about propagating it
4558 into everyone's system when I do a "cvs update".) When no
4559 customize-target-features.lisp file exists, the target *FEATURES* list
4560 should be constructed the same way as before.
4561 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4562 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4563 warn about behavior which is undefined under the ANSI spec. Note: This
4564 is specified by ANSI, but it's not very popular with programmers.
4565 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4566 macro in the SBCL sources for an example of a workaround which you
4567 might use to make portable ANSI-standard code which does what you want.
4568 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4569 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4570 when it's used as a non-toplevel form. (This is required in order
4571 to implement the DEFCONSTANT-EQX macro.)
4572 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4573 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4574 funny things, due to bugs in the implementation of EVAL-WHEN
4575 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4576 non-toplevel DEFCONSTANTs are uncommon.)
4577 * The core file version number and fasl file version number have been
4578 incremented, because the old noncompliant DEFCONSTANT behavior involved
4579 calling functions which no longer exist, and because I also took the
4580 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4581 * fixed bug 1 (error handling before read-eval-print loop starts), and
4582 redid debugger restarts and related debugger commands somewhat while
4584 ** The QUIT debugger command is gone, since it did something
4585 rather different than the SB-EXT:QUIT command, and since it never
4586 worked properly outside the main toplevel read/eval/print loop.
4587 Invoking the new TOPLEVEL restart provides the same functionality.
4588 ** The GO debugger command is also gone, since you can just invoke
4589 the CONTINUE restart directly instead.
4590 ** The TOP debugger command is also gone, since it's redundant with the
4591 FRAME 0 command, and since it interfered with abbreviations for the
4593 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4595 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4596 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4597 some porting work by Martin Atzmueller)
4598 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4599 with a boatload of patches which he ported to SBCL. Now that those
4601 ** The system tries to make sure that its low-priority messages
4602 are prefixed by semicolons, to help people who like to use
4603 syntax highlighting in their ILISP buffer. (This patch
4604 was originally due to Raymond Toy.)
4605 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4606 patches originally written by Raymond Toy.
4607 ** The compiler understands coercion between single-value and
4608 multiple-VALUES type expressions better, getting rid of some very
4609 weird behavior, thanks to patches originally by Robert MacLachlan
4610 and Douglas Crosher.
4611 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4612 lambda lists, thanks to a patch originally by Pierre Mai.
4613 ** The system no longer bogusly warns about "abbreviated type
4615 ** The compiler gets less confused by inlining and RETURN-FROM,
4616 thanks to some patches originally by Tim Moore.
4617 ** The system no longer hangs when dumping circular lists to fasl
4618 files, thanks to a patch originally from Douglas Crosher.
4619 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4620 undefined function error.
4621 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4622 in structure slot initforms) for now, documented workaround instead:-|
4623 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4624 * fixed bug 5: added stubs for various Gray stream functions called
4625 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4626 installed, at least appropriate type errors are generated
4627 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4628 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4629 consistently in DEFMETHOD forms.
4630 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4631 it doesn't seem to affect SBCL after all
4632 * The C runtime system now builds with better optimization and many
4633 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4635 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4637 * The system is now under CVS at SourceForge (instead of the
4638 CVS repository on my home machine).
4639 * The new signal handling code has been tweaked to treat register
4640 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4641 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4642 downstream, e.g. in debug-int.lisp, has implicit dependencies
4643 on the unsignedness of integer representation of machine words,
4644 and that caused the system to bomb out with infinite regress
4645 when trying to recover from type errors involving signed values,
4646 e.g. (BUTLAST '(1 2 3) -1).)
4647 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4648 (This was one of the bugs Peter Van Eynde reported back in July.)
4649 * The system now uses code inspired by Colin Walters' O(N)
4650 implementation of MAP (from the cmucl-imp@cons.org mailing
4651 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4652 inline the MAP operation, and there is more than one
4653 sequence argument to the MAP call (so that it can't just
4654 do ETYPECASE once and for all based on the type of the
4655 single sequence argument). (The old non-inline implementation
4656 of the general M-argument sequence-of-length-N case required
4657 O(M*N*N) time when any of the sequence arguments were LISTs.)
4658 * The QUIT :UNIX-CODE keyword argument has been renamed to
4659 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4660 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4662 * Raymond Wiker's patches to port dynamic loading from Linux to
4663 FreeBSD have been added.
4664 * The BUGS file is now more nearly up to date, thanks in large part
4665 to Martin Atzmueller's review of it.
4666 * The debugger now flushes standard output streams before it begins
4667 its output ("debugger invoked" and so forth).
4668 * The core version number and fasl file version number have both
4669 been incremented, because of incompatible changes in the layout
4671 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4672 used to not be called for a saved Lisp image.)
4673 * Martin Atzmueller's version of a patch to fix a compiler crash,
4674 as posted on sbcl-devel 13 September 2000, has been installed.
4675 * Instead of installing Martin Atzmueller's patch for the
4676 compiler transform for SUBSEQ, I deleted the compiler transform,
4677 and transforms for some similar consing operations.
4678 * A bug in signal handling which kept TRACE from working on OpenBSD
4680 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4681 away by constant folding
4682 * The system now defines its address space constants in one place
4683 (in the Lisp sources), and propagates them automatically elsewhere
4684 (through GENESIS and the sbcl.h file). Therefore, patching the
4685 address map is less unnecessarily tedious and error-prone. The
4686 Lisp names of address space constants have also been systematized.
4687 * CVS tags like dollar-Header-dollar have been removed from
4688 the sources, because they have never saved me trouble and
4689 they've been source of trouble working with patches and other
4690 diff-related operations.
4691 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4692 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4694 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4696 * The system has been ported to OpenBSD.
4697 * The system now compiles with a simple "sh make.sh" on the systems
4698 that it's supported on. I.e., now you no longer need to tweak
4699 text in the target-features.lisp-expr and symlinks in src/runtime/
4700 by hand, the make.sh takes care of it for you.
4701 * The system is no longer so grossly inefficient when compiling code
4702 involving vectors implemented as general (not simple) vectors (VECTOR T),
4703 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4704 which dares to use the various sequence functions on non-simple
4705 vectors, takes less of a performance hit.
4706 * There is now a primitive type predicate VECTOR-T-P
4707 to test for the (VECTOR T) type, so that e.g.
4708 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4709 can now be compiled with some semblance of efficiency. (The old code
4710 turned the type declaration into a full call to %TYPEP at runtime!)
4711 * AREF on (VECTOR T) is still not fast, since it's still compiled
4712 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4713 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4714 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4715 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4716 without making full calls to SUBTYPEP at runtime.
4717 (Some analogous efficiency issues for non-simple vectors specialized to
4718 element types other than T, or for non-simple multidimensional arrays,
4719 have not been addressed. They could almost certainly be handled the
4720 same way if anyone is motivated to do so.)
4721 * The changes in array handling break binary compatibility, so
4722 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4723 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4724 instead of (VALUES T T).
4725 * By following the instructions that Dan Barlow posted to sbcl-devel
4726 on 2 July 2000, I was able to enable primitive dynamic object
4727 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4728 functionality is not implemented (since it calls ld to resolve
4729 library references automatically, requiring RUN-PROGRAM for its
4730 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4731 try to resolve library references) is now supported.
4732 * The system now flushes the standard output streams when it terminates,
4733 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4734 them at several other probably-convenient times, e.g. in each pass of
4735 the toplevel read-eval-print loop, and after evaluating a form given
4736 as an "--eval" command-line option. (These changes were motivated by a
4737 discussion of stream flushing issues on cmucl-imp in August 2000.)
4738 * The source transform for TYPEP of array types no longer assumes
4739 that an array whose element type is a not-yet-defined type
4740 is implemented as an array of T, but instead punts, so that the
4741 type will be interpreted at runtime.
4742 * There is now some support for cross-compiling in make.sh: each of
4743 the phases of make.sh has its own script. (This should be transparent
4744 to people doing ordinary, non-cross-compile builds.)
4745 * Since my laptop doesn't have hundreds of megabytes of memory like
4746 my desktop machine, I became more motivated to do some items on
4747 my to-do list in order to reduce the size of the system a little:
4748 ** Arrange for various needed-only-at-cold-init things to be
4749 uninterned after cold init. To support this, those things have
4750 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4751 symbols with such names are now uninterned after cold init).
4752 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4753 fasl files for cold load.
4754 ** Remove the old compiler structure pooling code (which used to
4755 be conditional on the target feature :SB-ALLOC) completely.
4756 ** Redo the representation of some data in cold init to be more compact.
4757 (I also looked into supporting byte compiled code at bootstrap time,
4758 which would probably reduce the size of the system a lot, but that
4759 looked too complicated, so I punted for now.)
4760 * The maximum signal nesting depth in the src/runtime/ support code has
4761 been reduced from 4096 to 256. (I don't know any reason for the very
4762 large old value. If the new smaller value turns out to break something,
4763 I'll probably just bump it back up.)
4764 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4766 * Many, many bugs reported by Peter Van Eynde have been added to
4767 the BUGS list; some have even been fixed.
4768 * While enabling dynamic object file loading, I tried to make the
4769 code easier to understand, renaming various functions and variables
4770 with less ambiguous names, and changing some function calling
4771 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4772 * While trying to figure out how to do the OpenBSD port, I tried to
4773 clean up some of the code in src/runtime/. In particular, I dropped
4774 support for non-POSIX signal handling, added various comments,
4775 tweaked the code to reduce the number of compilation warnings, and
4776 renamed some files to increase consistency.
4777 * To support the new automatic configuration functionality in make.sh,
4778 the source file target-features.lisp-expr has been replaced with the
4779 source file base-target-features.lisp-expr and the machine-generated
4780 file local-target-features.lisp-expr.
4781 * fixed a stupid quoting error in make.sh so that using CMU CL
4782 "lisp -batch" as cross-compilation host works again
4784 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4786 * DESCRIBE no longer tries to call itself recursively to describe
4787 bound/fbound values, so that it no longer fails on symbols which are
4788 bound to themselves (like keywords, T, and NIL).
4789 * DESCRIBE now works on generic functions.
4790 * The printer now prints less-screwed-up representations of closures
4791 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4792 * A private symbol is used instead of the :EMPTY keyword previously
4793 used to mark empty slots in hash tables. Thus
4794 (DEFVAR *HT* (MAKE-HASH-TABLE))
4795 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4796 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4797 now does what ANSI says that it should. (You can still get
4798 similar noncompliant behavior if bang on the hash table
4799 implementation with all the symbols you get back from
4800 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4801 This breaks binary compatibility, since tests for equality to
4802 :EMPTY are wired into things like the macroexpansion of
4803 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4805 * There's now a minimal placeholder implementation for CL:STEP,
4806 as required by ANSI.
4807 * An obscure bug in the interaction of the normal compiler, the byte
4808 compiler, inlining, and structure predicates has been patched
4809 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4810 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4812 * Missing ordinary arguments in a macro call are now detected even
4813 when the macro lambda list contains &KEY or &REST.
4814 * The debugger no longer complains about encountering the top of the
4815 stack when you type "FRAME 0" to explicitly instruct it to go to
4816 the top of the stack. And it now prints the frame you request even
4817 if it's the current frame (instead of saying "You are here.").
4818 * As specified by ANSI, the system now always prints keywords
4819 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4821 * The default initial SIZE of HASH-TABLEs is now smaller.
4822 * Type information from CLOS class dispatch is now propagated
4823 into DEFMETHOD bodies, so that e.g.
4824 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4826 is now basically equivalent to
4827 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4828 (DECLARE (TYPE SINGLE-FLOAT X))
4830 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4831 operation, without having to do run-time type dispatch.
4832 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4833 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4834 * Since I don't seem to be making big file reorganizations very often
4835 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4836 is overflowing my ability to conveniently back them up), I've finally
4837 checked the system into CVS. (The CVS repository is on my home system,
4838 not at SourceForge -- putting it on SourceForge might come later.)
4839 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4840 high-level GC-NOTIFY-FOO functions send their output. (There's
4841 still very little control of where low-level verbose GC functions
4842 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4843 less than it used to -- the GC-NOTIFY-FOO functions are now under
4844 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4845 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4846 in only one place in the source code, and propagates it automatically
4847 everywhere that it's needed. Thus e.g. when I bump the version from
4848 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4849 * The C source files now include boilerplate legalese and documentation
4850 at the head of each file (just as the Lisp source files already did).
4851 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4852 to his page will be replaced with a link to his new CLiki service.
4854 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4856 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4857 * The build process now looks for GNU make under the default name "gmake",
4858 instead of "make" as it used to. If GNU make is not available as "gmake"
4859 on your system, you can change this default behavior by setting the
4860 GNUMAKE environment variable.
4861 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4862 can build without error under CMU CL.
4864 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4866 * There is now a partial SBCL user manual (with some new text and some
4867 text cribbed from the CMU CL manual).
4868 * The beginnings of a profiler have been added (starting with the
4869 CMU CL profiler and simplifying and cleaning up). Eventually the
4870 main interface should be through the TRACE macro, but for now,
4871 it's still accessed through vaguely CMU-CL-style functions and macros
4872 exported from the package SB-PROFILE.
4873 * Some problems left over from porting CMU CL to the new
4874 cross-compilation bootstrap process have been cleaned up:
4875 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4876 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4877 part when getting rid of the compiler *BACKEND* stuff.)
4878 ** The value of *NULL-TYPE* was screwed up, because it was
4879 being initialized before the type system knew the final
4880 definition of the 'NULL type. This screwed up several key
4881 optimizations in the compiler, causing inefficiency in all sorts
4882 of places. (I found it because I wanted to understand why
4883 GET-INTERNAL-RUN-TIME was consing.)
4884 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4885 PROCLAIM FTYPE information. Unfortunately this broke binary
4886 compatibility again, since now the forms output by DEFGENERIC
4887 to refer to functions which didn't exist in 0.6.3.
4888 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4889 can use the new (as of 0.6.3) transform for SOME into MAP into
4891 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4892 values to (INTEGER 0 1000000), so that the time code will no longer
4893 occasionally get blown up by Linux returning 1000000 microseconds
4894 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4895 its output conform to the ANSI spec. (Alas, this makes its output
4896 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4897 * A full call to MAP NIL with a single sequence argument no longer conses.
4898 * fixes to problems pointed out by Martin Atzmueller:
4899 * The manual page no longer talks about multiprocessing as though
4900 it were currently supported.
4901 * The ILISP support patches have been removed from the distribution,
4902 because as of version 5.10.1, ILISP now supports SBCL without us
4903 having to maintain patches.
4904 * added a modified version of Raymond Toy's recent CMU CL patch for
4905 EQUALP comparison of HASH-TABLE
4907 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4909 * The system still can't cross-compile itself with
4910 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4911 entails), but at least it can compile more of itself that way
4912 than it used to be able to, and various buglets which were uncovered
4913 by trying to cross-compile itself that way have now been fixed.
4914 * This release breaks binary compatibility again. This time
4915 at least I've incremented the FASL file format version to 2, so that the
4916 problem can be detected reliably instead of just causing weird errors.
4917 * various new style warnings:
4918 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4919 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4920 ** using the lexical binding of a variable named in the *FOO* style
4921 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4922 as specified by ANSI.
4923 * *RANDOM-STATE* is no longer automatically initialized from
4924 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4925 default behavior of the system is to repeat its behavior every time
4926 it's run. If you'd like to change this behavior, you can always
4927 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4928 old convention there was no comparably easy way to get the system to
4929 repeat its behavior every time it was run.
4930 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4931 FTYPE declarations has been removed, in favor of their ANSI
4932 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4933 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4934 the types of their sequence arguments can be determined at compile time.
4935 This is done through a new open code expansion for MAP which eliminates
4936 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4937 argument types can be determined at compile time.
4938 * The optimizer now transforms COERCE into an identity operation when it
4939 can prove that the coerced object is already of the correct type. (This
4940 can be a win for machine generated code, including the output of other
4941 optimization transforms, such as the MAP transform above.)
4942 * Credit information has been moved from source file headers into CREDITS.
4943 * Source file headers have been made more standard.
4944 * The CASE macro now compiles without complaining even when it has
4947 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4949 * (Note that the way that the PCL macroexpansions were rewritten
4950 to accommodate the change in DEFGENERIC below breaks binary
4951 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4952 not run under sbcl-0.6.2. Once we get out of alpha releases,
4953 i.e. hit release 1.0.0, we'll probably try to maintain binary
4954 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4955 and sbcl-1.4.4. Until then, however, it might be fairly common
4956 for maintenance releases to break binary compatibility.)
4957 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4959 * The reporting of unhandled signals has been changed to print some
4960 explanatory text as well as the report form. (Previously only
4961 the report form was printed.)
4962 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4963 it defines, so that the compiler no longer issues undefined function
4964 warnings for compiled-but-not-yet-loaded generic functions.
4965 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4966 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4967 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4968 section of the ANSI spec are supported.
4969 * The "" nickname for the "KEYWORD" package has been removed.
4970 The reader still handles symbol tokens which begin with a package marker
4971 as keywords, but it doesn't expose its mechanism for doing so in the
4972 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4973 * The system now issues STYLE-WARNINGs for contradictory TYPE
4974 proclamations. (Warnings for contradictory FTYPE proclamations would
4975 be nice too, but those can't be done usefully unless the type system
4976 is made smarter about FUNCTION types.)
4977 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4978 systematized, so that "*target-*.lisp is supposed to exist only on the
4979 target and imply that there's a related file which exists on the
4980 host, and *host-*.lisp is supposed to exist only on the host and imply
4981 that there's a related file which exists on the target. This involves a
4982 lot of renaming. Hopefully the acute confusion caused by the renaming
4983 will be justified by the reduction in chronic confusion..
4984 ** runtime-type.lisp -> early-target-type.lisp
4985 ** target-type.lisp -> late-target-type.lisp
4986 ** early-host-format.lisp -> early-format.lisp
4987 ** late-host-format.lisp -> late-format.lisp
4988 ** host-error.lisp -> misc-error.lisp
4989 ** early-error.lisp -> early-target-error.lisp
4990 ** late-error.lisp -> late-target-error.lisp
4991 ** host-defboot.lisp -> early-defboot.lisp
4992 ** code/misc.lisp -> code/target-misc.lisp
4993 ** code/host-misc.lisp -> code/misc.lisp
4994 ** code/numbers.lisp -> code/target-numbers.lisp
4995 ** code/early-numbers.lisp -> numbers.lisp
4996 ** early-host-type.lisp -> early-type.lisp
4997 ** late-host-type.lisp -> late-type.lisp
4998 ** host-typep.lisp -> typep.lisp
4999 ** load.lisp -> target-load.lisp
5000 ** host-load.lisp -> load.lisp
5001 ** host-disassem.lisp -> disassem.lisp
5002 ** host-insts.lisp -> insts.lisp
5003 ** byte-comp.lisp -> target-byte-comp.lisp
5004 ** host-byte-comp.lisp -> byte-comp.lisp
5005 ** host-signal.lisp -> signal.lisp
5006 ** host-defstruct.lisp -> defstruct.lisp
5007 ** late-target-type.lisp -> deftypes-for-target.lisp
5008 Furthermore, several other previously target-only files foo.lisp (e.g.
5009 hash-table.lisp and random.lisp) have been split into a target-and-host
5010 foo.lisp file and a target-only target-foo.lisp file, with their key type
5011 definitions in the target-and-host part, so that the cross-compiler will
5012 know more about target types.
5013 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5014 gone away. In their place are various *BACKEND-FOO* variables
5015 corresponding to the slots of the old structure.
5016 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5017 into the target SBCL has been fixed.
5018 * The chill.lisp system for loading cold code into a running SBCL
5020 * Support for the CMU CL "scavenger hook" extension has been removed.
5021 (It was undocumented and unused in the CMU CL sources that SBCL was
5022 derived from, and stale in sbcl-0.6.1.)
5023 * Various errors in the cross-compiler type system were detected
5024 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5025 (enabling various consistency checks). Many of them were fixed,
5026 but some hard problems remain, so the compiler is back to
5027 running without *TYPE-SYSTEM-INITIALIZED* for now.
5028 * As part of the cross-compiler type system cleanup, I implemented
5029 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5030 * I have started adding UNCROSS calls throughout the type system
5031 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5032 on cross-compiler input files will be able to go away, and various
5034 * CONSTANTP now returns true for quoted forms (as explicitly required
5037 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5039 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5040 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5041 (MAKE-LIST -1). In the longer term, it ought to become true
5042 that declarations are assertions even at SAFETY 1. For now, it's not
5043 quite true even at SAFETY 3, but it's at least more nearly true..
5044 (Note that this change seems to increases the size of the system by
5045 O(5%) and to decrease the speed of the compiler by 20% or more.)
5046 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5047 for the problem of printing dozens of lines of distracting information
5048 about low-level system machinery as part of the top stack frame
5049 on entry to the debugger when an undefined function was called.
5050 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5051 is not reset to COMMON-LISP-USER.
5052 * Compilation of stuff related to dyncount.lisp has been made conditional
5053 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5054 smaller. The various dyncount-related symbols have been moved into
5055 a new "SB-DYNCOUNT" package.
5056 * tty-inspect.lisp has been renamed to inspect.lisp.
5057 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5058 feature has gone away. (When we eventually port to other flavors of
5059 libc and/or Unix, we'll try to make the differences between flavors
5060 invisible at the user level.)
5061 * Various other *FEATURES* tags, and/or their associated conditionals,
5062 have been removed if obsolescent, or given better documentation, or
5063 sometimes given more-mnemonic names.
5065 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5067 * tidied up "make.sh" script
5068 * tidied up system directory structure
5069 * better "clean.sh" behavior
5070 * added doc/FOR-CMUCL-DEVELOPERS
5071 * many many small tweaks to output format, e.g. removing possibly-confusing
5072 trailing #\. character in DESCRIBE-INSTANCE
5073 * (EQUALP #\A 'A) no longer signals an error.
5074 * new hashing code, including EQUALP hashing
5075 * tidied up Lisp initialization and toplevel
5076 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5077 * command line argument processing
5078 * added POSIX-GETENV function to deal with Unix-ish environment variables
5079 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5080 terminating SBCL on EOF
5081 * non-verbose GC by default
5082 * There is no more "sbcl" shell script; the sbcl file is now the C
5083 runtime executable (just like CMU CL).
5084 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5086 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5087 symbol and package information as Lisp native symbol and package objects
5088 instead of strings naming symbols and strings naming packages. This way,
5089 whenever packages are renamed (as in warm init), debug information is
5090 transformed along with everything else.
5091 * tweaked the optimization policy declarations which control the building
5092 of SBCL itself. Now, among other things, the system no longer saves
5093 source location debugging information. (This helps two problems at once
5094 by reducing SBCL size and by keeping SBCL from trying to look for its
5095 sources -- which may not exist -- when reporting errors.)
5096 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5097 debugging and testing purposes
5098 * cleaned up printing, making the printer call PRINT-OBJECT for
5099 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5100 methods, giving nearly-ANSI behavior
5101 * converted almost all special variables to use *FOO* naming convention
5102 * deleted PARSE-TIME functionality, since it can be done portably
5103 * moved some files out of cold init into warm init
5104 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5106 * regularized formatting of source files
5107 * added an install.sh script
5108 * fixed ridiculous memory usage of cross-compiler by making
5109 compiler/alloc.lisp not try to do pooling unless it can hook
5110 itself into the GC of the cross-compilation host. Now the system
5111 builds nicely on my old laptop.
5112 * added :SB-ALLOC in target-features.lisp-expr
5113 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5114 was not implemented)
5115 * re-did condition handling and note reporting in the compiler. Notes
5116 are no longer handled by signalling conditions. Style warnings
5117 and warnings are handled more correctly and reported in such a way
5118 that it's easy to find one or the other in your output (so that you
5119 can e.g. figure out which of many problems caused COMPILE-FILE to
5121 * changed the severity of several compiler warnings from full WARNING
5122 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5123 compiler note reporting so that it doesn't use the condition system
5124 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5125 COMPILE-FILE command)
5126 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5127 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5128 no longer supported, since their implementation was deeply intertwingled
5129 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5130 * removed bogus "support" for compiler macros named (SETF FOO), and
5131 removed the compiler macro for SETF INFO (but only after making a fool
5132 of myself on the cmucl-imp mailing list by posting a bogus patch for
5133 DEFINE-COMPILER-MACRO..)
5134 * Compiled files containing forms which have side effects on the Lisp
5135 reader (such as DEFPACKAGE forms) are now handled more correctly.
5136 (Compiler queuing of top level lambdas has been suppressed by setting
5137 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5138 * deleted various currently-unused source files, e.g. gengc.lisp. They
5139 may be added back at some point e.g. when porting to other architectures,
5140 but until they are it's distracting to distribute them and to try to
5142 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5143 there were so many of them they're just distractions, and UNCROSS is
5144 known to be able to handle the current sources
5145 * moved PROFILE functionality into TRACE, so that it will be clear
5146 how the wrapping and unwrapping of functions when you profile them
5147 interacts with the wrapping and unwrapping of functions when you
5148 trace them. (Actually, the functionality isn't there yet, but at least
5149 the interface specification is there. Hopefully, the functionality will
5150 arrive with some future maintenance release.)
5151 * removed host-oops.lisp
5152 * changed signature of QUIT function to allow UNIX-CODE argument
5153 * fixed READ-SEQUENCE bug
5154 * tweaked verbose GC output so that it looks more like the progress
5155 output that ANSI specifies for functions like LOAD
5156 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5157 * added <http://sbcl.sourceforge.com> to the banner information printed by