1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 * optimization: compiler now generates faster typechecking code for
4 * optimization: multidimensional array accesses in the absence of type
5 information regarding array rank are approximately 10% faster due to
6 open coding of ARRAY-RANK.
7 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
8 result in their subforms being treated as top level forms. (reported by
10 * bug fix: disable address space randomization Linux/x86-64 as well,
11 not just x86-64. (reported by Ken Olum)
12 * bug fix: #201; type inference for CONS and ARRAY types could derive
13 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
14 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
15 could not be assigned to or bound with PROGV. (reported by Lorenz
18 changes in sbcl-1.0.28 relative to 1.0.27:
19 * a number of bugs in cross-compilation have been fixed, with the ultimate
20 result that building under (at least) clisp should be much more reliable.
21 * minor incompatible changes: echo-streams now propagate unread-char to the
22 underlying input stream, and no longer permit unreading more than one
24 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
25 frames, allowing for instance more reliable backtraces.
26 * improvement: the debugger REPL can now reference lexical variables
27 by name directly for code compiled with (DEBUG 3).
28 * improvement: errors from malformed declarations now have better source
29 paths associated with them. (thanks to Tobias Rittweiler)
30 * optimization: faster local calls on x86/x86-64
31 * bug fix: some error messages for out-of-bound array indexes confused the
32 index and the bound. (thanks to Stas Boukarev)
33 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
35 * bug fix: running regressions tests in shells without OSTYPE set now works.
36 (reported by Harald Hanche-Olsen)
37 * bug fix: more robust static space exhaustion signalling from
38 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
39 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
40 docstring away instead of storing it under names such as (LAMBDA (X)).
41 (reported by Leslie Polzer)
42 * bug fix: timers could go off in the wrong order, be delayed indefinitely
43 (thanks to Ole Arndt for the patch)
44 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
45 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
46 potentially-invalid effective methods in its cache.
47 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
48 instances as well (thanks to Paul Khuong)
49 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
50 SB-POSIX:WAITPID and their C equivalents.
51 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
53 changes in sbcl-1.0.27 relative to 1.0.26:
54 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
55 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
56 * improvement: the system either recovers from stack exhaustion or dies
57 properly as opposed to leaving the user uncertain of whether the handler
58 trampled on some random memory next to the stack or having to rely on
59 --lose-on-corruption (which is still a good idea to use in production
60 because stack exhaustion can happen in signal handlers which will likely
62 * bug fix: fix gc related interrupt handling bug on ppc (regression from
63 1.0.25.37, reported by Harald Hanche-Olsen)
64 * bug fix: work around signal delivery bug in darwin (regression from
65 1.0.25.44, reported by Sidney Markowitz)
66 * bug fix: fix ERROR leaking memory (reported by David Thompson)
68 changes in sbcl-1.0.26 relative to 1.0.25:
69 * incompatible change: an interruption (be it a function passed to
70 INTERRUPT-THREAD or a timer function) runs in an environment where
71 interrupts can be enabled. The interruption can use
72 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
73 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
74 potentially running out of stack. Keep in mind that in the absence
75 of WITHOUT-INTERRUPTS some potentially blocking operation such as
76 acquiring a lock can enable interrupts.
77 * incompatible change: GC-OFF and GC-ON are removed, as they were
78 always unsafe. Use WITHOUT-GCING instead.
79 * new feature: runtime option --disable-ldb
80 * new feature: runtime option --lose-on-corruption to die at the
81 slightest hint of possibly non-recoverable errors: running out of
82 memory, stack, alien stack, binding stack, encountering a memory
83 fault, etc. In the absence of --lose-on-corruption a warning is
85 * enhancement: detect binding stack exhaustion
86 * enhancement: detect alien stack exhaustion on x86/x86-64
87 * improvement: generally more stable and reliable interrupt handling
88 * improvement: there is a per thread interruption queue,
89 interruptions are executed in order of arrival
90 * improvement: a repeating timer reschedules itself when the it has
91 finished, but expiration times are spaced equally. If an
92 expiration time is in the past it will trigger after a short grace
93 period that may give a chance to other things to run.
94 * optimization: slightly faster gc on multithreaded builds
95 * optimization: faster WITHOUT-GCING
96 * bug fix: when JOIN-THREAD signals an error, do it when not holding
97 important locks so that the debugger/handler doesn't produce
98 recursive errors or deadlock.
99 * bug fix: real-time signals are not used anymore, so no more
100 hanging when the system wide real-time signal queue gets full.
101 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
103 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
104 * bug fix: fix random memory faults related to interrupts on alpha
105 * bug fix: fix random memory faults related to interrupts on sparc
106 * bug fix: fix deadlocks related to starting threads
107 * bug fix: fix deadlines on locks on futex platforms
108 * bug fix: restore errno in signal handlers
109 * bug fix: fix deadlocks related to hash tables
110 * bug fix: fix deadlocks in pcl
112 changes in sbcl-1.0.25 relative to 1.0.24:
113 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
114 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
115 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
116 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
117 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
118 better name and does not return values so stale on multiprocessor systems.
119 Also, HOLDING-MUTEX-P was added for about the only sane usage of
121 * improvement: unithread builds keep track of MUTEX-VALUE.
122 * improvement: reading from a TWO-WAY-STREAM does not touch the output
123 stream anymore making it thread safe to have a concurrent reader and
124 a writer, for instance, in a pipe.
125 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
126 arguments, making compiler-macros for SETF-functions able to inspect
127 their constant arguments.
128 * improvement: COMPILE-FILE reports times with millisecond accuracy
129 (thanks to Luis Oliveira)
130 * optimization: CHAR-CODE type derivation has been improved, making
131 TYPEP elimination on subtypes of CHARACTER work better. (reported
132 by Tobias Rittweiler, patch by Paul Khuong)
133 * bug fix: setting alien structure fields of type struct by value now
134 computes the right offset for the memory copy.
135 * bug fix: compilation problem involving inlined calls to aliens with
136 result type VOID. (reported by Ken Olum)
137 * bug fix: #235a; sequential inline expasion in different policies no
138 longer reuses the functional from the previous expansion site.
139 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
140 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
141 * bug fix: named ENUMs in multiply-referenced alien record types no longer
142 cause an implied type redefinition. Regression from 1.0.21.29.
143 * bug fix: sign-extension of small signed return values now works with C
144 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
145 * improvements to the Windows port:
146 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
147 Westfold, thanks to Rudi Schlatte)
149 changes in sbcl-1.0.24 relative to 1.0.23:
150 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
151 vector of a multidimensional SIMPLE-ARRAY.
152 * new feature: the system now signals a continuable error if standard
153 readtable modification is attempted. (thanks to Tobias Rittweiler)
154 * new feature: DIRECTORY has been extended with a non-standard keyword
155 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
156 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
157 has been robustified and documented better. (thanks to Robert Goldman)
158 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
159 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
160 * tradeoff: constant FORMAT control strings are now compiled unless
161 SPACE > SPEED (previously only when SPEED > SPACE.)
162 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
164 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
166 * bug fix: lexical type declarations are now correctly reported by
167 SB-CLTL2. (reported by Larry D'Anna)
168 * bug fix: STRING-TO-OCTETS did not handle :START properly when
169 using UTF-8 as external format. (thanks to Luis Oliveira)
170 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
171 are signalled correctly. (thanks to Stas Boukarev)
172 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
173 designator. (thanks to Tobias Rittweiler)
174 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
175 designator, and returns T instead of the function. (thanks to
177 * bug fix: direct superclasses of STANDARD-CLASS and
178 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
179 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
180 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
181 when no class with the named slot yet exists no longer causes a
182 compile-time style-warning.
183 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
184 in safe code. (reported by Didier Verna)
185 * bug fix: #430; nested structure constructors can stack allocate.
186 * bug fix: on some 64-bit platforms dynamic space size was truncated
187 to #xffff0000 bytes. (reported by Benjamin Lambert)
188 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
189 REPL unusable. (reported by Daniel Herring)
190 * bug fix: return values of READ-SEQUENCE did not take :START into
191 account on file streams, regressions since 1.0.12.22. (reported by
192 Thomas Russ, patch by Paul Khuong)
193 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
194 method specializer used to confuse permuation vector optimization.
195 * bug fix: system inserted bogus implicit type declarations for local
196 special variables in DEFMETHOD bodies.
197 * bug fix: #354; duplicated frames in backtraces due to
198 non-tail-call-optimized XEPs to functions with return type NIL
200 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
201 initializes structure object slots according to DEFSTRUCT initforms,
202 and DEFSTRUCT forms :INCLUDEind structure classes defined using
203 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
204 (reported by Bruno Haible and Stephen Wilson)
205 * bug fix: #395; fill-pointer output streams used now support
206 element-type BASE-CHAR as well.
207 * bug fix: compiler error when attempting to derive return value of
208 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
210 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
212 * bug fix: Mac OS X binaries should now be portable between Leopard
215 changes in sbcl-1.0.23 relative to 1.0.22:
216 * enhancement: when disassembling method functions, disassembly
217 for the associated fast function is also produced.
218 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
220 * optimization: printing with *PRINT-PRETTY* true is now more
221 efficient as long as the object being printed doesn't require
222 special handling by the pretty printer.
223 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
224 now interact correctly with type declarations.
225 * partial bug fix: PCL detects infinite recursion during wrapper
226 validation. (thanks to Attila Lendvai)
227 * bug fix: #426; nested function calls are inlined properly.
228 Previously if FOO was an inline function, in calls of the form
229 (FOO (FOO ...)) the outer call was not inlined.
230 * bug fix: long long arguments passed in registers to alien
231 functions on PowerPC Linux and NetBSD are now handled correctly
232 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
234 changes in sbcl-1.0.22 relative to 1.0.21:
235 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
236 for the shared object in the current directory, but passes the native
237 namestring of the designated pathname to the operation system's shared
238 object loading function as-it.
239 * minor incompatible change: --disable-debugger toplevel option now takes
240 effect before processing of initialization files and --eval or --load
242 * new feature: new commandline argument: --script, which supports
243 shebang lines. See documentation for details. (based on work by
245 * new feature: SAVE-LISP-AND-DIE can save current values of
246 --dynamic-space-size and --control-stack-size in the executable core,
247 causing it to skip normal runtime option processing. See documentation
248 for details. (thanks to Zach Beane)
249 * enhancement: inoccous calls to EVAL or generic functions dispatching
250 on subclasses of eg. STREAM no longer cause compiler notes to appear.
251 * enhancement: the system no longer resignals errors from --load and
252 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
253 associated with the original error to be lost. (thanks to Ariel
255 * enhancement: :DONT-SAVE keyword argument has been added to
256 LOAD-SHARED-OBJECT for controlling interaction with
257 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
258 the effects of an earlier LOAD-SHARED-OBJECT call.
259 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
260 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
261 provided. (reported by Cedric St-Jean)
262 * bug fix: circularity handling in the reader did not treat raw
263 structure slots correctly. (reported by Cedric St-Jean)
264 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
265 file descriptors when there were none.
266 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
267 was broken. (reporteed by Willem Broekema)
268 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
269 pathnames without a directory.
270 * bug fix: COMPILE-FILE did not warn about undefined variable
271 references at toplevel, and LOAD of the resulting fasl did
273 * bug fix: functions with non-required arguments used to end up with
274 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
275 * bug fix: redefining a function with non-required arguments didn't
276 update the system's knowledge about its call signature properly.
277 * bug fix: fixed #431; incompatible alien record type redefinitions
278 are detected and handled. (reported by Neil Haven)
279 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
280 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
282 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
283 after alien stack frames.
284 * bug fix: strings passed to mk{s,d}temp lacked null termination
286 changes in sbcl-1.0.21 relative to 1.0.20:
287 * new feature: the compiler is able to track the effective type of a
288 generic function across method addition and removal.
289 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
290 appropriately typed structure slots without locking.
291 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
292 information like those gathered by TIME using a programming-friendly
294 * new feature: TIME reports time taken even if the form performs a
295 non-local transfer of control.
296 * enhancement: reduced conservativism on GENCGC platforms: on
297 average 45% less pages pinned (measured from SBCL self build).
298 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
299 mutate constant symbols or violate declaimed type of the symbol.
300 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
301 sleeping on the semaphore in heavily contested semaphores.
302 * bug fix: semaphores and condition variables were not interrupt
304 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
305 owned by other threads anymore.
306 * bug fix: FIND on lists called KEY outside the specified
307 subsequence. (reported by budden)
308 * bug fix: LOG doesn't use single-float intermediate results when
309 given mixed integer and double-float arguments, leading to better
310 precision. (reported by Bob Felts)
311 * bug fix: LOG with base zero returned values of inconsistent type.
313 changes in sbcl-1.0.20 relative to 1.0.19:
314 * minor incompatible change: OPTIMIZE qualities
315 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
316 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
317 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
318 * documentation: some slot access efficiency guidelines have been
319 added to the user manual.
320 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
321 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
322 as ASSOC and MEMEBER.
323 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
324 to more efficient EQ-comparison versions more often.
325 * optimization: enhanced derivation of DOLIST iteration variable type
327 * optimization: constant folding of simple (LIST ...) forms as DOLIST
329 * optimization: runtime lookup of function definitions can be
330 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
331 * optimization: compiler is able to derive the return type of
332 (AREF (THE STRING X) Y) as being CHARACTER.
333 * optimization: CLRHASH on empty hash-tables no longer does pointless
334 work. (thanks to Alec Berryman)
335 * optimization: REPLACE deftransforms don't punt when :START1 or
337 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
338 no longer emits a code deletion note. (reported by Andrew
340 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
341 functions with an odd number of &OPTIONAL arguments, a &REST
342 argument, and one or more &KEY arguments at the call site.
343 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
344 arguments appeared at call sites as well.
345 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
346 READ-LINE &co can no longer read from them afterwards. (reported
348 * bug fix: fixed #427: unused local aliens no longer cause compiler
349 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
351 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
352 alien-stack leakage. (reported by Andy Hefner)
353 * bug fix: PROGV signals an error when an attempt to violate declared
354 type of a variable or bind a constant is made.
355 * bug fix: SET signals an error when an attempt to violate declared
356 type of a variable is made.
357 * bug fix: restart computation during the execution of a restart
358 test function no longer causes infinite recursion. (reported by
360 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
361 signals an error. (thanks to Pierre Mai)
362 * bug fix: EXPORT left symbol unexported in conflict situations.
363 (thanks to Michael Weber)
364 * bug fix: correctly handle name conflicts involving CL:NIL.
365 (thanks to Michael Weber)
366 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
367 conflicts arising from USEing package with conflicting symbols
368 correctly. (thanks to Michael Weber)
370 changes in sbcl-1.0.19 relative to 1.0.18:
371 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
372 warnings that go otherwise unhandled will be muffled if they are
373 of the type that's the value of this variable.
374 * optimization: stack allocation is slightly more efficient on x86
376 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
377 where a raw slot always is initialized using the initform whose
378 type is not know sufficiently well a compile-time are now compiled
379 correctly. (reported by John Morrison)
380 * bug fix: compiler no longer makes erronous assumptions in the
381 presence of non-foldable SATISFIES types.
382 * bug fix: stack analysis missed cleanups of dynamic-extent
383 arguments in non-let-converted calls to local functions.
384 * improvements to the Windows port:
385 ** adjusted address spaces for building on both Win32 and
386 Win64. (thanks for John Connors)
387 * fixed some bugs revealed by Paul Dietz' test suite:
388 ** interval arithmetic during type derivation used inexact integer
389 to single-float coercions.
390 ** arithmetic operations involving large integers and single
391 floats give the same results in compiled and interpreted code.
392 ** deriving the result type of COERCE no longer signals an error
393 if the derived type of the second argument is a MEMBER type
394 containing invalid type specifiers.
395 ** ADJOIN with constant NIL as second argument works correctly.
396 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
398 changes in sbcl-1.0.18 relative to 1.0.17:
399 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
400 profiles only the current thread.
401 * minor incompatible change: changes to SYMBOL-VALUE of constants
402 defined with DEFCONSTANT now signal an error.
403 * enhancement: SB-SPROF now has support for wallclock profiling,
404 and is also able to profile specific threads. REPORT output
405 has also additional sorting options.
406 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
408 * optimization: structure allocation has been improved
409 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
410 ** out of line constructors are ~10% faster.
411 ** inline constructors are ~15% faster.
412 ** inline constructors are capable of dynamic extent allocation
413 (generally on x86 and x86-64, in some cases on other platforms
415 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
417 * optimization: file compiler is now able to coalesce non-circular
418 lists, non-base strings, and bit-vectors. Additionally, constants
419 are never referenced through SYMBOL-VALUE at runtime.
420 * optimization: code defining methods on PRINT-OBJECT (and other
421 generic functions in the COMMON-LISP package) now loads faster.
422 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
424 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
425 fasl is now deleted. (reported by Attila Lendvai)
426 * bug fix: READ-LINE always returned NIL for the last line in files.
427 (reported by Yoshinori Tahara)
428 * bug fix: more accurate disassembly annotations of foreign function
429 calls. (thanks to Andy Hefner)
430 * bug fix: trimming non-simple strings and non-string string
431 designators when the there is nothing to trim works properly.
432 (thanks to James Knight)
433 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
434 (thanks to Travis Cross)
435 * fixed some bugs revealed by Paul Dietz' test suite:
436 ** NIL is a valid function name (regression at 1.0.13.38)
437 ** FILL on lists was missing its return value (regression at 1.0.12.27)
438 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
439 fill pointers properly (regression at 1.0.12.23)
440 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
441 displacement indices properly (regression at 1.0.12.23)
443 changes in sbcl-1.0.17 relative to 1.0.16:
444 * temporary regression: user code can no longer allocate closure
445 variable storage on stack, due to bug 419 without explicitly
446 requesting it. Please consult sbcl-devel for advice if you need to
447 use this feature in the meanwhile.
448 * new feature: runtime argument --control-stack-size can be used to
449 adjust thread default control stack size.
450 * enhancement: improved TIME output
451 ** all times are reported using the measured accuracy (milliseconds
452 for real and GC times, microseconds for everything else.)
453 ** processor cycle counts on x86 and x86-64.
454 ** interpreted forms are counted for both evaluator modes.
455 ** number of lambdas converted by the compiler is reported.
456 ** CPU percentage report (computed from real and total run time.)
457 ** more comprehensive run time reporting, using a condenced format
458 ** interperted form, lambda, and page fault counts are omitted
460 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
462 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
463 * optimization: two argument forms of LAST are upto ~10% faster
464 in normal SPEED policies.
465 * optimization: NCONC no longer needs to heap cons its &REST list
466 in normal SPEED policies.
467 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
469 * bug fix: UNION and NUNION work with :TEST-NOT once more,
470 regression since 1.0.9.1. (thanks to Eric Marsden)
471 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
472 since 1.0.15.36. (thanks to Paul Khuong)
473 * bug fix: LAST when always returned the whole list when given a bignum
474 as the second argument.
475 * bug fix: dynamic extent allocation of nested lists and vectors
476 could leak to otherwise accessible parts.
477 * bug fix: invalid optimization of heap-allocated alien variable
479 * bug fix: fasl header checking is less vulnerable to different
480 platform word lengths.
481 * bug fix: more correct assembler syntax for GNU binutils
482 2.18.50.0.4 support. (thanks to Marijn Schouten)
483 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
485 * bug fix: the fopcompiler can handle LOCALLY forms (with no
486 declarations) successfully. (reported by Attila Lendvai)
488 changes in sbcl-1.0.16 relative to 1.0.15:
489 * minor incompatible change: revert the changes to sb-posix's error
490 signaling added in 1.0.14.
491 * minor incompatible change: change PROBE-FILE back to returning
492 NIL whenever we can't get a truename, as was the case before 1.0.14.
493 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
494 inherits from ERROR instead of just CONDITION.
495 * new feature: SB-INTROSPECT can provide source locations for instances
496 as well. (thanks to Tobias Rittweiler)
497 * optimization: binding special variables now generates smaller code
498 on threaded platforms.
499 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
500 and cases where no :TEST is given but the compiler can infer that
501 the element to search is of type (OR FIXNUM (NOT NUMBER)).
502 * optimization: better LOGNOT on fixnums.
503 * optimization: modular arithmetic for a particular requested width
504 is implemented using a tagged representation unless a better
505 representation is available.
506 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
507 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
508 for wellformedness and returns a specialized vector. (reported by
509 Francois-Rene Rideau)
510 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
511 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
512 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
513 unscheduling timer at the same time as another timer fires.
514 * bug fix: don't reschedule timers for dead threads.
515 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
516 * bug fix: copying output from RUN-PROGRAM to a stream signalled
517 bogus errors if select() was interrupted.
518 * enhancement: add support for fcntl's struct flock to SB-POSIX.
520 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
521 * enhancement: cleaner backtraces for interactive interrupts, as
522 well as other cases where the interesting frames used to be
523 obscured by interrupt handling frames.
524 * enhancement: untracing a whole package using (UNTRACE "FOO") is
525 now supported, and tracing a whole package using (TRACE "FOO") now
526 traces SETF-functions as well.
527 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
528 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
529 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
530 * unreadably printed representation of hash-tables now includes
532 * bug fix: partially fixed #188: type propagation from assignments
533 is now more efficient.
534 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
535 'DOUBLE-FLOAT) are not flushable.
536 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
537 full address of the object, and none of the tag bits.
538 * bug fix: readably printing hash-tables now respects other printer
539 control variables. (reported by Cedric St-Jean)
540 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
541 keyword with MAKE-HASH-TABLE.
542 * bug fix: export SB-POSIX:MKSTEMP.
543 * bug fix: SORT was not interrupt safe.
544 * bug fix: XREF accounts for the last node of each basic-block as
546 * bug fix: MAKE-INSTANCE optimizations interacted badly with
547 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
548 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
549 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
550 create obsolete instances in the case of redefinition or
551 obsoletion of a superclass. (thanks to Andy Hefner)
552 * bug fix: Support for the Alpha architecture has been revived; it had
553 suffered somewhat from lack of maintenance since sbcl-1.0.
554 * improvements to the Windows port:
555 ** The system detects the codepage to use at startup, instead of
556 using the value from the saved core. (thanks to Kei Suzuki)
558 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
559 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
560 (see documentation for details.)
561 * revived support for OpenBSD (contributed by Josh Elsasser)
562 * partially fixed bug #108: ROOM no longer suffers from occasional
563 (AVER (SAP= CURRENT END)) failures.
564 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
565 bodies no longer cause a WARNING to be signalled. (reported by
567 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
568 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
569 the bug in TRUNCATE.) (reported by Michael Weber)
570 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
571 no samples. (reported by Andy Hefner)
572 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
573 no longer appear as (NIL ...) frames in backtraces.
574 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
575 is now actually exported from SB-EXT as documented. (reported by
577 * bug fix: sb-aclrepl now correctly understands how to inspect
578 single-floats on 64-bit platforms where single-floats are not boxed.
579 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
580 yet been finalized. (reported by Levente Meszaros)
581 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
582 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
583 SBCL-specific optimize qualities.
585 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
586 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
587 an executable in the search path, and does so in the child
588 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
589 been removed; it can be found in the CVS history, for somebody who
590 needs that search behavior (see the manual).
591 * minor incompatible change: compiler policy re. weakening type
592 checks has changed: now type checks are weakened only if SAFETY < 2
594 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
595 unparsing of directory pathnames as files. Analogously,
596 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
597 filename to parse into a directory pathname.
598 * enhancement: implicit generic function creation now signals a
599 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
600 users can bind handlers for and muffle around calls to LOAD.
601 * enhancement: RUN-PROGRAM allows unicode arguments and environments
602 to be used (using the default stream external format), and allows
603 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
604 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
605 strings and vectors whose element-type or simplicity is not fully
606 known at compile-time.
607 * optimization: STRING-TRIM and related functions no longer allocate
608 a new string when no trimming needs to be performed. These functions
609 are also faster than before when the input string has been declared
611 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
612 * optimization: READ-LINE is significantly faster for files containing
614 * optimization: non-open coded uses of character comparison operators
615 (e.g. char=) no longer cons when called with more than one parameter
616 on platforms supporting dynamic-extent allocation.
617 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
618 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
619 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
621 * bug fix: some sequence functions elided bounds checking when
623 * bug fix: too liberal weakening of union-type checks when SPEED >
625 * bug fix: more bogus fixnum declarations in ROOM implementation
628 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
629 * new feature: MAKE-HASH-TABLE now experimentally accepts a
630 :SYNCHRONIZED argument, which makes the hash-table safe for
631 concurrent accesses (but not iteration.) See also:
632 SB-EXT:WITH-LOCKED-HASH-TABLE, and
633 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
634 * optimization: CONCATENATE on strings is an order of magnitude faster
635 in code compiled with (> SPEED SPACE).
636 * optimization: SUBSEQ is ~50% faster on lists.
637 * bug fix: bug 417 fixed -- source location reporting is now more robust.
638 * bug fix: SUBSEQ on a list will now correctly signal an error if if
639 END is smaller then START.
640 * bug fix: SB-PROFILE will no longer report extra consing for nested
641 calls to profiled functions.
642 * bug fix: ROOM implementation had bogus fixnum declarations which
643 could cause type-errors when calling ROOM on large images.
644 * bug fix: if file compilation is aborted, the partial fasl is now
645 deleted, and COMPILE-FILE returns NIL as the primary value.
646 * bug fix: number of thread safety issues relating to SBCL's internal
647 hash-table usage have been fixed.
648 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
649 be returned from its body when the values were being returned
650 using unknown-values return convection and the W-P-O was wrapped
651 inside an UNWIND-PROTECT.
652 * bug fix: sb-posix should now compile again under Windows, enabling
655 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
656 * incompatible change: hash-table accessor functions are no longer
657 automatically protected by locks. Concurrent accesses on the same hash-table
658 from multiple threads can give inconsistent results or even corrupt the
659 hash-table completely. Multi-threaded applications should do their own
660 locking at the correct granularity. In the current implementation it is
661 still safe to have multiple readers access the same table, but it's not
662 guaranteed that this property will be maintained in future releases.
663 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
664 and will signal an error at runtime.
665 * enhancement: SB-THREAD package now exports a semaphore interface.
666 * enhancement: CONS can now stack-allocate on x86 and
667 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
669 * enhancement: nested lists can now be stack allocated on
670 platforms providing stack allocation support.
671 * enhancement: dynamic-extent support has been extended to support
672 cases where there are multiple possible sources for the stack
674 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
675 if the mutex is uncontested on Linux.
676 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
677 as specified by the CLHS. (thanks to Tobias Rittweiler)
678 * bug fix: NaN comparison now works on x86-64.
679 * bug fix: CLOSE :ABORT T on a stream with pending output now
681 * bug fix: instances of non-standard metaclasses using standard
682 instance structure protocol sometimes missed the slot type checks
684 * bug fix: known functions can be inlined in high-debug code.
686 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
687 * minor incompatible change: the MSI installer on Windows no longer
688 associates .lisp and .fasl files with the installed SBCL.
689 * minor incompatible change: :UNIX is no longer present in *FEATURES*
690 on Windows. (thanks to Luis Oliviera)
691 * new feature: SB-CLTL2 contrib module now implements
692 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
693 * optimization: scavenging weak pointers is now more efficient,
694 requiring O(1) instead of O(N) per weak pointer to identify
695 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
696 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
697 method bodies using constant slot-names when the first argument is
698 a specializer parameter for the method.
699 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
700 names, when the class of the instance is a direct instance
701 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
702 as slow as the constant slot-name case.)
703 * optimization: member type construction is now O(N) instead
705 * optimization: UNION and NUNION are now O(N+M) for large
706 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
708 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
709 now more readable in environments like Slime which display it.
710 (thanks to Tobias C. Rittweiler)
711 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
712 was able to infer the correct argument type for the object on which
713 the CAS operation was being performed.
714 * bug fix: copy propagation interfered with parallel assignment
715 semantics in local calls. (reported by Paul Khuong)
716 * bug fix: the signed modular fixnum shift compiled to wrong code on
717 x86 and x86-64. (spotted by a slight modification to some of
720 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
721 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
722 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
723 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
725 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
726 and generic function now print the object identity for anonymous
727 instances. (thanks to Attila Lendvai)
728 * enhancement: as an extension to MOP, SBCL now supports SETF of
729 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
730 (thanks to Attila Lendvai)
731 * enhancement: sb-cover has an alternate annotation mode which puts
732 the form annotations at the car of the form, instead of annotating
733 it completely. The benefit of this mode is that it shows better
734 which forms were instrumented by the compiler.
735 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
736 by printer control variables. (reported by Dan Corkill)
737 * bug fix: system leaked memory when delayed output was performed by
738 the OS in smaller chunks then expected. (thanks to David Smith)
739 * bug fix: system leaked memory when file streams were not closed
741 * bug fix: large objects written to slow streams that were modified
742 after the write could end up with the modified state written to
743 the underlying file descriptor.
744 * bug fix: multiple threads operating in parallel on the same stream
745 could cause buffer-overflows.
746 * bug fix: source location information is stored correctly
747 (broken since 1.0.6). This bug would generally show up as the
748 Slime debugger highlighting the wrong form when the sldb-show-source
750 * bug fix: Branch forms are again annotated as branches in the sb-cover
752 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
753 on x86-64. (reported by Gregory Vanuxem)
754 * bug fix: EQUALP could return wrong results for structures with raw
755 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
756 word). (reported by Vjacheslav Fyodorov)
757 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
760 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
761 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
762 atomic compare-and-swap operations on threaded platforms.
763 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
764 allows assigning a global minimum value to optimization qualities
765 (overriding proclamations and declarations).
766 * enhancement: closed over variables can be stack-allocated on x86
768 * performance bug fix: GETHASH and (SETF GETHASH) are once again
770 * optimization: slot definition lookup is now O(1). This speeds up
771 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
772 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
773 * optimization: ASSOC and MEMBER can now be open-coded for all
774 combinations of keyword arguments when second argument is constant
775 and SPEED >= SPACE. In other cases a specialized version is
777 * bug fix: using obsoleted structure instances with TYPEP and
778 generic functions now signals a sensible error.
779 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
780 (reported by Kristoffer Kvello)
781 * bug fix: backtrace construction is now more careful when making
782 lisp-objects from pointers on the stack, to avoid creating bogus
783 objects that can be seen by the GC.
784 * bug fix: defaulting of values in contexts expecting more than 7
785 variables now works on x86-64. (reported by Christopher Laux)
786 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
788 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
789 as the property-list of a symbol.
790 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
791 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
792 situations requested, are once again file-compileable. (reported
795 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
796 * MOP improvement: support for user-defined subclasses of
797 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
798 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
799 the expansion of DEFMETHOD, is responsible for generating a form
800 which creates a list of specializers when evaluated. Additional
801 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
802 debugging and introspective support.
803 * minor incompatible change: the (unsupported) spinlock interface
804 has changed: free spinlock now has the value NIL, and a held spinlock
805 has the owning thread as its value.
806 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
807 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
809 * enhancement: name of a socket-stream is now "a socket" instead of
811 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
812 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
813 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
814 for generic functions. (Thanks to Tobias C. Rittweiler)
815 * optimization: bignum printing speed has been improved by 20-40%
816 (depending on the bignum size.)
817 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
819 * bug fix: the cache used by the CLOS to store precomputed effective
820 methods, slot offsets, and constant return values is now thread and
822 * bug fix: generic function dispatch function updating is now thread
823 and interrupt safe (in the sense that the known issues have been
825 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
826 * bug fix: interning EQL-specializers is now thread and interrupt safe.
827 * bug fix: asdf systems with dependencies to the SB-POSIX or
828 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
829 * bug fix: interrupt safety of applicable method computation has been
832 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
833 * new contrib: sb-cover, an experimental code coverage tool, is included
835 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
836 significantly faster.
837 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
838 produce more efficient code on x86-64 if the compiler can determine
839 that (+ INDEX <constant>) does not require a bounds check and FOO
840 has an element type at least 8 bits wide.
841 * enhancement: a new, experimental synchronous timeout facility is
842 provided. Refer to SB-SYS:WITH-DEADLINE for details.
843 * enhancement: when a symbol name conflict error arises, the
844 conflicting symbols are always printed with a package prefix.
845 (thanks to Kevin Reid)
846 * enhancement: stepping is now once again supported on the SPARC and
847 MIPS platforms. (It is also now more likely to work on CheneyGC
849 * enhancement: sb-sprof can now also track and report accurate call
851 * bug fixes: the treatment of non-standard subclasses of
852 SB-MOP:SPECIALIZER is more correct.
853 * incompatible change: PURIFY no longer copies the data from the
854 dynamic space into the static and read-only spaces on platforms
855 that use the generational garbage collector
856 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
858 * bug fix: GC race condition occasionally resulting in crashes with
859 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
861 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
863 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
864 a STYLE-WARNING to be signalled (regression from 1.0.4.)
865 * bug fix: an asynchronous interrupt could previously leave the
866 system running with GC inhibited.
867 * bug fix: a DECLARE form evaluated at top-level now causes an error
868 rather than silently (or verbosely) returning NIL.
869 * bug fix: trying to dissassemble functions compiled at high DEBUG could
870 result in a error being signalled due to source form lookup errors.
871 (reported by Peter Graves)
873 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
874 * incompatible change: removed writer methods for host-ent-name,
875 host-ent-addresses -- changing the values did not update the DNS
877 * minor incompatible change: changed experimental JOIN-THREAD interface
878 * documentation: the manual now lists reader and writer methods
879 in class slot documentation sections. (thanks to Richard M Kreuter)
880 * documentation: unwinding from asyncronous events has been
881 documented as unsafe.
882 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
883 in multithreaded application code.
884 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
885 platforms. (thanks to James Anderson for the optimization hint)
886 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
888 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
889 variants no longer cons.
890 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
891 their NOT- variants no longer cons.
892 * optimization: EQUAL hash tables no longer use SXHASH for objects
893 of all data types, but instead use an EQL hash for types for which
894 EQUAL is the same as EQL.
895 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
896 are significantly faster.
897 * optimization: new STRING-OUTPUT-STREAM implementation is much
898 faster for large outputs and conses 30% less on average.
899 * enhancement: XREF information is now collected for references made
900 to global variables using SYMBOL-VALUE and a constant argument.
901 * enhancement: SIGINT now causes a specific condition
902 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
903 * bug fix: FILE-POSITION used to signal an error in some cases where
904 ANSI requires it to return NIL.
905 * bug fix: ADJUST-ARRAY is now interrupt-safe.
906 * bug fix: adding and removing fd-handlers is now interrupt-safe.
907 * bug fix: inlined calls to C now ensure 16byte stack alignment on
909 * bug fix: bad type declaration in the CLOS implementation has
910 been fixed. (reported by James Anderson)
911 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
912 (thanks to Lutz Euler)
913 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
914 required by the CLHS. (reported by Samium Gromoff)
915 * bug fix: dead unbound variable references now signal an error.
916 (reported by Marco Monteiro)
917 * bug fix: / with an unused value was being deleted in safe code.
918 (thanks to Marco Monteiro and Kevin Reid)
919 * bug fix: number of characters that can be written onto a single
920 line in a file is unlimited.
921 * bug fix: some GC deadlocks caused by asynchronous interrupts have
922 been fixed by inhibiting interrupts for when GC is disabled.
923 * bug fix: some interrupt safety issues with GETHASH, (SETF
924 GETHASH), CLRHASH and REMHASH have been fixed.
925 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
926 specifier no longer causes infinite recursion.
927 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
928 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
929 * bug fix: an &environment argument with the correct variable information
930 is passed to macros that are expanded during byte compilation.
931 (reported by Samium Gromoff)
932 * bug fix: a logic error was causing the hash function for bignums to
933 have excessive amounts of collisions. (reported by Faré Rideau)
934 * bug fix: modifying the contents of an array could change the return
935 value of SXHASH on that array, which is only allowed for strings
936 and bit vectors. (bug introduced in 0.9.16)
937 * improvement: the x86-64/darwin port now passes all tests (except
938 for the debugger tests) but should still be considered
939 experimental until this is fixed.
940 * improvement: a style-warning is signaled for CASE (etc) clauses with
941 duplicate keys. (thanks to Kevin Reid)
942 * improvement: macos/x86 and macos/x86-64 now use mach exception
943 handlers for dealing with illegal instructions (for trapping and
944 error handling) and memory protection violations (for GC).
946 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
947 * new platform: experimental support for x86-64/darwin (MacOS).
948 * incompatible change: the thread-safe (on most platforms) getaddrinfo
949 and getnameinfo sockets functions are used instead of gethostbyaddr
950 and gethostbyname, on platforms where the newer functions are available.
951 As a result, the ALIASES field of HOST-ENT will always be NIL on these
953 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
954 don't print the startup banner, but behave as if --noinform was passed
955 as a command line argument. (thanks to Kevin Reid)
956 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
957 * optimization: code using alien values with undeclared types is much faster.
958 * optimization: the compiler is now able to open code SEARCH in more cases.
959 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
960 * bug fix: using standardized COMMON-LISP special variables as loop
961 variables no longer signals bogus package lock violations. (reported
963 * bug fix: declaring local loop variables to be of a range-limited type
964 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
965 (reported by Andras Simon)
966 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
967 bugs remain on x86-64.)
968 * bug fix: the #= and ## reader macros now interact reasonably with
969 funcallable instances.
970 * bug fix: type-checks for function arguments were compiled using the
971 compiler policy settings of the wrong lexical environment (in 1.0.2
973 * bug fix: SHADOW accepts characters as string designators, as required
974 by the spec (thanks to Eric Marsden)
975 * bug fix: fixed GC safety issues when foreign functions are called with
976 non-base strings as arguments
977 * bug fix: more consistent error output for fatal-compiler-warnings, like
979 * bug fix: functions with proper names were showing up as NIL in
980 backtraces (reported by Edi Weitz, regression in 1.0.2)
982 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
983 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
984 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
985 produce more efficient code on the x86 if the compiler can determine
986 that (+ INDEX <constant>) does not require a bounds check and FOO
987 has an element type at least 8 bits wide.
988 * bug fix: references to undefined variables are handled the same way
989 in toplevel forms as in the normal compiler
990 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
992 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
993 (thanks to Jon Buller)
994 * improvement: faster compilation times for complex functions
995 * improvement: added readlink support to SB-POSIX (thanks to Richard
998 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
999 * improvement: experimental support for mach exception handling on
1000 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1002 * improvement: support for GBK external format.
1003 (thanks to Chun Tian (binghe))
1004 * improvement: the debugger now displays variables that have been closed
1005 over, in code compiled with (DEBUG 2) or higher
1006 * improvement: support for executable cores on NetBSD (thanks to
1008 * new feature: added a RESTART-FRAME debugger command
1009 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1010 be used to provide an implementation for FILE-POSITION on Gray streams
1011 (thanks to Eric Marsden)
1012 * new feature: add syslog, openlog and closelog support to SB-POSIX
1013 (thanks to Richard Kreuter)
1014 * optimization: the function call overhead in code compiled with
1015 a high DEBUG optimization setting is significantly
1016 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1017 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1018 (thanks to Tony Martinez)
1019 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1020 evaluated once (reported by Kevin Reid)
1021 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1022 works again (thanks to Kilian Sprotte)
1023 * bug fix: an error is signaled for tagbodies with duplicate tags
1024 (thanks to Stephen Wilson)
1025 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1026 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1027 stack frames from alien callbacks.
1028 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1029 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1030 65k pages (thanks to David Woodhouse)
1031 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1033 changes in sbcl-1.0.1 relative to sbcl-1.0:
1034 * new platform: FreeBSD/x86-64, including support for threading.
1035 * new feature: the compiler stores cross-referencing information
1036 about function calls (who-calls), macroexpansion (who-macroexpands)
1037 and special variables (who-binds, who-sets, who-references) for code
1038 compiled with (< SPACE 3). This information is available through the
1039 sb-introspect contrib.
1040 * new feature: users may subclass SEQUENCE, and have instances of
1041 these classes interoperate with standard Common Lisp functions if
1042 a number of methods are defined. (This feature is experimental
1043 and the interface subject to change based on feedback from SBCL
1044 users and the general community)
1045 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1046 x86/x86-64, rather than the previous fixed depth of 8
1047 * improvement: another pthread back-end of mutex "pthread-futex".
1048 * bug fix: non-ascii command-line arguments are processed correctly
1049 (thanks to Yaroslav Kavenchuk)
1050 * bug fix: non-required arguments were not passed correctly when a method
1051 defined using DEFMETHOD was called from a mop-generated method using
1052 CALL-NEXT-METHOD (reported by Pascal Costanza)
1053 * bug fix: recursion is now permitted in accessors through
1054 SLOT-UNBOUND. (reported by Pascal Costanza)
1055 * bug fix: an error was signaled at startup if the HOME environment
1056 variable was defined, but had an empty value (reported by Peter Van Eynde)
1057 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1058 signal a PROGRAM-ERROR, not a BUG.
1059 * bug fix: SB-POSIX:READDIR works when built with large file support.
1060 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1062 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1063 for better automatic indentation support. (thanks to Matt Pillsbury)
1064 * optimization: loading generic functions no longer takes O(n^2) time,
1065 proportional to the amount of methods in the generic function
1066 (reported by Todd Sabin and Jeremy Brown)
1067 * optimization: the FIND and POSITION family of sequence functions
1068 are significantly faster on arrays whose element types have been
1070 * improvements to the Windows port:
1071 ** Intermittent heap corruption problems have been fixed. (thanks
1072 to Alastair Bridgewater)
1073 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1075 ** Lisp is able to unwind foreign exception frames from alien
1076 callbacks. (thanks to Alastair Bridgewater)
1078 changes in sbcl-1.0 relative to sbcl-0.9.18:
1079 * improvement: experimental support for threading on FreeBSD/x86.
1080 (thanks to NIIMI Satoshi)
1081 * improvement: runtime option --dynamic-space-size can be used
1082 to set the size of the dynamic space reserved on startup.
1083 * improvement: floating point modes in effect are now saved in
1084 core, and restored on startup.
1085 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1086 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1087 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1088 threads are not running after *SAVE-HOOKS* have run.
1089 * improvement: writes to CLOS instance slots are type-checked in code
1090 compiled with (SAFETY 3)
1091 * improvement: floating-point exception handling on FreeBSD (thanks to
1093 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1094 (thanks to Zach Beane)
1095 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1097 * improvement: added support for the Shift-JIS external format.
1098 (contributed by NIIMI Satoshi)
1099 * improvement: callbacks are supported on Linux/PPC. (thanks to
1101 * bug fix: compiler bug triggered by a (non-standard) VALUES
1102 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1103 * bug fix: file compiler no longer confuses validated and already
1104 dumped structurres. (reported by Kaersten Poeck)
1105 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1106 fill-pointer signals a type-error as required. (thanks to
1108 * bug fix: disassemly of funcallable instances works.
1109 * bug fix: single stepping on PPC.
1110 * bug fix: fix thread-safety problems in the type system (generally
1111 manifesting as nonsensical errors like "STRING is a bad type specifier
1112 for sequences" or "The value 1 is not of type FIXNUM")
1113 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1114 (reported by Josip Gracin)
1115 * bug fix: an error is signaled for attempts to displace arrays with
1116 incompatible element types (thanks to Mario Mommer)
1117 * bug fix: more correct handling of wide characters in debug info
1118 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1119 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1120 and don't cause extra consing
1121 * optimization: MAP and MAP-INTO are significantly faster on vectors
1122 whose elements types have been declared.
1123 * Improvements to SB-SPROF:
1124 ** Support for allocation profiling
1125 ** Reduced profiling overhead, especially for long profiling runs
1126 * Improvements to the Windows port:
1127 ** floating point exceptions are now reported correctly.
1128 ** stack exhaustion detection works partially.
1129 ** more accurate GET-INTERNAL-REAL-TIME.
1130 ** better RUN-PROGRAM behaviour in case of errors in the child
1132 ** PROBE-FILE now simplifies pathnames correctly.
1133 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1134 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1135 1st 1970 (although time timezone data will be unreliable).
1136 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1138 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1139 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1140 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1142 * bug fix: two potential GC deadlocks affecting threaded builds.
1143 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1144 returns T (reported by Anton Kazennikov)
1145 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1146 with non-variable places
1147 * bug fix: remove a race condition in the setting of
1148 funcallable-instance functions, this should make threaded CLOS
1149 code more stable against memory faults.
1150 * bug fix: corruption of specials when unbinding is interrupted by an
1151 asynchronous unwind (reported by Hannu Koivisto)
1152 * improvement: the debugger will now also display local variables that
1153 are only used once, for code compiled with a DEBUG optimization quality
1156 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1157 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1158 * incompatible change: External-format support for FFI calls. The
1159 SB-ALIEN:C-STRING no longer implies an ASCII
1160 external-format. Instead the string is subject to external-format
1161 conversion. Additionally return values of type SB-ALIEN:C-STRING
1162 are by default of type (SIMPLE-ARRAY CHARACTER), not
1163 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1164 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1165 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1166 * incompatible change: SB-EXT package no longer contains the
1167 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1168 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1170 * incompatible change: the single-stepper is no longer available
1171 on Alpha, Mips and Sparc platforms.
1172 * minor incompatible change: the direct superclasses of
1173 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1174 not (STANDARD-OBJECT FUNCTION). This makes the
1175 class-precedence-lists of GENERIC-FUNCTION and
1176 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1178 * minor incompatible change: the default stream external format on
1179 non-unicode SBCL is no longer detected from the locale, but is always
1181 * new feature: Add a version of evaluator that uses an interpreter instead
1182 of the compiler. EVAL still uses the compiler by default, to switch it
1183 to use the interpreter, set the value of the variable
1184 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1185 * minor incompatible change: the single-stepper REPL has been merged
1186 with the normal debugger (see the "Stepping" heading of the debugger help
1187 for more details). The debugger command STEP will no longer switch
1188 to the single-stepper REPL.
1189 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1190 for a type now works.
1191 * bug fix: loading of default sysinit file works. (thanks to Leonid
1193 * bug fix: better detection of circularities in the file-compiler.
1194 (reported by Marco Monteiro)
1195 * bug fix: the CL pathname functions now work with files that have
1196 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1197 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1198 systems with Unix98 pty semantics.
1199 * bug fix: ASDF-INSTALL will now work with bsd tar.
1200 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1202 * bug fix: timers expiring in dead threads no longer cause a
1203 type-error (reported by Paul "Nonny Mouse").
1204 * bug fix: thanks to more lightweight single-stepper instrumentation,
1205 code compiled with (DEBUG 3) will compile and execute significantly faster,
1206 and will have more accurate type-inferencing than before
1207 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1208 whose bindings are modified
1209 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1210 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1211 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1212 CL:SOFTWARE-VERSION with proper version
1214 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1215 * feature: implemented the READER-METHOD-CLASS and
1216 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1217 as specified by AMOP.
1218 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1220 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1221 (thanks to Lutz Euler)
1222 * optimization: hashing of general arrays and vectors has been
1223 improved. (reported by Any Fingerhut)
1224 * enhancement: SB-INTROSPECT is now able to find definitions of
1225 profiled functions. (thanks to Troels Henriksen)
1226 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1227 as well, allowing compiler-macros for SETF-functions to expand.
1228 * enhancement: step-instrumentation no longer wraps known
1229 single-value functions with multiple-value context, allowing
1230 better type inference.
1231 * fixed bug #337: use of MAKE-METHOD in method combination now works
1232 even in the presence of user-defined method classes. (reported by
1233 Bruno Haible and Pascal Costanza)
1234 * fixed bug #339(c): if there are applicable methods not part of any
1235 long-form method-combination group, call INVALID-METHOD-ERROR.
1236 (reported by Bruno Haible)
1237 * fixed bug #361: the :FUNCTION initarg in the protocol for
1238 initialization of methods can now be used to override
1239 internally-produced optimized functions. (reported by Bruno
1241 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1242 system-provided lambda expression no longer cause warnings about
1243 unbound #:|pv-table| symbols.
1244 * bug fix: improved the handling of type declarations and the
1245 detection of violations for keyword arguments with non-constant
1247 * bug fix: potentially erronous calls to PATHNAME and
1248 MERGE-PATHNAMES were being flushed in some cases.
1249 (reported by Richard Kreuter)
1250 * bug fix: compiled calls to TYPEP were mishandling obsolete
1251 instances. (reported by James Bielman and Attila Lendvai)
1252 * bug fix: format strings with ~> without matching ~< no longer
1253 trigger an AVER, but signal an understandable error instead.
1254 (reported by Antonio Martinez)
1255 * bug fix: specifying an output-file without a pathname-type for
1256 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1257 of input-file instead of "fasl". (reported by Robert Dodier)
1258 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1259 (reported by James Y Knight).
1260 * bug fix: compiler-macros-function did not consider the environment
1261 argument for shadowing by local functions.
1262 * bug fix: compiler-macros expansion was inhibited by local INLINE
1264 * bug fix: inline expansions of known functions were subject to
1265 step-instrumentation in high debug policies, leading to problems
1266 with type-inference.
1267 * bug fix: compiler failed to differentiate between different CONS
1268 types in some cases.
1269 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1270 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1271 * bug fix: #368: incorrect use of expressed vs. upgraded array
1273 * bug fix: #306a: more precise unions of array types.
1274 * thread-safety improvements:
1275 ** CONDITION-WAIT could return early on Linux, if the thread was
1276 interrupted and subsequently continued with SIGCONT.
1277 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1279 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1280 * added support for the ucs-2 external format. (contributed by Ivan
1282 * minor incompatible change: pretty printing of objects of type
1283 (cons symbol) is, in the default pprint-dispatch-table, now
1284 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1286 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1287 called later in a class's lifetime, possibly as late as when the
1288 first instance of the class is created. Previously,
1289 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1290 class became finalizeable.
1291 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1292 for multibyte external-format streams. (thanks to "vbzoli")
1293 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1294 with :ALLOCATION :CLASS inherited from superclasses of the
1296 * fixed bug: anonymous classes can now be created using the :NAME
1297 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1298 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1299 * fixed bug: core-files saved with :EXECUTABLE T can again be
1300 executed when SBCL_HOME isn't set. (reported by James Knight)
1301 * fixed bug: toplevel LOCALLY forms with declarations could
1302 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1303 * fixed bug: printing from several different threads using different
1304 values of *print-case* could cause invalid output, due to
1305 some internal special variables of the printer not being bound
1306 thread-locally (reported by Max Mikhanosha)
1307 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1308 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1309 (reported by Pascal Costanza)
1310 * fixed bug: FINALIZE-INHERITANCE is called from
1311 REINITIALIZE-INSTANCE on classes when the class has previously
1312 been finalized, as required by AMOP.
1313 * minor code generation optimizations:
1314 ** better register allocation in CLOS dispatching functions
1315 ** overflow detection when coercing signed bytes to fixnums on x86-64
1316 ** is now implemented with one IMUL instruction instead of three shifts
1317 ** more efficient bit-vector access on x86 and x86-64
1318 ** more efficient access to raw structure slots on x86 and x86-64
1319 * fixed some bugs revealed by Paul Dietz' test suite:
1320 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1321 return its argument.
1323 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1324 * feature: thread support on Solaris/x86, and experimental thread support
1326 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1328 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1329 default method for SLOT-UNBOUND.
1330 * minor incompatible change: prevent the user from specializing the
1331 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1332 somewhat counter to the intent of the protocol, I (CSR) think, and
1333 additionally it just doesn't work in SBCL as currently
1334 implemented, thanks to optimizations (that are always valid for
1335 the other three SLOT-VALUEish functions, but not for the setter).
1336 * minor incompatible change: the :SB-LDB feature is now enabled by
1337 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1338 the low-level debugger.
1339 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1340 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1341 down to an SB-SYS:FD-STREAM, allowing direct communication in
1342 more cases. Temporary files and pipes are still used as a fallback
1344 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1345 on the non-reentrant C functions localtime(3) and gmtime(3).
1346 * bug fix: global optimization policy was not visible in LOCALLY and
1348 * bug fix: class objects can be used as specializers in methods.
1349 (reported by Pascal Costanza)
1350 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1351 with a type error. (reported by blitz_ on #lisp)
1352 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1353 failed with a type error. (reported by Pascal Bourguignon)
1354 * bug fix: merging pathnames against defaults with :DIRECTORY
1355 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1356 (reported by James Y Knight)
1357 * bug fix: saving large (>2GB) cores on x86-64 now works
1358 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1359 constant index and a (simple-array (signed-byte 32)) array
1360 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1361 error (patch by Robert J. Macomber)
1362 * feature: TIME now displays the amount of run-time spent in GC
1363 * bug fix: The debugger now does a better job of respecting
1364 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1365 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1366 workaround for bug 403.)
1367 * bug fix: floating point exception handling now works on FreeBSD
1368 (thanks to NIIMI Satoshi)
1369 * fixed some bugs revealed by Paul Dietz' test suite:
1370 ** MISC.641: LET-conversion were not supposed to work in late
1372 * improvements to the Win32/x86 port:
1373 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1374 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1376 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1377 * new feature: source path information is generated for macro-expansion
1378 errors for use in IDE's like Slime (thanks to Helmut Eller)
1379 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1380 * bug fix: compiler does not loop forever on an invalid type in
1382 * improvement: compilation of most CLOS applications is significantly
1384 * optimization: added a limited bytecode compiler for simple toplevel
1385 forms, speeding up compilation and FASL loading
1386 * bug fix: the statistical profiler now properly distinguishes anonymous
1389 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1390 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1391 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1392 into subclasses of the finalized class. Now user calls to
1393 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1394 system return before any subclasses are finalized.
1395 * minor incompatible change: The reader no longer ignores errors
1396 regarding non-existent packages in #+ and #- feature tests.
1397 * new feature: command line options --no-sysinit, --no-userinit to
1398 inhibit loading the corresponding init files
1399 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1400 for sending data through UDP sockets (thanks to François-René Rideau)
1401 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1402 error is signalled instead (thanks to François-René Rideau)
1403 * minor incompatible change: Error signalling behaviour of lexical
1404 operations violating package locks has changed slightly. Refer to
1405 documentation on package locks for details.
1406 * enhancement: EVAL can process IF-expressions without resorting to the
1408 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1409 constant-folded properly due to missing out-of-line functions.
1410 (reported by Mika Pihlajamäki)
1411 * bug fix: LISTEN sometimes returned T even in cases where no data was
1412 immediately available from the stream
1413 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1414 were reversed. (reported by Levente Mészáros)
1415 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1416 when given a vector displaced to an adjustable array. (reported
1418 * bug fix: derivation of float boundaries from numbers outside the
1419 appropriate float range (reported by John Wiseman)
1420 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1422 * bug fix: precomputing cache entries for generic functions with
1423 some subclasses of specializers as yet invalid does not attempt to
1424 fill a cache line with a negative offset. (reported by Levente
1426 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1427 allow condition class objects as arguments to DOCUMENTATION and
1428 (SETF DOCUMENTATION); only find and set documentation for
1429 structure names for the STRUCTURE doc-type. (suggested by Gary
1431 * improvements to the Win32/x86 port:
1432 ** (user-homedir-pathname) and default initialization file
1433 locations now know about the user's "Documents and Settings"
1434 directory (thanks to Yaroslav Kavenchuk)
1435 ** run-program is implemented (thanks to Mike Thomas)
1436 ** sockets support (thanks to Timothy Ritchey)
1437 ** better backtrace support (thanks to Alastair Bridgewater)
1438 ** sb-grovel supported
1439 ** asdf-install and sb-posix work somewhat
1440 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1441 * improvements to the Solaris/x86 port:
1442 ** works on Solaris 11/Solaris Express
1443 ** floating-point exception handling support
1444 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1445 * fixed some bugs revealed by Paul Dietz' test suite:
1446 ** REMOVE-METHOD returns its generic function argument even when
1447 the method is not one of the generic functions' methods.
1448 ** objects declared as MEMBER types can be admissible arguments to
1449 structure accessors.
1450 ** printing characters should simply be printed by the FORMAT ~:C
1452 ** compiler failure when compiling functions with hairy constant
1453 defaults for optional parameters.
1454 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1455 function, which is already optimized.
1457 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1458 * new platform: experimental support for SBCL x86/Darwin, including
1459 MacOS X 10.4.5 on Intel.
1460 * new feature: Unicode character names are now known to the system
1461 (through CHAR-NAME and NAME-CHAR).
1462 * new feature: the filesystem location of SBCL's core file is
1463 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1464 * minor incompatible change: the contrib modules SB-POSIX and
1465 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1466 this change is to make it easier to distribute
1467 location-independent binaries.
1468 * bug fix: as implied by AMOP, standardized classes no longer have
1469 slots named by external symbols of public packages. (reported by
1471 * optimization: calling functions via constant symbols -- as in
1472 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1473 via the function object as in (FUNCALL #'FOO).
1474 * optimization: on x86 and x86-64, the calling convention for the
1475 case of dealing with an unknown number of values has been altered
1476 to be friendlier to the prediction heuristics implemented,
1477 particularly with reference to CALL and RET pairing. (thanks to
1478 Alastair Bridgewater)
1479 * optimization: on x86, the code for access to thread-local symbol
1480 values and binding/unbinding of thread-local symbols is smaller.
1481 (thanks to Alastair Bridgewater)
1482 * enhancement: CONSTANTP is now able to determine constantness of
1483 more complex forms, including calls to constant-foldable standardized
1484 functions and some special forms beyond QUOTE.
1485 * fixed bug: occasional GC crashes on Solaris/x86
1486 * optimization: x86-64 supports stack allocation of results of simple
1487 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1488 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1489 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1490 (thanks to James Knight)
1491 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1492 by the standard (thanks to Stephen Compall)
1494 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1495 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1496 be used for bundling the runtime and the core file into one
1497 executable binary. This feature is not currently supported on all SBCL
1498 platforms. (thanks to James Bielman and NIIMI Satoshi)
1499 * new feature: a generational or ephemeral garbage collector is now
1500 the default on the PowerPC platform (both Linux and Darwin). The
1501 old Cheney (stop and copy) collector is a build-time option.
1502 (thanks to Cyrus Harmon, after Raymond Toy)
1503 * minor incompatible change: the method by which SBCL finds its
1504 contributed modules has changed; it no longer relies on symbolic
1505 links from an $SBCL_HOME/systems directory, but searches directly
1506 in the subdirectories of $SBCL_HOME.
1507 * enhancement: the dynamic heap size on the Linux/PPC platform is
1508 markedly larger, even using the older Cheney garbage collector.
1509 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1510 certain complicated string types. (reported by Gary King)
1511 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1512 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1513 * fixed bug: DOCUMENTATION on structures no longer signals an
1514 error if no structure type of the right name exists. (reported by
1516 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1517 applicable non-standard primary or :AROUND method on
1518 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1519 no longer cause unbound variable errors. (reported by Kevin Reid)
1520 * fixed bug: metacircle resolution in cases where methods have slots
1521 added before the slots from STANDARD-METHOD. (reported by Jean
1523 * fixed bug: the Power PC assembly code for calling into lisp
1524 sometimes computed its return address wrongly (depending on the
1525 15th bit of the address, fixed at link time). This used to
1526 manifest itself in a segmentation violation while building PCL.
1527 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1528 many others over the years)
1529 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1530 the stack pointer (treated as a signed integer) is negative.
1531 (thanks to Peter van Eynde)
1533 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1534 * new platform: experimental support for the Windows operating
1535 system has been added. (thanks to Alastair Bridgewater)
1536 * fixed several bugs in and robustified the PPC FFI (including
1537 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1538 * bug fix: allow non-simple string symbol names (reported by Paul
1540 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1542 * bug fix: saving a core corrupted callbacks on x86/x86-64
1543 * bug fix: closed a loophole in metacircularity detection and
1544 grounding in the PCL implementation of CLOS.
1545 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1546 (reported by tomppa on #lisp)
1547 * bug fix: clear the direction flag on Lisp -> C transitions, as
1548 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1549 (reported by Andrej Grozin and Hendrik Maryns)
1550 * optimization: major improvements to GC efficiency on GENCGC platforms
1551 * optimization: faster implementation of EQUAL
1552 * optimization: emit more efficient opcodes for some common
1553 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1554 * optimization: several other minor code-generation improvements on x86-64
1555 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1557 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1558 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1559 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1560 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1561 by ANSI as a generic function, and for consistency (SETF
1562 GENERIC-FUNCTION-NAME) is treated likewise.
1563 * fixed bug #233.b: make constraint propagation notice when a variable
1564 value is changed after it is referenced but before it is used
1565 * fixed bug #296: no more arbitrary behaviour from filenames with
1566 odd characters as --load arguments. (reported by Adam Warner)
1567 * fixed bug #390: :CHARACTER-SET pathname components now work as
1568 expected. (reported by Tim Daly Jr)
1569 * fixed bug #391: complicated :TYPE intersections in slot
1570 definitions no longer cause an error in PCL internals.
1571 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1572 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1573 returning the number of octets which would be written to the
1574 file-stream. (thanks to Robert J. Macomber)
1575 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1576 arguments to the internal stream functions. (thanks to David
1578 * bug fix: stack exhaustion is now handled on the x86/SunOS
1579 platform. (thanks to Daisuke Homma)
1580 * bug fix: more accurate ROOM results on GENCGC platforms
1581 * optimization: improved type inference for arithmetic-for
1582 index variables in LOOP
1583 * optimization: faster floating-point SQRT on x86-64
1584 * fixed some bugs revealed by Paul Dietz' test suite:
1585 ** DOCUMENTATION returns NIL instead of "" for method combinations
1586 that don't have a docstring
1588 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1589 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1590 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1591 therefore are not customizeable by user code (as seems to be at
1592 least permitted and maybe required by AMOP). As a consolation,
1593 however, the SBCL implementation of these functions now calls
1594 REINITIALIZE-INSTANCE as specified by AMOP.
1595 * bug fix: slot-definition documentation is propagated as per ANSI
1596 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1597 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1598 argument, as do eslotd initialization methods. (from Pascal
1599 Costanza's "Closer" project)
1600 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1601 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1603 * bug fix: it is now possible to have more than one subclass of
1604 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1605 (reported by Bruno Haible, Pascal Costanza and others)
1606 * bug fix: the dependent update protocol now works for generic
1607 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1608 and Pascal Costanza)
1609 * bug fix: condition-class instances corresponding to
1610 DEFINE-CONDITION forms are now created eagerly. (reported by
1611 Kalle Olavi Niemitalo on comp.lang.lisp)
1612 * bug fix: floating point printing is more accurate in some
1613 circumstances. (thanks to Simon Alexander)
1614 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1615 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1616 * bug fix: callbacks on OS X now preserve stack-alignment.
1617 (thanks to Cyrus Harmon)
1618 * enhancement: the x86-64 disassembler is much better at
1619 disassembling SSE instructions. (thanks to Lutz Euler)
1620 * enhancement: CHECK-TYPE on a variable now allows type-inference
1621 (thanks to Vincent Arkesteijn)
1622 * optimization: improved performance of EUC-JP external format.
1623 (thanks to NIIMI Satoshi)
1624 * optimization: performance improvements to IO on file streams of
1625 :ELEMENT-TYPE CHARACTER
1626 * optimization: much faster memory allocation on x86-64
1627 * optimization: faster garbage collections (latency and throughput)
1629 * optimization: faster bignum subtraction and division on x86 and x86-64
1630 (thanks to Lutz Euler)
1631 * optimization: more accurate type-inference for loops that use a
1632 floating point index variable or a negative step.
1634 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1635 * bug fix: add a workaround to SBCL looping infinitely at startup on
1636 Linux kernels with apparently buggy implementations of personality().
1637 (thanks to Svein Ove Aas)
1638 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1639 (thanks to David Lichteblau)
1640 * bug fix: local bindings shadow global symbol macros for
1641 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1642 * optimization: non-open coded uses of numeric comparison operators
1643 (e.g. >) no longer cons when called with more than one parameter
1644 on platforms supporting dynamic-extent allocation.
1645 * enhancement: saving cores with foreign code loaded is now supported
1646 on MIPS/Linux in addition to the previously supported platforms.
1647 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1648 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1649 * bug fix: division by zero in sb-sprof when no samples were collected
1650 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1651 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1652 (thanks to Alexander Kjeldaas)
1653 * new feature: ignoring the return values of destructive functions
1654 like NREVERSE signals a compile-time style-warning.
1655 (thanks to Kevin Reid)
1656 * new feature: the alignment of alien structure fields can be
1657 explicitly specified. (thanks to Cyrus Harmon)
1658 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1659 and dump core on SIGQUIT
1661 ** incompatible change: threads do not inherit values of specials
1662 from their parents (see manual)
1663 ** bug fix: threads stacks belonging to dead threads are freed by the
1664 next exiting thread, no need to gc to collect thread stacks anymore
1665 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1666 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1667 ** bug fix: dynamic variable and thread start related gc lossage
1668 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1670 * fixed some bugs revealed by Paul Dietz' test suite:
1671 ** SUBTYPEP is slightly more accurate on heinously complicated
1672 CONS types where some of the members have uncertain (in the
1673 NIL, NIL sense) type relationships to each other.
1675 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1676 no longer purified unless :PURIFY T is explicitly specified.
1677 ** Non-purified cores are significantly smaller than before
1679 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1680 * new feature: timers based on Zach Beane's excellent timer package
1681 * added support for the following external formats: koi8-u,
1682 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1683 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1684 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1685 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1686 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1687 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1688 * incompatible change: a threaded SBCL will no longer revert to
1689 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1690 * bug fix: interrupts are disabled until startup is complete; no
1691 more sigsegvs when receiving a signal to soon
1692 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1694 * bug fix: add a workaround for the memory randomization features in
1695 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1696 workaround will only be in effect on systems with the proc filesystem
1698 * bug fix: printing objects of type HASH-TABLE signals a
1699 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1701 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1702 have been running for over 50 days. (reported by Gilbert Baumann)
1703 * bug fix: the logic for getting names of functions gets less
1704 confused when confronded with alternate-metaclass
1705 funcallable-instances. (reported by Cyrus Harmon)
1706 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1707 consistent, even on internal alternate-metaclass objects.
1708 * bug fix: flush closure information collected by physical
1709 environment analysis prepass before the main pass. (bug reported
1711 * bug fix: compiler pack phase does not modify a hashtable, which is
1712 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1713 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1714 may be expanded. (reported by Alexander <malishev> on c.l.l)
1715 * bug fix: no more style-warnings when compiling code using LAST
1716 under high-SPEED low-SPACE settings. (reported by David Wallin
1718 * fixed some incompatibilities between SBCL's MOP and the MOP
1720 ** the METAOBJECT class is now implemented;
1721 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1722 STANDARD-OBJECT, as required;
1723 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1724 now compatible (as required by VALIDATE-SUPERCLASS); there
1725 remains a consistency requirement over the presence or absence
1726 of the FUNCTION class in the superclass list of finalized
1727 classes; see the manual for more details;
1728 ** the :AROUND method for COMPUTE-SLOTS on
1729 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1730 requested slot ordering.
1732 ** bug fix: parent thread now can be gc'ed even with a live
1734 ** bug fix: binding a special with PROGV to no value is not
1735 the same as that symbol not having been bound (thanks to
1737 ** bug fix: binding specials is thread safe (thanks to
1739 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1740 and SB-SPROF do not die with 'no handler for signal XX in
1741 interrupt_handle_now(..)' anymore
1742 ** bug fix: WITH-TIMEOUT works with multiple running threads
1743 * fixed some bugs revealed by Paul Dietz' test suite:
1744 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1745 the :method-class keyword argument.
1747 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1748 * new port: the Solaris operating system on x86 processors is now
1749 mostly supported, though some rough edges in the environment
1750 remain. (thanks to Daisuke Homma)
1751 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1752 space for its heap. (thanks to Thiemo Seufer)
1753 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1754 closures. (thanks to Thiemo Seufer)
1755 * minor incompatible change: eof selects abort in the debugger.
1756 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1757 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1759 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1760 deprecated, and will go away in a future revision of SBCL.
1761 * minor incompatible change: GC-ON and GC-OFF are no longer
1762 implemented with a counter, it does not matter how many times gc
1763 is switched on or off
1764 * bug fix: discriminating functions for generic function classes
1765 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1766 make invalid assumptions about method precedence order. (reported
1768 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1769 platforms. (thanks to Thiemo Seufer)
1770 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1771 * bug fix: correct pathname printing with printer escaping is on.
1772 (thanks to Kevin Reid)
1773 * bug fix: complex VOP definitions in "user-space" no longer trigger
1774 package locks. (reported by Zach Beane)
1775 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1776 for funcallable-instance functions, and is no different from
1778 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1779 characters in the current readtable. (reported by Nicholas Neuss)
1780 * bug fix: SUBTYPEP on various CONS types returns more a more
1781 accurate acknowledgment of its certainty.
1782 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1783 lists in some cases. This partially fixes bug 384.
1784 * flush all standard streams before prompting in the REPL and the
1786 * bug fix: signal handling and triggering gc do not conflict
1787 directly anymore, in particular a high frequency sb-sprof does
1788 not prevent gc from running
1789 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1790 approximation for timezone and DST information between the
1791 universal time epoch and the smallest negative 32-bit time_t.
1792 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1793 year 1899 so long as the encoded time is non-negative.
1794 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1795 the correct answer for non-constant rotate amounts when performing
1796 an inline 32-bit rotation.
1798 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1799 there is only one thread in the session
1800 ** bug fix: memory leak for streams created in one thread and
1801 written to in another
1802 ** bug fix: lockup when compiled with gcc4
1803 ** bug fix: race that allows the gc to be triggered when gc is
1805 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1806 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1808 ** bug fix: finished threads are gc'ed properly
1809 * fixed some bugs revealed by Paul Dietz' test suite:
1810 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1811 has arguments with a different set of applicable methods from
1812 the orignal arguments.
1813 ** The type error thrown by MAP now has a correct expected-type
1815 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1816 name a compiled function.
1817 ** Three MISC tests where a large bignum was improperly coerced to
1818 a float that couldn't represent that bignum during type
1819 derivation were fixed.
1820 ** SUBTYPEP can now handle types involving the negation of a
1821 list-form FUNCTION type.
1822 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1823 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1824 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1826 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1827 * New feature: Experimental support for bivalent streams: streams
1828 opened with :element-type :default now allow character and binary
1829 (unsigned-byte 8) I/O
1830 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1831 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1833 * bug fix: no more highly sporadic "couldn't check whether ~S is
1834 readable" when reading a stream and an interrupt hits in the middle
1835 of a select system call
1836 * compiler better recognizes complex arrays (reported by Eduardo
1838 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1839 for variables on SBCL built with linkage-tables. (reported by Luis
1841 * various error reporting improvements.
1842 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1843 (Thanks to Thiemo Seufer)
1844 * bug fix: nested reader invokations work correctly
1845 * bug fix: it is possible to have simultaneous references to foreign
1846 code and foreign data with the same name.
1848 ** added x86-64 support
1849 ** incompatible change: the threading api now works with thread
1850 objects instead of thread ids
1851 ** bug fix: threads are protected from signals and interruption when
1852 starting up or going down
1853 ** bug fix: a race where an exiting thread could lose its stack to gc
1854 ** bug fix: don't halt on infinite error in threads if possible
1855 ** fixed numerous gc deadlocks introduced in the pthread merge
1856 ** bug fix: fixed thread safety issues in read and print
1857 ** bug fix: debugger doesn't hang on session lock if interrupted at
1858 an inappropriate moment
1859 ** bug fix: run-program is now thread safe(r)
1860 ** bug fix: inner with-recursive-lock no longer releases the mutex
1861 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1862 * fixed some bugs revealed by Paul Dietz' test suite:
1863 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1864 EXPECTED-TYPE slots filled.
1865 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1867 ** MAKE-CONDITION accepts classes as type-designators.
1868 ** COMPILE may never return NIL.
1869 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1870 range before calling Unix time functions
1872 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1873 * numerous signal handling fixes to increase stability
1874 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1875 * minor incompatible change: we now correctly canonize default
1876 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1877 as per the MOP, rather than the historical (INITARG INITFUNCTION
1878 INITFORM). (reported by Bruno Haible)
1879 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1880 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1881 for more information.
1882 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1883 pathname is a directory pathname.
1884 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1885 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1887 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1888 :METACLASS argument in addition to a class name. (reported by
1889 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1890 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1891 (reported by Bruno Haible; patch from Gerd Moellmann)
1892 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1894 * bug fix: more cleanups to the floating point exception handling on
1895 x86-64 (thanks to James Knight)
1896 * bug fix: the compiler does not try to constant fold calls of
1897 COERCE and BIT-* functions when they return freshly constructed
1898 objects. (reported by Nikodemus Siivola and Paul Dietz)
1899 * optimization: improved the MIPS versions of generic (in the
1900 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1901 * optimization: direct conversion of (unsigned-byte 32) to floats on
1902 the PowerPC platform.
1903 * optimization: structure instances with raw slots now use less
1904 memory, and probably show better memory locality. (thanks to
1906 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1907 are treated as requests for stack allocation on the x86-64,
1908 Alpha-32, PPC and SPARC platforms.
1909 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1910 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1911 (thanks to Tony Martinez)
1913 ** gcing a dead thread can no longer lead to lockups
1914 ** threads block signals until they are set up properly
1915 ** errno is no longer shared by threads
1916 ** interrupt-thread restores the eflags register on x86
1917 ** fixed some lockups due to gc/thread interaction
1918 * fixed some bugs revealed by Paul Dietz' test suite:
1919 ** invalid dotted lists no longer raise a read error when
1920 *READ-SUPPRESS* is T
1921 ** signal an error if a symbol that names a declaration is used
1922 as the name of a type, or vice versa
1923 ** allow using the (declare (typespec var*)) abbreviation for
1924 (declare (type typespec var*)) with all type specifiers
1925 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1926 ** always return NIL from PPRINT-POP when OBJECT is NIL
1927 ** don't signal errors when pretty-printing malformed LABELS,
1928 FLET or MACROLET forms
1929 ** declarations in a DOLIST body are also in scope for the
1931 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1933 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1936 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1937 * fixed cross-compiler leakages that prevented building a 32-bit
1938 target with a 64-bit host compiler.
1939 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1940 opened with :IF-EXISTS :OVERWRITE.
1941 * fixed bug 281: error for an invalid qualifier in a short-form method
1942 combination method is not signalled until the faulty method is called.
1943 * bug fix: iteration variable type inferrer failed to deal with open
1944 intervals. (reported by Alan Shields)
1945 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1946 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1947 or more runtime options were provided to the sbcl binary.
1948 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1949 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1950 * optimization: there's now a fast-path for fixnum arguments in the
1951 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1952 * optimization: the code generated on x86-64 is more compact thanks
1953 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1954 * bug fix: floating-point exception handling now partly works on
1955 x86-64. (Thanks to James Knight)
1956 * improvement to the MIPS backend from Thiemo Seufer: C-style
1957 64-bit long long arguments and return values to alien functions
1959 * fixed some bugs revealed by Paul Dietz' test suite:
1960 ** the type-error signalled from WARN has a filled-in DATUM slot.
1961 ** the type-error required when a stream is not associated with
1962 a file has the stream as its datum.
1963 ** type-errors on single-floats on x86-64 no longer have
1964 :INVALID-OBJECT as the datum
1965 ** the type-errors signalled for invalid function names now have
1966 a correct expected type
1967 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1968 for structures defined with a :TYPE. Documentation strings for
1969 typed structures are no longer immediately discarded
1970 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1971 broadcast streams and synonym streams. FILE-LENGTH now also works
1972 on broadcast streams.
1974 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1975 * incompatible change: the --noprogrammer option, deprecated since
1976 version 0.7.5, has been removed. Please use the equivalent
1977 --disable-debugger option instead.
1978 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1980 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1981 inoperational for a while now) has been completely removed.
1982 * null lexical environments are now printed as #<NULL-LEXENV>,
1983 significantly reducing the amount of clutter in typical backtraces.
1984 * documentation on weak pointers, finalization, and after GC hooks
1985 has been added to the manual.
1986 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1987 as other specialized array types, is much faster. SUBSEQ and
1988 COPY-SEQ on such arrays have also been sped up.
1989 * optimization: EQL is now more efficient when at least other argument
1990 is known to be of type (OR FIXNUM (NOT NUMBER)).
1991 * fixed bug: the runtime is now less vulnerable to changes in the
1992 size of the SBCL object on OS X, and virtual memory is reserved for
1993 all spaces that need to be at a fixed address.
1994 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1995 * fixed bug: finalizers and after GC hooks that cause consing are now
1997 * fixed bug: compiler error messages and summaries are now printed to
1998 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1999 * fixed inference of the upper bound of an iteration variable.
2000 (reported by Rajat Datta).
2001 * fixed bug 211e: calling local functions with duplicated constant
2002 keyword argument no longer causes a bogus style warning about an
2004 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2005 ftype declaration to be disregarded. (reported by Dave Roberts)
2006 * fixed bug 373: caused by erronous compilation of references to alien
2007 variables in the runtime on ppc/darwin.
2008 * fixed bug 376: CONJUGATE type deriver.
2009 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2010 (reported by Baughn on #lisp)
2011 * a cross-compiler bug on non-x86 platforms has been identified and
2012 fixed. (thanks to Bruno Haible)
2013 * improvements to the MIPS runtime code for increased stability.
2014 (thanks to Thiemo Seufer)
2015 * increased the maximimum compact environment size to allow
2016 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2017 * partial workaround for bug 135: don't name the function we're
2018 calling for hairy cases of EVAL, so as not to accumulate one environment
2019 entry per call to EVAL. (reported by Kevin Reid)
2020 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2021 * optimization: type testing for non-vector arrays should be faster.
2022 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2024 * bug fix: setting 31st element of a bit vector to zero did not work
2026 * bug fix: redefining a class definition which failed due to a
2027 previous accessor / function clash now works (but see BUGS entry
2028 #380 for more problems in this area). (thanks to Zach Beane)
2029 * the long form of DEFINE-METHOD-COMBINATION disables method group
2030 checking when given a single method group with pattern *. (thanks
2032 * on x86 compiler supports stack allocation of results of simple
2033 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2034 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2035 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2036 * fixed some bugs related to Unicode integration:
2037 ** the restarts for recovering from input and output encoding
2038 errors only appear when there is in fact such an error to
2040 * fixed some bugs revealed by Paul Dietz' test suite:
2041 ** MISC.549 and similar: late transformation of unsafe type
2042 assertions into derived types caused unexpected code
2044 ** SCALE-FLOAT type deriver is less wrong.
2045 ** type derivers for EXP, LOG and similar functions compute result
2046 types for complex arguments better.
2047 ** (MISC.563) CONJUGATE type deriver works for very restricted
2049 ** out-of-line type testers for character strings are available.
2050 ** EQUAL compiler transform understands specialness of objects
2052 ** accessing double-floats stored on the stack now works on x86-64.
2053 ** debugger internals could sometimes create invalid lispobjs,
2054 resulting in GC crashes.
2055 ** MISC.548: type check weakening can convert required type into
2057 ** initialization forms for bindings are not in scope of free special
2060 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2061 * incompatible change: thread support for non-NPTL systems has
2062 been removed - locking is buggy and unreliable. A threaded
2063 SBCL build will now warn at startup and refuse to create
2064 new threads, unless futex support is detected in the kernel
2065 * incompatible change: the top level REPL now has only an ABORT
2066 restart associated with it, not TOPLEVEL and ABORT as it used to.
2067 TOP and TOPLEVEL are now available as debugger commands for
2068 returning to the top level.
2069 * incompatible change: forms evaluated in the REPL now use the
2070 global optimization policy.
2071 * incompatible change: user- and system-initialization files are
2072 no longer processed with LOAD, but by READ and EVAL; hence the
2073 global optimization policy, startup package, readtable, etc,
2075 * The .fasl file format number has been incremented because of
2076 various incompatible changes.
2077 * internal entry point details and argument counts no longer appear
2078 in backtraces unless explicitly requested by setting
2079 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2080 * built-in and standard functions no longer have names like "top
2081 level local call to FOO".
2082 * fixed bug 32: functions defined in non-null lexical environments
2083 now have more legible printed representation
2084 * fixed bug 33: functions defined in non-null lexical environemnts
2085 are now more amenable to inspection by INSPECT.
2086 * workaround for bug 354: XEPs no longer appear in backtraces unless
2087 explicitly requested.
2088 * fixed bug: receiving the signal which results from attempting to
2089 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2090 SIGBUS on other *BSDs) on architectures where the C stack is also
2091 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2092 to debugger confusion or wild instability. (reported by Cheuksan
2094 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2095 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2096 (reported by Lutz Euler)
2097 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2098 compiled code. (reported by Paul Dietz)
2099 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2100 specialized parameter to a method to something that is not TYPEP
2101 the specializer is now possible.
2102 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2103 face of package deletion.
2104 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2105 pathname. (reported by David Tolpin on c.l.l)
2106 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2107 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2108 STRING-STREAM and FILE-STREAM, respectively.
2109 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2110 than previously; in addition, its test suite is now run on build.
2111 (thanks to Luke Gorrie)
2112 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2113 * optimization: added a immediate representation for single-floats
2115 * fixed some bugs revealed by Paul Dietz' test suite:
2116 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2117 ** attempting to create a package with a colliding nickname causes
2118 correctable errors to be signalled.
2119 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2120 ** several x86-64 backend bugs related to sign-extension of immediate
2123 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2124 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2125 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2127 * fixed loading of multiply forward-referenced layouts.
2128 (thanks to Cheuksan Wang)
2129 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2130 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2131 (see COMPILE-FILE documentation for details of :PRINT option)
2132 * fixed bugs 19 and 317: fixed-format floating point printing is
2133 more accurate. This also fixes a bug reported by Adam Warner
2134 related to the ~@F format directive.
2135 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2137 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2138 dispatch table if the from-char is a dispatch macro character.
2139 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2140 by the wordsize no longer ignore the last word. (reported by Lutz
2142 * fixed bug in type checking of dynamic-extent variables. (reported
2144 * optimization: sequence traversal functions use their freedom to
2145 coerce function designators to functions.
2146 * optimization: code with many calls to CLOS methods specialized on
2147 CLOS classes has had redundant type checks removed.
2148 * optimization: added declarations to speed up operations that access
2149 the internal character database (for example STRING-UPCASE)
2150 * optimization: comparison operations between floats and sufficiently small
2151 fixnums no longer create extra rationals
2152 * fixed some bugs related to Unicode integration:
2153 ** portions of multibyte characters at the end of buffers for
2154 character-based file input are correctly transferred to the
2155 start of the buffer at the next read.
2156 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2157 passing it through to OPEN.
2158 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2159 argument given to internal calls to OPEN.
2160 ** the built-in comment readers (introduced by character sequences
2161 ";" and "#|") are more forgiving to encoding errors; they will
2162 STYLE-WARN and then attempt to resync the stream at a character
2163 boundary. (thanks to Teemu Kalvas)
2164 * fixed some bugs in the x86-64 port:
2165 ** Negative short int return values from c-calls are sign-extended
2167 ** The stack is aligned to 16-bytes for c-calls, as required by
2168 the ABI. (thanks to Cheuksan Wang)
2169 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2170 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2171 for enums. (thanks to Vincent Arkesteijn)
2172 ** Multiple small optimizations and bugfixes for floating point
2174 * fixed some bugs revealed by Paul Dietz' test suite:
2175 ** Space, Tab, Linefeed, Return and Page have the invalid
2176 secondary constituent character trait.
2177 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2179 ** WITH-INPUT-FROM-STRING should only update the index place on
2181 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2182 works more reliably.
2183 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2184 with one-dimensional arrays with fill pointers.
2185 ** TYPE-OF failed on a complex with an integer realpart and a
2187 ** compiler failure during type inference for the code of form
2188 (IF (EQL X (THE ...)) ...) (MISC.535).
2190 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2191 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2192 platform. The port supports 61-bit fixnums, large memory spaces
2193 and reloading shared object files.
2194 * enhancement: saving cores with foreign code loaded is now
2195 supported on ppc/Darwin in addition to the previously supported
2197 * enhancement: the statistical profiler now walks deeper into the
2198 call stack for more meaningful call-graphs and accrued time
2199 reports (x86/x86-64 only). It also now reports time spent in
2201 * enhancement: it is now possible to trace most individual methods
2202 of a generic function in addition to tracing the generic function
2204 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2206 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2207 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2208 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2209 produces an error. (thanks to Vincent Arkesteijn)
2210 * bug fix: NAMESTRING on pathnames with :WILD components in their
2211 directories works correctly. (thanks to Artem V. Andreev)
2212 * fixed bug 125: compiler preserves identity of closures. (reported
2214 * bug fixed: functions with &REST arguments sometimes failed with
2215 "Undefined function" when compiled with (DEBUG 3). (reported by
2217 * bug fix: overflow during compiling of setting element of a bit
2218 vector with constant index and value. (reported by Timmy Douglas)
2219 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2220 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2221 * on x86 compiler partially supports stack allocation of dynamic-extent
2223 * GO and RETURN-FROM do not check the extent of their exit points
2224 when compiled with SAFETY 0.
2225 * fixed some bugs related to Unicode integration:
2226 ** encoding and decoding errors are now much more robustly
2227 handled; it should now be possible to recover even from invalid
2228 input or output to the terminal. (thanks to Teemu Kalvas)
2229 ** provided a first cut at implementing STRING-TO-OCTETS and
2230 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2231 ** altered the SB-MD5 contributed module slightly, changing the
2232 interface just enough for it to be supportable for builds where
2233 lisp characters are not eight bits.
2234 * fixed some bugs revealed by Paul Dietz' test suite:
2235 ** the FORMATTER-generated functions for ~V[ conditionals require
2236 the correct number of arguments.
2237 ** READ-FROM-STRING returns the mandated second value when applied
2238 to displaced strings.
2239 ** the #\Rubout and #\Backspace characters are treated as invalid
2240 constituent characters by the tokenizer.
2242 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2243 * new feature: reloading changed shared object files with
2244 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2245 * new feature: references to foreign variables and functions
2246 can now be compiled and loaded before the corresponding shared
2247 object file is loaded, as long as the foreign definitions are
2248 available at runtime.
2249 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2250 supported. (thanks to Dan Debertin)
2251 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2252 just keywords. (thanks to Vincent Arkesteijn)
2253 * fixed bug #331: structure-class instances corresponding to
2254 DEFSTRUCT forms are now created eagerly.
2255 * fixed bug #345: backtraces from calls to undefined functions work
2256 on x86 as well. Related bug #61 is now also partially fixed on x86
2257 (backtraces from throws to unknown catch tags.)
2258 * bug fix: lambda-list parsing is now stricter vrt. order and number
2259 of lambda-list keywords.
2260 * bug fix: as specified by AMOP, an error is signalled if a
2261 class-option appears multiple times in a DEFCLASS form. (reported
2263 * bug fix: RANDOM can be compiled when the compiler derives the type
2264 of its numeric argument as a disjoint set of small integers.
2265 (reported by Paul Dietz)
2266 * bug fix: starting a core saved with shared objects loaded when
2267 those objects are not available no longer causes threaded SBCL to
2268 hang. (reported by Sean Ross)
2269 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2271 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2272 argument, which must always be NIL. (reported by Kalle Niemitalo)
2273 * bug fix: printing 1.0d+23 no longer results in an error.
2274 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2275 * bug fix: structure slot setters preserve evaluation order. (thanks
2277 * fixed some bugs related to Unicode integration:
2278 ** RUN-PROGRAM can allow its child to take input from a Lisp
2279 stream. (reported by Stefan Scholl)
2280 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2281 locales. (reported by Ken Causey)
2282 * fixed some bugs revealed by Paul Dietz' test suite:
2283 ** INCF, DECF and REMF evaluate their place form as specified in
2285 ** FORMATTER expands ~{ ~} iteration directives with V or #
2286 parameters correctly.
2287 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2288 consequent uses no arguments correctly.
2289 ** the system has a partial understanding of the (COMPLEX RATIO)
2291 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2294 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2295 * new feature: a build-time option (controlled by the :SB-UNICODE
2296 keyword feature, enabled by default) for building the system with
2297 support for the entire 21-bit character space defined by the
2299 * new feature: the system now has rudimentary external-format
2300 support; the primary user-visible change at this time is that
2301 characters with the high bit set (such as lower-case-e-acute) will
2302 print correctly to a terminal in a UTF-8 environment.
2303 * minor incompatible change: BASE-CHAR no longer names a class;
2304 however, CHARACTER continues to do so, as required by ANSI.
2305 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2306 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2307 has been moved to the SB-EXT package (temporarily re-exported from
2309 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2310 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2312 * fixed bug #351: better error-handlind and reporting for malformed
2314 * fixed bug #350: bignum-printing is now more memory-efficient,
2315 allowing printing of very large bignums, eg. (expt 2 10000000).
2316 (reported by Bruno Haible)
2317 * fixed bug #302: better primitive-type selection for intersection
2319 * fixed bug #308: non-graphic characters now all have names, as
2320 required. (reported by Bruno Haible)
2321 * bug fix: redefining a class with different superclasses now correctly
2322 removes it from the direct-subclasses of its previous superclasses.
2323 (reported by David Morse)
2324 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2325 new value now works. (reported by Bruno Haible)
2326 * bug fix: correct canonicalization of multiple non-standard slot
2327 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2328 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2329 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2330 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2331 now exists, an signals an error.
2332 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2333 during macroexpansion for non-iteration variables as well. (reported
2334 by Bruno Haible for CMUCL)
2335 * bug fix: Cyclic structures and unprintable objects in compiler
2336 messages no longer cause errors. (reported by Bruno Haible)
2337 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2338 and READ-FROM-STRING all now return a primary value of NIL if
2339 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2340 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2341 (reported by Bruno Haible for CMUCL)
2342 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2343 do not designate a condition. (reported by Bruno Haible for
2345 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2346 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2347 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2348 permitting resolution in favour of any of the conflicting symbols.
2349 (reported by Bruno Haible for CMUCL)
2350 * bug fix: EQUAL compiler optimizations is less aggressive on
2351 strings which can potentially compare true despite having distinct
2352 specialized array element types.
2353 * bug fix: unit enumerations can be defined without dividing by
2354 zero. (thanks to Vincent Arkesteijn)
2355 * FORMAT compile-time argument count checking has been enhanced.
2356 (report from Bruno Haible for CMUCL)
2357 * a partial workaround for the bug 262: the compiler does not try to
2358 inline-expand a local function doing RETURN-FROM from a deleted
2359 BLOCK. (thanks to Peter Denno for the bug report and to David
2360 Wragg for the simple test case)
2361 * fixed some bugs revealed by Paul Dietz' test suite:
2362 ** RENAME-PACKAGE allows all package designators as new package
2364 ** constraint propagation and lambda variable substitution are
2365 more cautious in dealing with partially deleted code.
2366 ** compiler sometimes lost reoptimization passes.
2367 ** CERROR, given a condition as condition designator, passes its
2368 remaining arguments to the continue format control without
2370 ** Case-altering FORMAT directives work correctly on non-ASCII
2372 ** The REAL type specifier handles bounds outside the floating
2373 point ranges without signalling FLOATING-POINT-OVERFLOW.
2374 ** Functions with IR1-transformations can create intercomponent
2375 references to global functions.
2376 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2378 ** FORMAT ~R treats a nil value for its first parameter correctly.
2380 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2381 * enhancement: saving cores with foreign code loaded is now
2382 supported on x86/NetBSD and sparc/Linux in addition to the previously
2383 supported platforms.
2384 * bug fix: on some platforms repeated installations caused multiple
2385 copies of HTML documentation to be installed -- should not happen
2386 any more. (reported by Stefan Scholl)
2387 * bug fix: parsing self-recursive alien record types multiple times
2388 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2389 original patch by Helmut Eller for CMUCL)
2390 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2391 (thanks to Richard Kreuter)
2392 * bug fix: defining classes whose accessors are methods on existing
2393 generic functions in other (locked) packages no longer signals
2394 bogus package lock violations. (reported by François-René Rideau)
2395 * bug fix: special variables as DEFMETHOD parameters no longer have
2396 associated bogus type declarations. (reported by David Wragg and
2398 * bug fix: read-write consistency on streams of element-type
2399 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2400 * bug fix: redefiniton of the only method of a generic function with
2401 no DEFGENERIC no longer emits a full WARNING. In addition,
2402 redefinition of generic functions with no DEFGENERIC to an
2403 incompatible lambda list now signals an error. (thanks to Zach
2405 * bug fix: DEFGENERIC now works even when there's a function of the
2406 same name in an enclosing lexical environment. (thanks to Zach
2408 * fixed compiler failure, caused by instrumenting code during
2409 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2410 * optimization: added loop analysis and improved register allocation
2411 * fixed some bugs revealed by Paul Dietz' test suite:
2412 ** POSITION on displaced vectors with non-zero displacement
2413 returns the right answer.
2414 ** (SIMPLE-STRING) is a valid type specifier for sequence
2416 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2418 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2419 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2421 ** The justification version of the FORMAT ~< directive treats
2422 non-zero minpad parameter correctly.
2424 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2425 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2426 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2427 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2428 the supported interface.
2429 * new feature: Single-stepping of code compiled with DEBUG 2 or
2430 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2431 * new feature: saving cores with foreign code loaded is now
2432 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2433 Timothy Moore's work for CMUCL)
2434 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2435 parameters to * instead of NIL if no initform is supplied.
2436 (reported by Johan Bockgaard)
2437 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2438 correctly when FUNCALL appears as the car of the form. Note:
2439 despite this FUNCALL forms are not currently subject to
2440 compiler-macro expansion. (port of Raymond Toy's fix for the
2441 same from CMUCL, reported by Johan Bockgaard)
2442 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2443 (thanks for Teemu Kalvas)
2444 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2445 case instead of terminating the iteration (thanks for Julian
2446 Squires, Sean Champ and Raymond Toy)
2447 * bug fix: incorrect expansion of defgeneric that caused a style
2448 warning. (thanks for Zach Beane)
2449 * bug fix: slot accessor effective method computation works properly
2450 for classes with multiple non-standard applicable methods on
2451 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2452 * on x86 compiler supports stack allocation of results of LIST and
2453 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2454 CMUCL implementation by Gerd Moellmann)
2455 * fixed some bugs revealed by Paul Dietz' test suite:
2456 ** FORMAT strings with both the ~<~:;~> form of the justification
2457 directive and pretty-printing directives cause an error.
2459 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2460 * incompatible change: the internal functions
2461 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2462 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2463 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2464 instead of the old functions.
2465 * new feature: on platforms where "dladdr" is available foreign
2466 function names now appear in backtraces. (based on Helmut Eller's
2468 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2469 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2471 * documentation: ASDF manual is now distributed with SBCL.
2472 * bug fix: SBCL can now load its contributed modules with REQUIRE
2473 even if the system-provided entries have been removed from
2474 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2476 * bug fix: backtraces involving undefined functions or assembly
2477 routines are more informative. (thanks to Brian Downing)
2478 * bug fix: mutually referent alien structures now work correctly.
2479 (reported by Rick Taube)
2480 * bug fix: structures defined by WITH-ALIEN can be referred to
2481 within other definitions in the same WITH-ALIEN.
2482 * bug fix: division operators (MOD, TRUNCATE and the like) with
2483 constant zero divisors and integer dividends no longer generate
2485 * bug fix: provide default methods for INPUT-STREAM-P and
2486 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2487 * bug fix: improve the detection and resolution of MOP metacycles
2488 (where effective method computation depends on the generic function
2489 having its effective method computed). (reported by Bruno Haible)
2490 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2491 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2492 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2493 * bug fix: pathnames with relative directory components can be
2494 represented relative to default pathnames.
2495 * optimization: in taking the GCD of bignums, reduce the two bignums
2496 to approximately the same size (using Euclid's algorithm) before
2497 applying the more sophisticated binary GCD. (thanks to Juho
2499 * optimization: COUNT on bitvectors now operates word-at-a-time.
2500 * optimization: ASH with a positive, but not necessarily constant,
2501 (leftwards) shift, when in a modular context, is compiled to a
2503 * fixed some bugs revealed by Paul Dietz' test suite:
2504 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2505 if the corresponding argument is NIL.
2506 ** FORMAT directives accepting a minpad parameter treat negative
2508 ** literal commas as character parameters to FORMAT directives are
2510 ** literal spaces directly after ~<Newline> directives within a
2511 format-logical-block (~:< ~@:>) do not induce :FILL-style
2512 conditional newlines.
2513 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2514 as specified: it no longer includes conditional newlines).
2515 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2517 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2518 :PER-LINE-PREFIX argument does not evaluate to a string.
2520 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2521 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2522 the manual for details; for now, package locks can be disabled by
2523 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2524 if no major problems are found then it is likely that they will be
2525 compiled in unconditionally.
2526 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2527 unsupported operators on all platforms. To load a shared library
2528 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2529 object file, link it into a shared library outside of SBCL and
2530 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2532 * bug workaround: ROOM T and the GC were not working together
2533 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2534 reporting facility aren't preserved. That reporting has been
2535 disabled, so now until and unless someone figures out how to make it
2536 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2537 * minor incompatible change: as threatened around sbcl-0.8.0, the
2538 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2539 The symbols are also exported from SB-PCL for backwards
2540 compatibility, but more so than before SB-PCL should be treated as
2541 an implementation-internal package.
2542 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2544 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2545 It is known as of this release not to work on the Alpha, however.
2546 * fixed bug #167: errors signalled due to illegal syntax in method
2547 bodies are now more legible.
2548 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2549 designators and can hence be used with TYPEP.
2550 * fixed bug #333: CHECK-TYPE now ensures that the type error
2551 signalled, if any, has the right object to be accessed by
2552 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2553 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2554 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2556 * fixed bug #334: programmatic addition of slots using specialized
2557 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2558 and :ALLOCATION :CLASS slots.
2559 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2560 float exponents. (rereported by Peter Seibel)
2561 * fixed a bug: #\Space (and other whitespace characters) are no
2562 longer considered to be macro characters in standard syntax by
2563 GET-MACRO-CHARACTER.
2564 * fixed bug: initialization of condition class metaobjects no longer
2565 causes an instance of the condition to be created. (reported by
2567 * fixed bug: it is now possible to have slots such that
2568 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2569 description is neither :INSTANCE nor :CLASS.
2570 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2571 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2572 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2573 non-standard SLOT-DEFINITION-ALLOCATION values.
2574 * fixed bugs: various race conditions handling exiting threads.
2575 CL-PPCRE's thread torture test now passes.
2576 * fixed arguably-a-bug: GC time proportional to number of threads in
2577 system even when most of them are idle
2578 * optimization: improved performance of BIT and SBIT on bit-vectors.
2579 * .fasl file incompatibility: The fasl file version number has
2580 been incremented because of changes associated with package locks.
2582 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2583 * minor incompatible change: the system no longer provides
2584 optimization hints (conditions which are TYPEP
2585 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2586 mode; these hints are still emitted when higher SPEED optimization
2588 * new contrib module: a sampling profiler (profiling by statistical
2589 sampling, rather than by instrumenting functions) is available as
2590 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2591 * the behaviour of the standard function ED is now customizeable by
2592 third parties through a hook variable: see ED's documentation
2593 string for information on the protocol.
2594 * the compiler no longer emits efficiency notes for (FUNCALL X)
2595 when the type of X is uncertain under default optimization
2597 * fixed bug 276: mutating a binding of a specialized parameter to a
2598 method to something that is not TYPEP the specializer is now
2600 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2601 exhibit the required behaviour on the x86 platform. (thanks to
2602 Peter van Eynde, Eric Marsden and Bruno Haible)
2603 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2604 even for difficult arguments. (reported by Peter Graves)
2605 * fixed bug 141a: the backquote printer now descends quoted
2607 * fixed another bug in backquote printing: no more destructive
2608 modification of the form's list structure. (reported by Brian
2610 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2611 move between its address being taken and the call to
2612 interrupt_thread, fixing a crashing race condition.
2613 * the SB-POSIX contrib implementation has been adjusted so that it
2614 no longer exhibits ridiculously poor performance when constructing
2615 instances corresponding to C structs.
2617 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2618 * minor incompatible change: the sb-grovel contrib now treats C
2619 structures as alien (in the sense of SB-ALIEN) objects rather than
2620 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2621 has implications for memory management of client code
2622 (sb-grovel-returned objects must now be manually managed) and for
2623 type safety (alien objects now have full types).
2624 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2625 used to control emission of compiler diagnostics, rather than the
2626 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2627 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2628 quality should be considered deprecated.
2629 * install.sh now installs the user manual as well
2630 * (not quite a new documentable feature, but worth considering in
2631 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2632 beginnings of a semantically meaningful condition hierarchy is
2633 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2635 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2636 designator as the defaults argument.
2637 * fixed bug: Displaced arrays whose displaced-to array has become
2638 too small now cause ARRAY-DIMENSION to signal an error, providing
2639 for safer bounds-checking. (reported by Bruno Haible)
2640 * fixed bug: DEFCLASS slot definitions with identical :READER and
2641 :WRITER names now signal a reasonable error. (reported by Thomas
2643 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2645 * fixed bug: Invalid slot specification errors now print correctly.
2646 (thanks to Zach Beane)
2647 * fixed bug 320: Shared to local slot value transfers in class
2648 redefinitions now happen corrently from superclasses as
2649 well. (reported by Bruno Haible)
2650 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2652 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2653 improper lists correctly. (reported by Bruno Haible)
2654 * fixed bug 313: source-transform for <fun-name> was erroneously
2655 applied to a call of a value of a variable with name <fun-name>.
2656 (reported by Antonio Menezes Leitao)
2657 * fixed bug 307: The obsolete instance protocol ensures that
2658 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2659 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2660 remove dead unknown-values globs from the middle of the stack.
2661 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2662 easier to resume long computations after using *BREAK-ON-SIGNALS*
2663 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2664 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2665 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2666 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2667 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2668 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2670 * fixed a (fairly theoretical) bug in string printing: if
2671 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2672 does not have array-element-type equal to the most general string
2674 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2675 function name. (thanks to Nikodemus Siivola)
2676 * fixed bug: compile-time format string checker failed on
2677 non-closed ~{. (reported by Thomas F Burdick)
2678 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2679 don't warn on legal loop code involving more than one aggregate
2680 boolean. (thanks to Nikodemus Siivola)
2681 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2682 in keyword position to MAKE-PACKAGE was not being reported as
2684 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2685 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2687 * fixed bugs 280 and 312: the checking for multiple definitions in a
2688 file is less likely to become confused by uses of inline
2690 * fixed bug: the #S reader macro performs the keyword coercion
2691 specified for slot names. (reported by Kalle Niemitalo)
2692 * fixed bug: lambda lists may contain symbols whose names start with
2693 & but are not lambda-list-keywords; their occurrence triggers a
2695 * fixed bug 321: define-method-combination argument lambda lists do
2696 not cause invalid code to be generated when &optional or &aux
2697 variables have default values. (reported by Bruno Haible)
2698 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2699 have CLOS classes; this appears also to have fixed reported
2700 instability in stack exhaustion detection.
2701 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2702 MUFFLE-WARNING when no associated restart is present is now
2704 * optimization: rearranged the expansion of various defining macros
2705 so that each expands into only one top-level form in a
2706 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2708 * optimization: used a previously little-used slot in symbols to
2709 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2711 * fixed some bugs revealed by Paul Dietz' test suite:
2712 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2714 ** Fixed an optimization bug related to inheritance of initforms
2715 from local to shared slots.
2716 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2717 ** CAST optimizer forgot to flush argument derived type.
2718 ** print/read consistency on floats is now orders of magnitude
2719 more likely. (thanks also to Bruno Haible for a similar report
2721 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2722 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2723 for objects of type REAL. Make it so.
2724 ** SXHASH is specified (infelicitously) to respect similarity,
2725 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2726 it so. (thanks to Markus Ziegler)
2727 ** on the Alpha, the compiler succeeds in compiling functions
2728 returning a known number of arguments greater than 63.
2729 ** fixed handling of invalid NIL arguments in keyword position
2731 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2733 ** provided a readably-printable representation for RANDOM-STATE
2735 ** ensured that pathnames, where they have a namestring, always
2736 print using #P"..." syntax.
2738 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2739 * Support for the forthcoming 2.0 version of the NetBSD kernel
2740 running on x86 hardware has been added. (thanks to Perry
2741 E. Metzger most immediately, and others for their past work)
2742 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2743 current release version; SBCL's OpenBSD support had been broken
2744 since about the time of OpenBSD's switch to ELF binary format.)
2745 * [placeholder for DX summary]
2746 ** user code with &REST lists declared dynamic-extent, under high
2747 speed or space and low safety and debug optimization policy.
2748 * The manual has been converted to Texinfo format and the debugger
2749 chapter from the cmucl manual has been added.
2750 * A facility has been added to extract documentation strings from
2751 sbcl and store them as Texinfo-formatted snippets for inclusion in
2752 the manual (via Texinfo's @include directive)
2753 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2754 the test case to Dave Roberts)
2755 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2756 pointers. (reported by Sean Ross)
2757 * bug fix: PROFILE output is printed nicely even for large numerical
2758 values. (thanks to Zach Beane)
2759 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2760 greater than 32 handle EOF correctly.
2761 * bug fix: on X86 an immediate argument of the IMUL instruction is
2762 correctly printed in disassembly. (thanks to Lutz Euler)
2763 * bug fix: class slots in redefined classes preserve their old
2764 values. (thanks to Bruno Haible and Nikodemus Siivola)
2765 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2766 succeeds. (reported by Marco Baringer)
2767 * fixed some bugs revealed by Paul Dietz' test suite:
2768 ** READ-SEQUENCE now works on ECHO-STREAMs.
2769 ** RATIONALIZE works more according to its specification. (thanks
2772 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2773 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2774 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2775 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2776 should matter to you only if you rebind the printer control
2777 variables and then find you want different bindings in the
2778 debugger than in the ordinary execution of your program.)
2779 * The runtime build system has been tweaked to support building
2780 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2781 assembler and linker. (thanks to Nikodemus Siivola)
2782 * Unbound, undefined, undeclared variables now trigger full
2783 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2784 more often programmer error than deliberate exploitation of undefined
2786 * optimization: the hash algorithm for strings has changed to one
2787 that is less vulnerable to spurious collisions. (thanks to Juho
2789 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2790 needless bounds checking. (thanks to Juho Snellman)
2791 * optimization: implemented multiplication as a modular
2792 (UNSIGNED-BYTE 32) operation on the PPC backend.
2793 * fixed some bugs revealed by Paul Dietz' test suite:
2794 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2796 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2797 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2798 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2800 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2802 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2804 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2805 their output stream on EOF from read.
2806 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2807 have been read to end-of-file.
2808 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2810 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2811 description of determination of which consecutive characters
2813 ** Printing the "Space" character with escaping on now yields "#\\ ",
2814 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2815 ** Reading floating-point numbers with *READ-BASE* set to a number
2816 less than 10 works correctly.
2817 ** Reading floating-point numbers with *READ-BASE* set to a number
2818 more than 10 works correctly.
2819 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2820 the readtable currently in effect.
2822 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2823 * minor incompatible change: parsing of namestrings on a physical
2824 (Unix) host has changed; numbers after the final #\. in a
2825 namestring are no longer interpreted as a version field. This is
2826 intented to be largely invisible to the user, except that the
2827 meaning of the namestring "*.*.*" has changed: it now refers to a
2828 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2829 should usually be replaced by
2830 (make-pathname :name :wild :type :wild :version :wild)
2831 with the added benefit that this is more likely to be portable.
2832 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2833 option to OPEN now signals an error if the file being opened
2834 exists; this may have an impact on existing code.
2835 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2836 signals received, on the PowerPC platforms (both Linux and
2837 Darwin). (thanks to Pierre Mai for pointing out the location of
2839 * several fixes on OS X: The system now builds and runs cleanly on
2840 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2841 (thanks to Brian Mastenbrook)
2842 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2843 second-resolution: integer multiples of 1/3600 between -24 and 24.
2844 (thanks to Vincent Arkesteijn)
2845 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2846 type when called with 1 argument; PEEK-CHAR checked type of
2847 PEEK-TYPE only after having read first character from a
2848 stream. (reported by Peter Graves)
2849 * bug fix: the garbage collector now has much better locality
2850 behaviour, and in particular no longer treats all memory as being
2851 exhausted when many small objects point to each other in a deeply
2853 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2854 recognized as being TYPEP their class.
2855 * bug fix: the PUSHNEW documentation string has been corrected.
2856 (thanks to Vincent Arkesteijn)
2857 * bug fix: defaulting of the value for the last of an atypically
2858 large number of multiple values being bound was not being
2859 performed correctly on the Alpha or PPC platforms
2860 * optimization: implemented multiplication as a modular
2861 (UNSIGNED-BYTE 32) operation on the x86 backend.
2862 * optimization: SEARCH on simple-base-strings can now be open-coded.
2863 (see also contrib/compiler-extras.lisp for inspiration for
2864 teaching the compiler about the Boyer-Moore algorithm).
2865 * value, returned by MAX (and MIN) called with several EQUALP, but
2866 not EQL, arguments now does not depend on compiler settings.
2867 * fixed some bugs revealed by Paul Dietz' test suite:
2868 ** in stack analysis liveness information is propagated from
2869 non-local entry points.
2870 ** pathwise CAST removing failed when the CAST node did not start
2872 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2873 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2875 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2876 host is already defined.
2877 ** RENAME-FILE works on streams instead of signalling an internal
2879 ** PEEK-CHAR uses the current readtable when determining whether
2880 or not a character is whitespace.
2881 ** MERGE-PATHNAMES handles the case when the pathname does not
2882 specify a name while the default-pathname specifies a version.
2883 ** Pathnames now stand a better chance of respecting print/read
2885 ** Attempting to use standardized file system operators with a
2886 pathname with invalid :DIRECTORY components signals a
2888 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2890 ** DIRECTORY on logical pathnames is more correct.
2891 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2892 signal a TYPE-ERROR if their argument is not a stream.
2893 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2894 designator argument does not designate a stream.
2895 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2896 examining the synonym.
2897 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2899 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2900 element-type larger than ([UN]SIGNED-BYTE 32).
2902 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2903 * When built with the :SB-FUTEX feature, threaded builds now take
2904 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2905 for faster/more reliable mutex and condition variable support.
2906 * Incompatible change (but one you probably shouldn't have been using
2907 anyway): the interface and code for arbitrating between multiple
2908 threads in the same user session has been redesigned.
2909 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2910 error when called without an explicit environment argument.
2911 (thanks to Nikodemus Siivola)
2912 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2913 confused about their position. (thanks to Adam Warner and Gerd
2915 * bug fix: Pretty printing backquoted forms with unquotations in the
2916 argument list position of various code constructs such as LAMBDA
2917 now works correctly. (reported by Paul Dietz)
2918 * bug fix: Pretty printing unquotations no longer loses all
2919 stream position information.
2920 * optimization: performance of string output streams is now less
2921 poor for multiple small sequence writes.
2922 * optimization: performance of CSUBTYPEP in the presence of complex
2923 expressions involving CONS and NOT many times has been improved.
2924 (reported by Paul Dietz)
2925 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2927 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2928 streams for the REPL, for the debugger, and so on. (thanks to
2930 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2931 * fixed some bugs revealed by Paul Dietz' test suite:
2932 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2933 is ignored if it is too small, rather than propagating through
2934 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2935 ** extremely complex negations of CONS types were not being
2936 sufficiently canonicalized, leading to inconsistencies in
2938 ** VALUES tranformer lost derived type.
2940 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2941 * fixed a bootstrapping bug: the build process no longer assumes
2942 that the various BOOLE-related constants have the same value in
2943 host and target lisps. (noted by Paul Dietz' test suite on an
2944 SBCL binary built from CLISP)
2945 * The system can now be dynamically linked on the MIPS platform,
2946 which enables dynamic loading of foreign code from Lisp. (thanks
2947 to Ralf Baechle for discussions on the MIPS ABI)
2948 * The system now records debugging information for its own source
2949 files in a filesystem-position-independent manner, relative to
2950 the "SYS" logical host.
2951 * fixed a compiler bug: MV-LET convertion did not check references
2952 to the "max args" entry point. (reported by Brian Downing)
2953 * tweaked disassembly notes to be less confident about proclaiming
2954 some instruction as an LRA. (thanks to Brian Downing)
2955 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2956 now each have their own history, command character, and other
2957 characteristics. (thanks to David Lichteblau)
2958 * fixed some bugs revealed by Paul Dietz' test suite:
2959 ** compiler failure in compiling LOGAND expressions including a
2961 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2962 shift greater than 32.
2963 ** FLUSH-DEST did not mark blocks for type check regeneration.
2964 ** HANDLER-CASE failed to accept declarations in handler clauses
2965 in some circumstances.
2967 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2968 * New code in contrib/sb-introspect (still probably not entirely
2969 stable yet) provides some support for smart Lisp development
2970 environments like SLIME.
2971 * The conditions signalled for errors occurring when loading .fasl
2972 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2973 in a way which should help ASDF recover gracefully.
2974 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2975 stuff has been cleaned up. If you code contrib/ stuff, this might
2976 affect you, and you can look at contrib/README, contrib/STANDARDS,
2977 and/or the 0.8.4.27 diff to check.
2978 * In full calls the compiler now does not generate checks for declared
2979 argument types for all arguments.
2980 * various threading fixes
2981 ** and some experimental patches which didn't make it into
2982 the main tree for this release, but which are shipped in
2983 contrib/experimental-thread.patch as a possible fix for some
2984 failures (deadlock, spinning...) in GC-intensive multithreaded
2986 * fixed PPC build problem (source code incompatibility of different
2987 library versions): added offsetof() hackery which attempts to divine
2988 where glibc maintainers put uc_mcontext today
2989 * fixed bug 282: compiler does not trust type assertions while passing
2990 arguments to a full call.
2991 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2992 &OPTIONAL VALUES type parameter.
2993 * fix bug 214: algorithm for noting rejected templates is now more
2994 similar to that of template seletion. (also reported by rydis on
2996 * fixed bug 141b: printing backquoted information readably and prettily
2997 inserts a space where necessary.
2998 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2999 (CHARACTER 10) are now reported as errors, rather than propagated
3000 as unknown types. (reported by piso on #lisp)
3001 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3002 with values NIL and :ERROR. (thanks to Milan Zamazal)
3003 * fixed bug 191c: CLOS now does proper keyword argument checking as
3004 described in CLHS 7.6.5 and 7.6.5.1.
3005 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3006 longer raise an error; further, using a list as a for-as-arithmetic
3007 counter now raises a meaningful error.
3008 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3009 now understood by sequence creation functions such as MAKE-SEQUENCE
3011 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3012 a STRING-INPUT-STREAM.
3013 * compiler enhancement: SIGNUM is now better able to derive the type
3015 * type declarations inside WITH-SLOTS are checked. (reported by
3017 * fixed some bugs revealed by Paul Dietz' test suite:
3018 ** incorrect optimization of TRUNCATE for a positive first
3019 argument and negative second.
3020 ** compiler failure in let-convertion during flushing dead code.
3021 ** compiler failure while deriving type of TRUNCATE on an
3022 interval, containing 0.
3023 ** ASH of a negative bignum by a negative bignum count now returns
3025 ** intersection of CONS types now canonicalizes properly, fixing
3026 inconsistencies in SUBTYPEP.
3028 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3029 * incompatible change: The --disable-debugger command line
3030 option now clobbers the debugger at a more fundamental
3031 level, by redefining #'INVOKE-DEBUGGER instead of by
3032 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3033 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3034 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3035 enter the debugger REPL and then suffer recursive errors
3036 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3037 while under the new system, BREAK in --disable-debugger mode
3038 terminates the system just as an unhandled error would.
3039 * fixed compiler performance when processing loops with a step >1;
3040 * bug fix: DOCUMENTATION now retrieves generic function
3041 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3042 support has been systematized, and now supports the methods
3043 specified by ANSI, along with a default method and a method for
3044 slot documentation. (reported by Nathan Froyd)
3045 * bug fix: effective methods associated with a generic function are
3046 no longer cached over a change of that generic function's method
3047 combination. (reported by Andreas Fuchs)
3048 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3049 names a non-existent directory. (thanks to Andreas Fuchs)
3050 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3051 the PPC platform, lead to stack corruption; this has been fixed.
3052 (reported by Rainer Joswig)
3053 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3054 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3055 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3056 arithmetic in general, is now much more reliable on the PPC
3058 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3059 platform now returns the right answer.
3060 * optimization: restored some effective method precomputation in
3061 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3062 precomputation is now tunable.
3063 * optimization: compiler-internal data structure use has been
3064 reviewed, and changes have been made that should improve the
3065 performance of the compiler by about 20%.
3066 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3067 simple-base-strings and simple-bit-vectors is improved.
3068 * optimization: the optimization of 32-bit logical and arithmetic
3069 functions introduced in version 0.8.3 on the x86 has been
3070 implemented on the mips, ppc and sparc platforms; an
3071 implementation of the same facility, but for 64-bit arithmetic,
3072 has been added for the alpha.
3073 * microoptimization: the compiler is better able to make use of the
3074 x86 LEA instruction for multiplication by constants.
3075 * bug fix: in some situations compiler did not report usage of
3076 generic arithmetic in (SPEED 3) policy.
3077 * bug 145b fix: compiler used wrong type specifier while converting
3078 MEMBER-types to numeric.
3079 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3081 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3082 index argument now works properly on the Alpha platform.
3083 * bug fix: floating point exception treatment on the Alpha platform
3085 * bug fix: FILE-POSITION works much better on string input and
3086 output streams. (thanks to Nikodemus Siivola)
3087 * bug fix: many threading/garbage collection symptoms sorted.
3088 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3089 might be pseudo-atomic.
3090 * internal change: Stopping for GC is now done with signals not
3091 ptrace. GC is now done in whichever thread wanted it, instead of
3093 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3095 * bug fix: result form in DO is not contained in the implicit
3097 * incompatible change: ICR structure is changed; the value part of
3098 CONTINUATION is now called LVAR; corresponding functions are
3099 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3100 * added type deriver for ISQRT (thanks to Robert E. Brown).
3101 * bug fix: better support for loading from the command line when an
3102 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3104 * fixed some bugs revealed by Paul Dietz' test suite:
3105 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3106 ** ROUND and FROUND now give the right answer when given very
3107 small float arguments.
3108 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3110 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3111 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3112 ** LCM should return a non-negative integer.
3113 ** PARSE-INTEGER returned the index of a terminator instead of the
3114 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3115 ** PARSE-INTEGER returned an incorrect index being applied to a
3117 ** LCM with two arguments of 0 returns 0 rather than signalling
3119 ** unsigned addition of a 32-bit constant with the high bit set no
3120 longer causes an internal compiler error on the x86.
3121 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3122 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3123 with negative last argument.
3124 ** byte specifiers with zero size and position no longer cause
3125 an error during type derivation.
3126 ** bignum multiplication on the Alpha platform now returns the
3128 * porting: The system now builds on SuSE AMD64, although it still
3129 generates a 32-bit binary.
3130 * .fasl file incompatibility: The fasl file version number has
3131 been incremented (because of the changes to internal compiler
3132 data structures referred to above).
3134 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3135 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3136 more accurately, on the Darwin kernel running on PowerPC hardware.
3137 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3138 * Compiler code deletion notes now signal a condition of type
3139 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3140 an associated MUFFLE-WARNING restart.
3141 * The compiler now performs limited argument count validation of
3142 constant format strings in FORMAT, and where appropriate in ERROR,
3143 CERROR and WARN. (thanks to Gerd Moellmann)
3144 * New ASDF-INSTALL contrib can be used for automatic download and
3145 installation of third-party Lisp code from CCLAN or other sites
3147 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3148 which forces another thread to execute a function supplied by the
3150 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3151 now accept and act upon their :ELEMENT-TYPE keyword argument.
3152 (reported by Martin Atzmueller, Edi Weitz)
3153 * bug fix: FILE-POSITION now accepts position designators up to
3154 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3155 is the greater. (thanks to Patrik Nordebo)
3156 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3157 Kalle Olavi Niemitalo)
3158 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3159 printable. (reported by Eric Marsden)
3160 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3161 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3162 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3163 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3164 (thanks to Stig Sandoe)
3165 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3166 UNREAD-CHAR now correctly includes the unread character in the
3167 target sequence. (thanks to Gerd Moellmann)
3168 * bug fix in threaded builds: the system can now be suspended and
3169 resumed by shell job control with minimal disruption.
3170 * bug fixes in times and timezones >2038AD
3171 * better handling of "where is GNU make?" problem in build scripts
3172 (thanks to Nikodemus Siivola)
3173 * new optimization: inside a named function any reference to a
3174 function with the same name is considered to be a self-reference;
3175 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3176 optimization quality.
3177 * new optimization on x86: logical functions and + now have
3178 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3179 used when the result is truncated to 32 bits.
3180 * VALUES declaration is partially enabled.
3181 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3182 * fixed some bugs revealed by Paul Dietz' test suite:
3183 ** The system now obeys the constraint imposed by
3184 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3185 types form a lattice under type intersection.
3186 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3187 ** ASSOC now ignores NIL elements in an alist.
3188 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3189 and (1+ MOST-POSITIVE-FIXNUM) answers.
3190 ** The addition of a method with invalid qualifiers to a generic
3191 function does not cause an error to be signalled immediately;
3192 a warning is signalled, and the error is generated only on
3193 calling the generic function.
3194 * changed .fasl file version number, in part to add type codes for
3195 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3196 obscure ANSI requirements
3198 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3199 * fixed bug 148: failure to inline-expand a local function left
3200 garbage, confusing the compiler.
3201 * fixed bugs 3cd: structure slot readers perform type check if the
3202 slot can have an invalid value (i.e. it is either not initialized
3203 or can be written with a less specific slot writer).
3204 * bug fix: the compiler now traps array references to elements off
3205 the end of an array; previously, the bounds checking in some
3206 circumstances could go off-by-one.
3207 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3209 * type declarations for array element types now obey the description
3210 on the CLHS page "Declaration TYPE", as per discussions on
3211 sbcl-help around 2003-05-08. This means that a declaration
3212 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3213 declaration, all references to BAR will be asserted or assumed
3214 (with THE, so dependent on compiler policy) to involve objects of
3215 type FOO. Note that no such declaration is implied in
3216 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3217 * declared types of functions from the "Conditions"
3218 chapter. (reported by Paul Dietz)
3219 * bug fix: CERROR accepts a function as its first argument.
3220 * bug fix: NTH an NTHCDR accept a bignum as index
3221 arguments. (reported by Adam Warner)
3222 * optimization: character compare routines now optimize comparing
3223 against a constant character. (reported by Gilbert Baumann)
3224 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3225 argument now works properly on the MIPS platform.
3226 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3227 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3228 declared to have a functional type, does not cause an error
3230 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3231 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3232 * fixed bug: if last continuation of a deleted block has a
3233 destination, this destination should be deleted too. (reported by
3235 * fixed a bug in the bootstrap process: the host compiler's values
3236 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3237 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3238 test case from Patrik Nordebo)
3239 * improved the ability of the disassembler on the PPC platform to
3240 provide helpful disassembly notes.
3241 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3242 the class in more cases than previously.
3243 * bug fix: FILE-POSITION now understands :START and :END for
3244 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3245 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3246 8) data. (thanks to David Lichteblau)
3247 * bug fix: it is possible to add a method to a generic function
3248 without lambda list.
3249 * bug fix: reader failed to signal END-OF-FILE inside an
3250 object representation. (reported by Nikodemus Siivola)
3251 * fixed some bugs revealed by Paul Dietz' test suite:
3252 ** LAST and [N]BUTLAST should accept a bignum.
3253 ** condition slot accessors are methods.
3254 ** (VECTOR NIL) is a subtype of STRING.
3256 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3257 * minor incompatible change: some nonsensical specialized lambda
3258 lists (used in DEFMETHOD) which were previously ignored now signal
3260 * minor incompatible change: the system is now aware of the types of
3261 variables in the COMMON-LISP package, and will signal errors for
3262 most violations of these type constraints (where previously they
3263 were silently accepted).
3264 * minor incompatible change: COMPILE-FILE now uses the freedom
3265 afforded (ANSI 3.2.2.3) to use derived function types for
3266 functions defined in the same file. This also permits the system
3267 to warn on static type mismatches and function
3268 redefinition. (Currently it does not work with high DEBUG level.)
3269 * minor incompatible change: VALUES declaration is disabled.
3270 * When issuing notes, the compiler now signals a condition of type
3271 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3272 restart for use in user handlers. It is expected that the
3273 COMPILER-NOTE condition will eventually become a condition
3274 supertype to a hierarchy of note types, which will then be
3275 handleable in a similar fashion. However, at the moment, no such
3276 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3277 but it's an implementation detail, not a classification for the
3279 * Changes in type checking closed the following bugs:
3280 ** type checking of unused values (192b, 194d, 203);
3281 ** template selection based on unsafe type assertions (192c, 236);
3282 ** type checking in branches (194bc).
3283 * A short form of VALUES type specifier has ANSI meaning (it has
3284 increased the number of situations when SBCL cannot perform type
3286 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3287 keywords or constants is permissible.
3288 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3289 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3290 classes. (thanks to Antonio Martinez)
3291 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3292 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3293 argument to a pretty-print stream. Instead, it leaves any such
3294 operations to DESCRIBE-OBJECT methods.
3295 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3297 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3298 lambda lists. (thanks to David Lichteblau)
3299 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3300 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3301 (thanks to Gerd Moellmann)
3302 * fixed bug 246: increased compilation speed of long
3303 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3305 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3307 * DEFCONSTANT now throws a condition of type
3308 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3309 constant to a non-EQL value; CONTINUE and ABORT restarts
3310 respectively change and preserve the value.
3311 * fixed bug 63: The code walker, part of the implementation of CLOS,
3312 is now better at handling symbol macros.
3313 * bug fix: There is no longer an internal implementation type named
3314 CL:LENGTH. (reported by Raymond Toy)
3315 * bug fix: In macro-like defining macros/special operators the
3316 implicit block does not enclose the lambda list.
3317 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3318 enclosed in parentheses) are not suitable as type specifiers, and
3319 their use properly signals an error now.
3320 * bug fix: An argument count mismatch for a type specifier in code
3321 being compiled no longer causes an unhandled error at compile
3322 time, but signals a compile-time warning.
3323 * fixed simple vector readable printing
3324 * bug fix: DESCRIBE takes more care over whether the class
3325 precedence list slot of a class is bound before accessing it.
3326 (reported by Markus Krummenacker)
3327 * bug fix: FORMATTER can successfully compile pretty-printer format
3328 strings which use variants of the ~* directive inside.
3329 * bug fix: SEARCH now applies its TEST predicate to the elements of
3330 the arguments in the correct order. (thanks to Wolfhard Buss)
3331 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3332 in type inference. (thanks to Robert E. Brown)
3333 * bug fix: Reading in symbols with an explicit package name of ""
3334 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3335 * fixed some bugs revealed by Paul Dietz' test suite:
3336 ** NIL is now allowed as a structure slot name.
3337 ** Arbitrary numbers, not just REALs, are allowed in certain
3338 circumstances in LOOP for-as-arithmetic clauses.
3339 ** Multiple class redefinitions before slot access no longer
3340 causes a type error.
3341 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3342 association between the name and a class.
3343 ** Generic functions with non-standard method-combination and over
3344 five methods all of which return constants no longer return NIL
3345 after the first few invocations. (thanks to Gerd Moellmann)
3346 ** CALL-NEXT-METHOD with no arguments now passes the original
3347 values of the arguments, even in the presence of assignment.
3348 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3350 ** DISASSEMBLE works with closures and funcallable instances.
3351 ** ADD-METHOD now returns the generic function, not the new method.
3352 ** FIND-METHOD signals an error if the lengths of the specializers
3353 is incompatible with the generic function, even if the ERRORP
3355 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3356 which its argument is a member.
3357 ** DEFCLASS only redefines the class named by its class-name
3358 argument if that name is the proper name of the class;
3359 otherwise, it creates a new class.
3360 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3361 of the UNBOUND-SLOT condition to the name of the slot.
3362 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3363 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3364 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3365 SLOT-MAKUNBOUND in the specified fashion.
3367 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3368 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3369 cross-compilation host. As a consequence, we can now bootstrap our
3370 way up to SBCL starting with a bare gcc toolchain and human-readable
3371 source code (first the source to CLISP, then the source to SBCL).
3372 * A contributed module containing a partial implementation of the
3373 simple-streams interface has been included. (thanks to Rudi
3375 * A contributed module implementing the RFC1321 Message Digest
3376 Algorithm, known as MD5, has been included.
3377 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3378 no longer has any effect, as the code controlled by this feature
3379 has been deleted. (As far as we know, no-one has ever built using
3380 this feature, and its semantics were confused in any case).
3381 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3382 work on conditions (as required by the ANSI specification),
3383 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3384 expected behaviour on conditions. Users should note, however,
3385 that such behaviour is not required by the ANSI specification,
3386 and so use of this behaviour may render their code unportable.
3387 * fixed some bugs revealed by Paul Dietz' test suite:
3388 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3390 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3391 ** SXHASH on condition objects no longer returns NIL.
3392 ** :ALLOCATION :CLASS slots are better treated; their values are
3393 updated on class redefinition, and initforms inherited from
3394 superclasses are applied.
3395 ** REMOVE-METHOD returns its generic function argument even when
3396 no method was removed.
3397 ** SHARED-INITIALIZE now initializes the values of the requested
3398 slots, including those with :ALLOCATION :CLASS.
3399 ** ALLOCATE-INSTANCE now works on structure classes defined via
3400 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3402 ** SLOT-EXISTS-P now works on conditions, as well as structures
3404 ** MAKE-LOAD-FORM now has the required methods on
3405 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3406 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3407 keyword, and now implements the SLOT-NAMES argument.
3408 ** methods with &OPTIONAL arguments no longer allow too many
3409 arguments to be passed in the call without error.
3410 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3411 option is consistent with the required arguments of the generic
3412 function lambda list.
3413 * bug fix: REQUIRE accepts a string designator. (Thanks to
3415 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3416 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3417 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3418 * bug fix: The compiler now checks for duplicated variables in macro
3420 * bug fix: SETQ on globals returns the correct value.
3421 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3422 not a COMPILER-ERROR (followed by some other strange error on
3423 choosing the CONTINUE restart).
3424 * bug fix: make.sh and friends are now more consistent in the way that
3425 they look for GNU "make".
3427 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3428 * experimental native threads support (on x86 Linux >=2.4 only).
3429 This is not compiled in by default: you need to add :SB-THREAD to
3430 the target features. See the "Beyond ANSI" chapter of the manual
3432 * fix for longstanding nonANSIism: The old distinction between
3433 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3434 The return value from CL:FIND-CLASS is now a CLOS class, and
3435 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3436 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3437 * An interface to the MetaObject Protocol, as described in Kiczales,
3438 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3439 MIT Press, 1991, is available from the SB-MOP package.
3440 * incompatible change: the SB-PCL package should now be considered
3441 a private implementation detail, and no longer a semi-private MOP
3443 * minor incompatible change: due to rearrangement for threads, the
3444 control stack and binding stack are now allocated at arbitrary
3445 addresses instead of being hardcoded per-port. Users affected by
3446 this probably have to be doing advanced things with shared
3447 libraries, and will know who they are.
3448 * minor incompatible change: Previously, all --eval forms used were
3449 processed with READ before any of them were processed with EVAL.
3450 Now each --eval form is processed with both READ and EVAL before
3451 the next --eval form is processed. (Thus package operations like
3452 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3453 work as the user might reasonably expect.)
3454 * minor incompatible change: *STANDARD-INPUT* is now only an
3455 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3457 * minor incompatible change: Y-OR-N-P is now character-oriented,
3458 not line oriented. Also, YES-OR-NO-P now works without errors.
3459 (thanks to Antonio Martinez)
3460 * sb-aclrepl module improvements: an integrated inspector, added
3461 repl features, and a bug fix to :trace command.
3462 * Known functions, which cannot be open coded by the backend, are
3463 considered to be able to check types of their arguments. (fixing
3464 a bug report by Nathan J. Froyd)
3465 * fixed a bug in computing method discriminating functions: It is
3466 now possible to define methods specialized on classes which have
3467 forward-referenced superclasses. (thanks to Gerd Moellmann)
3468 * fixed evaluation order in optional entries (reported by Gilbert
3470 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3471 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3472 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3473 required-by-AMOP NAME argument, as well as CLASS and
3474 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3475 * fixed bug 20: DEFMETHOD can define methods using names that are
3476 not the proper names of classes to designate class specializers.
3477 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3478 Unix file descriptors, instead of blowing up. (thanks to Antonio
3480 * Garbage collection refactoring: user-visible change is that a
3481 call to the GC function during WITHOUT-GCING will not do garbage
3482 collection until the end of the WITHOUT-GCING. If you were doing
3483 this you were probably losing anyway.
3484 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3485 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3486 TYPEP the latter but not the former.
3487 * The compiler issues a full WARNING for calls to undefined functions
3488 with names from the CL package.
3489 * MAP-INTO for a vector destination is open coded. (reported by
3490 Brian Downing on c.l.l)
3491 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3492 documentation string.
3493 * fixed some bugs revealed by Paul Dietz' test suite:
3494 ** COPY-ALIST now signals an error if its argument is a dotted
3496 ** Condition slots are now accessed more correctly in the presence
3497 of multiple initargs for a given slot.
3498 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3499 exclude restarts of the same name associated with a different
3501 ** DEFCLASS of forward-referenced classes with another
3502 forward-referenced class in the superclasses list no longer
3504 ** Condition slots are now initialized once each, not multiple
3505 times. (thanks to Gerd Moellmann)
3506 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3508 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3509 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3510 arguments contain duplicated elements.
3511 ** RESTART-CASE understands local macros.
3512 ** RESTART-CASE associates exactly its own restarts with a condition.
3513 ** ENDP in safe mode checks its argument to be of type LIST.
3514 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3515 in question is unbound.
3516 ** Optimized MAKE-INSTANCE functions no longer cause internal
3517 assertion failures in the presence of duplicate initargs.
3518 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3519 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3520 propagated even in the ctor optimized implementation of
3522 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3524 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3525 * a better implementation of SXHASH on (simple) bit vectors,
3526 measured both in execution speed and in distribution of results
3527 over the positive fixnums, has been installed. Likewise, a better
3528 implementation of EQUAL for simple bit vectors is now available.
3529 * fixed CEILING optimization for a divisor of form 2^k.
3530 * fixed bug 240 (emitting extra style warnings "using the lexical
3531 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3532 by Antonio Martinez)
3533 * fixed SXHASH, giving different results for NIL depending on type
3534 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3535 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3536 variable. (found by Rolf Wester)
3537 * SBCL does not ignore type declarations for special
3538 variables. (reported by rif on c.l.l 2003-03-05)
3539 * some bug fixes in contrib/sb-aclrepl/
3540 * fixed some bugs revealed by Paul Dietz' test suite:
3541 ** a bug in the CONS type specifier, whereby the CAR and CDR
3542 types got intertwined, has been fixed;
3543 ** the type system is now able to reason about the interaction
3544 between INTEGER and RATIO types more completely;
3545 ** APPEND, [N]REVERSE and NRECONC check that those their
3546 arguments, which must be proper lists, are really so;
3547 ** An array specialized to be unable to hold elements has been
3548 implemented, as required -- yes, really -- by ANSI;
3549 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3550 on malformed property lists;
3552 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3553 * incompatible packaging change: in line with Unix convention,
3554 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3555 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3556 * REQUIRE and PROVIDE are now optionally capable of doing something
3557 useful. See the documentation string for REQUIRE.
3558 * infrastructure for a managed SBCL contrib system: contributed
3559 modules in this release include:
3560 ** the ASDF system definition facility;
3561 ** an interface to the BSD Sockets API;
3562 ** an ACL-like convenience interface to the repl;
3563 (thanks to Kevin Rosenberg)
3564 ** an implementation of ROTATE-BYTE, with efficient implementation
3566 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3567 gives a closer approximation to the right answer than previously.
3568 (thanks to Raymond Toy)
3569 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3570 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3571 optional environment arguments, as required by ANSI.
3572 * fixed bugs in other functions taking environment objects, allowing
3573 calls with an explicit NIL environment argument to be compiled
3575 * fixed bug 228: primary return values from
3576 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3577 COMPILE or FUNCTION.
3578 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3579 :TYPE will no longer signal errors on innocuous objects.
3580 * fixed bug 231b: SETQ is better at respecting type declarations in
3581 the lexical environment.
3582 * fixed a bug in DEFCLASS: classes named by symbols with no or
3583 unprintable packages can now be defined.
3584 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3585 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3586 * fixed compiler failure related to checking types of functions.
3587 (reported by Robert E. Brown)
3588 * the compiler is now much more consistent in its error-checking
3589 treatment of bounding index arguments to sequence functions: in
3590 (SAFETY 3) code, errors will be signalled in almost all cases if
3591 invalid sequence bounding indices are passed to functions defined
3592 by ANSI to operate on sequences.
3593 * fixed a bug in the build procedure: documentation of SBCL-specific
3594 packages is now preserved and available in the final Lisp image.
3595 * lifted FDEFINITION lookup out of loops in the implementation of
3596 many list operations. (thanks to Robert E. Brown)
3597 * fixed a bug in the reader: the #n# reader macro now works for
3598 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3599 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3600 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3601 E. Brown and Tony Martinez respectively)
3602 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3603 SLOT-BOUNDP: in particular, it is now able to optimize them much
3604 better, and is now not vulnerable to having packages renamed.
3605 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3606 always called when a slot is not present in an instance. (thanks
3608 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3609 longer signal an error on unreading a character following EOF on
3610 the previous constituent stream. (thanks to Tony Martinez)
3611 * fixed some bugs revealed by Paul Dietz' test suite:
3612 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3613 not just nonnegative fixnums;
3614 ** the logical bit-array operators such as BIT-AND now accept an
3615 explicit NIL for their "opt-arg" argument (to indicate a
3616 freshly-consed result bit-array);
3617 ** ELT now signals an error on an invalid sequence index in safe
3619 ** the type system is now cleverer about negations of numeric
3620 types, and consequently understands the BIGNUM and RATIO types
3622 ** the type system is now cleverer about the interaction between
3623 INTEGER and RATIO types: while bugs still remain, many more
3624 cases are accurately computed;
3625 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3626 if it is in the last clause;
3627 ** CONSTANTLY now correctly returns a side-effect-free function in
3629 ** DECLARE is no longer treated as a special-operator; in
3630 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3631 * incremented fasl file version number due to the change in the
3632 DEFSTRUCT-SLOT-DESCRIPTION structure.
3634 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3635 * minor incompatible change: code processed by the "interpreter" or
3636 EVAL now has a compilation optimization policy of (DEBUG 2)
3637 (changed from (DEBUG 1)) to improve debuggability of interactive
3638 development, and to allow the use of the debug RETURN command in
3640 * an experimental implementation of the RETURN command for the
3641 debugger has been included. (thanks to Frederik Kuivinen)
3642 * fixed bug 62: constraints were not propagated into a loop.
3643 * fixed bug in embedded calls of SORT (reported and investigated by
3645 * fixed some bugs revealed by Paul Dietz' test suite:
3646 ** printing and reading of arrays with some dimensions having
3647 length 0 (thanks to Gerd Moellmann);
3648 ** BOA constructor with &AUX argument without a default value does
3649 not cause a type error;
3650 ** CONSTANTP now returns true for all self-evaluating objects.
3652 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3653 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3654 accessors that are related by inheritance, as specified in the
3655 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3656 Valtteri Vuorikoski)
3657 * The compiler is now able to inline functions that were defined in
3658 a complex lexical environment (e.g. inside a MACROLET).
3659 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3661 * Support for the upcoming FreeBSD-5.0 release has been included.
3662 (thanks to Dag-Erling Smorgrav)
3663 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3664 effect when it is not in a toplevel context.
3665 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3666 stand a better chance of being correct. (thanks to Gerd
3668 * fixed bug in COERCE, which now signals an error on coercing a
3669 rational to a bounded real type which excludes the expected
3671 * The compiler is now able to derive types more accurately from the
3672 COERCE and COMPILE functions.
3673 * fixed bug 223: functional binding is considered to be constant
3674 only for symbols in the CL package.
3675 * fixed bug 231: SETQ did not check the type of a variable being set
3676 (reported by Robert E. Brown)
3677 * A new optimization for MAKE-INSTANCE has been included, fixing
3678 various bugs (including relating to :ALLOCATION :CLASS slots and
3679 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3681 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3682 ** As required by ANSI, LOOP now disallows anonymous collection
3683 clauses such as COLLECT I in conjunction with aggregate boolean
3684 clauses such as THEREIS (= I 1);
3685 ** LOOP now signals an error when any variable is reused in the
3686 same loop (including the potentially useful construct analogous
3687 to WITH A = 1 WITH A = (1+ A);
3688 ** IT is only a special loop symbol within the first clause of a
3689 conditional loop clause;
3690 ** LOOP with a typed iteration variable over a hashtable now
3691 signals a type error iff it should.
3692 * fixed some other bugs revealed by Paul Dietz' test suite:
3693 ** FILE-STREAM now names the class previously known as FD-STREAM;
3694 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3695 argument) no longer signals an error;
3696 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3697 of the slot symbol, rather than using the current package
3698 ((:CONC-NAME "") continues to intern the slot's name in the
3700 * incremented fasl file version number, because of the incompatible
3701 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3702 of the new implementation of DEFINE-COMPILER-MACRO.
3704 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3705 * Support for building SBCL for MIPS platforms running in
3706 little-endian mode has now been checked in, and basic
3707 functionality on said platforms verified.
3708 * minor incompatible change: PCL now records the pathname of a file
3709 in which methods and the like are defined, rather than its
3711 * minor incompatible change: TRUENAME now considers the truename of
3712 a file naming a directory to be the pathname with :DIRECTORY
3713 component indicating that directory.
3714 * minor incompatible change: a NAMED clause in the extended form of
3715 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3716 reason for the previous behaviour is unclear.
3717 * more systematization and improvement of CLOS and MOP conformance
3718 in PCL (thanks to Gerd Moellman and Pierre Mai):
3719 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3721 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3722 primary methods with no specializers;
3723 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3725 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3726 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3727 FORWARD-REFERENCED-CLASSes; error reporting on
3728 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3730 ** SXHASH on CLOS instances now uses a slot internal to the
3731 instance to return different numbers on distinct instances,
3732 while preserving the same return value through invocations of
3734 ** DEFMETHOD signals errors when methods with longer incongruent
3735 lambda lists are added to generic functions;
3736 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3737 CLASS, as specified in AMOP;
3738 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3739 on the order returned by the primary method for classes of
3740 class STANDARD-CLASS;
3741 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3742 * fixed some bugs shown by Paul Dietz' test suite:
3743 ** DOLIST puts its body in TAGBODY;
3744 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3746 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3747 value producing form;
3748 ** if more variables are given to PROGV than values, extra
3749 variables are bound and made to have no value;
3750 ** NSUBSTITUTE on list arguments gets the right answer with
3752 ** ELT signals an error of type TYPE-ERROR when the index argument
3753 is not a valid sequence index;
3754 ** LOOP signals (at macroexpansion time) an error of type
3755 PROGRAM-ERROR when duplicate variable names are found;
3756 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3757 ** FUNCALL of special-operators now cause an error of type
3759 ** PSETQ now works as required in the presence of side-effecting
3760 symbol-macro places;
3761 ** NCONC accepts any object as its last argument;
3762 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3764 ** loop-for-as-package does not require a package to be explicitely
3766 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3768 * fixed bug 166: compiler preserves "there is a way to go"
3769 invariant when deleting code.
3770 * fixed bug 172: macro lambda lists with required arguments after
3771 &REST arguments now cause an error to be signalled. (thanks to
3773 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3774 bug. (thanks to Matthew Danish)
3775 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3777 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3778 arguments are lexically rebound. (thanks to Gerd Moellmann and
3780 * fixed bug 194: error messages are now more informative when there
3781 is no primary method applicable in a call to a generic
3782 function. (thanks to Gerd Moellmann)
3783 * fixed bug in command line argument checking (thanks to Julian
3785 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3786 (thanks to Matthew Danish)
3787 * incremented fasl file version number, because of the
3788 SXHASH-related changes in the layout of CLOS data structures
3790 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3791 * minor incompatible change: The runtime (the Unix executable named
3792 "sbcl") is now much pickier about the .core files it will load.
3793 Essentially it now requires .core files to descend from the same
3794 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3795 as the runtime does. (The intent is to prevent the crashes which
3796 can occur, and which can even be reported as mysterious failures,
3797 when people patch the sources or change the build parameters
3798 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3799 sbcl and .core files.)
3800 * fixed bug: VALUES-LIST is no longer optimized away.
3801 * fixed bug 142: The FFI conversion of C string values to Lisp
3802 string values no longer conses excessively. (thanks to Nathan
3803 Froyd porting Raymond Toy's fix to CMU CL)
3804 * began to systematize and improve MOP conformance in PCL (thanks to
3805 Nathan Froyd, Gerd Moellman and Pierre Mai):
3806 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3808 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3809 ** FINALIZE-INHERITANCE is now called on class finalization;
3810 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3811 argument precedence order.
3812 * fixed bug 202: The compiler no longer fails on functions whose
3813 derived types contradict their declared type.
3814 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3815 so it can be non-toplevel.
3816 * The fasl file version number has changed (because of the new
3817 implementation of DEFMACRO).
3818 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3819 safe code, that any length requirement by their type-specifier
3820 argument is valid. The exceptions to this are described in bug
3822 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3823 functions now check the types of their inputs as required by ANSI.
3824 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3825 introduced symbol is DECLAREd to be SPECIAL.
3826 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3827 * fixed bug: :COUNT argument to sequence functions may be negative
3828 * fixed bug: body of DO-SYMBOLS may contain declarations
3829 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3830 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3831 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3832 symbol macro only once
3833 * fixed printing of call frame when argument list is unavailable
3834 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3835 * compiler no longer signals WARNING on unknown keyword
3838 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3839 * A beta-quality port to the mips architecture running Linux,
3840 based on the old CMUCL backend, has been made. It has been tested
3841 on a big-endian kernel, and works sufficiently well to be able to
3842 rebuild itself; it has not been tested in little-endian mode.
3843 * fixed an inconsistency between gencgc.c and purify.c which made
3844 dumping/loading .core files unreliable
3845 * fixed bug 120a: The compiler now deals correctly with IFs where
3846 the consequent is the same as the alternative, instead of
3847 misderiving the return type. (thanks to Alexey Dejneka)
3848 * fixed bug 113: Logical pathnames are now dumpable (the logical
3849 host is resolved at load-time, throwing an error if it is not
3851 * fixed bug 174: FORMAT's error message is slightly clearer when a
3852 non-printing character is used in a format directive.
3853 * fixed several bugs in compiler checking of type declarations, i.e.
3854 violations of the Python "declarations are assertions" principle
3855 (thanks to Alexey Dejneka)
3856 * fixed several bugs in PCL's error checking (thanks to Gerd
3858 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3860 * fixed bug in compilation of functions as first class values
3861 (thanks to Antonio Martinez-Shotton)
3862 * The compiler's handling TYPE-ERRORs which it can prove will
3863 inevitably happen at runtime has been cleaned up and corrected
3864 in several ways. (thanks to Alexey Dejneka)
3865 * improved argument type checking for various basic arithmetic
3866 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3867 lavished on them in the past that they can be compiled in many
3868 ways in different special cases
3869 * fixed bug 181: compiler checks validity of user supplied type
3871 * cleaned up code flushing in optimization: Function calls which
3872 should signal errors for safety purposes (e.g. which ANSI says
3873 should signal errors when their arguments are of incorrect type)
3874 are no longer optimized away.
3875 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3876 * incremented fasl file version number, because changes in the
3877 implementation of sequence functions like COERCE caused
3878 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3879 in old inline expansions) to become undefined. (Actually these
3880 changes were later undone, so we might very well be binary
3881 compatible with 0.7.7 after all, but leaving the version number
3882 incremented seemed like the simplest and most conservative
3885 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3886 * An alpha-quality port to the parisc architecture running Linux,
3887 based on the old CMUCL backend, has been made. This, even more so
3888 than the other backends, should be considered still a work in
3889 progress; known problems include that the Linux kernel in 64-bit
3890 mode does not propagate the correct sigcontext structure to
3891 userspace, and consequently SBCL on a parisc64 kernel will not
3893 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3894 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3895 suppressed.) Also now that the compiler is looking at declarations
3896 in the environment, it checks optimization declarations as well,
3897 and suppresses inlining when (> DEBUG SPEED).
3898 * More fixes have been made to treatment of floating point exception
3899 treatment and other Unix signals. In particular, floating point
3900 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3901 * The detection and handling of control stack exhaustion (infinite
3902 or very deeply nested recursion) has changed. Stack exhaustion
3903 detection is now done by write-protecting pages at the OS level
3904 and applies at all optimization settings; when found, a
3905 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3906 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3907 be caught using IGNORE-ERRORS.
3908 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3909 correctly when transforming recursive LABELS functions to LETs.
3910 (thanks to Alexey Dejneka)
3911 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3912 that are names of constants or global variables.
3913 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3914 alien routines with docstrings.
3915 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3916 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3918 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3919 (thanks to Eric Marsden)
3920 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3921 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3922 * Bug fix: The SPARC backend can now compile functions involving
3923 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3924 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3925 to a functions expecting a PATHNAME-DESIGNATOR.
3926 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3927 lambda lists. (thanks to Alexey Dejneka)
3928 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3929 functions so that EXPORTing the name of the class doesn't cause
3930 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3931 to Antonio Martinez for reporting this)
3932 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3933 constant-folded without failing an assertion. (thanks to Einar
3934 Floystad Dorum for reporting this)
3935 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3936 types are now dealt with more correctly by the compiler.
3937 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3938 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3939 behaviour analogous to RENAME-FILE. This puts its behaviour more
3940 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3942 * The fasl file version number has changed again. (because of the
3943 bug fix involving the names of PCL MAKE-INSTANCE functions)
3945 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3946 * bug fix: Floating point exceptions are treated much more
3947 consistently on the x86/Linux and PPC/Linux platforms.
3948 * Array initialization with :INITIAL-ELEMENT is now much faster for
3949 cases when the compiler cannot open code the array creation, but
3950 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3951 array accesses have also seen a speed increase.
3952 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3953 specified. (This was at the root of some bad interactions between
3954 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3956 * bug fix: Internal error arguments for undefined functions are now
3957 computed correctly on the PPC/Linux platform.
3958 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3959 Raymond Toy's patch for CMU CL)
3960 * Support for the Solaris 9 operating environment has been included
3961 (thanks to Daniel Merritt)
3962 * A very ugly but hopefully complete draft of the missing FFI chapter
3963 of the manual has been created by reformatting the corresponding
3964 CMU CL manual chapter into (currently very ugly and incoherent)
3965 DocBook and bringing it up to date for SBCL behavior. Thus, the
3966 manual is now essentially complete, at least by my extreme
3967 once-and-only-once standards, whereby it's acceptable to refer to
3968 the doc strings of SB-EXT functions as the primary documentation.
3969 * The fasl file version number has changed again, due to cleanup of
3970 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3971 is no longer a static symbol.)
3973 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3974 * SBCL now builds with OpenMCL (version 0.12) as the
3975 cross-compilation host; also, more progress has been made toward
3976 bootstrapping under CLISP.
3977 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3979 * bug 158 fixed: The compiler can now deal with integer loop
3980 increments different from 1; fixing this turned out also to fix
3982 * bug 169 fixed: no more bogus warnings about using lexical bindings
3983 despite the presence of perfectly good SPECIAL declarations (thanks
3984 to David Lichteblau)
3985 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3986 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3987 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3989 * bug 180 fixed: Method combination specifications no longer ignore
3990 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3991 * bug fix: Structure type predicate functions now check their argument
3992 count as they should.
3993 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3994 correctly. (thanks to Pierre Mai)
3995 * minor incompatible change: The --noprogrammer option is deprecated
3996 in favor of the new --disable-debugger option, which is very similar.
3997 (The major difference is that it takes effect at a slightly different
3998 time at startup, causing handling of errors in --sysinit and
3999 --userinit files will be affected differently.) The
4000 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4001 been added to allow this functionality to be controlled from ordinary
4002 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4003 maintainers, who might want to run non-interactive scripts to
4004 build SBCL cores which will later be used interactively.)
4005 * minor incompatible change: The LOAD function no longer, when given
4006 a wild pathname to load, loads all files matching that pathname.
4007 Instead, an error of type FILE-ERROR is signalled.
4009 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4010 * bug 147 fixed: The compiler preserves its block link/count
4011 invariants more correctly now so that it doesn't crash. (thanks
4013 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4015 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4016 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4017 no-value-for-this-character as NIL (as specified by ANSI).
4018 * HOST-NAMESTRING on physical pathnames now returns a string that is
4019 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4020 (thanks to Christophe Rhodes)
4021 * The Alpha port handles icache flushing more correctly. (thanks to
4023 * More progress has been made toward bootstrapping under CLISP. (thanks
4024 to Christophe Rhodes)
4025 * The fasl file format has changed again, because dynamic loading
4026 on OpenBSD (which has non-ELF object files) motivated some cleanups
4027 in the way that foreign symbols are transformed and passed around.
4028 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4029 is no longer treated as whitespace by the reader, but instead as
4030 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4031 |A
\7fB|, instead of A as it used to.
4033 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4034 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4035 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4036 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4037 this as of 0.7.1.45, but was left out of the previous news section
4038 (thanks to Dan Barlow)
4039 * SBCL now runs on the Solaris operating system on SPARC architectures
4040 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4041 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4042 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4043 * SPARC backend cleanups, allowing builds of cores optimized for V8
4044 and V9 SPARCS, and also emission of code targeted to a particular
4045 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4047 * SBCL is closer to bootstrapping under CLISP, thanks to various
4048 fixes by Christophe Rhodes.
4049 * The fasl file format has changed again, to allow the compiler's
4050 INFO database to support symbol macros.
4051 * The user manual (in doc/) is formatted into HTML more nicely.
4052 (thanks to coreythomas)
4053 * The system is smarter about SUBTYPEP relationships, especially
4054 those involving NOT types (including types such as ATOM which are
4055 represented internally using NOT types). Thus SUBTYPEP is less
4056 likely to return (VALUES NIL NIL) in general, and in particular
4057 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4059 * The fasl file format has changed again, because the internal
4060 representation of types now includes a new slot to support the new
4061 SUBTYPEP-of-NOT-types logic.
4062 * (not a change in the main branch of SBCL, but a related prototype
4063 which can hopefully be merged into the main branch of SBCL in the
4064 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4065 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4066 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4067 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4068 dispatch macro character combinations. (thanks to Alexey Dejneka)
4069 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4070 that superficially look like logical namestrings correctly.
4071 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4073 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4074 default supplied-p) for &optional and &key arguments. (thanks to
4077 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4078 * incompatible change: The compiler is now less aggressive about
4079 tail call optimization, doing it only when (> SPACE DEBUG) or
4080 (> SPEED DEBUG). (This is an incompatible change because there are
4081 programs which relied on the old CMU-CL-style behavior to optimize
4082 away their unbounded recursion which will now die of stack overflow.)
4083 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4084 for non-GENCGC systems has been increased to 20M (since that
4085 seems much closer to the likely performance optimum for modern
4086 systems than the old 4M value was)
4087 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4088 and *DEBUG-PRINT-LEVEL*
4089 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4090 of CMU CL's support for SPARC, and various endianness and other
4091 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4092 * new syntactic sugar for the Unix command line: --load foo.bar is now
4093 an alternate notation for --eval '(load "foo.bar")'.
4095 ** The system now detects stack overflow and handles it gracefully,
4096 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4097 optimization settings. (This is a good thing in general, and
4098 its introduction in this version should be particularly timely
4099 for anyone whose code fails because of suppression of tail
4101 ** The system now hunts for the C variable "environ" in a more
4102 devious way, to avoid segfaults when the C library version differs
4103 between compile time and run time. (thanks to Christophe Rhodes)
4104 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4105 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4106 ** The compiler no longer issues bogus style warnings for undefined
4107 classes in the same source file as the DEFCLASSes which defined
4108 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4110 ** fixes in CONDITION class precedence list for undefined function
4111 errors (thanks to Alexei Dejneka)
4112 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4113 correctly. (thanks to Dan Barlow)
4114 ** portability fixes aiming at bootstrapping under CLISP (thanks
4115 to Dave McDonald and Christophe Rhodes)
4116 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4117 ** fixes in type translation and and type inference (thanks to
4119 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4120 ** various fixes in the new ports (thanks to Dan Barlow)
4121 * several changes related to debugging:
4122 ** suppression of tail recursion, as noted above
4123 ** stack overflow detection, as noted above
4124 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4125 is now the default. (For some time encapsulation has been more
4126 reliable than the breakpoint-based :ENCAPSULATE NIL
4127 implementation, at least on X86 systems; and I just noticed that
4128 encapsulation also seems closer to the spirit of the ANSI
4131 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4133 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4134 up properly again. (There was a packaging bug in 0.7.0 which
4135 left their definitions in SB-SYS::LOAD-FOREIGN and
4136 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4137 vital for most things which interface to C-level interfaces,
4138 like extensions working with sockets or databases or
4139 Perl-compatible regexes or whatever, and the need to fix
4140 this bug is the main reason that 0.7.1 was released so
4142 ** DEFGENERIC is now choosier about the methods it redefines, so that
4143 reLOADing a previously-LOADed file containing DEFGENERICs does
4144 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4145 cycle now works as it should. (thanks to Alexey Dejneka)
4146 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4147 fixed by Christophe Rhodes. (He actually submitted this patch
4148 months ago, and I delayed until after 0.7.0.)
4149 ** Bug 111 (internal compiler confusion about runtime checks on
4150 FUNCTION types) was fixed by Alexey Dejneka.
4151 * Some internal cleanups (getting rid of variables which aren't
4152 needed now that the byte interpreter is gone) caused the fasl
4153 file format number to change again.
4155 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4156 * major incompatible change: The default fasl file extension, i.e. the
4157 default extension for files produced by COMPILE-FILE, has changed
4158 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4160 ** There are many changes in the implementation of the compiler.
4161 SBCL is now essentially a compiler-only implementation of ANSI
4162 Common Lisp. EVAL still "interprets" a few special cases, but
4163 almost all the interesting cases are handled by creating
4164 a LAMBDA expression, calling COMPILE on it, then calling
4165 FUNCALL on the result.
4166 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4167 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4168 Since the code is newer, there might still be some new bugs
4169 (though not as many as before Martin Atzmueller's fixes:-). But
4170 the new code is substantially simpler and clearer, and hopefully
4171 any remaining bugs will be simpler, less fundamental, and more
4172 fixable then the bugs in the old code.
4173 ** The revised compiler is still a little unsteady on its feet.
4175 *** The debugging information it produces (particularly the names
4176 of FUNCTION objects) is sometimes much less useful than what
4177 the old compiler produced.
4178 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4179 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4180 within a MACROLET) has been temporarily weakened.
4181 ** There are new compiler optimizations for various functions:
4182 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4183 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4184 *** the math functions TRUNCATE, FLOOR, and CEILING
4185 *** the function-of-all-trades COERCE
4186 Mostly these should be transparent, but there's one
4187 potentially-annoying problem (bug 117): when the compiler
4188 inline-expands a function and does type analysis on the result,
4189 it can create control paths which have type mismatches, and
4190 when it can't prove that those control paths aren't taken,
4191 it will issue WARNINGs about the type mismatches. This is
4192 a particular problem in practice for the new sequence functions.
4193 It's not clear how this should be fixed, and for now, a
4194 workaround is given in the entry for 117 in the BUGS file.
4195 ** (Because of the interaction between the two previous items --
4196 occasional inlining problems and new inline expansions -- some
4197 of the new sequence function optimizations won't really kick in
4198 completely until debugging information, and then inlining, are
4199 straightened out in some future version.)
4200 * minor incompatible changes:
4201 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4202 DIRECTORY no longer implicitly promotes NIL slots of its
4203 pathname argument to :WILD. In particular, when you ask for the
4204 contents of a directory (which you used to be able to do without
4205 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4206 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4207 ** changes in behavior that ANSI explicitly defines to be
4208 implementation dependent:
4209 *** The new compiler-only implementation still conforms with ANSI,
4210 but acts a little different than before. Besides the obvious
4211 changes in performance tradeoffs (that the cost per form passed
4212 to EVAL has gone up, and the cost per form executed by EVAL
4213 has gone down), the behavior of the system changes a little
4214 because there are no longer any interpreted function objects.
4215 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4216 e.g. doing COMPILE on the output of interactive DEFUN is
4218 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4220 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4221 has changed from (:CL) to NIL.
4222 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4223 e.g. #\Newline and #\Tab, don't have names specified in the
4224 ANSI Common Lisp standard, is now based on their ASCII symbolic
4225 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4226 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4227 are no longer used for output.
4228 ** changes in internal implementation constants:
4229 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4230 4 million. (If your application spends a lot of time GCing and
4231 you have a lot of RAM, you might want to experiment with
4232 increasing it even more.)
4233 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4234 However, almost all old code should still continue to work without
4235 immediate update, as SB-C-CALL is now a (deprecated) nickname
4237 ** Old operator names in the style DEF-FOO are now deprecated in
4238 favor of new corresponding names DEFINE-FOO, for consistency with
4239 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4240 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4241 internal symbols, but a few supported extensions like
4242 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4243 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4244 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4245 etc. as you get deeper into recursive calls to the debugger
4246 command loop, instead of the old "5]", "5]]", "5]]]"
4247 sequence. (I was motivated to do this when squabbles between
4248 ILISP and SBCL left me very deeply nested in the debugger. In the
4249 short term, this change will probably provoke more ILISP/SBCL
4250 squabbles, but hopefully it will be an improvement in the long run.)
4251 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4252 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4253 function, since declaiming return types involving aliens is
4254 (1) annoyingly messy to do by hand and (2) vital to efficient
4255 compilation of code which calls such functions.
4256 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4257 longer reexported by the SB-EXT package. They're solely useful
4258 for alien code, so it seems more logical that you should get
4259 them from the SB-ALIEN package, not in SB-EXT.
4260 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4261 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4262 features. Instead, the code that they used to control is always
4263 built into the system.
4264 * many other bug fixes
4265 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4266 advantage of the new EVAL-WHEN stuff and to clean them up in
4267 general, and they are now more ANSI-compliant in a number of
4268 ways. Martin Atzmueller is responsible for a lot of this.
4269 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4271 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4272 *** correct ERROR type for various file operations
4273 *** some fixes for Lisp streams
4274 *** DEFMETHOD syntax checking
4275 *** changing old weird representation of debug information as
4276 strings (which, among their other deficiencies, don't transform
4277 correctly when you rename packages, and don't change their
4278 print representation when you change things like *PACKAGE*
4279 and *PRINT-LENGTH*) to symbols and lists of symbols
4280 He also made several improvements and fixed several bugs in DESCRIBE.
4281 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4283 *** misbehavior of WRITE-STRING/WRITE-LINE
4284 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4285 and several other LOOP problems as well
4286 *** DIRECTORY when similar filenames are present
4287 *** DEFGENERIC with :METHOD options
4288 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4289 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4290 *** argument ordering in FIND with :TEST option
4291 *** mishandled package designator argument in APROPOS-LIST
4292 *** various problems in the backquote readmacro
4293 *** a bug in APROPOS
4294 *** probably some others that I'm not describing very well here,
4295 since the CVS log documents them by reference to sbcl-devel
4296 messages, and the SourceForge archives aren't working well.:-(
4297 ** Dan Barlow improved the Alpha port (and is making progress on the
4298 PPC port, for those of you who think different).
4299 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4300 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4301 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4302 continues to work on the SPARC port (for those of you in a position
4303 to look down upon our little PC-compatible boxes from a great height).
4304 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4305 and uses that copy, rather than the current dynamic value, when
4306 it's trying to decide whether to truncate output. Thus e.g.
4307 (let ((*print-lines* 50))
4308 (pprint-logical-block (stream nil)
4310 (let ((*print-lines* 8))
4311 (print (aref possiblybigthings i) stream)))))
4312 should now truncate the logical block only at 50 lines, instead of
4313 often truncating it at 8 lines, as it did before.
4314 * The doc/cmucl/ directory, containing old CMU CL documentation
4315 from the time of the fork, is no longer part of the base system.
4316 SourceForge has shut down its anonymous FTP service, and with it
4317 my original plan for distributing the old CMU CL documentation
4318 there. For now, if you need these files you can download an old
4319 SBCL source release and extract them from it.
4320 * The fasl file version number changed again, for dozens of reasons,
4321 some of which are apparent above.
4323 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4324 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4325 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4326 patch, so that DISASSEMBLE output is much nicer.
4327 * The code in the SB-PROFILE package now seems reasonably stable.
4328 I still haven't decided what the final interface should look like
4329 (I'd like PROFILE to interact cleanly with TRACE, since both
4330 facilities use function encapsulation) but if you have a need
4331 for profiling now, you can probably use it successfully with
4332 the current CMU-CL-style interface.
4333 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4334 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4335 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4336 link now returns the dangling link itself, and for similar
4337 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4338 link itself. (In these cases the old code signalled an error and
4339 looped endlessly, respectively.) Thus, DIRECTORY now works even
4340 in the presence of dangling and cyclic symbolic links.
4341 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4342 is now a supported extension again, since the consensus on
4343 sbcl-devel was that it can be useful for ordinary development
4344 work, not just for debugging SBCL itself.
4345 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4346 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4347 that currently-defined functions might be redefined later with
4348 different return types.
4349 * Hash tables can be printed readably, as inspired by CMU CL code
4350 of Eric Marsden and SBCL code of Martin Atzmueller.
4351 * better error handling in CLOS method combination, thanks to
4352 Martin Atzmueller porting Pierre Mai's CMU CL patches
4353 * more overflow fixes for >16Mbyte I/O buffers
4354 * A bug in READ has been fixed, so that now a single Ctrl-D
4355 character suffices to cause end-of-file on character streams.
4356 In particular, now you only need one Ctrl-D at the command
4357 line (not two) to exit SBCL.
4358 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4360 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4361 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4362 does the right thing.
4363 * The implementation of some type tests, especially for CONDITION
4364 types, is now tidier and maybe faster, due to CMU CL code
4365 originally by Douglas Crosher, ported by Martin Atzmueller.
4366 * Some math functions have been fixed, and there are new
4367 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4368 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4369 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4370 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4371 likely to be built into sbcl-0.7.0.)
4372 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4373 CL, ported to SBCL by Martin Atzmueller)
4374 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4375 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4376 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4377 * There's a new slam.sh hack to shorten the edit/compile/debug
4378 cycle for low-level changes to SBCL itself, and a new
4379 :SB-AFTER-XC-CORE target feature to control the generation of
4380 the after-xc.core file needed by slam.sh.
4381 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4382 COMPILE-FILE is no longer supported, so that now every function
4383 gets an entry point, so that block compilation looks a little
4384 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4385 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4386 returns the number of bytes consed since the system started,
4387 rather than the number consed since the first time the function
4388 was called. (The new definition parallels ANSI functions like
4389 CL:GET-INTERNAL-RUN-TIME.)
4390 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4391 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4392 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4393 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4394 * Fasl file version numbers are now independent of the target CPU,
4395 since historically most system changes which required version
4396 number changes have affected all CPUs equally. Similarly,
4397 the byte fasl file version is now equal to the ordinary
4400 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4401 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4402 is no longer recognized. I apologize for this, because it was
4403 listed in SB-EXT as a supported extension, but I found that
4404 its existing behavior was poorly specified, as well as incorrectly
4405 specified, and it looked like too much of a mess to straighten it
4406 out. I have enough on my hands trying to get ANSI stuff to work..
4407 * many patches ported from CMU CL by Martin Atzmueller, with
4408 half a dozen bug fixes in pretty-printing and the debugger, and
4409 half a dozen others elsewhere
4410 * fixed bug 13: Floating point infinities are now supported again.
4411 They might still be a little bit flaky, but thanks to bug reports
4412 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4413 as flaky as they were.
4414 * The --noprogrammer command line option is now supported. (Its
4415 behavior is slightly different in detail from what the old man
4416 page claimed it would do, but it's still appropriate under the
4417 same circumstances that the man page talks about.)
4418 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4419 are now supported, and enabled by default. Thus, the compiler can
4420 handle many floating point and complex operations much less
4421 inefficiently. (Thus e.g. you can implement a complex FFT
4423 * The compiler now detects type mismatches between DECLAIM FTYPE
4424 and DEFUN better, and implements CHECK-TYPE more correctly, and
4425 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4426 thanks to patches from Martin Atzmueller.
4427 * various fixes to make the cross-compiler more portable to
4428 ANSI-conforming-but-different cross-compilation hosts (notably
4429 Lispworks for Windows, following bug reports from Arthur Lemmens)
4430 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4431 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4432 fixed thanks to Pierre Mai's CMU CL patches.
4433 * fixes to keep the system from overflowing internal counters when
4434 it tries to use i/o buffers larger than 16M bytes
4435 * fixed bug 45a: Various internal functions required to support
4436 complex special functions have been merged from CMU CL sources.
4437 (When I was first setting up SBCL, I misunderstood a compile-time
4438 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4439 * improved support for type intersection and union, fixing bug 12
4440 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4441 more obscure bugs as well
4442 * some steps toward byte-compiling non-performance-critical
4443 parts of the system, courtesy of patches from Martin Atzmueller
4444 * Christophe Rhodes has made some debian packages of sbcl at
4445 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4446 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4447 stable, but are nonetheless usable. When he's ready, I'd be happy
4448 to add them to the SourceForge "File Releases" section. (And if
4449 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4450 * new fasl file format version number (because of changes in
4451 internal representation of (OR ..) types to accommodate the new
4452 support for (AND ..) types, among other things)
4454 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4455 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4457 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4459 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4460 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4461 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4462 * DESCRIBE now works on structure objects again.
4463 * Most function call argument type mismatches are now handled as
4464 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4465 know whether the function will be redefined before the call is
4466 executed. (The compiler could flag local calls with full WARNINGs,
4467 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4468 it doesn't keep track of enough information to know whether calls
4469 are local in this sense.)
4470 * Compiler output is now more verbose, with messages truncated
4471 later than before. (There should be some supported way for users
4472 to override the default verbosity, but I haven't decided how to
4473 provide it yet, so this behavior is still controlled by the internal
4474 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4475 src/compiler/ir1util.lisp.)
4476 * Fasl file format version numbers have increased again, because
4477 support for the Gray streams extension changes the layout of the
4478 system's STREAM objects.
4479 * The Gray subclassable streams extension now works, thanks to a
4480 patch from Martin Atzmueller.
4481 * The full LOAD-FOREIGN extension (not just the primitive
4482 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4483 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4484 but like most other programs, it defaults to copying the Unix
4485 environment from the original process instead of starting the
4486 new process in an empty environment.
4487 * Extensions which manipulate the Unix environment now support
4488 an :ENVIRONMENT keyword option which doesn't smash case or
4489 do other bad things. The CMU-CL-style :ENV option is retained
4490 for porting convenience.
4491 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4492 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4494 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4496 * A patch from Martin Atzmueller seems to have solved the SIGINT
4497 problem, and as far as we know, signal-handling now works cleanly.
4498 (If you find any new bugs, please report them!)
4499 * The system no longer defaults Lisp source file names to types
4500 ".l", ".cl", or ".lsp", but only to ".lisp".
4501 * The compiler no longer uses special default file extensions for
4502 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4503 seems to expect a single default extension for all compiled code,
4504 and there's no compelling reason to try to stretch the standard
4505 to allow two different extensions.) Instead, byte-compiled files
4506 default to the same extension as native-compiled files.
4507 * Fasl file format version numbers have increased again, because
4508 a rearrangement of internal implementation packages made some
4509 dumped symbols in old fasl files unreadable in new cores.
4510 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4511 many fewer weird special cases.
4512 * Bug #17 (differing COMPILE-FILE behavior between logical and
4513 physical pathnames) has been fixed, and some related misbehavior too,
4514 thanks to a patch from Martin Atzmueller.
4515 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4516 by Tim Moore, ported to SBCL by Martin Atzmueller.
4517 * Martin Atzmueller fixed several filesystem-related problems,
4518 including bug #36, in part by porting CMU CL patches, which were
4519 written in part by Paul Werkowski.
4520 * More compiler warnings in src/runtime/ are gone, thanks to
4521 more patches from Martin Atzmueller.
4522 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4525 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4527 * DESCRIBE now works on CONDITION objects.
4528 * The debugger now handles errors which arise when trying to print
4529 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4531 * The build system now uses an additional file, customize-target-features.lisp,
4532 to allow local modifications to the target *FEATURES* list. (The point of
4533 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4534 debugging features enabled, without having to worry about propagating it
4535 into everyone's system when I do a "cvs update".) When no
4536 customize-target-features.lisp file exists, the target *FEATURES* list
4537 should be constructed the same way as before.
4538 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4539 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4540 warn about behavior which is undefined under the ANSI spec. Note: This
4541 is specified by ANSI, but it's not very popular with programmers.
4542 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4543 macro in the SBCL sources for an example of a workaround which you
4544 might use to make portable ANSI-standard code which does what you want.
4545 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4546 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4547 when it's used as a non-toplevel form. (This is required in order
4548 to implement the DEFCONSTANT-EQX macro.)
4549 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4550 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4551 funny things, due to bugs in the implementation of EVAL-WHEN
4552 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4553 non-toplevel DEFCONSTANTs are uncommon.)
4554 * The core file version number and fasl file version number have been
4555 incremented, because the old noncompliant DEFCONSTANT behavior involved
4556 calling functions which no longer exist, and because I also took the
4557 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4558 * fixed bug 1 (error handling before read-eval-print loop starts), and
4559 redid debugger restarts and related debugger commands somewhat while
4561 ** The QUIT debugger command is gone, since it did something
4562 rather different than the SB-EXT:QUIT command, and since it never
4563 worked properly outside the main toplevel read/eval/print loop.
4564 Invoking the new TOPLEVEL restart provides the same functionality.
4565 ** The GO debugger command is also gone, since you can just invoke
4566 the CONTINUE restart directly instead.
4567 ** The TOP debugger command is also gone, since it's redundant with the
4568 FRAME 0 command, and since it interfered with abbreviations for the
4570 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4572 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4573 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4574 some porting work by Martin Atzmueller)
4575 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4576 with a boatload of patches which he ported to SBCL. Now that those
4578 ** The system tries to make sure that its low-priority messages
4579 are prefixed by semicolons, to help people who like to use
4580 syntax highlighting in their ILISP buffer. (This patch
4581 was originally due to Raymond Toy.)
4582 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4583 patches originally written by Raymond Toy.
4584 ** The compiler understands coercion between single-value and
4585 multiple-VALUES type expressions better, getting rid of some very
4586 weird behavior, thanks to patches originally by Robert MacLachlan
4587 and Douglas Crosher.
4588 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4589 lambda lists, thanks to a patch originally by Pierre Mai.
4590 ** The system no longer bogusly warns about "abbreviated type
4592 ** The compiler gets less confused by inlining and RETURN-FROM,
4593 thanks to some patches originally by Tim Moore.
4594 ** The system no longer hangs when dumping circular lists to fasl
4595 files, thanks to a patch originally from Douglas Crosher.
4596 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4597 undefined function error.
4598 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4599 in structure slot initforms) for now, documented workaround instead:-|
4600 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4601 * fixed bug 5: added stubs for various Gray stream functions called
4602 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4603 installed, at least appropriate type errors are generated
4604 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4605 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4606 consistently in DEFMETHOD forms.
4607 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4608 it doesn't seem to affect SBCL after all
4609 * The C runtime system now builds with better optimization and many
4610 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4612 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4614 * The system is now under CVS at SourceForge (instead of the
4615 CVS repository on my home machine).
4616 * The new signal handling code has been tweaked to treat register
4617 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4618 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4619 downstream, e.g. in debug-int.lisp, has implicit dependencies
4620 on the unsignedness of integer representation of machine words,
4621 and that caused the system to bomb out with infinite regress
4622 when trying to recover from type errors involving signed values,
4623 e.g. (BUTLAST '(1 2 3) -1).)
4624 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4625 (This was one of the bugs Peter Van Eynde reported back in July.)
4626 * The system now uses code inspired by Colin Walters' O(N)
4627 implementation of MAP (from the cmucl-imp@cons.org mailing
4628 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4629 inline the MAP operation, and there is more than one
4630 sequence argument to the MAP call (so that it can't just
4631 do ETYPECASE once and for all based on the type of the
4632 single sequence argument). (The old non-inline implementation
4633 of the general M-argument sequence-of-length-N case required
4634 O(M*N*N) time when any of the sequence arguments were LISTs.)
4635 * The QUIT :UNIX-CODE keyword argument has been renamed to
4636 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4637 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4639 * Raymond Wiker's patches to port dynamic loading from Linux to
4640 FreeBSD have been added.
4641 * The BUGS file is now more nearly up to date, thanks in large part
4642 to Martin Atzmueller's review of it.
4643 * The debugger now flushes standard output streams before it begins
4644 its output ("debugger invoked" and so forth).
4645 * The core version number and fasl file version number have both
4646 been incremented, because of incompatible changes in the layout
4648 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4649 used to not be called for a saved Lisp image.)
4650 * Martin Atzmueller's version of a patch to fix a compiler crash,
4651 as posted on sbcl-devel 13 September 2000, has been installed.
4652 * Instead of installing Martin Atzmueller's patch for the
4653 compiler transform for SUBSEQ, I deleted the compiler transform,
4654 and transforms for some similar consing operations.
4655 * A bug in signal handling which kept TRACE from working on OpenBSD
4657 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4658 away by constant folding
4659 * The system now defines its address space constants in one place
4660 (in the Lisp sources), and propagates them automatically elsewhere
4661 (through GENESIS and the sbcl.h file). Therefore, patching the
4662 address map is less unnecessarily tedious and error-prone. The
4663 Lisp names of address space constants have also been systematized.
4664 * CVS tags like dollar-Header-dollar have been removed from
4665 the sources, because they have never saved me trouble and
4666 they've been source of trouble working with patches and other
4667 diff-related operations.
4668 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4669 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4671 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4673 * The system has been ported to OpenBSD.
4674 * The system now compiles with a simple "sh make.sh" on the systems
4675 that it's supported on. I.e., now you no longer need to tweak
4676 text in the target-features.lisp-expr and symlinks in src/runtime/
4677 by hand, the make.sh takes care of it for you.
4678 * The system is no longer so grossly inefficient when compiling code
4679 involving vectors implemented as general (not simple) vectors (VECTOR T),
4680 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4681 which dares to use the various sequence functions on non-simple
4682 vectors, takes less of a performance hit.
4683 * There is now a primitive type predicate VECTOR-T-P
4684 to test for the (VECTOR T) type, so that e.g.
4685 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4686 can now be compiled with some semblance of efficiency. (The old code
4687 turned the type declaration into a full call to %TYPEP at runtime!)
4688 * AREF on (VECTOR T) is still not fast, since it's still compiled
4689 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4690 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4691 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4692 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4693 without making full calls to SUBTYPEP at runtime.
4694 (Some analogous efficiency issues for non-simple vectors specialized to
4695 element types other than T, or for non-simple multidimensional arrays,
4696 have not been addressed. They could almost certainly be handled the
4697 same way if anyone is motivated to do so.)
4698 * The changes in array handling break binary compatibility, so
4699 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4700 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4701 instead of (VALUES T T).
4702 * By following the instructions that Dan Barlow posted to sbcl-devel
4703 on 2 July 2000, I was able to enable primitive dynamic object
4704 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4705 functionality is not implemented (since it calls ld to resolve
4706 library references automatically, requiring RUN-PROGRAM for its
4707 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4708 try to resolve library references) is now supported.
4709 * The system now flushes the standard output streams when it terminates,
4710 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4711 them at several other probably-convenient times, e.g. in each pass of
4712 the toplevel read-eval-print loop, and after evaluating a form given
4713 as an "--eval" command-line option. (These changes were motivated by a
4714 discussion of stream flushing issues on cmucl-imp in August 2000.)
4715 * The source transform for TYPEP of array types no longer assumes
4716 that an array whose element type is a not-yet-defined type
4717 is implemented as an array of T, but instead punts, so that the
4718 type will be interpreted at runtime.
4719 * There is now some support for cross-compiling in make.sh: each of
4720 the phases of make.sh has its own script. (This should be transparent
4721 to people doing ordinary, non-cross-compile builds.)
4722 * Since my laptop doesn't have hundreds of megabytes of memory like
4723 my desktop machine, I became more motivated to do some items on
4724 my to-do list in order to reduce the size of the system a little:
4725 ** Arrange for various needed-only-at-cold-init things to be
4726 uninterned after cold init. To support this, those things have
4727 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4728 symbols with such names are now uninterned after cold init).
4729 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4730 fasl files for cold load.
4731 ** Remove the old compiler structure pooling code (which used to
4732 be conditional on the target feature :SB-ALLOC) completely.
4733 ** Redo the representation of some data in cold init to be more compact.
4734 (I also looked into supporting byte compiled code at bootstrap time,
4735 which would probably reduce the size of the system a lot, but that
4736 looked too complicated, so I punted for now.)
4737 * The maximum signal nesting depth in the src/runtime/ support code has
4738 been reduced from 4096 to 256. (I don't know any reason for the very
4739 large old value. If the new smaller value turns out to break something,
4740 I'll probably just bump it back up.)
4741 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4743 * Many, many bugs reported by Peter Van Eynde have been added to
4744 the BUGS list; some have even been fixed.
4745 * While enabling dynamic object file loading, I tried to make the
4746 code easier to understand, renaming various functions and variables
4747 with less ambiguous names, and changing some function calling
4748 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4749 * While trying to figure out how to do the OpenBSD port, I tried to
4750 clean up some of the code in src/runtime/. In particular, I dropped
4751 support for non-POSIX signal handling, added various comments,
4752 tweaked the code to reduce the number of compilation warnings, and
4753 renamed some files to increase consistency.
4754 * To support the new automatic configuration functionality in make.sh,
4755 the source file target-features.lisp-expr has been replaced with the
4756 source file base-target-features.lisp-expr and the machine-generated
4757 file local-target-features.lisp-expr.
4758 * fixed a stupid quoting error in make.sh so that using CMU CL
4759 "lisp -batch" as cross-compilation host works again
4761 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4763 * DESCRIBE no longer tries to call itself recursively to describe
4764 bound/fbound values, so that it no longer fails on symbols which are
4765 bound to themselves (like keywords, T, and NIL).
4766 * DESCRIBE now works on generic functions.
4767 * The printer now prints less-screwed-up representations of closures
4768 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4769 * A private symbol is used instead of the :EMPTY keyword previously
4770 used to mark empty slots in hash tables. Thus
4771 (DEFVAR *HT* (MAKE-HASH-TABLE))
4772 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4773 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4774 now does what ANSI says that it should. (You can still get
4775 similar noncompliant behavior if bang on the hash table
4776 implementation with all the symbols you get back from
4777 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4778 This breaks binary compatibility, since tests for equality to
4779 :EMPTY are wired into things like the macroexpansion of
4780 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4782 * There's now a minimal placeholder implementation for CL:STEP,
4783 as required by ANSI.
4784 * An obscure bug in the interaction of the normal compiler, the byte
4785 compiler, inlining, and structure predicates has been patched
4786 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4787 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4789 * Missing ordinary arguments in a macro call are now detected even
4790 when the macro lambda list contains &KEY or &REST.
4791 * The debugger no longer complains about encountering the top of the
4792 stack when you type "FRAME 0" to explicitly instruct it to go to
4793 the top of the stack. And it now prints the frame you request even
4794 if it's the current frame (instead of saying "You are here.").
4795 * As specified by ANSI, the system now always prints keywords
4796 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4798 * The default initial SIZE of HASH-TABLEs is now smaller.
4799 * Type information from CLOS class dispatch is now propagated
4800 into DEFMETHOD bodies, so that e.g.
4801 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4803 is now basically equivalent to
4804 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4805 (DECLARE (TYPE SINGLE-FLOAT X))
4807 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4808 operation, without having to do run-time type dispatch.
4809 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4810 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4811 * Since I don't seem to be making big file reorganizations very often
4812 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4813 is overflowing my ability to conveniently back them up), I've finally
4814 checked the system into CVS. (The CVS repository is on my home system,
4815 not at SourceForge -- putting it on SourceForge might come later.)
4816 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4817 high-level GC-NOTIFY-FOO functions send their output. (There's
4818 still very little control of where low-level verbose GC functions
4819 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4820 less than it used to -- the GC-NOTIFY-FOO functions are now under
4821 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4822 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4823 in only one place in the source code, and propagates it automatically
4824 everywhere that it's needed. Thus e.g. when I bump the version from
4825 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4826 * The C source files now include boilerplate legalese and documentation
4827 at the head of each file (just as the Lisp source files already did).
4828 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4829 to his page will be replaced with a link to his new CLiki service.
4831 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4833 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4834 * The build process now looks for GNU make under the default name "gmake",
4835 instead of "make" as it used to. If GNU make is not available as "gmake"
4836 on your system, you can change this default behavior by setting the
4837 GNUMAKE environment variable.
4838 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4839 can build without error under CMU CL.
4841 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4843 * There is now a partial SBCL user manual (with some new text and some
4844 text cribbed from the CMU CL manual).
4845 * The beginnings of a profiler have been added (starting with the
4846 CMU CL profiler and simplifying and cleaning up). Eventually the
4847 main interface should be through the TRACE macro, but for now,
4848 it's still accessed through vaguely CMU-CL-style functions and macros
4849 exported from the package SB-PROFILE.
4850 * Some problems left over from porting CMU CL to the new
4851 cross-compilation bootstrap process have been cleaned up:
4852 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4853 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4854 part when getting rid of the compiler *BACKEND* stuff.)
4855 ** The value of *NULL-TYPE* was screwed up, because it was
4856 being initialized before the type system knew the final
4857 definition of the 'NULL type. This screwed up several key
4858 optimizations in the compiler, causing inefficiency in all sorts
4859 of places. (I found it because I wanted to understand why
4860 GET-INTERNAL-RUN-TIME was consing.)
4861 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4862 PROCLAIM FTYPE information. Unfortunately this broke binary
4863 compatibility again, since now the forms output by DEFGENERIC
4864 to refer to functions which didn't exist in 0.6.3.
4865 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4866 can use the new (as of 0.6.3) transform for SOME into MAP into
4868 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4869 values to (INTEGER 0 1000000), so that the time code will no longer
4870 occasionally get blown up by Linux returning 1000000 microseconds
4871 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4872 its output conform to the ANSI spec. (Alas, this makes its output
4873 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4874 * A full call to MAP NIL with a single sequence argument no longer conses.
4875 * fixes to problems pointed out by Martin Atzmueller:
4876 * The manual page no longer talks about multiprocessing as though
4877 it were currently supported.
4878 * The ILISP support patches have been removed from the distribution,
4879 because as of version 5.10.1, ILISP now supports SBCL without us
4880 having to maintain patches.
4881 * added a modified version of Raymond Toy's recent CMU CL patch for
4882 EQUALP comparison of HASH-TABLE
4884 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4886 * The system still can't cross-compile itself with
4887 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4888 entails), but at least it can compile more of itself that way
4889 than it used to be able to, and various buglets which were uncovered
4890 by trying to cross-compile itself that way have now been fixed.
4891 * This release breaks binary compatibility again. This time
4892 at least I've incremented the FASL file format version to 2, so that the
4893 problem can be detected reliably instead of just causing weird errors.
4894 * various new style warnings:
4895 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4896 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4897 ** using the lexical binding of a variable named in the *FOO* style
4898 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4899 as specified by ANSI.
4900 * *RANDOM-STATE* is no longer automatically initialized from
4901 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4902 default behavior of the system is to repeat its behavior every time
4903 it's run. If you'd like to change this behavior, you can always
4904 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4905 old convention there was no comparably easy way to get the system to
4906 repeat its behavior every time it was run.
4907 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4908 FTYPE declarations has been removed, in favor of their ANSI
4909 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4910 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4911 the types of their sequence arguments can be determined at compile time.
4912 This is done through a new open code expansion for MAP which eliminates
4913 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4914 argument types can be determined at compile time.
4915 * The optimizer now transforms COERCE into an identity operation when it
4916 can prove that the coerced object is already of the correct type. (This
4917 can be a win for machine generated code, including the output of other
4918 optimization transforms, such as the MAP transform above.)
4919 * Credit information has been moved from source file headers into CREDITS.
4920 * Source file headers have been made more standard.
4921 * The CASE macro now compiles without complaining even when it has
4924 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4926 * (Note that the way that the PCL macroexpansions were rewritten
4927 to accommodate the change in DEFGENERIC below breaks binary
4928 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4929 not run under sbcl-0.6.2. Once we get out of alpha releases,
4930 i.e. hit release 1.0.0, we'll probably try to maintain binary
4931 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4932 and sbcl-1.4.4. Until then, however, it might be fairly common
4933 for maintenance releases to break binary compatibility.)
4934 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4936 * The reporting of unhandled signals has been changed to print some
4937 explanatory text as well as the report form. (Previously only
4938 the report form was printed.)
4939 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4940 it defines, so that the compiler no longer issues undefined function
4941 warnings for compiled-but-not-yet-loaded generic functions.
4942 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4943 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4944 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4945 section of the ANSI spec are supported.
4946 * The "" nickname for the "KEYWORD" package has been removed.
4947 The reader still handles symbol tokens which begin with a package marker
4948 as keywords, but it doesn't expose its mechanism for doing so in the
4949 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4950 * The system now issues STYLE-WARNINGs for contradictory TYPE
4951 proclamations. (Warnings for contradictory FTYPE proclamations would
4952 be nice too, but those can't be done usefully unless the type system
4953 is made smarter about FUNCTION types.)
4954 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4955 systematized, so that "*target-*.lisp is supposed to exist only on the
4956 target and imply that there's a related file which exists on the
4957 host, and *host-*.lisp is supposed to exist only on the host and imply
4958 that there's a related file which exists on the target. This involves a
4959 lot of renaming. Hopefully the acute confusion caused by the renaming
4960 will be justified by the reduction in chronic confusion..
4961 ** runtime-type.lisp -> early-target-type.lisp
4962 ** target-type.lisp -> late-target-type.lisp
4963 ** early-host-format.lisp -> early-format.lisp
4964 ** late-host-format.lisp -> late-format.lisp
4965 ** host-error.lisp -> misc-error.lisp
4966 ** early-error.lisp -> early-target-error.lisp
4967 ** late-error.lisp -> late-target-error.lisp
4968 ** host-defboot.lisp -> early-defboot.lisp
4969 ** code/misc.lisp -> code/target-misc.lisp
4970 ** code/host-misc.lisp -> code/misc.lisp
4971 ** code/numbers.lisp -> code/target-numbers.lisp
4972 ** code/early-numbers.lisp -> numbers.lisp
4973 ** early-host-type.lisp -> early-type.lisp
4974 ** late-host-type.lisp -> late-type.lisp
4975 ** host-typep.lisp -> typep.lisp
4976 ** load.lisp -> target-load.lisp
4977 ** host-load.lisp -> load.lisp
4978 ** host-disassem.lisp -> disassem.lisp
4979 ** host-insts.lisp -> insts.lisp
4980 ** byte-comp.lisp -> target-byte-comp.lisp
4981 ** host-byte-comp.lisp -> byte-comp.lisp
4982 ** host-signal.lisp -> signal.lisp
4983 ** host-defstruct.lisp -> defstruct.lisp
4984 ** late-target-type.lisp -> deftypes-for-target.lisp
4985 Furthermore, several other previously target-only files foo.lisp (e.g.
4986 hash-table.lisp and random.lisp) have been split into a target-and-host
4987 foo.lisp file and a target-only target-foo.lisp file, with their key type
4988 definitions in the target-and-host part, so that the cross-compiler will
4989 know more about target types.
4990 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4991 gone away. In their place are various *BACKEND-FOO* variables
4992 corresponding to the slots of the old structure.
4993 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4994 into the target SBCL has been fixed.
4995 * The chill.lisp system for loading cold code into a running SBCL
4997 * Support for the CMU CL "scavenger hook" extension has been removed.
4998 (It was undocumented and unused in the CMU CL sources that SBCL was
4999 derived from, and stale in sbcl-0.6.1.)
5000 * Various errors in the cross-compiler type system were detected
5001 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5002 (enabling various consistency checks). Many of them were fixed,
5003 but some hard problems remain, so the compiler is back to
5004 running without *TYPE-SYSTEM-INITIALIZED* for now.
5005 * As part of the cross-compiler type system cleanup, I implemented
5006 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5007 * I have started adding UNCROSS calls throughout the type system
5008 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5009 on cross-compiler input files will be able to go away, and various
5011 * CONSTANTP now returns true for quoted forms (as explicitly required
5014 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5016 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5017 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5018 (MAKE-LIST -1). In the longer term, it ought to become true
5019 that declarations are assertions even at SAFETY 1. For now, it's not
5020 quite true even at SAFETY 3, but it's at least more nearly true..
5021 (Note that this change seems to increases the size of the system by
5022 O(5%) and to decrease the speed of the compiler by 20% or more.)
5023 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5024 for the problem of printing dozens of lines of distracting information
5025 about low-level system machinery as part of the top stack frame
5026 on entry to the debugger when an undefined function was called.
5027 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5028 is not reset to COMMON-LISP-USER.
5029 * Compilation of stuff related to dyncount.lisp has been made conditional
5030 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5031 smaller. The various dyncount-related symbols have been moved into
5032 a new "SB-DYNCOUNT" package.
5033 * tty-inspect.lisp has been renamed to inspect.lisp.
5034 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5035 feature has gone away. (When we eventually port to other flavors of
5036 libc and/or Unix, we'll try to make the differences between flavors
5037 invisible at the user level.)
5038 * Various other *FEATURES* tags, and/or their associated conditionals,
5039 have been removed if obsolescent, or given better documentation, or
5040 sometimes given more-mnemonic names.
5042 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5044 * tidied up "make.sh" script
5045 * tidied up system directory structure
5046 * better "clean.sh" behavior
5047 * added doc/FOR-CMUCL-DEVELOPERS
5048 * many many small tweaks to output format, e.g. removing possibly-confusing
5049 trailing #\. character in DESCRIBE-INSTANCE
5050 * (EQUALP #\A 'A) no longer signals an error.
5051 * new hashing code, including EQUALP hashing
5052 * tidied up Lisp initialization and toplevel
5053 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5054 * command line argument processing
5055 * added POSIX-GETENV function to deal with Unix-ish environment variables
5056 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5057 terminating SBCL on EOF
5058 * non-verbose GC by default
5059 * There is no more "sbcl" shell script; the sbcl file is now the C
5060 runtime executable (just like CMU CL).
5061 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5063 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5064 symbol and package information as Lisp native symbol and package objects
5065 instead of strings naming symbols and strings naming packages. This way,
5066 whenever packages are renamed (as in warm init), debug information is
5067 transformed along with everything else.
5068 * tweaked the optimization policy declarations which control the building
5069 of SBCL itself. Now, among other things, the system no longer saves
5070 source location debugging information. (This helps two problems at once
5071 by reducing SBCL size and by keeping SBCL from trying to look for its
5072 sources -- which may not exist -- when reporting errors.)
5073 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5074 debugging and testing purposes
5075 * cleaned up printing, making the printer call PRINT-OBJECT for
5076 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5077 methods, giving nearly-ANSI behavior
5078 * converted almost all special variables to use *FOO* naming convention
5079 * deleted PARSE-TIME functionality, since it can be done portably
5080 * moved some files out of cold init into warm init
5081 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5083 * regularized formatting of source files
5084 * added an install.sh script
5085 * fixed ridiculous memory usage of cross-compiler by making
5086 compiler/alloc.lisp not try to do pooling unless it can hook
5087 itself into the GC of the cross-compilation host. Now the system
5088 builds nicely on my old laptop.
5089 * added :SB-ALLOC in target-features.lisp-expr
5090 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5091 was not implemented)
5092 * re-did condition handling and note reporting in the compiler. Notes
5093 are no longer handled by signalling conditions. Style warnings
5094 and warnings are handled more correctly and reported in such a way
5095 that it's easy to find one or the other in your output (so that you
5096 can e.g. figure out which of many problems caused COMPILE-FILE to
5098 * changed the severity of several compiler warnings from full WARNING
5099 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5100 compiler note reporting so that it doesn't use the condition system
5101 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5102 COMPILE-FILE command)
5103 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5104 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5105 no longer supported, since their implementation was deeply intertwingled
5106 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5107 * removed bogus "support" for compiler macros named (SETF FOO), and
5108 removed the compiler macro for SETF INFO (but only after making a fool
5109 of myself on the cmucl-imp mailing list by posting a bogus patch for
5110 DEFINE-COMPILER-MACRO..)
5111 * Compiled files containing forms which have side effects on the Lisp
5112 reader (such as DEFPACKAGE forms) are now handled more correctly.
5113 (Compiler queuing of top level lambdas has been suppressed by setting
5114 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5115 * deleted various currently-unused source files, e.g. gengc.lisp. They
5116 may be added back at some point e.g. when porting to other architectures,
5117 but until they are it's distracting to distribute them and to try to
5119 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5120 there were so many of them they're just distractions, and UNCROSS is
5121 known to be able to handle the current sources
5122 * moved PROFILE functionality into TRACE, so that it will be clear
5123 how the wrapping and unwrapping of functions when you profile them
5124 interacts with the wrapping and unwrapping of functions when you
5125 trace them. (Actually, the functionality isn't there yet, but at least
5126 the interface specification is there. Hopefully, the functionality will
5127 arrive with some future maintenance release.)
5128 * removed host-oops.lisp
5129 * changed signature of QUIT function to allow UNIX-CODE argument
5130 * fixed READ-SEQUENCE bug
5131 * tweaked verbose GC output so that it looks more like the progress
5132 output that ANSI specifies for functions like LOAD
5133 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5134 * added <http://sbcl.sourceforge.com> to the banner information printed by