1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 * bug fix: disable address space randomization Linux/x86-64 as well,
3 not just x86-64. (reported by Ken Olum)
5 changes in sbcl-1.0.28 relative to 1.0.27:
6 * a number of bugs in cross-compilation have been fixed, with the ultimate
7 result that building under (at least) clisp should be much more reliable.
8 * minor incompatible changes: echo-streams now propagate unread-char to the
9 underlying input stream, and no longer permit unreading more than one
11 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
12 frames, allowing for instance more reliable backtraces.
13 * improvement: the debugger REPL can now reference lexical variables
14 by name directly for code compiled with (DEBUG 3).
15 * improvement: errors from malformed declarations now have better source
16 paths associated with them. (thanks to Tobias Rittweiler)
17 * optimization: faster local calls on x86/x86-64
18 * bug fix: some error messages for out-of-bound array indexes confused the
19 index and the bound. (thanks to Stas Boukarev)
20 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
22 * bug fix: running regressions tests in shells without OSTYPE set now works.
23 (reported by Harald Hanche-Olsen)
24 * bug fix: more robust static space exhaustion signalling from
25 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
26 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
27 docstring away instead of storing it under names such as (LAMBDA (X)).
28 (reported by Leslie Polzer)
29 * bug fix: timers could go off in the wrong order, be delayed indefinitely
30 (thanks to Ole Arndt for the patch)
31 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
32 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
33 potentially-invalid effective methods in its cache.
34 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
35 instances as well (thanks to Paul Khuong)
36 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
37 SB-POSIX:WAITPID and their C equivalents.
38 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
40 changes in sbcl-1.0.27 relative to 1.0.26:
41 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
42 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
43 * improvement: the system either recovers from stack exhaustion or dies
44 properly as opposed to leaving the user uncertain of whether the handler
45 trampled on some random memory next to the stack or having to rely on
46 --lose-on-corruption (which is still a good idea to use in production
47 because stack exhaustion can happen in signal handlers which will likely
49 * bug fix: fix gc related interrupt handling bug on ppc (regression from
50 1.0.25.37, reported by Harald Hanche-Olsen)
51 * bug fix: work around signal delivery bug in darwin (regression from
52 1.0.25.44, reported by Sidney Markowitz)
53 * bug fix: fix ERROR leaking memory (reported by David Thompson)
55 changes in sbcl-1.0.26 relative to 1.0.25:
56 * incompatible change: an interruption (be it a function passed to
57 INTERRUPT-THREAD or a timer function) runs in an environment where
58 interrupts can be enabled. The interruption can use
59 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
60 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
61 potentially running out of stack. Keep in mind that in the absence
62 of WITHOUT-INTERRUPTS some potentially blocking operation such as
63 acquiring a lock can enable interrupts.
64 * incompatible change: GC-OFF and GC-ON are removed, as they were
65 always unsafe. Use WITHOUT-GCING instead.
66 * new feature: runtime option --disable-ldb
67 * new feature: runtime option --lose-on-corruption to die at the
68 slightest hint of possibly non-recoverable errors: running out of
69 memory, stack, alien stack, binding stack, encountering a memory
70 fault, etc. In the absence of --lose-on-corruption a warning is
72 * enhancement: detect binding stack exhaustion
73 * enhancement: detect alien stack exhaustion on x86/x86-64
74 * improvement: generally more stable and reliable interrupt handling
75 * improvement: there is a per thread interruption queue,
76 interruptions are executed in order of arrival
77 * improvement: a repeating timer reschedules itself when the it has
78 finished, but expiration times are spaced equally. If an
79 expiration time is in the past it will trigger after a short grace
80 period that may give a chance to other things to run.
81 * optimization: slightly faster gc on multithreaded builds
82 * optimization: faster WITHOUT-GCING
83 * bug fix: when JOIN-THREAD signals an error, do it when not holding
84 important locks so that the debugger/handler doesn't produce
85 recursive errors or deadlock.
86 * bug fix: real-time signals are not used anymore, so no more
87 hanging when the system wide real-time signal queue gets full.
88 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
90 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
91 * bug fix: fix random memory faults related to interrupts on alpha
92 * bug fix: fix random memory faults related to interrupts on sparc
93 * bug fix: fix deadlocks related to starting threads
94 * bug fix: fix deadlines on locks on futex platforms
95 * bug fix: restore errno in signal handlers
96 * bug fix: fix deadlocks related to hash tables
97 * bug fix: fix deadlocks in pcl
99 changes in sbcl-1.0.25 relative to 1.0.24:
100 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
101 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
102 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
103 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
104 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
105 better name and does not return values so stale on multiprocessor systems.
106 Also, HOLDING-MUTEX-P was added for about the only sane usage of
108 * improvement: unithread builds keep track of MUTEX-VALUE.
109 * improvement: reading from a TWO-WAY-STREAM does not touch the output
110 stream anymore making it thread safe to have a concurrent reader and
111 a writer, for instance, in a pipe.
112 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
113 arguments, making compiler-macros for SETF-functions able to inspect
114 their constant arguments.
115 * improvement: COMPILE-FILE reports times with millisecond accuracy
116 (thanks to Luis Oliveira)
117 * optimization: CHAR-CODE type derivation has been improved, making
118 TYPEP elimination on subtypes of CHARACTER work better. (reported
119 by Tobias Rittweiler, patch by Paul Khuong)
120 * bug fix: setting alien structure fields of type struct by value now
121 computes the right offset for the memory copy.
122 * bug fix: compilation problem involving inlined calls to aliens with
123 result type VOID. (reported by Ken Olum)
124 * bug fix: #235a; sequential inline expasion in different policies no
125 longer reuses the functional from the previous expansion site.
126 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
127 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
128 * bug fix: named ENUMs in multiply-referenced alien record types no longer
129 cause an implied type redefinition. Regression from 1.0.21.29.
130 * bug fix: sign-extension of small signed return values now works with C
131 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
132 * improvements to the Windows port:
133 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
134 Westfold, thanks to Rudi Schlatte)
136 changes in sbcl-1.0.24 relative to 1.0.23:
137 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
138 vector of a multidimensional SIMPLE-ARRAY.
139 * new feature: the system now signals a continuable error if standard
140 readtable modification is attempted. (thanks to Tobias Rittweiler)
141 * new feature: DIRECTORY has been extended with a non-standard keyword
142 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
143 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
144 has been robustified and documented better. (thanks to Robert Goldman)
145 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
146 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
147 * tradeoff: constant FORMAT control strings are now compiled unless
148 SPACE > SPEED (previously only when SPEED > SPACE.)
149 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
151 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
153 * bug fix: lexical type declarations are now correctly reported by
154 SB-CLTL2. (reported by Larry D'Anna)
155 * bug fix: STRING-TO-OCTETS did not handle :START properly when
156 using UTF-8 as external format. (thanks to Luis Oliveira)
157 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
158 are signalled correctly. (thanks to Stas Boukarev)
159 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
160 designator. (thanks to Tobias Rittweiler)
161 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
162 designator, and returns T instead of the function. (thanks to
164 * bug fix: direct superclasses of STANDARD-CLASS and
165 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
166 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
167 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
168 when no class with the named slot yet exists no longer causes a
169 compile-time style-warning.
170 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
171 in safe code. (reported by Didier Verna)
172 * bug fix: #430; nested structure constructors can stack allocate.
173 * bug fix: on some 64-bit platforms dynamic space size was truncated
174 to #xffff0000 bytes. (reported by Benjamin Lambert)
175 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
176 REPL unusable. (reported by Daniel Herring)
177 * bug fix: return values of READ-SEQUENCE did not take :START into
178 account on file streams, regressions since 1.0.12.22. (reported by
179 Thomas Russ, patch by Paul Khuong)
180 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
181 method specializer used to confuse permuation vector optimization.
182 * bug fix: system inserted bogus implicit type declarations for local
183 special variables in DEFMETHOD bodies.
184 * bug fix: #354; duplicated frames in backtraces due to
185 non-tail-call-optimized XEPs to functions with return type NIL
187 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
188 initializes structure object slots according to DEFSTRUCT initforms,
189 and DEFSTRUCT forms :INCLUDEind structure classes defined using
190 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
191 (reported by Bruno Haible and Stephen Wilson)
192 * bug fix: #395; fill-pointer output streams used now support
193 element-type BASE-CHAR as well.
194 * bug fix: compiler error when attempting to derive return value of
195 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
197 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
199 * bug fix: Mac OS X binaries should now be portable between Leopard
202 changes in sbcl-1.0.23 relative to 1.0.22:
203 * enhancement: when disassembling method functions, disassembly
204 for the associated fast function is also produced.
205 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
207 * optimization: printing with *PRINT-PRETTY* true is now more
208 efficient as long as the object being printed doesn't require
209 special handling by the pretty printer.
210 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
211 now interact correctly with type declarations.
212 * partial bug fix: PCL detects infinite recursion during wrapper
213 validation. (thanks to Attila Lendvai)
214 * bug fix: #426; nested function calls are inlined properly.
215 Previously if FOO was an inline function, in calls of the form
216 (FOO (FOO ...)) the outer call was not inlined.
217 * bug fix: long long arguments passed in registers to alien
218 functions on PowerPC Linux and NetBSD are now handled correctly
219 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
221 changes in sbcl-1.0.22 relative to 1.0.21:
222 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
223 for the shared object in the current directory, but passes the native
224 namestring of the designated pathname to the operation system's shared
225 object loading function as-it.
226 * minor incompatible change: --disable-debugger toplevel option now takes
227 effect before processing of initialization files and --eval or --load
229 * new feature: new commandline argument: --script, which supports
230 shebang lines. See documentation for details. (based on work by
232 * new feature: SAVE-LISP-AND-DIE can save current values of
233 --dynamic-space-size and --control-stack-size in the executable core,
234 causing it to skip normal runtime option processing. See documentation
235 for details. (thanks to Zach Beane)
236 * enhancement: inoccous calls to EVAL or generic functions dispatching
237 on subclasses of eg. STREAM no longer cause compiler notes to appear.
238 * enhancement: the system no longer resignals errors from --load and
239 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
240 associated with the original error to be lost. (thanks to Ariel
242 * enhancement: :DONT-SAVE keyword argument has been added to
243 LOAD-SHARED-OBJECT for controlling interaction with
244 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
245 the effects of an earlier LOAD-SHARED-OBJECT call.
246 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
247 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
248 provided. (reported by Cedric St-Jean)
249 * bug fix: circularity handling in the reader did not treat raw
250 structure slots correctly. (reported by Cedric St-Jean)
251 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
252 file descriptors when there were none.
253 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
254 was broken. (reporteed by Willem Broekema)
255 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
256 pathnames without a directory.
257 * bug fix: COMPILE-FILE did not warn about undefined variable
258 references at toplevel, and LOAD of the resulting fasl did
260 * bug fix: functions with non-required arguments used to end up with
261 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
262 * bug fix: redefining a function with non-required arguments didn't
263 update the system's knowledge about its call signature properly.
264 * bug fix: fixed #431; incompatible alien record type redefinitions
265 are detected and handled. (reported by Neil Haven)
266 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
267 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
269 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
270 after alien stack frames.
271 * bug fix: strings passed to mk{s,d}temp lacked null termination
273 changes in sbcl-1.0.21 relative to 1.0.20:
274 * new feature: the compiler is able to track the effective type of a
275 generic function across method addition and removal.
276 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
277 appropriately typed structure slots without locking.
278 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
279 information like those gathered by TIME using a programming-friendly
281 * new feature: TIME reports time taken even if the form performs a
282 non-local transfer of control.
283 * enhancement: reduced conservativism on GENCGC platforms: on
284 average 45% less pages pinned (measured from SBCL self build).
285 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
286 mutate constant symbols or violate declaimed type of the symbol.
287 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
288 sleeping on the semaphore in heavily contested semaphores.
289 * bug fix: semaphores and condition variables were not interrupt
291 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
292 owned by other threads anymore.
293 * bug fix: FIND on lists called KEY outside the specified
294 subsequence. (reported by budden)
295 * bug fix: LOG doesn't use single-float intermediate results when
296 given mixed integer and double-float arguments, leading to better
297 precision. (reported by Bob Felts)
298 * bug fix: LOG with base zero returned values of inconsistent type.
300 changes in sbcl-1.0.20 relative to 1.0.19:
301 * minor incompatible change: OPTIMIZE qualities
302 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
303 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
304 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
305 * documentation: some slot access efficiency guidelines have been
306 added to the user manual.
307 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
308 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
309 as ASSOC and MEMEBER.
310 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
311 to more efficient EQ-comparison versions more often.
312 * optimization: enhanced derivation of DOLIST iteration variable type
314 * optimization: constant folding of simple (LIST ...) forms as DOLIST
316 * optimization: runtime lookup of function definitions can be
317 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
318 * optimization: compiler is able to derive the return type of
319 (AREF (THE STRING X) Y) as being CHARACTER.
320 * optimization: CLRHASH on empty hash-tables no longer does pointless
321 work. (thanks to Alec Berryman)
322 * optimization: REPLACE deftransforms don't punt when :START1 or
324 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
325 no longer emits a code deletion note. (reported by Andrew
327 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
328 functions with an odd number of &OPTIONAL arguments, a &REST
329 argument, and one or more &KEY arguments at the call site.
330 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
331 arguments appeared at call sites as well.
332 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
333 READ-LINE &co can no longer read from them afterwards. (reported
335 * bug fix: fixed #427: unused local aliens no longer cause compiler
336 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
338 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
339 alien-stack leakage. (reported by Andy Hefner)
340 * bug fix: PROGV signals an error when an attempt to violate declared
341 type of a variable or bind a constant is made.
342 * bug fix: SET signals an error when an attempt to violate declared
343 type of a variable is made.
344 * bug fix: restart computation during the execution of a restart
345 test function no longer causes infinite recursion. (reported by
347 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
348 signals an error. (thanks to Pierre Mai)
349 * bug fix: EXPORT left symbol unexported in conflict situations.
350 (thanks to Michael Weber)
351 * bug fix: correctly handle name conflicts involving CL:NIL.
352 (thanks to Michael Weber)
353 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
354 conflicts arising from USEing package with conflicting symbols
355 correctly. (thanks to Michael Weber)
357 changes in sbcl-1.0.19 relative to 1.0.18:
358 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
359 warnings that go otherwise unhandled will be muffled if they are
360 of the type that's the value of this variable.
361 * optimization: stack allocation is slightly more efficient on x86
363 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
364 where a raw slot always is initialized using the initform whose
365 type is not know sufficiently well a compile-time are now compiled
366 correctly. (reported by John Morrison)
367 * bug fix: compiler no longer makes erronous assumptions in the
368 presence of non-foldable SATISFIES types.
369 * bug fix: stack analysis missed cleanups of dynamic-extent
370 arguments in non-let-converted calls to local functions.
371 * improvements to the Windows port:
372 ** adjusted address spaces for building on both Win32 and
373 Win64. (thanks for John Connors)
374 * fixed some bugs revealed by Paul Dietz' test suite:
375 ** interval arithmetic during type derivation used inexact integer
376 to single-float coercions.
377 ** arithmetic operations involving large integers and single
378 floats give the same results in compiled and interpreted code.
379 ** deriving the result type of COERCE no longer signals an error
380 if the derived type of the second argument is a MEMBER type
381 containing invalid type specifiers.
382 ** ADJOIN with constant NIL as second argument works correctly.
383 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
385 changes in sbcl-1.0.18 relative to 1.0.17:
386 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
387 profiles only the current thread.
388 * minor incompatible change: changes to SYMBOL-VALUE of constants
389 defined with DEFCONSTANT now signal an error.
390 * enhancement: SB-SPROF now has support for wallclock profiling,
391 and is also able to profile specific threads. REPORT output
392 has also additional sorting options.
393 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
395 * optimization: structure allocation has been improved
396 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
397 ** out of line constructors are ~10% faster.
398 ** inline constructors are ~15% faster.
399 ** inline constructors are capable of dynamic extent allocation
400 (generally on x86 and x86-64, in some cases on other platforms
402 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
404 * optimization: file compiler is now able to coalesce non-circular
405 lists, non-base strings, and bit-vectors. Additionally, constants
406 are never referenced through SYMBOL-VALUE at runtime.
407 * optimization: code defining methods on PRINT-OBJECT (and other
408 generic functions in the COMMON-LISP package) now loads faster.
409 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
411 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
412 fasl is now deleted. (reported by Attila Lendvai)
413 * bug fix: READ-LINE always returned NIL for the last line in files.
414 (reported by Yoshinori Tahara)
415 * bug fix: more accurate disassembly annotations of foreign function
416 calls. (thanks to Andy Hefner)
417 * bug fix: trimming non-simple strings and non-string string
418 designators when the there is nothing to trim works properly.
419 (thanks to James Knight)
420 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
421 (thanks to Travis Cross)
422 * fixed some bugs revealed by Paul Dietz' test suite:
423 ** NIL is a valid function name (regression at 1.0.13.38)
424 ** FILL on lists was missing its return value (regression at 1.0.12.27)
425 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
426 fill pointers properly (regression at 1.0.12.23)
427 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
428 displacement indices properly (regression at 1.0.12.23)
430 changes in sbcl-1.0.17 relative to 1.0.16:
431 * temporary regression: user code can no longer allocate closure
432 variable storage on stack, due to bug 419 without explicitly
433 requesting it. Please consult sbcl-devel for advice if you need to
434 use this feature in the meanwhile.
435 * new feature: runtime argument --control-stack-size can be used to
436 adjust thread default control stack size.
437 * enhancement: improved TIME output
438 ** all times are reported using the measured accuracy (milliseconds
439 for real and GC times, microseconds for everything else.)
440 ** processor cycle counts on x86 and x86-64.
441 ** interpreted forms are counted for both evaluator modes.
442 ** number of lambdas converted by the compiler is reported.
443 ** CPU percentage report (computed from real and total run time.)
444 ** more comprehensive run time reporting, using a condenced format
445 ** interperted form, lambda, and page fault counts are omitted
447 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
449 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
450 * optimization: two argument forms of LAST are upto ~10% faster
451 in normal SPEED policies.
452 * optimization: NCONC no longer needs to heap cons its &REST list
453 in normal SPEED policies.
454 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
456 * bug fix: UNION and NUNION work with :TEST-NOT once more,
457 regression since 1.0.9.1. (thanks to Eric Marsden)
458 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
459 since 1.0.15.36. (thanks to Paul Khuong)
460 * bug fix: LAST when always returned the whole list when given a bignum
461 as the second argument.
462 * bug fix: dynamic extent allocation of nested lists and vectors
463 could leak to otherwise accessible parts.
464 * bug fix: invalid optimization of heap-allocated alien variable
466 * bug fix: fasl header checking is less vulnerable to different
467 platform word lengths.
468 * bug fix: more correct assembler syntax for GNU binutils
469 2.18.50.0.4 support. (thanks to Marijn Schouten)
470 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
472 * bug fix: the fopcompiler can handle LOCALLY forms (with no
473 declarations) successfully. (reported by Attila Lendvai)
475 changes in sbcl-1.0.16 relative to 1.0.15:
476 * minor incompatible change: revert the changes to sb-posix's error
477 signaling added in 1.0.14.
478 * minor incompatible change: change PROBE-FILE back to returning
479 NIL whenever we can't get a truename, as was the case before 1.0.14.
480 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
481 inherits from ERROR instead of just CONDITION.
482 * new feature: SB-INTROSPECT can provide source locations for instances
483 as well. (thanks to Tobias Rittweiler)
484 * optimization: binding special variables now generates smaller code
485 on threaded platforms.
486 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
487 and cases where no :TEST is given but the compiler can infer that
488 the element to search is of type (OR FIXNUM (NOT NUMBER)).
489 * optimization: better LOGNOT on fixnums.
490 * optimization: modular arithmetic for a particular requested width
491 is implemented using a tagged representation unless a better
492 representation is available.
493 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
494 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
495 for wellformedness and returns a specialized vector. (reported by
496 Francois-Rene Rideau)
497 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
498 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
499 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
500 unscheduling timer at the same time as another timer fires.
501 * bug fix: don't reschedule timers for dead threads.
502 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
503 * bug fix: copying output from RUN-PROGRAM to a stream signalled
504 bogus errors if select() was interrupted.
505 * enhancement: add support for fcntl's struct flock to SB-POSIX.
507 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
508 * enhancement: cleaner backtraces for interactive interrupts, as
509 well as other cases where the interesting frames used to be
510 obscured by interrupt handling frames.
511 * enhancement: untracing a whole package using (UNTRACE "FOO") is
512 now supported, and tracing a whole package using (TRACE "FOO") now
513 traces SETF-functions as well.
514 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
515 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
516 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
517 * unreadably printed representation of hash-tables now includes
519 * bug fix: partially fixed #188: type propagation from assignments
520 is now more efficient.
521 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
522 'DOUBLE-FLOAT) are not flushable.
523 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
524 full address of the object, and none of the tag bits.
525 * bug fix: readably printing hash-tables now respects other printer
526 control variables. (reported by Cedric St-Jean)
527 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
528 keyword with MAKE-HASH-TABLE.
529 * bug fix: export SB-POSIX:MKSTEMP.
530 * bug fix: SORT was not interrupt safe.
531 * bug fix: XREF accounts for the last node of each basic-block as
533 * bug fix: MAKE-INSTANCE optimizations interacted badly with
534 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
535 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
536 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
537 create obsolete instances in the case of redefinition or
538 obsoletion of a superclass. (thanks to Andy Hefner)
539 * bug fix: Support for the Alpha architecture has been revived; it had
540 suffered somewhat from lack of maintenance since sbcl-1.0.
541 * improvements to the Windows port:
542 ** The system detects the codepage to use at startup, instead of
543 using the value from the saved core. (thanks to Kei Suzuki)
545 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
546 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
547 (see documentation for details.)
548 * revived support for OpenBSD (contributed by Josh Elsasser)
549 * partially fixed bug #108: ROOM no longer suffers from occasional
550 (AVER (SAP= CURRENT END)) failures.
551 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
552 bodies no longer cause a WARNING to be signalled. (reported by
554 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
555 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
556 the bug in TRUNCATE.) (reported by Michael Weber)
557 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
558 no samples. (reported by Andy Hefner)
559 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
560 no longer appear as (NIL ...) frames in backtraces.
561 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
562 is now actually exported from SB-EXT as documented. (reported by
564 * bug fix: sb-aclrepl now correctly understands how to inspect
565 single-floats on 64-bit platforms where single-floats are not boxed.
566 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
567 yet been finalized. (reported by Levente Meszaros)
568 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
569 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
570 SBCL-specific optimize qualities.
572 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
573 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
574 an executable in the search path, and does so in the child
575 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
576 been removed; it can be found in the CVS history, for somebody who
577 needs that search behavior (see the manual).
578 * minor incompatible change: compiler policy re. weakening type
579 checks has changed: now type checks are weakened only if SAFETY < 2
581 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
582 unparsing of directory pathnames as files. Analogously,
583 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
584 filename to parse into a directory pathname.
585 * enhancement: implicit generic function creation now signals a
586 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
587 users can bind handlers for and muffle around calls to LOAD.
588 * enhancement: RUN-PROGRAM allows unicode arguments and environments
589 to be used (using the default stream external format), and allows
590 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
591 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
592 strings and vectors whose element-type or simplicity is not fully
593 known at compile-time.
594 * optimization: STRING-TRIM and related functions no longer allocate
595 a new string when no trimming needs to be performed. These functions
596 are also faster than before when the input string has been declared
598 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
599 * optimization: READ-LINE is significantly faster for files containing
601 * optimization: non-open coded uses of character comparison operators
602 (e.g. char=) no longer cons when called with more than one parameter
603 on platforms supporting dynamic-extent allocation.
604 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
605 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
606 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
608 * bug fix: some sequence functions elided bounds checking when
610 * bug fix: too liberal weakening of union-type checks when SPEED >
612 * bug fix: more bogus fixnum declarations in ROOM implementation
615 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
616 * new feature: MAKE-HASH-TABLE now experimentally accepts a
617 :SYNCHRONIZED argument, which makes the hash-table safe for
618 concurrent accesses (but not iteration.) See also:
619 SB-EXT:WITH-LOCKED-HASH-TABLE, and
620 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
621 * optimization: CONCATENATE on strings is an order of magnitude faster
622 in code compiled with (> SPEED SPACE).
623 * optimization: SUBSEQ is ~50% faster on lists.
624 * bug fix: bug 417 fixed -- source location reporting is now more robust.
625 * bug fix: SUBSEQ on a list will now correctly signal an error if if
626 END is smaller then START.
627 * bug fix: SB-PROFILE will no longer report extra consing for nested
628 calls to profiled functions.
629 * bug fix: ROOM implementation had bogus fixnum declarations which
630 could cause type-errors when calling ROOM on large images.
631 * bug fix: if file compilation is aborted, the partial fasl is now
632 deleted, and COMPILE-FILE returns NIL as the primary value.
633 * bug fix: number of thread safety issues relating to SBCL's internal
634 hash-table usage have been fixed.
635 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
636 be returned from its body when the values were being returned
637 using unknown-values return convection and the W-P-O was wrapped
638 inside an UNWIND-PROTECT.
639 * bug fix: sb-posix should now compile again under Windows, enabling
642 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
643 * incompatible change: hash-table accessor functions are no longer
644 automatically protected by locks. Concurrent accesses on the same hash-table
645 from multiple threads can give inconsistent results or even corrupt the
646 hash-table completely. Multi-threaded applications should do their own
647 locking at the correct granularity. In the current implementation it is
648 still safe to have multiple readers access the same table, but it's not
649 guaranteed that this property will be maintained in future releases.
650 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
651 and will signal an error at runtime.
652 * enhancement: SB-THREAD package now exports a semaphore interface.
653 * enhancement: CONS can now stack-allocate on x86 and
654 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
656 * enhancement: nested lists can now be stack allocated on
657 platforms providing stack allocation support.
658 * enhancement: dynamic-extent support has been extended to support
659 cases where there are multiple possible sources for the stack
661 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
662 if the mutex is uncontested on Linux.
663 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
664 as specified by the CLHS. (thanks to Tobias Rittweiler)
665 * bug fix: NaN comparison now works on x86-64.
666 * bug fix: CLOSE :ABORT T on a stream with pending output now
668 * bug fix: instances of non-standard metaclasses using standard
669 instance structure protocol sometimes missed the slot type checks
671 * bug fix: known functions can be inlined in high-debug code.
673 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
674 * minor incompatible change: the MSI installer on Windows no longer
675 associates .lisp and .fasl files with the installed SBCL.
676 * minor incompatible change: :UNIX is no longer present in *FEATURES*
677 on Windows. (thanks to Luis Oliviera)
678 * new feature: SB-CLTL2 contrib module now implements
679 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
680 * optimization: scavenging weak pointers is now more efficient,
681 requiring O(1) instead of O(N) per weak pointer to identify
682 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
683 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
684 method bodies using constant slot-names when the first argument is
685 a specializer parameter for the method.
686 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
687 names, when the class of the instance is a direct instance
688 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
689 as slow as the constant slot-name case.)
690 * optimization: member type construction is now O(N) instead
692 * optimization: UNION and NUNION are now O(N+M) for large
693 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
695 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
696 now more readable in environments like Slime which display it.
697 (thanks to Tobias C. Rittweiler)
698 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
699 was able to infer the correct argument type for the object on which
700 the CAS operation was being performed.
701 * bug fix: copy propagation interfered with parallel assignment
702 semantics in local calls. (reported by Paul Khuong)
703 * bug fix: the signed modular fixnum shift compiled to wrong code on
704 x86 and x86-64. (spotted by a slight modification to some of
707 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
708 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
709 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
710 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
712 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
713 and generic function now print the object identity for anonymous
714 instances. (thanks to Attila Lendvai)
715 * enhancement: as an extension to MOP, SBCL now supports SETF of
716 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
717 (thanks to Attila Lendvai)
718 * enhancement: sb-cover has an alternate annotation mode which puts
719 the form annotations at the car of the form, instead of annotating
720 it completely. The benefit of this mode is that it shows better
721 which forms were instrumented by the compiler.
722 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
723 by printer control variables. (reported by Dan Corkill)
724 * bug fix: system leaked memory when delayed output was performed by
725 the OS in smaller chunks then expected. (thanks to David Smith)
726 * bug fix: system leaked memory when file streams were not closed
728 * bug fix: large objects written to slow streams that were modified
729 after the write could end up with the modified state written to
730 the underlying file descriptor.
731 * bug fix: multiple threads operating in parallel on the same stream
732 could cause buffer-overflows.
733 * bug fix: source location information is stored correctly
734 (broken since 1.0.6). This bug would generally show up as the
735 Slime debugger highlighting the wrong form when the sldb-show-source
737 * bug fix: Branch forms are again annotated as branches in the sb-cover
739 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
740 on x86-64. (reported by Gregory Vanuxem)
741 * bug fix: EQUALP could return wrong results for structures with raw
742 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
743 word). (reported by Vjacheslav Fyodorov)
744 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
747 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
748 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
749 atomic compare-and-swap operations on threaded platforms.
750 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
751 allows assigning a global minimum value to optimization qualities
752 (overriding proclamations and declarations).
753 * enhancement: closed over variables can be stack-allocated on x86
755 * performance bug fix: GETHASH and (SETF GETHASH) are once again
757 * optimization: slot definition lookup is now O(1). This speeds up
758 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
759 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
760 * optimization: ASSOC and MEMBER can now be open-coded for all
761 combinations of keyword arguments when second argument is constant
762 and SPEED >= SPACE. In other cases a specialized version is
764 * bug fix: using obsoleted structure instances with TYPEP and
765 generic functions now signals a sensible error.
766 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
767 (reported by Kristoffer Kvello)
768 * bug fix: backtrace construction is now more careful when making
769 lisp-objects from pointers on the stack, to avoid creating bogus
770 objects that can be seen by the GC.
771 * bug fix: defaulting of values in contexts expecting more than 7
772 variables now works on x86-64. (reported by Christopher Laux)
773 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
775 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
776 as the property-list of a symbol.
777 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
778 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
779 situations requested, are once again file-compileable. (reported
782 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
783 * MOP improvement: support for user-defined subclasses of
784 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
785 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
786 the expansion of DEFMETHOD, is responsible for generating a form
787 which creates a list of specializers when evaluated. Additional
788 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
789 debugging and introspective support.
790 * minor incompatible change: the (unsupported) spinlock interface
791 has changed: free spinlock now has the value NIL, and a held spinlock
792 has the owning thread as its value.
793 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
794 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
796 * enhancement: name of a socket-stream is now "a socket" instead of
798 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
799 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
800 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
801 for generic functions. (Thanks to Tobias C. Rittweiler)
802 * optimization: bignum printing speed has been improved by 20-40%
803 (depending on the bignum size.)
804 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
806 * bug fix: the cache used by the CLOS to store precomputed effective
807 methods, slot offsets, and constant return values is now thread and
809 * bug fix: generic function dispatch function updating is now thread
810 and interrupt safe (in the sense that the known issues have been
812 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
813 * bug fix: interning EQL-specializers is now thread and interrupt safe.
814 * bug fix: asdf systems with dependencies to the SB-POSIX or
815 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
816 * bug fix: interrupt safety of applicable method computation has been
819 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
820 * new contrib: sb-cover, an experimental code coverage tool, is included
822 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
823 significantly faster.
824 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
825 produce more efficient code on x86-64 if the compiler can determine
826 that (+ INDEX <constant>) does not require a bounds check and FOO
827 has an element type at least 8 bits wide.
828 * enhancement: a new, experimental synchronous timeout facility is
829 provided. Refer to SB-SYS:WITH-DEADLINE for details.
830 * enhancement: when a symbol name conflict error arises, the
831 conflicting symbols are always printed with a package prefix.
832 (thanks to Kevin Reid)
833 * enhancement: stepping is now once again supported on the SPARC and
834 MIPS platforms. (It is also now more likely to work on CheneyGC
836 * enhancement: sb-sprof can now also track and report accurate call
838 * bug fixes: the treatment of non-standard subclasses of
839 SB-MOP:SPECIALIZER is more correct.
840 * incompatible change: PURIFY no longer copies the data from the
841 dynamic space into the static and read-only spaces on platforms
842 that use the generational garbage collector
843 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
845 * bug fix: GC race condition occasionally resulting in crashes with
846 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
848 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
850 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
851 a STYLE-WARNING to be signalled (regression from 1.0.4.)
852 * bug fix: an asynchronous interrupt could previously leave the
853 system running with GC inhibited.
854 * bug fix: a DECLARE form evaluated at top-level now causes an error
855 rather than silently (or verbosely) returning NIL.
856 * bug fix: trying to dissassemble functions compiled at high DEBUG could
857 result in a error being signalled due to source form lookup errors.
858 (reported by Peter Graves)
860 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
861 * incompatible change: removed writer methods for host-ent-name,
862 host-ent-addresses -- changing the values did not update the DNS
864 * minor incompatible change: changed experimental JOIN-THREAD interface
865 * documentation: the manual now lists reader and writer methods
866 in class slot documentation sections. (thanks to Richard M Kreuter)
867 * documentation: unwinding from asyncronous events has been
868 documented as unsafe.
869 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
870 in multithreaded application code.
871 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
872 platforms. (thanks to James Anderson for the optimization hint)
873 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
875 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
876 variants no longer cons.
877 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
878 their NOT- variants no longer cons.
879 * optimization: EQUAL hash tables no longer use SXHASH for objects
880 of all data types, but instead use an EQL hash for types for which
881 EQUAL is the same as EQL.
882 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
883 are significantly faster.
884 * optimization: new STRING-OUTPUT-STREAM implementation is much
885 faster for large outputs and conses 30% less on average.
886 * enhancement: XREF information is now collected for references made
887 to global variables using SYMBOL-VALUE and a constant argument.
888 * enhancement: SIGINT now causes a specific condition
889 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
890 * bug fix: FILE-POSITION used to signal an error in some cases where
891 ANSI requires it to return NIL.
892 * bug fix: ADJUST-ARRAY is now interrupt-safe.
893 * bug fix: adding and removing fd-handlers is now interrupt-safe.
894 * bug fix: inlined calls to C now ensure 16byte stack alignment on
896 * bug fix: bad type declaration in the CLOS implementation has
897 been fixed. (reported by James Anderson)
898 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
899 (thanks to Lutz Euler)
900 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
901 required by the CLHS. (reported by Samium Gromoff)
902 * bug fix: dead unbound variable references now signal an error.
903 (reported by Marco Monteiro)
904 * bug fix: / with an unused value was being deleted in safe code.
905 (thanks to Marco Monteiro and Kevin Reid)
906 * bug fix: number of characters that can be written onto a single
907 line in a file is unlimited.
908 * bug fix: some GC deadlocks caused by asynchronous interrupts have
909 been fixed by inhibiting interrupts for when GC is disabled.
910 * bug fix: some interrupt safety issues with GETHASH, (SETF
911 GETHASH), CLRHASH and REMHASH have been fixed.
912 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
913 specifier no longer causes infinite recursion.
914 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
915 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
916 * bug fix: an &environment argument with the correct variable information
917 is passed to macros that are expanded during byte compilation.
918 (reported by Samium Gromoff)
919 * bug fix: a logic error was causing the hash function for bignums to
920 have excessive amounts of collisions. (reported by Faré Rideau)
921 * bug fix: modifying the contents of an array could change the return
922 value of SXHASH on that array, which is only allowed for strings
923 and bit vectors. (bug introduced in 0.9.16)
924 * improvement: the x86-64/darwin port now passes all tests (except
925 for the debugger tests) but should still be considered
926 experimental until this is fixed.
927 * improvement: a style-warning is signaled for CASE (etc) clauses with
928 duplicate keys. (thanks to Kevin Reid)
929 * improvement: macos/x86 and macos/x86-64 now use mach exception
930 handlers for dealing with illegal instructions (for trapping and
931 error handling) and memory protection violations (for GC).
933 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
934 * new platform: experimental support for x86-64/darwin (MacOS).
935 * incompatible change: the thread-safe (on most platforms) getaddrinfo
936 and getnameinfo sockets functions are used instead of gethostbyaddr
937 and gethostbyname, on platforms where the newer functions are available.
938 As a result, the ALIASES field of HOST-ENT will always be NIL on these
940 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
941 don't print the startup banner, but behave as if --noinform was passed
942 as a command line argument. (thanks to Kevin Reid)
943 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
944 * optimization: code using alien values with undeclared types is much faster.
945 * optimization: the compiler is now able to open code SEARCH in more cases.
946 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
947 * bug fix: using standardized COMMON-LISP special variables as loop
948 variables no longer signals bogus package lock violations. (reported
950 * bug fix: declaring local loop variables to be of a range-limited type
951 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
952 (reported by Andras Simon)
953 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
954 bugs remain on x86-64.)
955 * bug fix: the #= and ## reader macros now interact reasonably with
956 funcallable instances.
957 * bug fix: type-checks for function arguments were compiled using the
958 compiler policy settings of the wrong lexical environment (in 1.0.2
960 * bug fix: SHADOW accepts characters as string designators, as required
961 by the spec (thanks to Eric Marsden)
962 * bug fix: fixed GC safety issues when foreign functions are called with
963 non-base strings as arguments
964 * bug fix: more consistent error output for fatal-compiler-warnings, like
966 * bug fix: functions with proper names were showing up as NIL in
967 backtraces (reported by Edi Weitz, regression in 1.0.2)
969 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
970 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
971 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
972 produce more efficient code on the x86 if the compiler can determine
973 that (+ INDEX <constant>) does not require a bounds check and FOO
974 has an element type at least 8 bits wide.
975 * bug fix: references to undefined variables are handled the same way
976 in toplevel forms as in the normal compiler
977 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
979 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
980 (thanks to Jon Buller)
981 * improvement: faster compilation times for complex functions
982 * improvement: added readlink support to SB-POSIX (thanks to Richard
985 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
986 * improvement: experimental support for mach exception handling on
987 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
989 * improvement: support for GBK external format.
990 (thanks to Chun Tian (binghe))
991 * improvement: the debugger now displays variables that have been closed
992 over, in code compiled with (DEBUG 2) or higher
993 * improvement: support for executable cores on NetBSD (thanks to
995 * new feature: added a RESTART-FRAME debugger command
996 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
997 be used to provide an implementation for FILE-POSITION on Gray streams
998 (thanks to Eric Marsden)
999 * new feature: add syslog, openlog and closelog support to SB-POSIX
1000 (thanks to Richard Kreuter)
1001 * optimization: the function call overhead in code compiled with
1002 a high DEBUG optimization setting is significantly
1003 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1004 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1005 (thanks to Tony Martinez)
1006 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1007 evaluated once (reported by Kevin Reid)
1008 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1009 works again (thanks to Kilian Sprotte)
1010 * bug fix: an error is signaled for tagbodies with duplicate tags
1011 (thanks to Stephen Wilson)
1012 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1013 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1014 stack frames from alien callbacks.
1015 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1016 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1017 65k pages (thanks to David Woodhouse)
1018 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1020 changes in sbcl-1.0.1 relative to sbcl-1.0:
1021 * new platform: FreeBSD/x86-64, including support for threading.
1022 * new feature: the compiler stores cross-referencing information
1023 about function calls (who-calls), macroexpansion (who-macroexpands)
1024 and special variables (who-binds, who-sets, who-references) for code
1025 compiled with (< SPACE 3). This information is available through the
1026 sb-introspect contrib.
1027 * new feature: users may subclass SEQUENCE, and have instances of
1028 these classes interoperate with standard Common Lisp functions if
1029 a number of methods are defined. (This feature is experimental
1030 and the interface subject to change based on feedback from SBCL
1031 users and the general community)
1032 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1033 x86/x86-64, rather than the previous fixed depth of 8
1034 * improvement: another pthread back-end of mutex "pthread-futex".
1035 * bug fix: non-ascii command-line arguments are processed correctly
1036 (thanks to Yaroslav Kavenchuk)
1037 * bug fix: non-required arguments were not passed correctly when a method
1038 defined using DEFMETHOD was called from a mop-generated method using
1039 CALL-NEXT-METHOD (reported by Pascal Costanza)
1040 * bug fix: recursion is now permitted in accessors through
1041 SLOT-UNBOUND. (reported by Pascal Costanza)
1042 * bug fix: an error was signaled at startup if the HOME environment
1043 variable was defined, but had an empty value (reported by Peter Van Eynde)
1044 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1045 signal a PROGRAM-ERROR, not a BUG.
1046 * bug fix: SB-POSIX:READDIR works when built with large file support.
1047 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1049 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1050 for better automatic indentation support. (thanks to Matt Pillsbury)
1051 * optimization: loading generic functions no longer takes O(n^2) time,
1052 proportional to the amount of methods in the generic function
1053 (reported by Todd Sabin and Jeremy Brown)
1054 * optimization: the FIND and POSITION family of sequence functions
1055 are significantly faster on arrays whose element types have been
1057 * improvements to the Windows port:
1058 ** Intermittent heap corruption problems have been fixed. (thanks
1059 to Alastair Bridgewater)
1060 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1062 ** Lisp is able to unwind foreign exception frames from alien
1063 callbacks. (thanks to Alastair Bridgewater)
1065 changes in sbcl-1.0 relative to sbcl-0.9.18:
1066 * improvement: experimental support for threading on FreeBSD/x86.
1067 (thanks to NIIMI Satoshi)
1068 * improvement: runtime option --dynamic-space-size can be used
1069 to set the size of the dynamic space reserved on startup.
1070 * improvement: floating point modes in effect are now saved in
1071 core, and restored on startup.
1072 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1073 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1074 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1075 threads are not running after *SAVE-HOOKS* have run.
1076 * improvement: writes to CLOS instance slots are type-checked in code
1077 compiled with (SAFETY 3)
1078 * improvement: floating-point exception handling on FreeBSD (thanks to
1080 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1081 (thanks to Zach Beane)
1082 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1084 * improvement: added support for the Shift-JIS external format.
1085 (contributed by NIIMI Satoshi)
1086 * improvement: callbacks are supported on Linux/PPC. (thanks to
1088 * bug fix: compiler bug triggered by a (non-standard) VALUES
1089 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1090 * bug fix: file compiler no longer confuses validated and already
1091 dumped structurres. (reported by Kaersten Poeck)
1092 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1093 fill-pointer signals a type-error as required. (thanks to
1095 * bug fix: disassemly of funcallable instances works.
1096 * bug fix: single stepping on PPC.
1097 * bug fix: fix thread-safety problems in the type system (generally
1098 manifesting as nonsensical errors like "STRING is a bad type specifier
1099 for sequences" or "The value 1 is not of type FIXNUM")
1100 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1101 (reported by Josip Gracin)
1102 * bug fix: an error is signaled for attempts to displace arrays with
1103 incompatible element types (thanks to Mario Mommer)
1104 * bug fix: more correct handling of wide characters in debug info
1105 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1106 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1107 and don't cause extra consing
1108 * optimization: MAP and MAP-INTO are significantly faster on vectors
1109 whose elements types have been declared.
1110 * Improvements to SB-SPROF:
1111 ** Support for allocation profiling
1112 ** Reduced profiling overhead, especially for long profiling runs
1113 * Improvements to the Windows port:
1114 ** floating point exceptions are now reported correctly.
1115 ** stack exhaustion detection works partially.
1116 ** more accurate GET-INTERNAL-REAL-TIME.
1117 ** better RUN-PROGRAM behaviour in case of errors in the child
1119 ** PROBE-FILE now simplifies pathnames correctly.
1120 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1121 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1122 1st 1970 (although time timezone data will be unreliable).
1123 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1125 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1126 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1127 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1129 * bug fix: two potential GC deadlocks affecting threaded builds.
1130 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1131 returns T (reported by Anton Kazennikov)
1132 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1133 with non-variable places
1134 * bug fix: remove a race condition in the setting of
1135 funcallable-instance functions, this should make threaded CLOS
1136 code more stable against memory faults.
1137 * bug fix: corruption of specials when unbinding is interrupted by an
1138 asynchronous unwind (reported by Hannu Koivisto)
1139 * improvement: the debugger will now also display local variables that
1140 are only used once, for code compiled with a DEBUG optimization quality
1143 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1144 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1145 * incompatible change: External-format support for FFI calls. The
1146 SB-ALIEN:C-STRING no longer implies an ASCII
1147 external-format. Instead the string is subject to external-format
1148 conversion. Additionally return values of type SB-ALIEN:C-STRING
1149 are by default of type (SIMPLE-ARRAY CHARACTER), not
1150 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1151 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1152 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1153 * incompatible change: SB-EXT package no longer contains the
1154 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1155 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1157 * incompatible change: the single-stepper is no longer available
1158 on Alpha, Mips and Sparc platforms.
1159 * minor incompatible change: the direct superclasses of
1160 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1161 not (STANDARD-OBJECT FUNCTION). This makes the
1162 class-precedence-lists of GENERIC-FUNCTION and
1163 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1165 * minor incompatible change: the default stream external format on
1166 non-unicode SBCL is no longer detected from the locale, but is always
1168 * new feature: Add a version of evaluator that uses an interpreter instead
1169 of the compiler. EVAL still uses the compiler by default, to switch it
1170 to use the interpreter, set the value of the variable
1171 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1172 * minor incompatible change: the single-stepper REPL has been merged
1173 with the normal debugger (see the "Stepping" heading of the debugger help
1174 for more details). The debugger command STEP will no longer switch
1175 to the single-stepper REPL.
1176 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1177 for a type now works.
1178 * bug fix: loading of default sysinit file works. (thanks to Leonid
1180 * bug fix: better detection of circularities in the file-compiler.
1181 (reported by Marco Monteiro)
1182 * bug fix: the CL pathname functions now work with files that have
1183 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1184 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1185 systems with Unix98 pty semantics.
1186 * bug fix: ASDF-INSTALL will now work with bsd tar.
1187 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1189 * bug fix: timers expiring in dead threads no longer cause a
1190 type-error (reported by Paul "Nonny Mouse").
1191 * bug fix: thanks to more lightweight single-stepper instrumentation,
1192 code compiled with (DEBUG 3) will compile and execute significantly faster,
1193 and will have more accurate type-inferencing than before
1194 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1195 whose bindings are modified
1196 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1197 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1198 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1199 CL:SOFTWARE-VERSION with proper version
1201 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1202 * feature: implemented the READER-METHOD-CLASS and
1203 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1204 as specified by AMOP.
1205 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1207 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1208 (thanks to Lutz Euler)
1209 * optimization: hashing of general arrays and vectors has been
1210 improved. (reported by Any Fingerhut)
1211 * enhancement: SB-INTROSPECT is now able to find definitions of
1212 profiled functions. (thanks to Troels Henriksen)
1213 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1214 as well, allowing compiler-macros for SETF-functions to expand.
1215 * enhancement: step-instrumentation no longer wraps known
1216 single-value functions with multiple-value context, allowing
1217 better type inference.
1218 * fixed bug #337: use of MAKE-METHOD in method combination now works
1219 even in the presence of user-defined method classes. (reported by
1220 Bruno Haible and Pascal Costanza)
1221 * fixed bug #339(c): if there are applicable methods not part of any
1222 long-form method-combination group, call INVALID-METHOD-ERROR.
1223 (reported by Bruno Haible)
1224 * fixed bug #361: the :FUNCTION initarg in the protocol for
1225 initialization of methods can now be used to override
1226 internally-produced optimized functions. (reported by Bruno
1228 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1229 system-provided lambda expression no longer cause warnings about
1230 unbound #:|pv-table| symbols.
1231 * bug fix: improved the handling of type declarations and the
1232 detection of violations for keyword arguments with non-constant
1234 * bug fix: potentially erronous calls to PATHNAME and
1235 MERGE-PATHNAMES were being flushed in some cases.
1236 (reported by Richard Kreuter)
1237 * bug fix: compiled calls to TYPEP were mishandling obsolete
1238 instances. (reported by James Bielman and Attila Lendvai)
1239 * bug fix: format strings with ~> without matching ~< no longer
1240 trigger an AVER, but signal an understandable error instead.
1241 (reported by Antonio Martinez)
1242 * bug fix: specifying an output-file without a pathname-type for
1243 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1244 of input-file instead of "fasl". (reported by Robert Dodier)
1245 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1246 (reported by James Y Knight).
1247 * bug fix: compiler-macros-function did not consider the environment
1248 argument for shadowing by local functions.
1249 * bug fix: compiler-macros expansion was inhibited by local INLINE
1251 * bug fix: inline expansions of known functions were subject to
1252 step-instrumentation in high debug policies, leading to problems
1253 with type-inference.
1254 * bug fix: compiler failed to differentiate between different CONS
1255 types in some cases.
1256 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1257 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1258 * bug fix: #368: incorrect use of expressed vs. upgraded array
1260 * bug fix: #306a: more precise unions of array types.
1261 * thread-safety improvements:
1262 ** CONDITION-WAIT could return early on Linux, if the thread was
1263 interrupted and subsequently continued with SIGCONT.
1264 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1266 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1267 * added support for the ucs-2 external format. (contributed by Ivan
1269 * minor incompatible change: pretty printing of objects of type
1270 (cons symbol) is, in the default pprint-dispatch-table, now
1271 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1273 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1274 called later in a class's lifetime, possibly as late as when the
1275 first instance of the class is created. Previously,
1276 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1277 class became finalizeable.
1278 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1279 for multibyte external-format streams. (thanks to "vbzoli")
1280 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1281 with :ALLOCATION :CLASS inherited from superclasses of the
1283 * fixed bug: anonymous classes can now be created using the :NAME
1284 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1285 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1286 * fixed bug: core-files saved with :EXECUTABLE T can again be
1287 executed when SBCL_HOME isn't set. (reported by James Knight)
1288 * fixed bug: toplevel LOCALLY forms with declarations could
1289 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1290 * fixed bug: printing from several different threads using different
1291 values of *print-case* could cause invalid output, due to
1292 some internal special variables of the printer not being bound
1293 thread-locally (reported by Max Mikhanosha)
1294 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1295 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1296 (reported by Pascal Costanza)
1297 * fixed bug: FINALIZE-INHERITANCE is called from
1298 REINITIALIZE-INSTANCE on classes when the class has previously
1299 been finalized, as required by AMOP.
1300 * minor code generation optimizations:
1301 ** better register allocation in CLOS dispatching functions
1302 ** overflow detection when coercing signed bytes to fixnums on x86-64
1303 ** is now implemented with one IMUL instruction instead of three shifts
1304 ** more efficient bit-vector access on x86 and x86-64
1305 ** more efficient access to raw structure slots on x86 and x86-64
1306 * fixed some bugs revealed by Paul Dietz' test suite:
1307 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1308 return its argument.
1310 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1311 * feature: thread support on Solaris/x86, and experimental thread support
1313 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1315 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1316 default method for SLOT-UNBOUND.
1317 * minor incompatible change: prevent the user from specializing the
1318 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1319 somewhat counter to the intent of the protocol, I (CSR) think, and
1320 additionally it just doesn't work in SBCL as currently
1321 implemented, thanks to optimizations (that are always valid for
1322 the other three SLOT-VALUEish functions, but not for the setter).
1323 * minor incompatible change: the :SB-LDB feature is now enabled by
1324 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1325 the low-level debugger.
1326 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1327 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1328 down to an SB-SYS:FD-STREAM, allowing direct communication in
1329 more cases. Temporary files and pipes are still used as a fallback
1331 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1332 on the non-reentrant C functions localtime(3) and gmtime(3).
1333 * bug fix: global optimization policy was not visible in LOCALLY and
1335 * bug fix: class objects can be used as specializers in methods.
1336 (reported by Pascal Costanza)
1337 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1338 with a type error. (reported by blitz_ on #lisp)
1339 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1340 failed with a type error. (reported by Pascal Bourguignon)
1341 * bug fix: merging pathnames against defaults with :DIRECTORY
1342 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1343 (reported by James Y Knight)
1344 * bug fix: saving large (>2GB) cores on x86-64 now works
1345 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1346 constant index and a (simple-array (signed-byte 32)) array
1347 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1348 error (patch by Robert J. Macomber)
1349 * feature: TIME now displays the amount of run-time spent in GC
1350 * bug fix: The debugger now does a better job of respecting
1351 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1352 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1353 workaround for bug 403.)
1354 * bug fix: floating point exception handling now works on FreeBSD
1355 (thanks to NIIMI Satoshi)
1356 * fixed some bugs revealed by Paul Dietz' test suite:
1357 ** MISC.641: LET-conversion were not supposed to work in late
1359 * improvements to the Win32/x86 port:
1360 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1361 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1363 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1364 * new feature: source path information is generated for macro-expansion
1365 errors for use in IDE's like Slime (thanks to Helmut Eller)
1366 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1367 * bug fix: compiler does not loop forever on an invalid type in
1369 * improvement: compilation of most CLOS applications is significantly
1371 * optimization: added a limited bytecode compiler for simple toplevel
1372 forms, speeding up compilation and FASL loading
1373 * bug fix: the statistical profiler now properly distinguishes anonymous
1376 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1377 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1378 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1379 into subclasses of the finalized class. Now user calls to
1380 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1381 system return before any subclasses are finalized.
1382 * minor incompatible change: The reader no longer ignores errors
1383 regarding non-existent packages in #+ and #- feature tests.
1384 * new feature: command line options --no-sysinit, --no-userinit to
1385 inhibit loading the corresponding init files
1386 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1387 for sending data through UDP sockets (thanks to François-René Rideau)
1388 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1389 error is signalled instead (thanks to François-René Rideau)
1390 * minor incompatible change: Error signalling behaviour of lexical
1391 operations violating package locks has changed slightly. Refer to
1392 documentation on package locks for details.
1393 * enhancement: EVAL can process IF-expressions without resorting to the
1395 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1396 constant-folded properly due to missing out-of-line functions.
1397 (reported by Mika Pihlajamäki)
1398 * bug fix: LISTEN sometimes returned T even in cases where no data was
1399 immediately available from the stream
1400 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1401 were reversed. (reported by Levente Mészáros)
1402 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1403 when given a vector displaced to an adjustable array. (reported
1405 * bug fix: derivation of float boundaries from numbers outside the
1406 appropriate float range (reported by John Wiseman)
1407 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1409 * bug fix: precomputing cache entries for generic functions with
1410 some subclasses of specializers as yet invalid does not attempt to
1411 fill a cache line with a negative offset. (reported by Levente
1413 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1414 allow condition class objects as arguments to DOCUMENTATION and
1415 (SETF DOCUMENTATION); only find and set documentation for
1416 structure names for the STRUCTURE doc-type. (suggested by Gary
1418 * improvements to the Win32/x86 port:
1419 ** (user-homedir-pathname) and default initialization file
1420 locations now know about the user's "Documents and Settings"
1421 directory (thanks to Yaroslav Kavenchuk)
1422 ** run-program is implemented (thanks to Mike Thomas)
1423 ** sockets support (thanks to Timothy Ritchey)
1424 ** better backtrace support (thanks to Alastair Bridgewater)
1425 ** sb-grovel supported
1426 ** asdf-install and sb-posix work somewhat
1427 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1428 * improvements to the Solaris/x86 port:
1429 ** works on Solaris 11/Solaris Express
1430 ** floating-point exception handling support
1431 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1432 * fixed some bugs revealed by Paul Dietz' test suite:
1433 ** REMOVE-METHOD returns its generic function argument even when
1434 the method is not one of the generic functions' methods.
1435 ** objects declared as MEMBER types can be admissible arguments to
1436 structure accessors.
1437 ** printing characters should simply be printed by the FORMAT ~:C
1439 ** compiler failure when compiling functions with hairy constant
1440 defaults for optional parameters.
1441 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1442 function, which is already optimized.
1444 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1445 * new platform: experimental support for SBCL x86/Darwin, including
1446 MacOS X 10.4.5 on Intel.
1447 * new feature: Unicode character names are now known to the system
1448 (through CHAR-NAME and NAME-CHAR).
1449 * new feature: the filesystem location of SBCL's core file is
1450 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1451 * minor incompatible change: the contrib modules SB-POSIX and
1452 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1453 this change is to make it easier to distribute
1454 location-independent binaries.
1455 * bug fix: as implied by AMOP, standardized classes no longer have
1456 slots named by external symbols of public packages. (reported by
1458 * optimization: calling functions via constant symbols -- as in
1459 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1460 via the function object as in (FUNCALL #'FOO).
1461 * optimization: on x86 and x86-64, the calling convention for the
1462 case of dealing with an unknown number of values has been altered
1463 to be friendlier to the prediction heuristics implemented,
1464 particularly with reference to CALL and RET pairing. (thanks to
1465 Alastair Bridgewater)
1466 * optimization: on x86, the code for access to thread-local symbol
1467 values and binding/unbinding of thread-local symbols is smaller.
1468 (thanks to Alastair Bridgewater)
1469 * enhancement: CONSTANTP is now able to determine constantness of
1470 more complex forms, including calls to constant-foldable standardized
1471 functions and some special forms beyond QUOTE.
1472 * fixed bug: occasional GC crashes on Solaris/x86
1473 * optimization: x86-64 supports stack allocation of results of simple
1474 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1475 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1476 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1477 (thanks to James Knight)
1478 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1479 by the standard (thanks to Stephen Compall)
1481 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1482 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1483 be used for bundling the runtime and the core file into one
1484 executable binary. This feature is not currently supported on all SBCL
1485 platforms. (thanks to James Bielman and NIIMI Satoshi)
1486 * new feature: a generational or ephemeral garbage collector is now
1487 the default on the PowerPC platform (both Linux and Darwin). The
1488 old Cheney (stop and copy) collector is a build-time option.
1489 (thanks to Cyrus Harmon, after Raymond Toy)
1490 * minor incompatible change: the method by which SBCL finds its
1491 contributed modules has changed; it no longer relies on symbolic
1492 links from an $SBCL_HOME/systems directory, but searches directly
1493 in the subdirectories of $SBCL_HOME.
1494 * enhancement: the dynamic heap size on the Linux/PPC platform is
1495 markedly larger, even using the older Cheney garbage collector.
1496 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1497 certain complicated string types. (reported by Gary King)
1498 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1499 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1500 * fixed bug: DOCUMENTATION on structures no longer signals an
1501 error if no structure type of the right name exists. (reported by
1503 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1504 applicable non-standard primary or :AROUND method on
1505 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1506 no longer cause unbound variable errors. (reported by Kevin Reid)
1507 * fixed bug: metacircle resolution in cases where methods have slots
1508 added before the slots from STANDARD-METHOD. (reported by Jean
1510 * fixed bug: the Power PC assembly code for calling into lisp
1511 sometimes computed its return address wrongly (depending on the
1512 15th bit of the address, fixed at link time). This used to
1513 manifest itself in a segmentation violation while building PCL.
1514 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1515 many others over the years)
1516 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1517 the stack pointer (treated as a signed integer) is negative.
1518 (thanks to Peter van Eynde)
1520 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1521 * new platform: experimental support for the Windows operating
1522 system has been added. (thanks to Alastair Bridgewater)
1523 * fixed several bugs in and robustified the PPC FFI (including
1524 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1525 * bug fix: allow non-simple string symbol names (reported by Paul
1527 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1529 * bug fix: saving a core corrupted callbacks on x86/x86-64
1530 * bug fix: closed a loophole in metacircularity detection and
1531 grounding in the PCL implementation of CLOS.
1532 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1533 (reported by tomppa on #lisp)
1534 * bug fix: clear the direction flag on Lisp -> C transitions, as
1535 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1536 (reported by Andrej Grozin and Hendrik Maryns)
1537 * optimization: major improvements to GC efficiency on GENCGC platforms
1538 * optimization: faster implementation of EQUAL
1539 * optimization: emit more efficient opcodes for some common
1540 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1541 * optimization: several other minor code-generation improvements on x86-64
1542 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1544 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1545 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1546 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1547 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1548 by ANSI as a generic function, and for consistency (SETF
1549 GENERIC-FUNCTION-NAME) is treated likewise.
1550 * fixed bug #233.b: make constraint propagation notice when a variable
1551 value is changed after it is referenced but before it is used
1552 * fixed bug #296: no more arbitrary behaviour from filenames with
1553 odd characters as --load arguments. (reported by Adam Warner)
1554 * fixed bug #390: :CHARACTER-SET pathname components now work as
1555 expected. (reported by Tim Daly Jr)
1556 * fixed bug #391: complicated :TYPE intersections in slot
1557 definitions no longer cause an error in PCL internals.
1558 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1559 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1560 returning the number of octets which would be written to the
1561 file-stream. (thanks to Robert J. Macomber)
1562 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1563 arguments to the internal stream functions. (thanks to David
1565 * bug fix: stack exhaustion is now handled on the x86/SunOS
1566 platform. (thanks to Daisuke Homma)
1567 * bug fix: more accurate ROOM results on GENCGC platforms
1568 * optimization: improved type inference for arithmetic-for
1569 index variables in LOOP
1570 * optimization: faster floating-point SQRT on x86-64
1571 * fixed some bugs revealed by Paul Dietz' test suite:
1572 ** DOCUMENTATION returns NIL instead of "" for method combinations
1573 that don't have a docstring
1575 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1576 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1577 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1578 therefore are not customizeable by user code (as seems to be at
1579 least permitted and maybe required by AMOP). As a consolation,
1580 however, the SBCL implementation of these functions now calls
1581 REINITIALIZE-INSTANCE as specified by AMOP.
1582 * bug fix: slot-definition documentation is propagated as per ANSI
1583 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1584 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1585 argument, as do eslotd initialization methods. (from Pascal
1586 Costanza's "Closer" project)
1587 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1588 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1590 * bug fix: it is now possible to have more than one subclass of
1591 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1592 (reported by Bruno Haible, Pascal Costanza and others)
1593 * bug fix: the dependent update protocol now works for generic
1594 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1595 and Pascal Costanza)
1596 * bug fix: condition-class instances corresponding to
1597 DEFINE-CONDITION forms are now created eagerly. (reported by
1598 Kalle Olavi Niemitalo on comp.lang.lisp)
1599 * bug fix: floating point printing is more accurate in some
1600 circumstances. (thanks to Simon Alexander)
1601 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1602 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1603 * bug fix: callbacks on OS X now preserve stack-alignment.
1604 (thanks to Cyrus Harmon)
1605 * enhancement: the x86-64 disassembler is much better at
1606 disassembling SSE instructions. (thanks to Lutz Euler)
1607 * enhancement: CHECK-TYPE on a variable now allows type-inference
1608 (thanks to Vincent Arkesteijn)
1609 * optimization: improved performance of EUC-JP external format.
1610 (thanks to NIIMI Satoshi)
1611 * optimization: performance improvements to IO on file streams of
1612 :ELEMENT-TYPE CHARACTER
1613 * optimization: much faster memory allocation on x86-64
1614 * optimization: faster garbage collections (latency and throughput)
1616 * optimization: faster bignum subtraction and division on x86 and x86-64
1617 (thanks to Lutz Euler)
1618 * optimization: more accurate type-inference for loops that use a
1619 floating point index variable or a negative step.
1621 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1622 * bug fix: add a workaround to SBCL looping infinitely at startup on
1623 Linux kernels with apparently buggy implementations of personality().
1624 (thanks to Svein Ove Aas)
1625 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1626 (thanks to David Lichteblau)
1627 * bug fix: local bindings shadow global symbol macros for
1628 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1629 * optimization: non-open coded uses of numeric comparison operators
1630 (e.g. >) no longer cons when called with more than one parameter
1631 on platforms supporting dynamic-extent allocation.
1632 * enhancement: saving cores with foreign code loaded is now supported
1633 on MIPS/Linux in addition to the previously supported platforms.
1634 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1635 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1636 * bug fix: division by zero in sb-sprof when no samples were collected
1637 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1638 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1639 (thanks to Alexander Kjeldaas)
1640 * new feature: ignoring the return values of destructive functions
1641 like NREVERSE signals a compile-time style-warning.
1642 (thanks to Kevin Reid)
1643 * new feature: the alignment of alien structure fields can be
1644 explicitly specified. (thanks to Cyrus Harmon)
1645 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1646 and dump core on SIGQUIT
1648 ** incompatible change: threads do not inherit values of specials
1649 from their parents (see manual)
1650 ** bug fix: threads stacks belonging to dead threads are freed by the
1651 next exiting thread, no need to gc to collect thread stacks anymore
1652 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1653 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1654 ** bug fix: dynamic variable and thread start related gc lossage
1655 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1657 * fixed some bugs revealed by Paul Dietz' test suite:
1658 ** SUBTYPEP is slightly more accurate on heinously complicated
1659 CONS types where some of the members have uncertain (in the
1660 NIL, NIL sense) type relationships to each other.
1662 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1663 no longer purified unless :PURIFY T is explicitly specified.
1664 ** Non-purified cores are significantly smaller than before
1666 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1667 * new feature: timers based on Zach Beane's excellent timer package
1668 * added support for the following external formats: koi8-u,
1669 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1670 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1671 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1672 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1673 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1674 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1675 * incompatible change: a threaded SBCL will no longer revert to
1676 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1677 * bug fix: interrupts are disabled until startup is complete; no
1678 more sigsegvs when receiving a signal to soon
1679 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1681 * bug fix: add a workaround for the memory randomization features in
1682 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1683 workaround will only be in effect on systems with the proc filesystem
1685 * bug fix: printing objects of type HASH-TABLE signals a
1686 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1688 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1689 have been running for over 50 days. (reported by Gilbert Baumann)
1690 * bug fix: the logic for getting names of functions gets less
1691 confused when confronded with alternate-metaclass
1692 funcallable-instances. (reported by Cyrus Harmon)
1693 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1694 consistent, even on internal alternate-metaclass objects.
1695 * bug fix: flush closure information collected by physical
1696 environment analysis prepass before the main pass. (bug reported
1698 * bug fix: compiler pack phase does not modify a hashtable, which is
1699 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1700 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1701 may be expanded. (reported by Alexander <malishev> on c.l.l)
1702 * bug fix: no more style-warnings when compiling code using LAST
1703 under high-SPEED low-SPACE settings. (reported by David Wallin
1705 * fixed some incompatibilities between SBCL's MOP and the MOP
1707 ** the METAOBJECT class is now implemented;
1708 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1709 STANDARD-OBJECT, as required;
1710 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1711 now compatible (as required by VALIDATE-SUPERCLASS); there
1712 remains a consistency requirement over the presence or absence
1713 of the FUNCTION class in the superclass list of finalized
1714 classes; see the manual for more details;
1715 ** the :AROUND method for COMPUTE-SLOTS on
1716 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1717 requested slot ordering.
1719 ** bug fix: parent thread now can be gc'ed even with a live
1721 ** bug fix: binding a special with PROGV to no value is not
1722 the same as that symbol not having been bound (thanks to
1724 ** bug fix: binding specials is thread safe (thanks to
1726 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1727 and SB-SPROF do not die with 'no handler for signal XX in
1728 interrupt_handle_now(..)' anymore
1729 ** bug fix: WITH-TIMEOUT works with multiple running threads
1730 * fixed some bugs revealed by Paul Dietz' test suite:
1731 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1732 the :method-class keyword argument.
1734 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1735 * new port: the Solaris operating system on x86 processors is now
1736 mostly supported, though some rough edges in the environment
1737 remain. (thanks to Daisuke Homma)
1738 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1739 space for its heap. (thanks to Thiemo Seufer)
1740 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1741 closures. (thanks to Thiemo Seufer)
1742 * minor incompatible change: eof selects abort in the debugger.
1743 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1744 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1746 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1747 deprecated, and will go away in a future revision of SBCL.
1748 * minor incompatible change: GC-ON and GC-OFF are no longer
1749 implemented with a counter, it does not matter how many times gc
1750 is switched on or off
1751 * bug fix: discriminating functions for generic function classes
1752 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1753 make invalid assumptions about method precedence order. (reported
1755 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1756 platforms. (thanks to Thiemo Seufer)
1757 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1758 * bug fix: correct pathname printing with printer escaping is on.
1759 (thanks to Kevin Reid)
1760 * bug fix: complex VOP definitions in "user-space" no longer trigger
1761 package locks. (reported by Zach Beane)
1762 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1763 for funcallable-instance functions, and is no different from
1765 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1766 characters in the current readtable. (reported by Nicholas Neuss)
1767 * bug fix: SUBTYPEP on various CONS types returns more a more
1768 accurate acknowledgment of its certainty.
1769 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1770 lists in some cases. This partially fixes bug 384.
1771 * flush all standard streams before prompting in the REPL and the
1773 * bug fix: signal handling and triggering gc do not conflict
1774 directly anymore, in particular a high frequency sb-sprof does
1775 not prevent gc from running
1776 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1777 approximation for timezone and DST information between the
1778 universal time epoch and the smallest negative 32-bit time_t.
1779 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1780 year 1899 so long as the encoded time is non-negative.
1781 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1782 the correct answer for non-constant rotate amounts when performing
1783 an inline 32-bit rotation.
1785 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1786 there is only one thread in the session
1787 ** bug fix: memory leak for streams created in one thread and
1788 written to in another
1789 ** bug fix: lockup when compiled with gcc4
1790 ** bug fix: race that allows the gc to be triggered when gc is
1792 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1793 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1795 ** bug fix: finished threads are gc'ed properly
1796 * fixed some bugs revealed by Paul Dietz' test suite:
1797 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1798 has arguments with a different set of applicable methods from
1799 the orignal arguments.
1800 ** The type error thrown by MAP now has a correct expected-type
1802 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1803 name a compiled function.
1804 ** Three MISC tests where a large bignum was improperly coerced to
1805 a float that couldn't represent that bignum during type
1806 derivation were fixed.
1807 ** SUBTYPEP can now handle types involving the negation of a
1808 list-form FUNCTION type.
1809 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1810 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1811 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1813 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1814 * New feature: Experimental support for bivalent streams: streams
1815 opened with :element-type :default now allow character and binary
1816 (unsigned-byte 8) I/O
1817 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1818 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1820 * bug fix: no more highly sporadic "couldn't check whether ~S is
1821 readable" when reading a stream and an interrupt hits in the middle
1822 of a select system call
1823 * compiler better recognizes complex arrays (reported by Eduardo
1825 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1826 for variables on SBCL built with linkage-tables. (reported by Luis
1828 * various error reporting improvements.
1829 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1830 (Thanks to Thiemo Seufer)
1831 * bug fix: nested reader invokations work correctly
1832 * bug fix: it is possible to have simultaneous references to foreign
1833 code and foreign data with the same name.
1835 ** added x86-64 support
1836 ** incompatible change: the threading api now works with thread
1837 objects instead of thread ids
1838 ** bug fix: threads are protected from signals and interruption when
1839 starting up or going down
1840 ** bug fix: a race where an exiting thread could lose its stack to gc
1841 ** bug fix: don't halt on infinite error in threads if possible
1842 ** fixed numerous gc deadlocks introduced in the pthread merge
1843 ** bug fix: fixed thread safety issues in read and print
1844 ** bug fix: debugger doesn't hang on session lock if interrupted at
1845 an inappropriate moment
1846 ** bug fix: run-program is now thread safe(r)
1847 ** bug fix: inner with-recursive-lock no longer releases the mutex
1848 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1849 * fixed some bugs revealed by Paul Dietz' test suite:
1850 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1851 EXPECTED-TYPE slots filled.
1852 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1854 ** MAKE-CONDITION accepts classes as type-designators.
1855 ** COMPILE may never return NIL.
1856 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1857 range before calling Unix time functions
1859 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1860 * numerous signal handling fixes to increase stability
1861 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1862 * minor incompatible change: we now correctly canonize default
1863 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1864 as per the MOP, rather than the historical (INITARG INITFUNCTION
1865 INITFORM). (reported by Bruno Haible)
1866 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1867 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1868 for more information.
1869 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1870 pathname is a directory pathname.
1871 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1872 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1874 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1875 :METACLASS argument in addition to a class name. (reported by
1876 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1877 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1878 (reported by Bruno Haible; patch from Gerd Moellmann)
1879 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1881 * bug fix: more cleanups to the floating point exception handling on
1882 x86-64 (thanks to James Knight)
1883 * bug fix: the compiler does not try to constant fold calls of
1884 COERCE and BIT-* functions when they return freshly constructed
1885 objects. (reported by Nikodemus Siivola and Paul Dietz)
1886 * optimization: improved the MIPS versions of generic (in the
1887 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1888 * optimization: direct conversion of (unsigned-byte 32) to floats on
1889 the PowerPC platform.
1890 * optimization: structure instances with raw slots now use less
1891 memory, and probably show better memory locality. (thanks to
1893 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1894 are treated as requests for stack allocation on the x86-64,
1895 Alpha-32, PPC and SPARC platforms.
1896 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1897 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1898 (thanks to Tony Martinez)
1900 ** gcing a dead thread can no longer lead to lockups
1901 ** threads block signals until they are set up properly
1902 ** errno is no longer shared by threads
1903 ** interrupt-thread restores the eflags register on x86
1904 ** fixed some lockups due to gc/thread interaction
1905 * fixed some bugs revealed by Paul Dietz' test suite:
1906 ** invalid dotted lists no longer raise a read error when
1907 *READ-SUPPRESS* is T
1908 ** signal an error if a symbol that names a declaration is used
1909 as the name of a type, or vice versa
1910 ** allow using the (declare (typespec var*)) abbreviation for
1911 (declare (type typespec var*)) with all type specifiers
1912 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1913 ** always return NIL from PPRINT-POP when OBJECT is NIL
1914 ** don't signal errors when pretty-printing malformed LABELS,
1915 FLET or MACROLET forms
1916 ** declarations in a DOLIST body are also in scope for the
1918 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1920 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1923 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1924 * fixed cross-compiler leakages that prevented building a 32-bit
1925 target with a 64-bit host compiler.
1926 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1927 opened with :IF-EXISTS :OVERWRITE.
1928 * fixed bug 281: error for an invalid qualifier in a short-form method
1929 combination method is not signalled until the faulty method is called.
1930 * bug fix: iteration variable type inferrer failed to deal with open
1931 intervals. (reported by Alan Shields)
1932 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1933 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1934 or more runtime options were provided to the sbcl binary.
1935 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1936 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1937 * optimization: there's now a fast-path for fixnum arguments in the
1938 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1939 * optimization: the code generated on x86-64 is more compact thanks
1940 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1941 * bug fix: floating-point exception handling now partly works on
1942 x86-64. (Thanks to James Knight)
1943 * improvement to the MIPS backend from Thiemo Seufer: C-style
1944 64-bit long long arguments and return values to alien functions
1946 * fixed some bugs revealed by Paul Dietz' test suite:
1947 ** the type-error signalled from WARN has a filled-in DATUM slot.
1948 ** the type-error required when a stream is not associated with
1949 a file has the stream as its datum.
1950 ** type-errors on single-floats on x86-64 no longer have
1951 :INVALID-OBJECT as the datum
1952 ** the type-errors signalled for invalid function names now have
1953 a correct expected type
1954 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1955 for structures defined with a :TYPE. Documentation strings for
1956 typed structures are no longer immediately discarded
1957 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1958 broadcast streams and synonym streams. FILE-LENGTH now also works
1959 on broadcast streams.
1961 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1962 * incompatible change: the --noprogrammer option, deprecated since
1963 version 0.7.5, has been removed. Please use the equivalent
1964 --disable-debugger option instead.
1965 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1967 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1968 inoperational for a while now) has been completely removed.
1969 * null lexical environments are now printed as #<NULL-LEXENV>,
1970 significantly reducing the amount of clutter in typical backtraces.
1971 * documentation on weak pointers, finalization, and after GC hooks
1972 has been added to the manual.
1973 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1974 as other specialized array types, is much faster. SUBSEQ and
1975 COPY-SEQ on such arrays have also been sped up.
1976 * optimization: EQL is now more efficient when at least other argument
1977 is known to be of type (OR FIXNUM (NOT NUMBER)).
1978 * fixed bug: the runtime is now less vulnerable to changes in the
1979 size of the SBCL object on OS X, and virtual memory is reserved for
1980 all spaces that need to be at a fixed address.
1981 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1982 * fixed bug: finalizers and after GC hooks that cause consing are now
1984 * fixed bug: compiler error messages and summaries are now printed to
1985 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1986 * fixed inference of the upper bound of an iteration variable.
1987 (reported by Rajat Datta).
1988 * fixed bug 211e: calling local functions with duplicated constant
1989 keyword argument no longer causes a bogus style warning about an
1991 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1992 ftype declaration to be disregarded. (reported by Dave Roberts)
1993 * fixed bug 373: caused by erronous compilation of references to alien
1994 variables in the runtime on ppc/darwin.
1995 * fixed bug 376: CONJUGATE type deriver.
1996 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1997 (reported by Baughn on #lisp)
1998 * a cross-compiler bug on non-x86 platforms has been identified and
1999 fixed. (thanks to Bruno Haible)
2000 * improvements to the MIPS runtime code for increased stability.
2001 (thanks to Thiemo Seufer)
2002 * increased the maximimum compact environment size to allow
2003 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2004 * partial workaround for bug 135: don't name the function we're
2005 calling for hairy cases of EVAL, so as not to accumulate one environment
2006 entry per call to EVAL. (reported by Kevin Reid)
2007 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2008 * optimization: type testing for non-vector arrays should be faster.
2009 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2011 * bug fix: setting 31st element of a bit vector to zero did not work
2013 * bug fix: redefining a class definition which failed due to a
2014 previous accessor / function clash now works (but see BUGS entry
2015 #380 for more problems in this area). (thanks to Zach Beane)
2016 * the long form of DEFINE-METHOD-COMBINATION disables method group
2017 checking when given a single method group with pattern *. (thanks
2019 * on x86 compiler supports stack allocation of results of simple
2020 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2021 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2022 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2023 * fixed some bugs related to Unicode integration:
2024 ** the restarts for recovering from input and output encoding
2025 errors only appear when there is in fact such an error to
2027 * fixed some bugs revealed by Paul Dietz' test suite:
2028 ** MISC.549 and similar: late transformation of unsafe type
2029 assertions into derived types caused unexpected code
2031 ** SCALE-FLOAT type deriver is less wrong.
2032 ** type derivers for EXP, LOG and similar functions compute result
2033 types for complex arguments better.
2034 ** (MISC.563) CONJUGATE type deriver works for very restricted
2036 ** out-of-line type testers for character strings are available.
2037 ** EQUAL compiler transform understands specialness of objects
2039 ** accessing double-floats stored on the stack now works on x86-64.
2040 ** debugger internals could sometimes create invalid lispobjs,
2041 resulting in GC crashes.
2042 ** MISC.548: type check weakening can convert required type into
2044 ** initialization forms for bindings are not in scope of free special
2047 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2048 * incompatible change: thread support for non-NPTL systems has
2049 been removed - locking is buggy and unreliable. A threaded
2050 SBCL build will now warn at startup and refuse to create
2051 new threads, unless futex support is detected in the kernel
2052 * incompatible change: the top level REPL now has only an ABORT
2053 restart associated with it, not TOPLEVEL and ABORT as it used to.
2054 TOP and TOPLEVEL are now available as debugger commands for
2055 returning to the top level.
2056 * incompatible change: forms evaluated in the REPL now use the
2057 global optimization policy.
2058 * incompatible change: user- and system-initialization files are
2059 no longer processed with LOAD, but by READ and EVAL; hence the
2060 global optimization policy, startup package, readtable, etc,
2062 * The .fasl file format number has been incremented because of
2063 various incompatible changes.
2064 * internal entry point details and argument counts no longer appear
2065 in backtraces unless explicitly requested by setting
2066 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2067 * built-in and standard functions no longer have names like "top
2068 level local call to FOO".
2069 * fixed bug 32: functions defined in non-null lexical environments
2070 now have more legible printed representation
2071 * fixed bug 33: functions defined in non-null lexical environemnts
2072 are now more amenable to inspection by INSPECT.
2073 * workaround for bug 354: XEPs no longer appear in backtraces unless
2074 explicitly requested.
2075 * fixed bug: receiving the signal which results from attempting to
2076 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2077 SIGBUS on other *BSDs) on architectures where the C stack is also
2078 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2079 to debugger confusion or wild instability. (reported by Cheuksan
2081 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2082 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2083 (reported by Lutz Euler)
2084 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2085 compiled code. (reported by Paul Dietz)
2086 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2087 specialized parameter to a method to something that is not TYPEP
2088 the specializer is now possible.
2089 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2090 face of package deletion.
2091 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2092 pathname. (reported by David Tolpin on c.l.l)
2093 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2094 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2095 STRING-STREAM and FILE-STREAM, respectively.
2096 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2097 than previously; in addition, its test suite is now run on build.
2098 (thanks to Luke Gorrie)
2099 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2100 * optimization: added a immediate representation for single-floats
2102 * fixed some bugs revealed by Paul Dietz' test suite:
2103 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2104 ** attempting to create a package with a colliding nickname causes
2105 correctable errors to be signalled.
2106 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2107 ** several x86-64 backend bugs related to sign-extension of immediate
2110 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2111 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2112 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2114 * fixed loading of multiply forward-referenced layouts.
2115 (thanks to Cheuksan Wang)
2116 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2117 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2118 (see COMPILE-FILE documentation for details of :PRINT option)
2119 * fixed bugs 19 and 317: fixed-format floating point printing is
2120 more accurate. This also fixes a bug reported by Adam Warner
2121 related to the ~@F format directive.
2122 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2124 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2125 dispatch table if the from-char is a dispatch macro character.
2126 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2127 by the wordsize no longer ignore the last word. (reported by Lutz
2129 * fixed bug in type checking of dynamic-extent variables. (reported
2131 * optimization: sequence traversal functions use their freedom to
2132 coerce function designators to functions.
2133 * optimization: code with many calls to CLOS methods specialized on
2134 CLOS classes has had redundant type checks removed.
2135 * optimization: added declarations to speed up operations that access
2136 the internal character database (for example STRING-UPCASE)
2137 * optimization: comparison operations between floats and sufficiently small
2138 fixnums no longer create extra rationals
2139 * fixed some bugs related to Unicode integration:
2140 ** portions of multibyte characters at the end of buffers for
2141 character-based file input are correctly transferred to the
2142 start of the buffer at the next read.
2143 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2144 passing it through to OPEN.
2145 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2146 argument given to internal calls to OPEN.
2147 ** the built-in comment readers (introduced by character sequences
2148 ";" and "#|") are more forgiving to encoding errors; they will
2149 STYLE-WARN and then attempt to resync the stream at a character
2150 boundary. (thanks to Teemu Kalvas)
2151 * fixed some bugs in the x86-64 port:
2152 ** Negative short int return values from c-calls are sign-extended
2154 ** The stack is aligned to 16-bytes for c-calls, as required by
2155 the ABI. (thanks to Cheuksan Wang)
2156 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2157 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2158 for enums. (thanks to Vincent Arkesteijn)
2159 ** Multiple small optimizations and bugfixes for floating point
2161 * fixed some bugs revealed by Paul Dietz' test suite:
2162 ** Space, Tab, Linefeed, Return and Page have the invalid
2163 secondary constituent character trait.
2164 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2166 ** WITH-INPUT-FROM-STRING should only update the index place on
2168 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2169 works more reliably.
2170 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2171 with one-dimensional arrays with fill pointers.
2172 ** TYPE-OF failed on a complex with an integer realpart and a
2174 ** compiler failure during type inference for the code of form
2175 (IF (EQL X (THE ...)) ...) (MISC.535).
2177 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2178 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2179 platform. The port supports 61-bit fixnums, large memory spaces
2180 and reloading shared object files.
2181 * enhancement: saving cores with foreign code loaded is now
2182 supported on ppc/Darwin in addition to the previously supported
2184 * enhancement: the statistical profiler now walks deeper into the
2185 call stack for more meaningful call-graphs and accrued time
2186 reports (x86/x86-64 only). It also now reports time spent in
2188 * enhancement: it is now possible to trace most individual methods
2189 of a generic function in addition to tracing the generic function
2191 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2193 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2194 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2195 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2196 produces an error. (thanks to Vincent Arkesteijn)
2197 * bug fix: NAMESTRING on pathnames with :WILD components in their
2198 directories works correctly. (thanks to Artem V. Andreev)
2199 * fixed bug 125: compiler preserves identity of closures. (reported
2201 * bug fixed: functions with &REST arguments sometimes failed with
2202 "Undefined function" when compiled with (DEBUG 3). (reported by
2204 * bug fix: overflow during compiling of setting element of a bit
2205 vector with constant index and value. (reported by Timmy Douglas)
2206 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2207 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2208 * on x86 compiler partially supports stack allocation of dynamic-extent
2210 * GO and RETURN-FROM do not check the extent of their exit points
2211 when compiled with SAFETY 0.
2212 * fixed some bugs related to Unicode integration:
2213 ** encoding and decoding errors are now much more robustly
2214 handled; it should now be possible to recover even from invalid
2215 input or output to the terminal. (thanks to Teemu Kalvas)
2216 ** provided a first cut at implementing STRING-TO-OCTETS and
2217 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2218 ** altered the SB-MD5 contributed module slightly, changing the
2219 interface just enough for it to be supportable for builds where
2220 lisp characters are not eight bits.
2221 * fixed some bugs revealed by Paul Dietz' test suite:
2222 ** the FORMATTER-generated functions for ~V[ conditionals require
2223 the correct number of arguments.
2224 ** READ-FROM-STRING returns the mandated second value when applied
2225 to displaced strings.
2226 ** the #\Rubout and #\Backspace characters are treated as invalid
2227 constituent characters by the tokenizer.
2229 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2230 * new feature: reloading changed shared object files with
2231 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2232 * new feature: references to foreign variables and functions
2233 can now be compiled and loaded before the corresponding shared
2234 object file is loaded, as long as the foreign definitions are
2235 available at runtime.
2236 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2237 supported. (thanks to Dan Debertin)
2238 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2239 just keywords. (thanks to Vincent Arkesteijn)
2240 * fixed bug #331: structure-class instances corresponding to
2241 DEFSTRUCT forms are now created eagerly.
2242 * fixed bug #345: backtraces from calls to undefined functions work
2243 on x86 as well. Related bug #61 is now also partially fixed on x86
2244 (backtraces from throws to unknown catch tags.)
2245 * bug fix: lambda-list parsing is now stricter vrt. order and number
2246 of lambda-list keywords.
2247 * bug fix: as specified by AMOP, an error is signalled if a
2248 class-option appears multiple times in a DEFCLASS form. (reported
2250 * bug fix: RANDOM can be compiled when the compiler derives the type
2251 of its numeric argument as a disjoint set of small integers.
2252 (reported by Paul Dietz)
2253 * bug fix: starting a core saved with shared objects loaded when
2254 those objects are not available no longer causes threaded SBCL to
2255 hang. (reported by Sean Ross)
2256 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2258 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2259 argument, which must always be NIL. (reported by Kalle Niemitalo)
2260 * bug fix: printing 1.0d+23 no longer results in an error.
2261 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2262 * bug fix: structure slot setters preserve evaluation order. (thanks
2264 * fixed some bugs related to Unicode integration:
2265 ** RUN-PROGRAM can allow its child to take input from a Lisp
2266 stream. (reported by Stefan Scholl)
2267 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2268 locales. (reported by Ken Causey)
2269 * fixed some bugs revealed by Paul Dietz' test suite:
2270 ** INCF, DECF and REMF evaluate their place form as specified in
2272 ** FORMATTER expands ~{ ~} iteration directives with V or #
2273 parameters correctly.
2274 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2275 consequent uses no arguments correctly.
2276 ** the system has a partial understanding of the (COMPLEX RATIO)
2278 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2281 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2282 * new feature: a build-time option (controlled by the :SB-UNICODE
2283 keyword feature, enabled by default) for building the system with
2284 support for the entire 21-bit character space defined by the
2286 * new feature: the system now has rudimentary external-format
2287 support; the primary user-visible change at this time is that
2288 characters with the high bit set (such as lower-case-e-acute) will
2289 print correctly to a terminal in a UTF-8 environment.
2290 * minor incompatible change: BASE-CHAR no longer names a class;
2291 however, CHARACTER continues to do so, as required by ANSI.
2292 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2293 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2294 has been moved to the SB-EXT package (temporarily re-exported from
2296 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2297 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2299 * fixed bug #351: better error-handlind and reporting for malformed
2301 * fixed bug #350: bignum-printing is now more memory-efficient,
2302 allowing printing of very large bignums, eg. (expt 2 10000000).
2303 (reported by Bruno Haible)
2304 * fixed bug #302: better primitive-type selection for intersection
2306 * fixed bug #308: non-graphic characters now all have names, as
2307 required. (reported by Bruno Haible)
2308 * bug fix: redefining a class with different superclasses now correctly
2309 removes it from the direct-subclasses of its previous superclasses.
2310 (reported by David Morse)
2311 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2312 new value now works. (reported by Bruno Haible)
2313 * bug fix: correct canonicalization of multiple non-standard slot
2314 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2315 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2316 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2317 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2318 now exists, an signals an error.
2319 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2320 during macroexpansion for non-iteration variables as well. (reported
2321 by Bruno Haible for CMUCL)
2322 * bug fix: Cyclic structures and unprintable objects in compiler
2323 messages no longer cause errors. (reported by Bruno Haible)
2324 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2325 and READ-FROM-STRING all now return a primary value of NIL if
2326 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2327 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2328 (reported by Bruno Haible for CMUCL)
2329 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2330 do not designate a condition. (reported by Bruno Haible for
2332 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2333 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2334 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2335 permitting resolution in favour of any of the conflicting symbols.
2336 (reported by Bruno Haible for CMUCL)
2337 * bug fix: EQUAL compiler optimizations is less aggressive on
2338 strings which can potentially compare true despite having distinct
2339 specialized array element types.
2340 * bug fix: unit enumerations can be defined without dividing by
2341 zero. (thanks to Vincent Arkesteijn)
2342 * FORMAT compile-time argument count checking has been enhanced.
2343 (report from Bruno Haible for CMUCL)
2344 * a partial workaround for the bug 262: the compiler does not try to
2345 inline-expand a local function doing RETURN-FROM from a deleted
2346 BLOCK. (thanks to Peter Denno for the bug report and to David
2347 Wragg for the simple test case)
2348 * fixed some bugs revealed by Paul Dietz' test suite:
2349 ** RENAME-PACKAGE allows all package designators as new package
2351 ** constraint propagation and lambda variable substitution are
2352 more cautious in dealing with partially deleted code.
2353 ** compiler sometimes lost reoptimization passes.
2354 ** CERROR, given a condition as condition designator, passes its
2355 remaining arguments to the continue format control without
2357 ** Case-altering FORMAT directives work correctly on non-ASCII
2359 ** The REAL type specifier handles bounds outside the floating
2360 point ranges without signalling FLOATING-POINT-OVERFLOW.
2361 ** Functions with IR1-transformations can create intercomponent
2362 references to global functions.
2363 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2365 ** FORMAT ~R treats a nil value for its first parameter correctly.
2367 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2368 * enhancement: saving cores with foreign code loaded is now
2369 supported on x86/NetBSD and sparc/Linux in addition to the previously
2370 supported platforms.
2371 * bug fix: on some platforms repeated installations caused multiple
2372 copies of HTML documentation to be installed -- should not happen
2373 any more. (reported by Stefan Scholl)
2374 * bug fix: parsing self-recursive alien record types multiple times
2375 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2376 original patch by Helmut Eller for CMUCL)
2377 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2378 (thanks to Richard Kreuter)
2379 * bug fix: defining classes whose accessors are methods on existing
2380 generic functions in other (locked) packages no longer signals
2381 bogus package lock violations. (reported by François-René Rideau)
2382 * bug fix: special variables as DEFMETHOD parameters no longer have
2383 associated bogus type declarations. (reported by David Wragg and
2385 * bug fix: read-write consistency on streams of element-type
2386 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2387 * bug fix: redefiniton of the only method of a generic function with
2388 no DEFGENERIC no longer emits a full WARNING. In addition,
2389 redefinition of generic functions with no DEFGENERIC to an
2390 incompatible lambda list now signals an error. (thanks to Zach
2392 * bug fix: DEFGENERIC now works even when there's a function of the
2393 same name in an enclosing lexical environment. (thanks to Zach
2395 * fixed compiler failure, caused by instrumenting code during
2396 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2397 * optimization: added loop analysis and improved register allocation
2398 * fixed some bugs revealed by Paul Dietz' test suite:
2399 ** POSITION on displaced vectors with non-zero displacement
2400 returns the right answer.
2401 ** (SIMPLE-STRING) is a valid type specifier for sequence
2403 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2405 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2406 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2408 ** The justification version of the FORMAT ~< directive treats
2409 non-zero minpad parameter correctly.
2411 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2412 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2413 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2414 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2415 the supported interface.
2416 * new feature: Single-stepping of code compiled with DEBUG 2 or
2417 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2418 * new feature: saving cores with foreign code loaded is now
2419 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2420 Timothy Moore's work for CMUCL)
2421 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2422 parameters to * instead of NIL if no initform is supplied.
2423 (reported by Johan Bockgaard)
2424 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2425 correctly when FUNCALL appears as the car of the form. Note:
2426 despite this FUNCALL forms are not currently subject to
2427 compiler-macro expansion. (port of Raymond Toy's fix for the
2428 same from CMUCL, reported by Johan Bockgaard)
2429 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2430 (thanks for Teemu Kalvas)
2431 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2432 case instead of terminating the iteration (thanks for Julian
2433 Squires, Sean Champ and Raymond Toy)
2434 * bug fix: incorrect expansion of defgeneric that caused a style
2435 warning. (thanks for Zach Beane)
2436 * bug fix: slot accessor effective method computation works properly
2437 for classes with multiple non-standard applicable methods on
2438 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2439 * on x86 compiler supports stack allocation of results of LIST and
2440 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2441 CMUCL implementation by Gerd Moellmann)
2442 * fixed some bugs revealed by Paul Dietz' test suite:
2443 ** FORMAT strings with both the ~<~:;~> form of the justification
2444 directive and pretty-printing directives cause an error.
2446 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2447 * incompatible change: the internal functions
2448 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2449 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2450 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2451 instead of the old functions.
2452 * new feature: on platforms where "dladdr" is available foreign
2453 function names now appear in backtraces. (based on Helmut Eller's
2455 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2456 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2458 * documentation: ASDF manual is now distributed with SBCL.
2459 * bug fix: SBCL can now load its contributed modules with REQUIRE
2460 even if the system-provided entries have been removed from
2461 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2463 * bug fix: backtraces involving undefined functions or assembly
2464 routines are more informative. (thanks to Brian Downing)
2465 * bug fix: mutually referent alien structures now work correctly.
2466 (reported by Rick Taube)
2467 * bug fix: structures defined by WITH-ALIEN can be referred to
2468 within other definitions in the same WITH-ALIEN.
2469 * bug fix: division operators (MOD, TRUNCATE and the like) with
2470 constant zero divisors and integer dividends no longer generate
2472 * bug fix: provide default methods for INPUT-STREAM-P and
2473 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2474 * bug fix: improve the detection and resolution of MOP metacycles
2475 (where effective method computation depends on the generic function
2476 having its effective method computed). (reported by Bruno Haible)
2477 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2478 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2479 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2480 * bug fix: pathnames with relative directory components can be
2481 represented relative to default pathnames.
2482 * optimization: in taking the GCD of bignums, reduce the two bignums
2483 to approximately the same size (using Euclid's algorithm) before
2484 applying the more sophisticated binary GCD. (thanks to Juho
2486 * optimization: COUNT on bitvectors now operates word-at-a-time.
2487 * optimization: ASH with a positive, but not necessarily constant,
2488 (leftwards) shift, when in a modular context, is compiled to a
2490 * fixed some bugs revealed by Paul Dietz' test suite:
2491 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2492 if the corresponding argument is NIL.
2493 ** FORMAT directives accepting a minpad parameter treat negative
2495 ** literal commas as character parameters to FORMAT directives are
2497 ** literal spaces directly after ~<Newline> directives within a
2498 format-logical-block (~:< ~@:>) do not induce :FILL-style
2499 conditional newlines.
2500 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2501 as specified: it no longer includes conditional newlines).
2502 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2504 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2505 :PER-LINE-PREFIX argument does not evaluate to a string.
2507 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2508 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2509 the manual for details; for now, package locks can be disabled by
2510 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2511 if no major problems are found then it is likely that they will be
2512 compiled in unconditionally.
2513 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2514 unsupported operators on all platforms. To load a shared library
2515 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2516 object file, link it into a shared library outside of SBCL and
2517 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2519 * bug workaround: ROOM T and the GC were not working together
2520 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2521 reporting facility aren't preserved. That reporting has been
2522 disabled, so now until and unless someone figures out how to make it
2523 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2524 * minor incompatible change: as threatened around sbcl-0.8.0, the
2525 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2526 The symbols are also exported from SB-PCL for backwards
2527 compatibility, but more so than before SB-PCL should be treated as
2528 an implementation-internal package.
2529 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2531 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2532 It is known as of this release not to work on the Alpha, however.
2533 * fixed bug #167: errors signalled due to illegal syntax in method
2534 bodies are now more legible.
2535 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2536 designators and can hence be used with TYPEP.
2537 * fixed bug #333: CHECK-TYPE now ensures that the type error
2538 signalled, if any, has the right object to be accessed by
2539 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2540 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2541 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2543 * fixed bug #334: programmatic addition of slots using specialized
2544 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2545 and :ALLOCATION :CLASS slots.
2546 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2547 float exponents. (rereported by Peter Seibel)
2548 * fixed a bug: #\Space (and other whitespace characters) are no
2549 longer considered to be macro characters in standard syntax by
2550 GET-MACRO-CHARACTER.
2551 * fixed bug: initialization of condition class metaobjects no longer
2552 causes an instance of the condition to be created. (reported by
2554 * fixed bug: it is now possible to have slots such that
2555 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2556 description is neither :INSTANCE nor :CLASS.
2557 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2558 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2559 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2560 non-standard SLOT-DEFINITION-ALLOCATION values.
2561 * fixed bugs: various race conditions handling exiting threads.
2562 CL-PPCRE's thread torture test now passes.
2563 * fixed arguably-a-bug: GC time proportional to number of threads in
2564 system even when most of them are idle
2565 * optimization: improved performance of BIT and SBIT on bit-vectors.
2566 * .fasl file incompatibility: The fasl file version number has
2567 been incremented because of changes associated with package locks.
2569 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2570 * minor incompatible change: the system no longer provides
2571 optimization hints (conditions which are TYPEP
2572 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2573 mode; these hints are still emitted when higher SPEED optimization
2575 * new contrib module: a sampling profiler (profiling by statistical
2576 sampling, rather than by instrumenting functions) is available as
2577 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2578 * the behaviour of the standard function ED is now customizeable by
2579 third parties through a hook variable: see ED's documentation
2580 string for information on the protocol.
2581 * the compiler no longer emits efficiency notes for (FUNCALL X)
2582 when the type of X is uncertain under default optimization
2584 * fixed bug 276: mutating a binding of a specialized parameter to a
2585 method to something that is not TYPEP the specializer is now
2587 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2588 exhibit the required behaviour on the x86 platform. (thanks to
2589 Peter van Eynde, Eric Marsden and Bruno Haible)
2590 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2591 even for difficult arguments. (reported by Peter Graves)
2592 * fixed bug 141a: the backquote printer now descends quoted
2594 * fixed another bug in backquote printing: no more destructive
2595 modification of the form's list structure. (reported by Brian
2597 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2598 move between its address being taken and the call to
2599 interrupt_thread, fixing a crashing race condition.
2600 * the SB-POSIX contrib implementation has been adjusted so that it
2601 no longer exhibits ridiculously poor performance when constructing
2602 instances corresponding to C structs.
2604 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2605 * minor incompatible change: the sb-grovel contrib now treats C
2606 structures as alien (in the sense of SB-ALIEN) objects rather than
2607 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2608 has implications for memory management of client code
2609 (sb-grovel-returned objects must now be manually managed) and for
2610 type safety (alien objects now have full types).
2611 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2612 used to control emission of compiler diagnostics, rather than the
2613 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2614 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2615 quality should be considered deprecated.
2616 * install.sh now installs the user manual as well
2617 * (not quite a new documentable feature, but worth considering in
2618 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2619 beginnings of a semantically meaningful condition hierarchy is
2620 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2622 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2623 designator as the defaults argument.
2624 * fixed bug: Displaced arrays whose displaced-to array has become
2625 too small now cause ARRAY-DIMENSION to signal an error, providing
2626 for safer bounds-checking. (reported by Bruno Haible)
2627 * fixed bug: DEFCLASS slot definitions with identical :READER and
2628 :WRITER names now signal a reasonable error. (reported by Thomas
2630 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2632 * fixed bug: Invalid slot specification errors now print correctly.
2633 (thanks to Zach Beane)
2634 * fixed bug 320: Shared to local slot value transfers in class
2635 redefinitions now happen corrently from superclasses as
2636 well. (reported by Bruno Haible)
2637 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2639 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2640 improper lists correctly. (reported by Bruno Haible)
2641 * fixed bug 313: source-transform for <fun-name> was erroneously
2642 applied to a call of a value of a variable with name <fun-name>.
2643 (reported by Antonio Menezes Leitao)
2644 * fixed bug 307: The obsolete instance protocol ensures that
2645 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2646 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2647 remove dead unknown-values globs from the middle of the stack.
2648 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2649 easier to resume long computations after using *BREAK-ON-SIGNALS*
2650 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2651 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2652 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2653 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2654 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2655 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2657 * fixed a (fairly theoretical) bug in string printing: if
2658 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2659 does not have array-element-type equal to the most general string
2661 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2662 function name. (thanks to Nikodemus Siivola)
2663 * fixed bug: compile-time format string checker failed on
2664 non-closed ~{. (reported by Thomas F Burdick)
2665 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2666 don't warn on legal loop code involving more than one aggregate
2667 boolean. (thanks to Nikodemus Siivola)
2668 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2669 in keyword position to MAKE-PACKAGE was not being reported as
2671 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2672 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2674 * fixed bugs 280 and 312: the checking for multiple definitions in a
2675 file is less likely to become confused by uses of inline
2677 * fixed bug: the #S reader macro performs the keyword coercion
2678 specified for slot names. (reported by Kalle Niemitalo)
2679 * fixed bug: lambda lists may contain symbols whose names start with
2680 & but are not lambda-list-keywords; their occurrence triggers a
2682 * fixed bug 321: define-method-combination argument lambda lists do
2683 not cause invalid code to be generated when &optional or &aux
2684 variables have default values. (reported by Bruno Haible)
2685 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2686 have CLOS classes; this appears also to have fixed reported
2687 instability in stack exhaustion detection.
2688 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2689 MUFFLE-WARNING when no associated restart is present is now
2691 * optimization: rearranged the expansion of various defining macros
2692 so that each expands into only one top-level form in a
2693 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2695 * optimization: used a previously little-used slot in symbols to
2696 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2698 * fixed some bugs revealed by Paul Dietz' test suite:
2699 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2701 ** Fixed an optimization bug related to inheritance of initforms
2702 from local to shared slots.
2703 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2704 ** CAST optimizer forgot to flush argument derived type.
2705 ** print/read consistency on floats is now orders of magnitude
2706 more likely. (thanks also to Bruno Haible for a similar report
2708 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2709 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2710 for objects of type REAL. Make it so.
2711 ** SXHASH is specified (infelicitously) to respect similarity,
2712 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2713 it so. (thanks to Markus Ziegler)
2714 ** on the Alpha, the compiler succeeds in compiling functions
2715 returning a known number of arguments greater than 63.
2716 ** fixed handling of invalid NIL arguments in keyword position
2718 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2720 ** provided a readably-printable representation for RANDOM-STATE
2722 ** ensured that pathnames, where they have a namestring, always
2723 print using #P"..." syntax.
2725 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2726 * Support for the forthcoming 2.0 version of the NetBSD kernel
2727 running on x86 hardware has been added. (thanks to Perry
2728 E. Metzger most immediately, and others for their past work)
2729 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2730 current release version; SBCL's OpenBSD support had been broken
2731 since about the time of OpenBSD's switch to ELF binary format.)
2732 * [placeholder for DX summary]
2733 ** user code with &REST lists declared dynamic-extent, under high
2734 speed or space and low safety and debug optimization policy.
2735 * The manual has been converted to Texinfo format and the debugger
2736 chapter from the cmucl manual has been added.
2737 * A facility has been added to extract documentation strings from
2738 sbcl and store them as Texinfo-formatted snippets for inclusion in
2739 the manual (via Texinfo's @include directive)
2740 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2741 the test case to Dave Roberts)
2742 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2743 pointers. (reported by Sean Ross)
2744 * bug fix: PROFILE output is printed nicely even for large numerical
2745 values. (thanks to Zach Beane)
2746 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2747 greater than 32 handle EOF correctly.
2748 * bug fix: on X86 an immediate argument of the IMUL instruction is
2749 correctly printed in disassembly. (thanks to Lutz Euler)
2750 * bug fix: class slots in redefined classes preserve their old
2751 values. (thanks to Bruno Haible and Nikodemus Siivola)
2752 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2753 succeeds. (reported by Marco Baringer)
2754 * fixed some bugs revealed by Paul Dietz' test suite:
2755 ** READ-SEQUENCE now works on ECHO-STREAMs.
2756 ** RATIONALIZE works more according to its specification. (thanks
2759 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2760 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2761 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2762 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2763 should matter to you only if you rebind the printer control
2764 variables and then find you want different bindings in the
2765 debugger than in the ordinary execution of your program.)
2766 * The runtime build system has been tweaked to support building
2767 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2768 assembler and linker. (thanks to Nikodemus Siivola)
2769 * Unbound, undefined, undeclared variables now trigger full
2770 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2771 more often programmer error than deliberate exploitation of undefined
2773 * optimization: the hash algorithm for strings has changed to one
2774 that is less vulnerable to spurious collisions. (thanks to Juho
2776 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2777 needless bounds checking. (thanks to Juho Snellman)
2778 * optimization: implemented multiplication as a modular
2779 (UNSIGNED-BYTE 32) operation on the PPC backend.
2780 * fixed some bugs revealed by Paul Dietz' test suite:
2781 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2783 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2784 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2785 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2787 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2789 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2791 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2792 their output stream on EOF from read.
2793 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2794 have been read to end-of-file.
2795 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2797 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2798 description of determination of which consecutive characters
2800 ** Printing the "Space" character with escaping on now yields "#\\ ",
2801 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2802 ** Reading floating-point numbers with *READ-BASE* set to a number
2803 less than 10 works correctly.
2804 ** Reading floating-point numbers with *READ-BASE* set to a number
2805 more than 10 works correctly.
2806 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2807 the readtable currently in effect.
2809 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2810 * minor incompatible change: parsing of namestrings on a physical
2811 (Unix) host has changed; numbers after the final #\. in a
2812 namestring are no longer interpreted as a version field. This is
2813 intented to be largely invisible to the user, except that the
2814 meaning of the namestring "*.*.*" has changed: it now refers to a
2815 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2816 should usually be replaced by
2817 (make-pathname :name :wild :type :wild :version :wild)
2818 with the added benefit that this is more likely to be portable.
2819 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2820 option to OPEN now signals an error if the file being opened
2821 exists; this may have an impact on existing code.
2822 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2823 signals received, on the PowerPC platforms (both Linux and
2824 Darwin). (thanks to Pierre Mai for pointing out the location of
2826 * several fixes on OS X: The system now builds and runs cleanly on
2827 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2828 (thanks to Brian Mastenbrook)
2829 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2830 second-resolution: integer multiples of 1/3600 between -24 and 24.
2831 (thanks to Vincent Arkesteijn)
2832 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2833 type when called with 1 argument; PEEK-CHAR checked type of
2834 PEEK-TYPE only after having read first character from a
2835 stream. (reported by Peter Graves)
2836 * bug fix: the garbage collector now has much better locality
2837 behaviour, and in particular no longer treats all memory as being
2838 exhausted when many small objects point to each other in a deeply
2840 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2841 recognized as being TYPEP their class.
2842 * bug fix: the PUSHNEW documentation string has been corrected.
2843 (thanks to Vincent Arkesteijn)
2844 * bug fix: defaulting of the value for the last of an atypically
2845 large number of multiple values being bound was not being
2846 performed correctly on the Alpha or PPC platforms
2847 * optimization: implemented multiplication as a modular
2848 (UNSIGNED-BYTE 32) operation on the x86 backend.
2849 * optimization: SEARCH on simple-base-strings can now be open-coded.
2850 (see also contrib/compiler-extras.lisp for inspiration for
2851 teaching the compiler about the Boyer-Moore algorithm).
2852 * value, returned by MAX (and MIN) called with several EQUALP, but
2853 not EQL, arguments now does not depend on compiler settings.
2854 * fixed some bugs revealed by Paul Dietz' test suite:
2855 ** in stack analysis liveness information is propagated from
2856 non-local entry points.
2857 ** pathwise CAST removing failed when the CAST node did not start
2859 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2860 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2862 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2863 host is already defined.
2864 ** RENAME-FILE works on streams instead of signalling an internal
2866 ** PEEK-CHAR uses the current readtable when determining whether
2867 or not a character is whitespace.
2868 ** MERGE-PATHNAMES handles the case when the pathname does not
2869 specify a name while the default-pathname specifies a version.
2870 ** Pathnames now stand a better chance of respecting print/read
2872 ** Attempting to use standardized file system operators with a
2873 pathname with invalid :DIRECTORY components signals a
2875 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2877 ** DIRECTORY on logical pathnames is more correct.
2878 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2879 signal a TYPE-ERROR if their argument is not a stream.
2880 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2881 designator argument does not designate a stream.
2882 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2883 examining the synonym.
2884 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2886 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2887 element-type larger than ([UN]SIGNED-BYTE 32).
2889 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2890 * When built with the :SB-FUTEX feature, threaded builds now take
2891 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2892 for faster/more reliable mutex and condition variable support.
2893 * Incompatible change (but one you probably shouldn't have been using
2894 anyway): the interface and code for arbitrating between multiple
2895 threads in the same user session has been redesigned.
2896 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2897 error when called without an explicit environment argument.
2898 (thanks to Nikodemus Siivola)
2899 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2900 confused about their position. (thanks to Adam Warner and Gerd
2902 * bug fix: Pretty printing backquoted forms with unquotations in the
2903 argument list position of various code constructs such as LAMBDA
2904 now works correctly. (reported by Paul Dietz)
2905 * bug fix: Pretty printing unquotations no longer loses all
2906 stream position information.
2907 * optimization: performance of string output streams is now less
2908 poor for multiple small sequence writes.
2909 * optimization: performance of CSUBTYPEP in the presence of complex
2910 expressions involving CONS and NOT many times has been improved.
2911 (reported by Paul Dietz)
2912 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2914 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2915 streams for the REPL, for the debugger, and so on. (thanks to
2917 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2918 * fixed some bugs revealed by Paul Dietz' test suite:
2919 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2920 is ignored if it is too small, rather than propagating through
2921 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2922 ** extremely complex negations of CONS types were not being
2923 sufficiently canonicalized, leading to inconsistencies in
2925 ** VALUES tranformer lost derived type.
2927 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2928 * fixed a bootstrapping bug: the build process no longer assumes
2929 that the various BOOLE-related constants have the same value in
2930 host and target lisps. (noted by Paul Dietz' test suite on an
2931 SBCL binary built from CLISP)
2932 * The system can now be dynamically linked on the MIPS platform,
2933 which enables dynamic loading of foreign code from Lisp. (thanks
2934 to Ralf Baechle for discussions on the MIPS ABI)
2935 * The system now records debugging information for its own source
2936 files in a filesystem-position-independent manner, relative to
2937 the "SYS" logical host.
2938 * fixed a compiler bug: MV-LET convertion did not check references
2939 to the "max args" entry point. (reported by Brian Downing)
2940 * tweaked disassembly notes to be less confident about proclaiming
2941 some instruction as an LRA. (thanks to Brian Downing)
2942 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2943 now each have their own history, command character, and other
2944 characteristics. (thanks to David Lichteblau)
2945 * fixed some bugs revealed by Paul Dietz' test suite:
2946 ** compiler failure in compiling LOGAND expressions including a
2948 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2949 shift greater than 32.
2950 ** FLUSH-DEST did not mark blocks for type check regeneration.
2951 ** HANDLER-CASE failed to accept declarations in handler clauses
2952 in some circumstances.
2954 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2955 * New code in contrib/sb-introspect (still probably not entirely
2956 stable yet) provides some support for smart Lisp development
2957 environments like SLIME.
2958 * The conditions signalled for errors occurring when loading .fasl
2959 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2960 in a way which should help ASDF recover gracefully.
2961 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2962 stuff has been cleaned up. If you code contrib/ stuff, this might
2963 affect you, and you can look at contrib/README, contrib/STANDARDS,
2964 and/or the 0.8.4.27 diff to check.
2965 * In full calls the compiler now does not generate checks for declared
2966 argument types for all arguments.
2967 * various threading fixes
2968 ** and some experimental patches which didn't make it into
2969 the main tree for this release, but which are shipped in
2970 contrib/experimental-thread.patch as a possible fix for some
2971 failures (deadlock, spinning...) in GC-intensive multithreaded
2973 * fixed PPC build problem (source code incompatibility of different
2974 library versions): added offsetof() hackery which attempts to divine
2975 where glibc maintainers put uc_mcontext today
2976 * fixed bug 282: compiler does not trust type assertions while passing
2977 arguments to a full call.
2978 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2979 &OPTIONAL VALUES type parameter.
2980 * fix bug 214: algorithm for noting rejected templates is now more
2981 similar to that of template seletion. (also reported by rydis on
2983 * fixed bug 141b: printing backquoted information readably and prettily
2984 inserts a space where necessary.
2985 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2986 (CHARACTER 10) are now reported as errors, rather than propagated
2987 as unknown types. (reported by piso on #lisp)
2988 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2989 with values NIL and :ERROR. (thanks to Milan Zamazal)
2990 * fixed bug 191c: CLOS now does proper keyword argument checking as
2991 described in CLHS 7.6.5 and 7.6.5.1.
2992 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2993 longer raise an error; further, using a list as a for-as-arithmetic
2994 counter now raises a meaningful error.
2995 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2996 now understood by sequence creation functions such as MAKE-SEQUENCE
2998 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2999 a STRING-INPUT-STREAM.
3000 * compiler enhancement: SIGNUM is now better able to derive the type
3002 * type declarations inside WITH-SLOTS are checked. (reported by
3004 * fixed some bugs revealed by Paul Dietz' test suite:
3005 ** incorrect optimization of TRUNCATE for a positive first
3006 argument and negative second.
3007 ** compiler failure in let-convertion during flushing dead code.
3008 ** compiler failure while deriving type of TRUNCATE on an
3009 interval, containing 0.
3010 ** ASH of a negative bignum by a negative bignum count now returns
3012 ** intersection of CONS types now canonicalizes properly, fixing
3013 inconsistencies in SUBTYPEP.
3015 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3016 * incompatible change: The --disable-debugger command line
3017 option now clobbers the debugger at a more fundamental
3018 level, by redefining #'INVOKE-DEBUGGER instead of by
3019 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3020 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3021 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3022 enter the debugger REPL and then suffer recursive errors
3023 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3024 while under the new system, BREAK in --disable-debugger mode
3025 terminates the system just as an unhandled error would.
3026 * fixed compiler performance when processing loops with a step >1;
3027 * bug fix: DOCUMENTATION now retrieves generic function
3028 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3029 support has been systematized, and now supports the methods
3030 specified by ANSI, along with a default method and a method for
3031 slot documentation. (reported by Nathan Froyd)
3032 * bug fix: effective methods associated with a generic function are
3033 no longer cached over a change of that generic function's method
3034 combination. (reported by Andreas Fuchs)
3035 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3036 names a non-existent directory. (thanks to Andreas Fuchs)
3037 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3038 the PPC platform, lead to stack corruption; this has been fixed.
3039 (reported by Rainer Joswig)
3040 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3041 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3042 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3043 arithmetic in general, is now much more reliable on the PPC
3045 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3046 platform now returns the right answer.
3047 * optimization: restored some effective method precomputation in
3048 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3049 precomputation is now tunable.
3050 * optimization: compiler-internal data structure use has been
3051 reviewed, and changes have been made that should improve the
3052 performance of the compiler by about 20%.
3053 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3054 simple-base-strings and simple-bit-vectors is improved.
3055 * optimization: the optimization of 32-bit logical and arithmetic
3056 functions introduced in version 0.8.3 on the x86 has been
3057 implemented on the mips, ppc and sparc platforms; an
3058 implementation of the same facility, but for 64-bit arithmetic,
3059 has been added for the alpha.
3060 * microoptimization: the compiler is better able to make use of the
3061 x86 LEA instruction for multiplication by constants.
3062 * bug fix: in some situations compiler did not report usage of
3063 generic arithmetic in (SPEED 3) policy.
3064 * bug 145b fix: compiler used wrong type specifier while converting
3065 MEMBER-types to numeric.
3066 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3068 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3069 index argument now works properly on the Alpha platform.
3070 * bug fix: floating point exception treatment on the Alpha platform
3072 * bug fix: FILE-POSITION works much better on string input and
3073 output streams. (thanks to Nikodemus Siivola)
3074 * bug fix: many threading/garbage collection symptoms sorted.
3075 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3076 might be pseudo-atomic.
3077 * internal change: Stopping for GC is now done with signals not
3078 ptrace. GC is now done in whichever thread wanted it, instead of
3080 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3082 * bug fix: result form in DO is not contained in the implicit
3084 * incompatible change: ICR structure is changed; the value part of
3085 CONTINUATION is now called LVAR; corresponding functions are
3086 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3087 * added type deriver for ISQRT (thanks to Robert E. Brown).
3088 * bug fix: better support for loading from the command line when an
3089 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3091 * fixed some bugs revealed by Paul Dietz' test suite:
3092 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3093 ** ROUND and FROUND now give the right answer when given very
3094 small float arguments.
3095 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3097 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3098 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3099 ** LCM should return a non-negative integer.
3100 ** PARSE-INTEGER returned the index of a terminator instead of the
3101 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3102 ** PARSE-INTEGER returned an incorrect index being applied to a
3104 ** LCM with two arguments of 0 returns 0 rather than signalling
3106 ** unsigned addition of a 32-bit constant with the high bit set no
3107 longer causes an internal compiler error on the x86.
3108 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3109 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3110 with negative last argument.
3111 ** byte specifiers with zero size and position no longer cause
3112 an error during type derivation.
3113 ** bignum multiplication on the Alpha platform now returns the
3115 * porting: The system now builds on SuSE AMD64, although it still
3116 generates a 32-bit binary.
3117 * .fasl file incompatibility: The fasl file version number has
3118 been incremented (because of the changes to internal compiler
3119 data structures referred to above).
3121 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3122 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3123 more accurately, on the Darwin kernel running on PowerPC hardware.
3124 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3125 * Compiler code deletion notes now signal a condition of type
3126 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3127 an associated MUFFLE-WARNING restart.
3128 * The compiler now performs limited argument count validation of
3129 constant format strings in FORMAT, and where appropriate in ERROR,
3130 CERROR and WARN. (thanks to Gerd Moellmann)
3131 * New ASDF-INSTALL contrib can be used for automatic download and
3132 installation of third-party Lisp code from CCLAN or other sites
3134 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3135 which forces another thread to execute a function supplied by the
3137 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3138 now accept and act upon their :ELEMENT-TYPE keyword argument.
3139 (reported by Martin Atzmueller, Edi Weitz)
3140 * bug fix: FILE-POSITION now accepts position designators up to
3141 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3142 is the greater. (thanks to Patrik Nordebo)
3143 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3144 Kalle Olavi Niemitalo)
3145 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3146 printable. (reported by Eric Marsden)
3147 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3148 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3149 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3150 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3151 (thanks to Stig Sandoe)
3152 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3153 UNREAD-CHAR now correctly includes the unread character in the
3154 target sequence. (thanks to Gerd Moellmann)
3155 * bug fix in threaded builds: the system can now be suspended and
3156 resumed by shell job control with minimal disruption.
3157 * bug fixes in times and timezones >2038AD
3158 * better handling of "where is GNU make?" problem in build scripts
3159 (thanks to Nikodemus Siivola)
3160 * new optimization: inside a named function any reference to a
3161 function with the same name is considered to be a self-reference;
3162 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3163 optimization quality.
3164 * new optimization on x86: logical functions and + now have
3165 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3166 used when the result is truncated to 32 bits.
3167 * VALUES declaration is partially enabled.
3168 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3169 * fixed some bugs revealed by Paul Dietz' test suite:
3170 ** The system now obeys the constraint imposed by
3171 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3172 types form a lattice under type intersection.
3173 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3174 ** ASSOC now ignores NIL elements in an alist.
3175 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3176 and (1+ MOST-POSITIVE-FIXNUM) answers.
3177 ** The addition of a method with invalid qualifiers to a generic
3178 function does not cause an error to be signalled immediately;
3179 a warning is signalled, and the error is generated only on
3180 calling the generic function.
3181 * changed .fasl file version number, in part to add type codes for
3182 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3183 obscure ANSI requirements
3185 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3186 * fixed bug 148: failure to inline-expand a local function left
3187 garbage, confusing the compiler.
3188 * fixed bugs 3cd: structure slot readers perform type check if the
3189 slot can have an invalid value (i.e. it is either not initialized
3190 or can be written with a less specific slot writer).
3191 * bug fix: the compiler now traps array references to elements off
3192 the end of an array; previously, the bounds checking in some
3193 circumstances could go off-by-one.
3194 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3196 * type declarations for array element types now obey the description
3197 on the CLHS page "Declaration TYPE", as per discussions on
3198 sbcl-help around 2003-05-08. This means that a declaration
3199 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3200 declaration, all references to BAR will be asserted or assumed
3201 (with THE, so dependent on compiler policy) to involve objects of
3202 type FOO. Note that no such declaration is implied in
3203 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3204 * declared types of functions from the "Conditions"
3205 chapter. (reported by Paul Dietz)
3206 * bug fix: CERROR accepts a function as its first argument.
3207 * bug fix: NTH an NTHCDR accept a bignum as index
3208 arguments. (reported by Adam Warner)
3209 * optimization: character compare routines now optimize comparing
3210 against a constant character. (reported by Gilbert Baumann)
3211 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3212 argument now works properly on the MIPS platform.
3213 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3214 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3215 declared to have a functional type, does not cause an error
3217 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3218 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3219 * fixed bug: if last continuation of a deleted block has a
3220 destination, this destination should be deleted too. (reported by
3222 * fixed a bug in the bootstrap process: the host compiler's values
3223 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3224 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3225 test case from Patrik Nordebo)
3226 * improved the ability of the disassembler on the PPC platform to
3227 provide helpful disassembly notes.
3228 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3229 the class in more cases than previously.
3230 * bug fix: FILE-POSITION now understands :START and :END for
3231 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3232 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3233 8) data. (thanks to David Lichteblau)
3234 * bug fix: it is possible to add a method to a generic function
3235 without lambda list.
3236 * bug fix: reader failed to signal END-OF-FILE inside an
3237 object representation. (reported by Nikodemus Siivola)
3238 * fixed some bugs revealed by Paul Dietz' test suite:
3239 ** LAST and [N]BUTLAST should accept a bignum.
3240 ** condition slot accessors are methods.
3241 ** (VECTOR NIL) is a subtype of STRING.
3243 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3244 * minor incompatible change: some nonsensical specialized lambda
3245 lists (used in DEFMETHOD) which were previously ignored now signal
3247 * minor incompatible change: the system is now aware of the types of
3248 variables in the COMMON-LISP package, and will signal errors for
3249 most violations of these type constraints (where previously they
3250 were silently accepted).
3251 * minor incompatible change: COMPILE-FILE now uses the freedom
3252 afforded (ANSI 3.2.2.3) to use derived function types for
3253 functions defined in the same file. This also permits the system
3254 to warn on static type mismatches and function
3255 redefinition. (Currently it does not work with high DEBUG level.)
3256 * minor incompatible change: VALUES declaration is disabled.
3257 * When issuing notes, the compiler now signals a condition of type
3258 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3259 restart for use in user handlers. It is expected that the
3260 COMPILER-NOTE condition will eventually become a condition
3261 supertype to a hierarchy of note types, which will then be
3262 handleable in a similar fashion. However, at the moment, no such
3263 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3264 but it's an implementation detail, not a classification for the
3266 * Changes in type checking closed the following bugs:
3267 ** type checking of unused values (192b, 194d, 203);
3268 ** template selection based on unsafe type assertions (192c, 236);
3269 ** type checking in branches (194bc).
3270 * A short form of VALUES type specifier has ANSI meaning (it has
3271 increased the number of situations when SBCL cannot perform type
3273 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3274 keywords or constants is permissible.
3275 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3276 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3277 classes. (thanks to Antonio Martinez)
3278 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3279 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3280 argument to a pretty-print stream. Instead, it leaves any such
3281 operations to DESCRIBE-OBJECT methods.
3282 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3284 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3285 lambda lists. (thanks to David Lichteblau)
3286 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3287 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3288 (thanks to Gerd Moellmann)
3289 * fixed bug 246: increased compilation speed of long
3290 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3292 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3294 * DEFCONSTANT now throws a condition of type
3295 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3296 constant to a non-EQL value; CONTINUE and ABORT restarts
3297 respectively change and preserve the value.
3298 * fixed bug 63: The code walker, part of the implementation of CLOS,
3299 is now better at handling symbol macros.
3300 * bug fix: There is no longer an internal implementation type named
3301 CL:LENGTH. (reported by Raymond Toy)
3302 * bug fix: In macro-like defining macros/special operators the
3303 implicit block does not enclose the lambda list.
3304 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3305 enclosed in parentheses) are not suitable as type specifiers, and
3306 their use properly signals an error now.
3307 * bug fix: An argument count mismatch for a type specifier in code
3308 being compiled no longer causes an unhandled error at compile
3309 time, but signals a compile-time warning.
3310 * fixed simple vector readable printing
3311 * bug fix: DESCRIBE takes more care over whether the class
3312 precedence list slot of a class is bound before accessing it.
3313 (reported by Markus Krummenacker)
3314 * bug fix: FORMATTER can successfully compile pretty-printer format
3315 strings which use variants of the ~* directive inside.
3316 * bug fix: SEARCH now applies its TEST predicate to the elements of
3317 the arguments in the correct order. (thanks to Wolfhard Buss)
3318 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3319 in type inference. (thanks to Robert E. Brown)
3320 * bug fix: Reading in symbols with an explicit package name of ""
3321 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3322 * fixed some bugs revealed by Paul Dietz' test suite:
3323 ** NIL is now allowed as a structure slot name.
3324 ** Arbitrary numbers, not just REALs, are allowed in certain
3325 circumstances in LOOP for-as-arithmetic clauses.
3326 ** Multiple class redefinitions before slot access no longer
3327 causes a type error.
3328 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3329 association between the name and a class.
3330 ** Generic functions with non-standard method-combination and over
3331 five methods all of which return constants no longer return NIL
3332 after the first few invocations. (thanks to Gerd Moellmann)
3333 ** CALL-NEXT-METHOD with no arguments now passes the original
3334 values of the arguments, even in the presence of assignment.
3335 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3337 ** DISASSEMBLE works with closures and funcallable instances.
3338 ** ADD-METHOD now returns the generic function, not the new method.
3339 ** FIND-METHOD signals an error if the lengths of the specializers
3340 is incompatible with the generic function, even if the ERRORP
3342 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3343 which its argument is a member.
3344 ** DEFCLASS only redefines the class named by its class-name
3345 argument if that name is the proper name of the class;
3346 otherwise, it creates a new class.
3347 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3348 of the UNBOUND-SLOT condition to the name of the slot.
3349 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3350 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3351 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3352 SLOT-MAKUNBOUND in the specified fashion.
3354 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3355 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3356 cross-compilation host. As a consequence, we can now bootstrap our
3357 way up to SBCL starting with a bare gcc toolchain and human-readable
3358 source code (first the source to CLISP, then the source to SBCL).
3359 * A contributed module containing a partial implementation of the
3360 simple-streams interface has been included. (thanks to Rudi
3362 * A contributed module implementing the RFC1321 Message Digest
3363 Algorithm, known as MD5, has been included.
3364 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3365 no longer has any effect, as the code controlled by this feature
3366 has been deleted. (As far as we know, no-one has ever built using
3367 this feature, and its semantics were confused in any case).
3368 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3369 work on conditions (as required by the ANSI specification),
3370 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3371 expected behaviour on conditions. Users should note, however,
3372 that such behaviour is not required by the ANSI specification,
3373 and so use of this behaviour may render their code unportable.
3374 * fixed some bugs revealed by Paul Dietz' test suite:
3375 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3377 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3378 ** SXHASH on condition objects no longer returns NIL.
3379 ** :ALLOCATION :CLASS slots are better treated; their values are
3380 updated on class redefinition, and initforms inherited from
3381 superclasses are applied.
3382 ** REMOVE-METHOD returns its generic function argument even when
3383 no method was removed.
3384 ** SHARED-INITIALIZE now initializes the values of the requested
3385 slots, including those with :ALLOCATION :CLASS.
3386 ** ALLOCATE-INSTANCE now works on structure classes defined via
3387 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3389 ** SLOT-EXISTS-P now works on conditions, as well as structures
3391 ** MAKE-LOAD-FORM now has the required methods on
3392 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3393 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3394 keyword, and now implements the SLOT-NAMES argument.
3395 ** methods with &OPTIONAL arguments no longer allow too many
3396 arguments to be passed in the call without error.
3397 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3398 option is consistent with the required arguments of the generic
3399 function lambda list.
3400 * bug fix: REQUIRE accepts a string designator. (Thanks to
3402 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3403 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3404 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3405 * bug fix: The compiler now checks for duplicated variables in macro
3407 * bug fix: SETQ on globals returns the correct value.
3408 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3409 not a COMPILER-ERROR (followed by some other strange error on
3410 choosing the CONTINUE restart).
3411 * bug fix: make.sh and friends are now more consistent in the way that
3412 they look for GNU "make".
3414 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3415 * experimental native threads support (on x86 Linux >=2.4 only).
3416 This is not compiled in by default: you need to add :SB-THREAD to
3417 the target features. See the "Beyond ANSI" chapter of the manual
3419 * fix for longstanding nonANSIism: The old distinction between
3420 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3421 The return value from CL:FIND-CLASS is now a CLOS class, and
3422 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3423 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3424 * An interface to the MetaObject Protocol, as described in Kiczales,
3425 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3426 MIT Press, 1991, is available from the SB-MOP package.
3427 * incompatible change: the SB-PCL package should now be considered
3428 a private implementation detail, and no longer a semi-private MOP
3430 * minor incompatible change: due to rearrangement for threads, the
3431 control stack and binding stack are now allocated at arbitrary
3432 addresses instead of being hardcoded per-port. Users affected by
3433 this probably have to be doing advanced things with shared
3434 libraries, and will know who they are.
3435 * minor incompatible change: Previously, all --eval forms used were
3436 processed with READ before any of them were processed with EVAL.
3437 Now each --eval form is processed with both READ and EVAL before
3438 the next --eval form is processed. (Thus package operations like
3439 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3440 work as the user might reasonably expect.)
3441 * minor incompatible change: *STANDARD-INPUT* is now only an
3442 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3444 * minor incompatible change: Y-OR-N-P is now character-oriented,
3445 not line oriented. Also, YES-OR-NO-P now works without errors.
3446 (thanks to Antonio Martinez)
3447 * sb-aclrepl module improvements: an integrated inspector, added
3448 repl features, and a bug fix to :trace command.
3449 * Known functions, which cannot be open coded by the backend, are
3450 considered to be able to check types of their arguments. (fixing
3451 a bug report by Nathan J. Froyd)
3452 * fixed a bug in computing method discriminating functions: It is
3453 now possible to define methods specialized on classes which have
3454 forward-referenced superclasses. (thanks to Gerd Moellmann)
3455 * fixed evaluation order in optional entries (reported by Gilbert
3457 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3458 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3459 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3460 required-by-AMOP NAME argument, as well as CLASS and
3461 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3462 * fixed bug 20: DEFMETHOD can define methods using names that are
3463 not the proper names of classes to designate class specializers.
3464 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3465 Unix file descriptors, instead of blowing up. (thanks to Antonio
3467 * Garbage collection refactoring: user-visible change is that a
3468 call to the GC function during WITHOUT-GCING will not do garbage
3469 collection until the end of the WITHOUT-GCING. If you were doing
3470 this you were probably losing anyway.
3471 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3472 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3473 TYPEP the latter but not the former.
3474 * The compiler issues a full WARNING for calls to undefined functions
3475 with names from the CL package.
3476 * MAP-INTO for a vector destination is open coded. (reported by
3477 Brian Downing on c.l.l)
3478 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3479 documentation string.
3480 * fixed some bugs revealed by Paul Dietz' test suite:
3481 ** COPY-ALIST now signals an error if its argument is a dotted
3483 ** Condition slots are now accessed more correctly in the presence
3484 of multiple initargs for a given slot.
3485 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3486 exclude restarts of the same name associated with a different
3488 ** DEFCLASS of forward-referenced classes with another
3489 forward-referenced class in the superclasses list no longer
3491 ** Condition slots are now initialized once each, not multiple
3492 times. (thanks to Gerd Moellmann)
3493 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3495 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3496 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3497 arguments contain duplicated elements.
3498 ** RESTART-CASE understands local macros.
3499 ** RESTART-CASE associates exactly its own restarts with a condition.
3500 ** ENDP in safe mode checks its argument to be of type LIST.
3501 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3502 in question is unbound.
3503 ** Optimized MAKE-INSTANCE functions no longer cause internal
3504 assertion failures in the presence of duplicate initargs.
3505 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3506 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3507 propagated even in the ctor optimized implementation of
3509 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3511 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3512 * a better implementation of SXHASH on (simple) bit vectors,
3513 measured both in execution speed and in distribution of results
3514 over the positive fixnums, has been installed. Likewise, a better
3515 implementation of EQUAL for simple bit vectors is now available.
3516 * fixed CEILING optimization for a divisor of form 2^k.
3517 * fixed bug 240 (emitting extra style warnings "using the lexical
3518 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3519 by Antonio Martinez)
3520 * fixed SXHASH, giving different results for NIL depending on type
3521 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3522 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3523 variable. (found by Rolf Wester)
3524 * SBCL does not ignore type declarations for special
3525 variables. (reported by rif on c.l.l 2003-03-05)
3526 * some bug fixes in contrib/sb-aclrepl/
3527 * fixed some bugs revealed by Paul Dietz' test suite:
3528 ** a bug in the CONS type specifier, whereby the CAR and CDR
3529 types got intertwined, has been fixed;
3530 ** the type system is now able to reason about the interaction
3531 between INTEGER and RATIO types more completely;
3532 ** APPEND, [N]REVERSE and NRECONC check that those their
3533 arguments, which must be proper lists, are really so;
3534 ** An array specialized to be unable to hold elements has been
3535 implemented, as required -- yes, really -- by ANSI;
3536 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3537 on malformed property lists;
3539 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3540 * incompatible packaging change: in line with Unix convention,
3541 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3542 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3543 * REQUIRE and PROVIDE are now optionally capable of doing something
3544 useful. See the documentation string for REQUIRE.
3545 * infrastructure for a managed SBCL contrib system: contributed
3546 modules in this release include:
3547 ** the ASDF system definition facility;
3548 ** an interface to the BSD Sockets API;
3549 ** an ACL-like convenience interface to the repl;
3550 (thanks to Kevin Rosenberg)
3551 ** an implementation of ROTATE-BYTE, with efficient implementation
3553 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3554 gives a closer approximation to the right answer than previously.
3555 (thanks to Raymond Toy)
3556 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3557 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3558 optional environment arguments, as required by ANSI.
3559 * fixed bugs in other functions taking environment objects, allowing
3560 calls with an explicit NIL environment argument to be compiled
3562 * fixed bug 228: primary return values from
3563 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3564 COMPILE or FUNCTION.
3565 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3566 :TYPE will no longer signal errors on innocuous objects.
3567 * fixed bug 231b: SETQ is better at respecting type declarations in
3568 the lexical environment.
3569 * fixed a bug in DEFCLASS: classes named by symbols with no or
3570 unprintable packages can now be defined.
3571 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3572 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3573 * fixed compiler failure related to checking types of functions.
3574 (reported by Robert E. Brown)
3575 * the compiler is now much more consistent in its error-checking
3576 treatment of bounding index arguments to sequence functions: in
3577 (SAFETY 3) code, errors will be signalled in almost all cases if
3578 invalid sequence bounding indices are passed to functions defined
3579 by ANSI to operate on sequences.
3580 * fixed a bug in the build procedure: documentation of SBCL-specific
3581 packages is now preserved and available in the final Lisp image.
3582 * lifted FDEFINITION lookup out of loops in the implementation of
3583 many list operations. (thanks to Robert E. Brown)
3584 * fixed a bug in the reader: the #n# reader macro now works for
3585 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3586 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3587 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3588 E. Brown and Tony Martinez respectively)
3589 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3590 SLOT-BOUNDP: in particular, it is now able to optimize them much
3591 better, and is now not vulnerable to having packages renamed.
3592 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3593 always called when a slot is not present in an instance. (thanks
3595 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3596 longer signal an error on unreading a character following EOF on
3597 the previous constituent stream. (thanks to Tony Martinez)
3598 * fixed some bugs revealed by Paul Dietz' test suite:
3599 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3600 not just nonnegative fixnums;
3601 ** the logical bit-array operators such as BIT-AND now accept an
3602 explicit NIL for their "opt-arg" argument (to indicate a
3603 freshly-consed result bit-array);
3604 ** ELT now signals an error on an invalid sequence index in safe
3606 ** the type system is now cleverer about negations of numeric
3607 types, and consequently understands the BIGNUM and RATIO types
3609 ** the type system is now cleverer about the interaction between
3610 INTEGER and RATIO types: while bugs still remain, many more
3611 cases are accurately computed;
3612 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3613 if it is in the last clause;
3614 ** CONSTANTLY now correctly returns a side-effect-free function in
3616 ** DECLARE is no longer treated as a special-operator; in
3617 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3618 * incremented fasl file version number due to the change in the
3619 DEFSTRUCT-SLOT-DESCRIPTION structure.
3621 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3622 * minor incompatible change: code processed by the "interpreter" or
3623 EVAL now has a compilation optimization policy of (DEBUG 2)
3624 (changed from (DEBUG 1)) to improve debuggability of interactive
3625 development, and to allow the use of the debug RETURN command in
3627 * an experimental implementation of the RETURN command for the
3628 debugger has been included. (thanks to Frederik Kuivinen)
3629 * fixed bug 62: constraints were not propagated into a loop.
3630 * fixed bug in embedded calls of SORT (reported and investigated by
3632 * fixed some bugs revealed by Paul Dietz' test suite:
3633 ** printing and reading of arrays with some dimensions having
3634 length 0 (thanks to Gerd Moellmann);
3635 ** BOA constructor with &AUX argument without a default value does
3636 not cause a type error;
3637 ** CONSTANTP now returns true for all self-evaluating objects.
3639 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3640 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3641 accessors that are related by inheritance, as specified in the
3642 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3643 Valtteri Vuorikoski)
3644 * The compiler is now able to inline functions that were defined in
3645 a complex lexical environment (e.g. inside a MACROLET).
3646 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3648 * Support for the upcoming FreeBSD-5.0 release has been included.
3649 (thanks to Dag-Erling Smorgrav)
3650 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3651 effect when it is not in a toplevel context.
3652 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3653 stand a better chance of being correct. (thanks to Gerd
3655 * fixed bug in COERCE, which now signals an error on coercing a
3656 rational to a bounded real type which excludes the expected
3658 * The compiler is now able to derive types more accurately from the
3659 COERCE and COMPILE functions.
3660 * fixed bug 223: functional binding is considered to be constant
3661 only for symbols in the CL package.
3662 * fixed bug 231: SETQ did not check the type of a variable being set
3663 (reported by Robert E. Brown)
3664 * A new optimization for MAKE-INSTANCE has been included, fixing
3665 various bugs (including relating to :ALLOCATION :CLASS slots and
3666 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3668 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3669 ** As required by ANSI, LOOP now disallows anonymous collection
3670 clauses such as COLLECT I in conjunction with aggregate boolean
3671 clauses such as THEREIS (= I 1);
3672 ** LOOP now signals an error when any variable is reused in the
3673 same loop (including the potentially useful construct analogous
3674 to WITH A = 1 WITH A = (1+ A);
3675 ** IT is only a special loop symbol within the first clause of a
3676 conditional loop clause;
3677 ** LOOP with a typed iteration variable over a hashtable now
3678 signals a type error iff it should.
3679 * fixed some other bugs revealed by Paul Dietz' test suite:
3680 ** FILE-STREAM now names the class previously known as FD-STREAM;
3681 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3682 argument) no longer signals an error;
3683 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3684 of the slot symbol, rather than using the current package
3685 ((:CONC-NAME "") continues to intern the slot's name in the
3687 * incremented fasl file version number, because of the incompatible
3688 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3689 of the new implementation of DEFINE-COMPILER-MACRO.
3691 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3692 * Support for building SBCL for MIPS platforms running in
3693 little-endian mode has now been checked in, and basic
3694 functionality on said platforms verified.
3695 * minor incompatible change: PCL now records the pathname of a file
3696 in which methods and the like are defined, rather than its
3698 * minor incompatible change: TRUENAME now considers the truename of
3699 a file naming a directory to be the pathname with :DIRECTORY
3700 component indicating that directory.
3701 * minor incompatible change: a NAMED clause in the extended form of
3702 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3703 reason for the previous behaviour is unclear.
3704 * more systematization and improvement of CLOS and MOP conformance
3705 in PCL (thanks to Gerd Moellman and Pierre Mai):
3706 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3708 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3709 primary methods with no specializers;
3710 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3712 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3713 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3714 FORWARD-REFERENCED-CLASSes; error reporting on
3715 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3717 ** SXHASH on CLOS instances now uses a slot internal to the
3718 instance to return different numbers on distinct instances,
3719 while preserving the same return value through invocations of
3721 ** DEFMETHOD signals errors when methods with longer incongruent
3722 lambda lists are added to generic functions;
3723 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3724 CLASS, as specified in AMOP;
3725 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3726 on the order returned by the primary method for classes of
3727 class STANDARD-CLASS;
3728 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3729 * fixed some bugs shown by Paul Dietz' test suite:
3730 ** DOLIST puts its body in TAGBODY;
3731 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3733 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3734 value producing form;
3735 ** if more variables are given to PROGV than values, extra
3736 variables are bound and made to have no value;
3737 ** NSUBSTITUTE on list arguments gets the right answer with
3739 ** ELT signals an error of type TYPE-ERROR when the index argument
3740 is not a valid sequence index;
3741 ** LOOP signals (at macroexpansion time) an error of type
3742 PROGRAM-ERROR when duplicate variable names are found;
3743 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3744 ** FUNCALL of special-operators now cause an error of type
3746 ** PSETQ now works as required in the presence of side-effecting
3747 symbol-macro places;
3748 ** NCONC accepts any object as its last argument;
3749 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3751 ** loop-for-as-package does not require a package to be explicitely
3753 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3755 * fixed bug 166: compiler preserves "there is a way to go"
3756 invariant when deleting code.
3757 * fixed bug 172: macro lambda lists with required arguments after
3758 &REST arguments now cause an error to be signalled. (thanks to
3760 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3761 bug. (thanks to Matthew Danish)
3762 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3764 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3765 arguments are lexically rebound. (thanks to Gerd Moellmann and
3767 * fixed bug 194: error messages are now more informative when there
3768 is no primary method applicable in a call to a generic
3769 function. (thanks to Gerd Moellmann)
3770 * fixed bug in command line argument checking (thanks to Julian
3772 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3773 (thanks to Matthew Danish)
3774 * incremented fasl file version number, because of the
3775 SXHASH-related changes in the layout of CLOS data structures
3777 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3778 * minor incompatible change: The runtime (the Unix executable named
3779 "sbcl") is now much pickier about the .core files it will load.
3780 Essentially it now requires .core files to descend from the same
3781 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3782 as the runtime does. (The intent is to prevent the crashes which
3783 can occur, and which can even be reported as mysterious failures,
3784 when people patch the sources or change the build parameters
3785 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3786 sbcl and .core files.)
3787 * fixed bug: VALUES-LIST is no longer optimized away.
3788 * fixed bug 142: The FFI conversion of C string values to Lisp
3789 string values no longer conses excessively. (thanks to Nathan
3790 Froyd porting Raymond Toy's fix to CMU CL)
3791 * began to systematize and improve MOP conformance in PCL (thanks to
3792 Nathan Froyd, Gerd Moellman and Pierre Mai):
3793 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3795 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3796 ** FINALIZE-INHERITANCE is now called on class finalization;
3797 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3798 argument precedence order.
3799 * fixed bug 202: The compiler no longer fails on functions whose
3800 derived types contradict their declared type.
3801 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3802 so it can be non-toplevel.
3803 * The fasl file version number has changed (because of the new
3804 implementation of DEFMACRO).
3805 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3806 safe code, that any length requirement by their type-specifier
3807 argument is valid. The exceptions to this are described in bug
3809 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3810 functions now check the types of their inputs as required by ANSI.
3811 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3812 introduced symbol is DECLAREd to be SPECIAL.
3813 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3814 * fixed bug: :COUNT argument to sequence functions may be negative
3815 * fixed bug: body of DO-SYMBOLS may contain declarations
3816 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3817 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3818 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3819 symbol macro only once
3820 * fixed printing of call frame when argument list is unavailable
3821 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3822 * compiler no longer signals WARNING on unknown keyword
3825 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3826 * A beta-quality port to the mips architecture running Linux,
3827 based on the old CMUCL backend, has been made. It has been tested
3828 on a big-endian kernel, and works sufficiently well to be able to
3829 rebuild itself; it has not been tested in little-endian mode.
3830 * fixed an inconsistency between gencgc.c and purify.c which made
3831 dumping/loading .core files unreliable
3832 * fixed bug 120a: The compiler now deals correctly with IFs where
3833 the consequent is the same as the alternative, instead of
3834 misderiving the return type. (thanks to Alexey Dejneka)
3835 * fixed bug 113: Logical pathnames are now dumpable (the logical
3836 host is resolved at load-time, throwing an error if it is not
3838 * fixed bug 174: FORMAT's error message is slightly clearer when a
3839 non-printing character is used in a format directive.
3840 * fixed several bugs in compiler checking of type declarations, i.e.
3841 violations of the Python "declarations are assertions" principle
3842 (thanks to Alexey Dejneka)
3843 * fixed several bugs in PCL's error checking (thanks to Gerd
3845 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3847 * fixed bug in compilation of functions as first class values
3848 (thanks to Antonio Martinez-Shotton)
3849 * The compiler's handling TYPE-ERRORs which it can prove will
3850 inevitably happen at runtime has been cleaned up and corrected
3851 in several ways. (thanks to Alexey Dejneka)
3852 * improved argument type checking for various basic arithmetic
3853 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3854 lavished on them in the past that they can be compiled in many
3855 ways in different special cases
3856 * fixed bug 181: compiler checks validity of user supplied type
3858 * cleaned up code flushing in optimization: Function calls which
3859 should signal errors for safety purposes (e.g. which ANSI says
3860 should signal errors when their arguments are of incorrect type)
3861 are no longer optimized away.
3862 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3863 * incremented fasl file version number, because changes in the
3864 implementation of sequence functions like COERCE caused
3865 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3866 in old inline expansions) to become undefined. (Actually these
3867 changes were later undone, so we might very well be binary
3868 compatible with 0.7.7 after all, but leaving the version number
3869 incremented seemed like the simplest and most conservative
3872 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3873 * An alpha-quality port to the parisc architecture running Linux,
3874 based on the old CMUCL backend, has been made. This, even more so
3875 than the other backends, should be considered still a work in
3876 progress; known problems include that the Linux kernel in 64-bit
3877 mode does not propagate the correct sigcontext structure to
3878 userspace, and consequently SBCL on a parisc64 kernel will not
3880 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3881 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3882 suppressed.) Also now that the compiler is looking at declarations
3883 in the environment, it checks optimization declarations as well,
3884 and suppresses inlining when (> DEBUG SPEED).
3885 * More fixes have been made to treatment of floating point exception
3886 treatment and other Unix signals. In particular, floating point
3887 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3888 * The detection and handling of control stack exhaustion (infinite
3889 or very deeply nested recursion) has changed. Stack exhaustion
3890 detection is now done by write-protecting pages at the OS level
3891 and applies at all optimization settings; when found, a
3892 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3893 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3894 be caught using IGNORE-ERRORS.
3895 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3896 correctly when transforming recursive LABELS functions to LETs.
3897 (thanks to Alexey Dejneka)
3898 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3899 that are names of constants or global variables.
3900 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3901 alien routines with docstrings.
3902 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3903 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3905 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3906 (thanks to Eric Marsden)
3907 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3908 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3909 * Bug fix: The SPARC backend can now compile functions involving
3910 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3911 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3912 to a functions expecting a PATHNAME-DESIGNATOR.
3913 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3914 lambda lists. (thanks to Alexey Dejneka)
3915 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3916 functions so that EXPORTing the name of the class doesn't cause
3917 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3918 to Antonio Martinez for reporting this)
3919 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3920 constant-folded without failing an assertion. (thanks to Einar
3921 Floystad Dorum for reporting this)
3922 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3923 types are now dealt with more correctly by the compiler.
3924 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3925 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3926 behaviour analogous to RENAME-FILE. This puts its behaviour more
3927 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3929 * The fasl file version number has changed again. (because of the
3930 bug fix involving the names of PCL MAKE-INSTANCE functions)
3932 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3933 * bug fix: Floating point exceptions are treated much more
3934 consistently on the x86/Linux and PPC/Linux platforms.
3935 * Array initialization with :INITIAL-ELEMENT is now much faster for
3936 cases when the compiler cannot open code the array creation, but
3937 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3938 array accesses have also seen a speed increase.
3939 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3940 specified. (This was at the root of some bad interactions between
3941 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3943 * bug fix: Internal error arguments for undefined functions are now
3944 computed correctly on the PPC/Linux platform.
3945 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3946 Raymond Toy's patch for CMU CL)
3947 * Support for the Solaris 9 operating environment has been included
3948 (thanks to Daniel Merritt)
3949 * A very ugly but hopefully complete draft of the missing FFI chapter
3950 of the manual has been created by reformatting the corresponding
3951 CMU CL manual chapter into (currently very ugly and incoherent)
3952 DocBook and bringing it up to date for SBCL behavior. Thus, the
3953 manual is now essentially complete, at least by my extreme
3954 once-and-only-once standards, whereby it's acceptable to refer to
3955 the doc strings of SB-EXT functions as the primary documentation.
3956 * The fasl file version number has changed again, due to cleanup of
3957 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3958 is no longer a static symbol.)
3960 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3961 * SBCL now builds with OpenMCL (version 0.12) as the
3962 cross-compilation host; also, more progress has been made toward
3963 bootstrapping under CLISP.
3964 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3966 * bug 158 fixed: The compiler can now deal with integer loop
3967 increments different from 1; fixing this turned out also to fix
3969 * bug 169 fixed: no more bogus warnings about using lexical bindings
3970 despite the presence of perfectly good SPECIAL declarations (thanks
3971 to David Lichteblau)
3972 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3973 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3974 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3976 * bug 180 fixed: Method combination specifications no longer ignore
3977 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3978 * bug fix: Structure type predicate functions now check their argument
3979 count as they should.
3980 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3981 correctly. (thanks to Pierre Mai)
3982 * minor incompatible change: The --noprogrammer option is deprecated
3983 in favor of the new --disable-debugger option, which is very similar.
3984 (The major difference is that it takes effect at a slightly different
3985 time at startup, causing handling of errors in --sysinit and
3986 --userinit files will be affected differently.) The
3987 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3988 been added to allow this functionality to be controlled from ordinary
3989 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3990 maintainers, who might want to run non-interactive scripts to
3991 build SBCL cores which will later be used interactively.)
3992 * minor incompatible change: The LOAD function no longer, when given
3993 a wild pathname to load, loads all files matching that pathname.
3994 Instead, an error of type FILE-ERROR is signalled.
3996 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3997 * bug 147 fixed: The compiler preserves its block link/count
3998 invariants more correctly now so that it doesn't crash. (thanks
4000 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4002 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4003 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4004 no-value-for-this-character as NIL (as specified by ANSI).
4005 * HOST-NAMESTRING on physical pathnames now returns a string that is
4006 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4007 (thanks to Christophe Rhodes)
4008 * The Alpha port handles icache flushing more correctly. (thanks to
4010 * More progress has been made toward bootstrapping under CLISP. (thanks
4011 to Christophe Rhodes)
4012 * The fasl file format has changed again, because dynamic loading
4013 on OpenBSD (which has non-ELF object files) motivated some cleanups
4014 in the way that foreign symbols are transformed and passed around.
4015 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4016 is no longer treated as whitespace by the reader, but instead as
4017 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4018 |A
\7fB|, instead of A as it used to.
4020 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4021 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4022 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4023 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4024 this as of 0.7.1.45, but was left out of the previous news section
4025 (thanks to Dan Barlow)
4026 * SBCL now runs on the Solaris operating system on SPARC architectures
4027 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4028 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4029 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4030 * SPARC backend cleanups, allowing builds of cores optimized for V8
4031 and V9 SPARCS, and also emission of code targeted to a particular
4032 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4034 * SBCL is closer to bootstrapping under CLISP, thanks to various
4035 fixes by Christophe Rhodes.
4036 * The fasl file format has changed again, to allow the compiler's
4037 INFO database to support symbol macros.
4038 * The user manual (in doc/) is formatted into HTML more nicely.
4039 (thanks to coreythomas)
4040 * The system is smarter about SUBTYPEP relationships, especially
4041 those involving NOT types (including types such as ATOM which are
4042 represented internally using NOT types). Thus SUBTYPEP is less
4043 likely to return (VALUES NIL NIL) in general, and in particular
4044 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4046 * The fasl file format has changed again, because the internal
4047 representation of types now includes a new slot to support the new
4048 SUBTYPEP-of-NOT-types logic.
4049 * (not a change in the main branch of SBCL, but a related prototype
4050 which can hopefully be merged into the main branch of SBCL in the
4051 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4052 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4053 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4054 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4055 dispatch macro character combinations. (thanks to Alexey Dejneka)
4056 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4057 that superficially look like logical namestrings correctly.
4058 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4060 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4061 default supplied-p) for &optional and &key arguments. (thanks to
4064 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4065 * incompatible change: The compiler is now less aggressive about
4066 tail call optimization, doing it only when (> SPACE DEBUG) or
4067 (> SPEED DEBUG). (This is an incompatible change because there are
4068 programs which relied on the old CMU-CL-style behavior to optimize
4069 away their unbounded recursion which will now die of stack overflow.)
4070 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4071 for non-GENCGC systems has been increased to 20M (since that
4072 seems much closer to the likely performance optimum for modern
4073 systems than the old 4M value was)
4074 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4075 and *DEBUG-PRINT-LEVEL*
4076 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4077 of CMU CL's support for SPARC, and various endianness and other
4078 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4079 * new syntactic sugar for the Unix command line: --load foo.bar is now
4080 an alternate notation for --eval '(load "foo.bar")'.
4082 ** The system now detects stack overflow and handles it gracefully,
4083 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4084 optimization settings. (This is a good thing in general, and
4085 its introduction in this version should be particularly timely
4086 for anyone whose code fails because of suppression of tail
4088 ** The system now hunts for the C variable "environ" in a more
4089 devious way, to avoid segfaults when the C library version differs
4090 between compile time and run time. (thanks to Christophe Rhodes)
4091 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4092 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4093 ** The compiler no longer issues bogus style warnings for undefined
4094 classes in the same source file as the DEFCLASSes which defined
4095 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4097 ** fixes in CONDITION class precedence list for undefined function
4098 errors (thanks to Alexei Dejneka)
4099 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4100 correctly. (thanks to Dan Barlow)
4101 ** portability fixes aiming at bootstrapping under CLISP (thanks
4102 to Dave McDonald and Christophe Rhodes)
4103 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4104 ** fixes in type translation and and type inference (thanks to
4106 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4107 ** various fixes in the new ports (thanks to Dan Barlow)
4108 * several changes related to debugging:
4109 ** suppression of tail recursion, as noted above
4110 ** stack overflow detection, as noted above
4111 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4112 is now the default. (For some time encapsulation has been more
4113 reliable than the breakpoint-based :ENCAPSULATE NIL
4114 implementation, at least on X86 systems; and I just noticed that
4115 encapsulation also seems closer to the spirit of the ANSI
4118 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4120 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4121 up properly again. (There was a packaging bug in 0.7.0 which
4122 left their definitions in SB-SYS::LOAD-FOREIGN and
4123 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4124 vital for most things which interface to C-level interfaces,
4125 like extensions working with sockets or databases or
4126 Perl-compatible regexes or whatever, and the need to fix
4127 this bug is the main reason that 0.7.1 was released so
4129 ** DEFGENERIC is now choosier about the methods it redefines, so that
4130 reLOADing a previously-LOADed file containing DEFGENERICs does
4131 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4132 cycle now works as it should. (thanks to Alexey Dejneka)
4133 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4134 fixed by Christophe Rhodes. (He actually submitted this patch
4135 months ago, and I delayed until after 0.7.0.)
4136 ** Bug 111 (internal compiler confusion about runtime checks on
4137 FUNCTION types) was fixed by Alexey Dejneka.
4138 * Some internal cleanups (getting rid of variables which aren't
4139 needed now that the byte interpreter is gone) caused the fasl
4140 file format number to change again.
4142 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4143 * major incompatible change: The default fasl file extension, i.e. the
4144 default extension for files produced by COMPILE-FILE, has changed
4145 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4147 ** There are many changes in the implementation of the compiler.
4148 SBCL is now essentially a compiler-only implementation of ANSI
4149 Common Lisp. EVAL still "interprets" a few special cases, but
4150 almost all the interesting cases are handled by creating
4151 a LAMBDA expression, calling COMPILE on it, then calling
4152 FUNCALL on the result.
4153 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4154 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4155 Since the code is newer, there might still be some new bugs
4156 (though not as many as before Martin Atzmueller's fixes:-). But
4157 the new code is substantially simpler and clearer, and hopefully
4158 any remaining bugs will be simpler, less fundamental, and more
4159 fixable then the bugs in the old code.
4160 ** The revised compiler is still a little unsteady on its feet.
4162 *** The debugging information it produces (particularly the names
4163 of FUNCTION objects) is sometimes much less useful than what
4164 the old compiler produced.
4165 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4166 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4167 within a MACROLET) has been temporarily weakened.
4168 ** There are new compiler optimizations for various functions:
4169 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4170 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4171 *** the math functions TRUNCATE, FLOOR, and CEILING
4172 *** the function-of-all-trades COERCE
4173 Mostly these should be transparent, but there's one
4174 potentially-annoying problem (bug 117): when the compiler
4175 inline-expands a function and does type analysis on the result,
4176 it can create control paths which have type mismatches, and
4177 when it can't prove that those control paths aren't taken,
4178 it will issue WARNINGs about the type mismatches. This is
4179 a particular problem in practice for the new sequence functions.
4180 It's not clear how this should be fixed, and for now, a
4181 workaround is given in the entry for 117 in the BUGS file.
4182 ** (Because of the interaction between the two previous items --
4183 occasional inlining problems and new inline expansions -- some
4184 of the new sequence function optimizations won't really kick in
4185 completely until debugging information, and then inlining, are
4186 straightened out in some future version.)
4187 * minor incompatible changes:
4188 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4189 DIRECTORY no longer implicitly promotes NIL slots of its
4190 pathname argument to :WILD. In particular, when you ask for the
4191 contents of a directory (which you used to be able to do without
4192 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4193 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4194 ** changes in behavior that ANSI explicitly defines to be
4195 implementation dependent:
4196 *** The new compiler-only implementation still conforms with ANSI,
4197 but acts a little different than before. Besides the obvious
4198 changes in performance tradeoffs (that the cost per form passed
4199 to EVAL has gone up, and the cost per form executed by EVAL
4200 has gone down), the behavior of the system changes a little
4201 because there are no longer any interpreted function objects.
4202 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4203 e.g. doing COMPILE on the output of interactive DEFUN is
4205 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4207 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4208 has changed from (:CL) to NIL.
4209 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4210 e.g. #\Newline and #\Tab, don't have names specified in the
4211 ANSI Common Lisp standard, is now based on their ASCII symbolic
4212 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4213 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4214 are no longer used for output.
4215 ** changes in internal implementation constants:
4216 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4217 4 million. (If your application spends a lot of time GCing and
4218 you have a lot of RAM, you might want to experiment with
4219 increasing it even more.)
4220 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4221 However, almost all old code should still continue to work without
4222 immediate update, as SB-C-CALL is now a (deprecated) nickname
4224 ** Old operator names in the style DEF-FOO are now deprecated in
4225 favor of new corresponding names DEFINE-FOO, for consistency with
4226 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4227 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4228 internal symbols, but a few supported extensions like
4229 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4230 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4231 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4232 etc. as you get deeper into recursive calls to the debugger
4233 command loop, instead of the old "5]", "5]]", "5]]]"
4234 sequence. (I was motivated to do this when squabbles between
4235 ILISP and SBCL left me very deeply nested in the debugger. In the
4236 short term, this change will probably provoke more ILISP/SBCL
4237 squabbles, but hopefully it will be an improvement in the long run.)
4238 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4239 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4240 function, since declaiming return types involving aliens is
4241 (1) annoyingly messy to do by hand and (2) vital to efficient
4242 compilation of code which calls such functions.
4243 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4244 longer reexported by the SB-EXT package. They're solely useful
4245 for alien code, so it seems more logical that you should get
4246 them from the SB-ALIEN package, not in SB-EXT.
4247 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4248 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4249 features. Instead, the code that they used to control is always
4250 built into the system.
4251 * many other bug fixes
4252 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4253 advantage of the new EVAL-WHEN stuff and to clean them up in
4254 general, and they are now more ANSI-compliant in a number of
4255 ways. Martin Atzmueller is responsible for a lot of this.
4256 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4258 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4259 *** correct ERROR type for various file operations
4260 *** some fixes for Lisp streams
4261 *** DEFMETHOD syntax checking
4262 *** changing old weird representation of debug information as
4263 strings (which, among their other deficiencies, don't transform
4264 correctly when you rename packages, and don't change their
4265 print representation when you change things like *PACKAGE*
4266 and *PRINT-LENGTH*) to symbols and lists of symbols
4267 He also made several improvements and fixed several bugs in DESCRIBE.
4268 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4270 *** misbehavior of WRITE-STRING/WRITE-LINE
4271 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4272 and several other LOOP problems as well
4273 *** DIRECTORY when similar filenames are present
4274 *** DEFGENERIC with :METHOD options
4275 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4276 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4277 *** argument ordering in FIND with :TEST option
4278 *** mishandled package designator argument in APROPOS-LIST
4279 *** various problems in the backquote readmacro
4280 *** a bug in APROPOS
4281 *** probably some others that I'm not describing very well here,
4282 since the CVS log documents them by reference to sbcl-devel
4283 messages, and the SourceForge archives aren't working well.:-(
4284 ** Dan Barlow improved the Alpha port (and is making progress on the
4285 PPC port, for those of you who think different).
4286 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4287 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4288 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4289 continues to work on the SPARC port (for those of you in a position
4290 to look down upon our little PC-compatible boxes from a great height).
4291 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4292 and uses that copy, rather than the current dynamic value, when
4293 it's trying to decide whether to truncate output. Thus e.g.
4294 (let ((*print-lines* 50))
4295 (pprint-logical-block (stream nil)
4297 (let ((*print-lines* 8))
4298 (print (aref possiblybigthings i) stream)))))
4299 should now truncate the logical block only at 50 lines, instead of
4300 often truncating it at 8 lines, as it did before.
4301 * The doc/cmucl/ directory, containing old CMU CL documentation
4302 from the time of the fork, is no longer part of the base system.
4303 SourceForge has shut down its anonymous FTP service, and with it
4304 my original plan for distributing the old CMU CL documentation
4305 there. For now, if you need these files you can download an old
4306 SBCL source release and extract them from it.
4307 * The fasl file version number changed again, for dozens of reasons,
4308 some of which are apparent above.
4310 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4311 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4312 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4313 patch, so that DISASSEMBLE output is much nicer.
4314 * The code in the SB-PROFILE package now seems reasonably stable.
4315 I still haven't decided what the final interface should look like
4316 (I'd like PROFILE to interact cleanly with TRACE, since both
4317 facilities use function encapsulation) but if you have a need
4318 for profiling now, you can probably use it successfully with
4319 the current CMU-CL-style interface.
4320 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4321 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4322 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4323 link now returns the dangling link itself, and for similar
4324 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4325 link itself. (In these cases the old code signalled an error and
4326 looped endlessly, respectively.) Thus, DIRECTORY now works even
4327 in the presence of dangling and cyclic symbolic links.
4328 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4329 is now a supported extension again, since the consensus on
4330 sbcl-devel was that it can be useful for ordinary development
4331 work, not just for debugging SBCL itself.
4332 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4333 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4334 that currently-defined functions might be redefined later with
4335 different return types.
4336 * Hash tables can be printed readably, as inspired by CMU CL code
4337 of Eric Marsden and SBCL code of Martin Atzmueller.
4338 * better error handling in CLOS method combination, thanks to
4339 Martin Atzmueller porting Pierre Mai's CMU CL patches
4340 * more overflow fixes for >16Mbyte I/O buffers
4341 * A bug in READ has been fixed, so that now a single Ctrl-D
4342 character suffices to cause end-of-file on character streams.
4343 In particular, now you only need one Ctrl-D at the command
4344 line (not two) to exit SBCL.
4345 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4347 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4348 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4349 does the right thing.
4350 * The implementation of some type tests, especially for CONDITION
4351 types, is now tidier and maybe faster, due to CMU CL code
4352 originally by Douglas Crosher, ported by Martin Atzmueller.
4353 * Some math functions have been fixed, and there are new
4354 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4355 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4356 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4357 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4358 likely to be built into sbcl-0.7.0.)
4359 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4360 CL, ported to SBCL by Martin Atzmueller)
4361 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4362 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4363 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4364 * There's a new slam.sh hack to shorten the edit/compile/debug
4365 cycle for low-level changes to SBCL itself, and a new
4366 :SB-AFTER-XC-CORE target feature to control the generation of
4367 the after-xc.core file needed by slam.sh.
4368 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4369 COMPILE-FILE is no longer supported, so that now every function
4370 gets an entry point, so that block compilation looks a little
4371 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4372 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4373 returns the number of bytes consed since the system started,
4374 rather than the number consed since the first time the function
4375 was called. (The new definition parallels ANSI functions like
4376 CL:GET-INTERNAL-RUN-TIME.)
4377 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4378 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4379 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4380 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4381 * Fasl file version numbers are now independent of the target CPU,
4382 since historically most system changes which required version
4383 number changes have affected all CPUs equally. Similarly,
4384 the byte fasl file version is now equal to the ordinary
4387 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4388 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4389 is no longer recognized. I apologize for this, because it was
4390 listed in SB-EXT as a supported extension, but I found that
4391 its existing behavior was poorly specified, as well as incorrectly
4392 specified, and it looked like too much of a mess to straighten it
4393 out. I have enough on my hands trying to get ANSI stuff to work..
4394 * many patches ported from CMU CL by Martin Atzmueller, with
4395 half a dozen bug fixes in pretty-printing and the debugger, and
4396 half a dozen others elsewhere
4397 * fixed bug 13: Floating point infinities are now supported again.
4398 They might still be a little bit flaky, but thanks to bug reports
4399 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4400 as flaky as they were.
4401 * The --noprogrammer command line option is now supported. (Its
4402 behavior is slightly different in detail from what the old man
4403 page claimed it would do, but it's still appropriate under the
4404 same circumstances that the man page talks about.)
4405 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4406 are now supported, and enabled by default. Thus, the compiler can
4407 handle many floating point and complex operations much less
4408 inefficiently. (Thus e.g. you can implement a complex FFT
4410 * The compiler now detects type mismatches between DECLAIM FTYPE
4411 and DEFUN better, and implements CHECK-TYPE more correctly, and
4412 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4413 thanks to patches from Martin Atzmueller.
4414 * various fixes to make the cross-compiler more portable to
4415 ANSI-conforming-but-different cross-compilation hosts (notably
4416 Lispworks for Windows, following bug reports from Arthur Lemmens)
4417 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4418 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4419 fixed thanks to Pierre Mai's CMU CL patches.
4420 * fixes to keep the system from overflowing internal counters when
4421 it tries to use i/o buffers larger than 16M bytes
4422 * fixed bug 45a: Various internal functions required to support
4423 complex special functions have been merged from CMU CL sources.
4424 (When I was first setting up SBCL, I misunderstood a compile-time
4425 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4426 * improved support for type intersection and union, fixing bug 12
4427 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4428 more obscure bugs as well
4429 * some steps toward byte-compiling non-performance-critical
4430 parts of the system, courtesy of patches from Martin Atzmueller
4431 * Christophe Rhodes has made some debian packages of sbcl at
4432 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4433 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4434 stable, but are nonetheless usable. When he's ready, I'd be happy
4435 to add them to the SourceForge "File Releases" section. (And if
4436 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4437 * new fasl file format version number (because of changes in
4438 internal representation of (OR ..) types to accommodate the new
4439 support for (AND ..) types, among other things)
4441 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4442 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4444 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4446 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4447 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4448 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4449 * DESCRIBE now works on structure objects again.
4450 * Most function call argument type mismatches are now handled as
4451 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4452 know whether the function will be redefined before the call is
4453 executed. (The compiler could flag local calls with full WARNINGs,
4454 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4455 it doesn't keep track of enough information to know whether calls
4456 are local in this sense.)
4457 * Compiler output is now more verbose, with messages truncated
4458 later than before. (There should be some supported way for users
4459 to override the default verbosity, but I haven't decided how to
4460 provide it yet, so this behavior is still controlled by the internal
4461 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4462 src/compiler/ir1util.lisp.)
4463 * Fasl file format version numbers have increased again, because
4464 support for the Gray streams extension changes the layout of the
4465 system's STREAM objects.
4466 * The Gray subclassable streams extension now works, thanks to a
4467 patch from Martin Atzmueller.
4468 * The full LOAD-FOREIGN extension (not just the primitive
4469 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4470 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4471 but like most other programs, it defaults to copying the Unix
4472 environment from the original process instead of starting the
4473 new process in an empty environment.
4474 * Extensions which manipulate the Unix environment now support
4475 an :ENVIRONMENT keyword option which doesn't smash case or
4476 do other bad things. The CMU-CL-style :ENV option is retained
4477 for porting convenience.
4478 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4479 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4481 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4483 * A patch from Martin Atzmueller seems to have solved the SIGINT
4484 problem, and as far as we know, signal-handling now works cleanly.
4485 (If you find any new bugs, please report them!)
4486 * The system no longer defaults Lisp source file names to types
4487 ".l", ".cl", or ".lsp", but only to ".lisp".
4488 * The compiler no longer uses special default file extensions for
4489 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4490 seems to expect a single default extension for all compiled code,
4491 and there's no compelling reason to try to stretch the standard
4492 to allow two different extensions.) Instead, byte-compiled files
4493 default to the same extension as native-compiled files.
4494 * Fasl file format version numbers have increased again, because
4495 a rearrangement of internal implementation packages made some
4496 dumped symbols in old fasl files unreadable in new cores.
4497 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4498 many fewer weird special cases.
4499 * Bug #17 (differing COMPILE-FILE behavior between logical and
4500 physical pathnames) has been fixed, and some related misbehavior too,
4501 thanks to a patch from Martin Atzmueller.
4502 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4503 by Tim Moore, ported to SBCL by Martin Atzmueller.
4504 * Martin Atzmueller fixed several filesystem-related problems,
4505 including bug #36, in part by porting CMU CL patches, which were
4506 written in part by Paul Werkowski.
4507 * More compiler warnings in src/runtime/ are gone, thanks to
4508 more patches from Martin Atzmueller.
4509 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4512 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4514 * DESCRIBE now works on CONDITION objects.
4515 * The debugger now handles errors which arise when trying to print
4516 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4518 * The build system now uses an additional file, customize-target-features.lisp,
4519 to allow local modifications to the target *FEATURES* list. (The point of
4520 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4521 debugging features enabled, without having to worry about propagating it
4522 into everyone's system when I do a "cvs update".) When no
4523 customize-target-features.lisp file exists, the target *FEATURES* list
4524 should be constructed the same way as before.
4525 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4526 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4527 warn about behavior which is undefined under the ANSI spec. Note: This
4528 is specified by ANSI, but it's not very popular with programmers.
4529 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4530 macro in the SBCL sources for an example of a workaround which you
4531 might use to make portable ANSI-standard code which does what you want.
4532 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4533 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4534 when it's used as a non-toplevel form. (This is required in order
4535 to implement the DEFCONSTANT-EQX macro.)
4536 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4537 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4538 funny things, due to bugs in the implementation of EVAL-WHEN
4539 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4540 non-toplevel DEFCONSTANTs are uncommon.)
4541 * The core file version number and fasl file version number have been
4542 incremented, because the old noncompliant DEFCONSTANT behavior involved
4543 calling functions which no longer exist, and because I also took the
4544 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4545 * fixed bug 1 (error handling before read-eval-print loop starts), and
4546 redid debugger restarts and related debugger commands somewhat while
4548 ** The QUIT debugger command is gone, since it did something
4549 rather different than the SB-EXT:QUIT command, and since it never
4550 worked properly outside the main toplevel read/eval/print loop.
4551 Invoking the new TOPLEVEL restart provides the same functionality.
4552 ** The GO debugger command is also gone, since you can just invoke
4553 the CONTINUE restart directly instead.
4554 ** The TOP debugger command is also gone, since it's redundant with the
4555 FRAME 0 command, and since it interfered with abbreviations for the
4557 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4559 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4560 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4561 some porting work by Martin Atzmueller)
4562 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4563 with a boatload of patches which he ported to SBCL. Now that those
4565 ** The system tries to make sure that its low-priority messages
4566 are prefixed by semicolons, to help people who like to use
4567 syntax highlighting in their ILISP buffer. (This patch
4568 was originally due to Raymond Toy.)
4569 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4570 patches originally written by Raymond Toy.
4571 ** The compiler understands coercion between single-value and
4572 multiple-VALUES type expressions better, getting rid of some very
4573 weird behavior, thanks to patches originally by Robert MacLachlan
4574 and Douglas Crosher.
4575 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4576 lambda lists, thanks to a patch originally by Pierre Mai.
4577 ** The system no longer bogusly warns about "abbreviated type
4579 ** The compiler gets less confused by inlining and RETURN-FROM,
4580 thanks to some patches originally by Tim Moore.
4581 ** The system no longer hangs when dumping circular lists to fasl
4582 files, thanks to a patch originally from Douglas Crosher.
4583 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4584 undefined function error.
4585 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4586 in structure slot initforms) for now, documented workaround instead:-|
4587 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4588 * fixed bug 5: added stubs for various Gray stream functions called
4589 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4590 installed, at least appropriate type errors are generated
4591 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4592 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4593 consistently in DEFMETHOD forms.
4594 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4595 it doesn't seem to affect SBCL after all
4596 * The C runtime system now builds with better optimization and many
4597 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4599 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4601 * The system is now under CVS at SourceForge (instead of the
4602 CVS repository on my home machine).
4603 * The new signal handling code has been tweaked to treat register
4604 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4605 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4606 downstream, e.g. in debug-int.lisp, has implicit dependencies
4607 on the unsignedness of integer representation of machine words,
4608 and that caused the system to bomb out with infinite regress
4609 when trying to recover from type errors involving signed values,
4610 e.g. (BUTLAST '(1 2 3) -1).)
4611 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4612 (This was one of the bugs Peter Van Eynde reported back in July.)
4613 * The system now uses code inspired by Colin Walters' O(N)
4614 implementation of MAP (from the cmucl-imp@cons.org mailing
4615 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4616 inline the MAP operation, and there is more than one
4617 sequence argument to the MAP call (so that it can't just
4618 do ETYPECASE once and for all based on the type of the
4619 single sequence argument). (The old non-inline implementation
4620 of the general M-argument sequence-of-length-N case required
4621 O(M*N*N) time when any of the sequence arguments were LISTs.)
4622 * The QUIT :UNIX-CODE keyword argument has been renamed to
4623 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4624 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4626 * Raymond Wiker's patches to port dynamic loading from Linux to
4627 FreeBSD have been added.
4628 * The BUGS file is now more nearly up to date, thanks in large part
4629 to Martin Atzmueller's review of it.
4630 * The debugger now flushes standard output streams before it begins
4631 its output ("debugger invoked" and so forth).
4632 * The core version number and fasl file version number have both
4633 been incremented, because of incompatible changes in the layout
4635 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4636 used to not be called for a saved Lisp image.)
4637 * Martin Atzmueller's version of a patch to fix a compiler crash,
4638 as posted on sbcl-devel 13 September 2000, has been installed.
4639 * Instead of installing Martin Atzmueller's patch for the
4640 compiler transform for SUBSEQ, I deleted the compiler transform,
4641 and transforms for some similar consing operations.
4642 * A bug in signal handling which kept TRACE from working on OpenBSD
4644 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4645 away by constant folding
4646 * The system now defines its address space constants in one place
4647 (in the Lisp sources), and propagates them automatically elsewhere
4648 (through GENESIS and the sbcl.h file). Therefore, patching the
4649 address map is less unnecessarily tedious and error-prone. The
4650 Lisp names of address space constants have also been systematized.
4651 * CVS tags like dollar-Header-dollar have been removed from
4652 the sources, because they have never saved me trouble and
4653 they've been source of trouble working with patches and other
4654 diff-related operations.
4655 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4656 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4658 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4660 * The system has been ported to OpenBSD.
4661 * The system now compiles with a simple "sh make.sh" on the systems
4662 that it's supported on. I.e., now you no longer need to tweak
4663 text in the target-features.lisp-expr and symlinks in src/runtime/
4664 by hand, the make.sh takes care of it for you.
4665 * The system is no longer so grossly inefficient when compiling code
4666 involving vectors implemented as general (not simple) vectors (VECTOR T),
4667 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4668 which dares to use the various sequence functions on non-simple
4669 vectors, takes less of a performance hit.
4670 * There is now a primitive type predicate VECTOR-T-P
4671 to test for the (VECTOR T) type, so that e.g.
4672 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4673 can now be compiled with some semblance of efficiency. (The old code
4674 turned the type declaration into a full call to %TYPEP at runtime!)
4675 * AREF on (VECTOR T) is still not fast, since it's still compiled
4676 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4677 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4678 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4679 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4680 without making full calls to SUBTYPEP at runtime.
4681 (Some analogous efficiency issues for non-simple vectors specialized to
4682 element types other than T, or for non-simple multidimensional arrays,
4683 have not been addressed. They could almost certainly be handled the
4684 same way if anyone is motivated to do so.)
4685 * The changes in array handling break binary compatibility, so
4686 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4687 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4688 instead of (VALUES T T).
4689 * By following the instructions that Dan Barlow posted to sbcl-devel
4690 on 2 July 2000, I was able to enable primitive dynamic object
4691 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4692 functionality is not implemented (since it calls ld to resolve
4693 library references automatically, requiring RUN-PROGRAM for its
4694 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4695 try to resolve library references) is now supported.
4696 * The system now flushes the standard output streams when it terminates,
4697 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4698 them at several other probably-convenient times, e.g. in each pass of
4699 the toplevel read-eval-print loop, and after evaluating a form given
4700 as an "--eval" command-line option. (These changes were motivated by a
4701 discussion of stream flushing issues on cmucl-imp in August 2000.)
4702 * The source transform for TYPEP of array types no longer assumes
4703 that an array whose element type is a not-yet-defined type
4704 is implemented as an array of T, but instead punts, so that the
4705 type will be interpreted at runtime.
4706 * There is now some support for cross-compiling in make.sh: each of
4707 the phases of make.sh has its own script. (This should be transparent
4708 to people doing ordinary, non-cross-compile builds.)
4709 * Since my laptop doesn't have hundreds of megabytes of memory like
4710 my desktop machine, I became more motivated to do some items on
4711 my to-do list in order to reduce the size of the system a little:
4712 ** Arrange for various needed-only-at-cold-init things to be
4713 uninterned after cold init. To support this, those things have
4714 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4715 symbols with such names are now uninterned after cold init).
4716 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4717 fasl files for cold load.
4718 ** Remove the old compiler structure pooling code (which used to
4719 be conditional on the target feature :SB-ALLOC) completely.
4720 ** Redo the representation of some data in cold init to be more compact.
4721 (I also looked into supporting byte compiled code at bootstrap time,
4722 which would probably reduce the size of the system a lot, but that
4723 looked too complicated, so I punted for now.)
4724 * The maximum signal nesting depth in the src/runtime/ support code has
4725 been reduced from 4096 to 256. (I don't know any reason for the very
4726 large old value. If the new smaller value turns out to break something,
4727 I'll probably just bump it back up.)
4728 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4730 * Many, many bugs reported by Peter Van Eynde have been added to
4731 the BUGS list; some have even been fixed.
4732 * While enabling dynamic object file loading, I tried to make the
4733 code easier to understand, renaming various functions and variables
4734 with less ambiguous names, and changing some function calling
4735 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4736 * While trying to figure out how to do the OpenBSD port, I tried to
4737 clean up some of the code in src/runtime/. In particular, I dropped
4738 support for non-POSIX signal handling, added various comments,
4739 tweaked the code to reduce the number of compilation warnings, and
4740 renamed some files to increase consistency.
4741 * To support the new automatic configuration functionality in make.sh,
4742 the source file target-features.lisp-expr has been replaced with the
4743 source file base-target-features.lisp-expr and the machine-generated
4744 file local-target-features.lisp-expr.
4745 * fixed a stupid quoting error in make.sh so that using CMU CL
4746 "lisp -batch" as cross-compilation host works again
4748 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4750 * DESCRIBE no longer tries to call itself recursively to describe
4751 bound/fbound values, so that it no longer fails on symbols which are
4752 bound to themselves (like keywords, T, and NIL).
4753 * DESCRIBE now works on generic functions.
4754 * The printer now prints less-screwed-up representations of closures
4755 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4756 * A private symbol is used instead of the :EMPTY keyword previously
4757 used to mark empty slots in hash tables. Thus
4758 (DEFVAR *HT* (MAKE-HASH-TABLE))
4759 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4760 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4761 now does what ANSI says that it should. (You can still get
4762 similar noncompliant behavior if bang on the hash table
4763 implementation with all the symbols you get back from
4764 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4765 This breaks binary compatibility, since tests for equality to
4766 :EMPTY are wired into things like the macroexpansion of
4767 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4769 * There's now a minimal placeholder implementation for CL:STEP,
4770 as required by ANSI.
4771 * An obscure bug in the interaction of the normal compiler, the byte
4772 compiler, inlining, and structure predicates has been patched
4773 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4774 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4776 * Missing ordinary arguments in a macro call are now detected even
4777 when the macro lambda list contains &KEY or &REST.
4778 * The debugger no longer complains about encountering the top of the
4779 stack when you type "FRAME 0" to explicitly instruct it to go to
4780 the top of the stack. And it now prints the frame you request even
4781 if it's the current frame (instead of saying "You are here.").
4782 * As specified by ANSI, the system now always prints keywords
4783 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4785 * The default initial SIZE of HASH-TABLEs is now smaller.
4786 * Type information from CLOS class dispatch is now propagated
4787 into DEFMETHOD bodies, so that e.g.
4788 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4790 is now basically equivalent to
4791 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4792 (DECLARE (TYPE SINGLE-FLOAT X))
4794 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4795 operation, without having to do run-time type dispatch.
4796 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4797 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4798 * Since I don't seem to be making big file reorganizations very often
4799 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4800 is overflowing my ability to conveniently back them up), I've finally
4801 checked the system into CVS. (The CVS repository is on my home system,
4802 not at SourceForge -- putting it on SourceForge might come later.)
4803 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4804 high-level GC-NOTIFY-FOO functions send their output. (There's
4805 still very little control of where low-level verbose GC functions
4806 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4807 less than it used to -- the GC-NOTIFY-FOO functions are now under
4808 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4809 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4810 in only one place in the source code, and propagates it automatically
4811 everywhere that it's needed. Thus e.g. when I bump the version from
4812 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4813 * The C source files now include boilerplate legalese and documentation
4814 at the head of each file (just as the Lisp source files already did).
4815 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4816 to his page will be replaced with a link to his new CLiki service.
4818 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4820 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4821 * The build process now looks for GNU make under the default name "gmake",
4822 instead of "make" as it used to. If GNU make is not available as "gmake"
4823 on your system, you can change this default behavior by setting the
4824 GNUMAKE environment variable.
4825 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4826 can build without error under CMU CL.
4828 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4830 * There is now a partial SBCL user manual (with some new text and some
4831 text cribbed from the CMU CL manual).
4832 * The beginnings of a profiler have been added (starting with the
4833 CMU CL profiler and simplifying and cleaning up). Eventually the
4834 main interface should be through the TRACE macro, but for now,
4835 it's still accessed through vaguely CMU-CL-style functions and macros
4836 exported from the package SB-PROFILE.
4837 * Some problems left over from porting CMU CL to the new
4838 cross-compilation bootstrap process have been cleaned up:
4839 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4840 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4841 part when getting rid of the compiler *BACKEND* stuff.)
4842 ** The value of *NULL-TYPE* was screwed up, because it was
4843 being initialized before the type system knew the final
4844 definition of the 'NULL type. This screwed up several key
4845 optimizations in the compiler, causing inefficiency in all sorts
4846 of places. (I found it because I wanted to understand why
4847 GET-INTERNAL-RUN-TIME was consing.)
4848 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4849 PROCLAIM FTYPE information. Unfortunately this broke binary
4850 compatibility again, since now the forms output by DEFGENERIC
4851 to refer to functions which didn't exist in 0.6.3.
4852 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4853 can use the new (as of 0.6.3) transform for SOME into MAP into
4855 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4856 values to (INTEGER 0 1000000), so that the time code will no longer
4857 occasionally get blown up by Linux returning 1000000 microseconds
4858 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4859 its output conform to the ANSI spec. (Alas, this makes its output
4860 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4861 * A full call to MAP NIL with a single sequence argument no longer conses.
4862 * fixes to problems pointed out by Martin Atzmueller:
4863 * The manual page no longer talks about multiprocessing as though
4864 it were currently supported.
4865 * The ILISP support patches have been removed from the distribution,
4866 because as of version 5.10.1, ILISP now supports SBCL without us
4867 having to maintain patches.
4868 * added a modified version of Raymond Toy's recent CMU CL patch for
4869 EQUALP comparison of HASH-TABLE
4871 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4873 * The system still can't cross-compile itself with
4874 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4875 entails), but at least it can compile more of itself that way
4876 than it used to be able to, and various buglets which were uncovered
4877 by trying to cross-compile itself that way have now been fixed.
4878 * This release breaks binary compatibility again. This time
4879 at least I've incremented the FASL file format version to 2, so that the
4880 problem can be detected reliably instead of just causing weird errors.
4881 * various new style warnings:
4882 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4883 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4884 ** using the lexical binding of a variable named in the *FOO* style
4885 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4886 as specified by ANSI.
4887 * *RANDOM-STATE* is no longer automatically initialized from
4888 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4889 default behavior of the system is to repeat its behavior every time
4890 it's run. If you'd like to change this behavior, you can always
4891 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4892 old convention there was no comparably easy way to get the system to
4893 repeat its behavior every time it was run.
4894 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4895 FTYPE declarations has been removed, in favor of their ANSI
4896 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4897 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4898 the types of their sequence arguments can be determined at compile time.
4899 This is done through a new open code expansion for MAP which eliminates
4900 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4901 argument types can be determined at compile time.
4902 * The optimizer now transforms COERCE into an identity operation when it
4903 can prove that the coerced object is already of the correct type. (This
4904 can be a win for machine generated code, including the output of other
4905 optimization transforms, such as the MAP transform above.)
4906 * Credit information has been moved from source file headers into CREDITS.
4907 * Source file headers have been made more standard.
4908 * The CASE macro now compiles without complaining even when it has
4911 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4913 * (Note that the way that the PCL macroexpansions were rewritten
4914 to accommodate the change in DEFGENERIC below breaks binary
4915 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4916 not run under sbcl-0.6.2. Once we get out of alpha releases,
4917 i.e. hit release 1.0.0, we'll probably try to maintain binary
4918 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4919 and sbcl-1.4.4. Until then, however, it might be fairly common
4920 for maintenance releases to break binary compatibility.)
4921 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4923 * The reporting of unhandled signals has been changed to print some
4924 explanatory text as well as the report form. (Previously only
4925 the report form was printed.)
4926 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4927 it defines, so that the compiler no longer issues undefined function
4928 warnings for compiled-but-not-yet-loaded generic functions.
4929 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4930 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4931 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4932 section of the ANSI spec are supported.
4933 * The "" nickname for the "KEYWORD" package has been removed.
4934 The reader still handles symbol tokens which begin with a package marker
4935 as keywords, but it doesn't expose its mechanism for doing so in the
4936 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4937 * The system now issues STYLE-WARNINGs for contradictory TYPE
4938 proclamations. (Warnings for contradictory FTYPE proclamations would
4939 be nice too, but those can't be done usefully unless the type system
4940 is made smarter about FUNCTION types.)
4941 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4942 systematized, so that "*target-*.lisp is supposed to exist only on the
4943 target and imply that there's a related file which exists on the
4944 host, and *host-*.lisp is supposed to exist only on the host and imply
4945 that there's a related file which exists on the target. This involves a
4946 lot of renaming. Hopefully the acute confusion caused by the renaming
4947 will be justified by the reduction in chronic confusion..
4948 ** runtime-type.lisp -> early-target-type.lisp
4949 ** target-type.lisp -> late-target-type.lisp
4950 ** early-host-format.lisp -> early-format.lisp
4951 ** late-host-format.lisp -> late-format.lisp
4952 ** host-error.lisp -> misc-error.lisp
4953 ** early-error.lisp -> early-target-error.lisp
4954 ** late-error.lisp -> late-target-error.lisp
4955 ** host-defboot.lisp -> early-defboot.lisp
4956 ** code/misc.lisp -> code/target-misc.lisp
4957 ** code/host-misc.lisp -> code/misc.lisp
4958 ** code/numbers.lisp -> code/target-numbers.lisp
4959 ** code/early-numbers.lisp -> numbers.lisp
4960 ** early-host-type.lisp -> early-type.lisp
4961 ** late-host-type.lisp -> late-type.lisp
4962 ** host-typep.lisp -> typep.lisp
4963 ** load.lisp -> target-load.lisp
4964 ** host-load.lisp -> load.lisp
4965 ** host-disassem.lisp -> disassem.lisp
4966 ** host-insts.lisp -> insts.lisp
4967 ** byte-comp.lisp -> target-byte-comp.lisp
4968 ** host-byte-comp.lisp -> byte-comp.lisp
4969 ** host-signal.lisp -> signal.lisp
4970 ** host-defstruct.lisp -> defstruct.lisp
4971 ** late-target-type.lisp -> deftypes-for-target.lisp
4972 Furthermore, several other previously target-only files foo.lisp (e.g.
4973 hash-table.lisp and random.lisp) have been split into a target-and-host
4974 foo.lisp file and a target-only target-foo.lisp file, with their key type
4975 definitions in the target-and-host part, so that the cross-compiler will
4976 know more about target types.
4977 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4978 gone away. In their place are various *BACKEND-FOO* variables
4979 corresponding to the slots of the old structure.
4980 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4981 into the target SBCL has been fixed.
4982 * The chill.lisp system for loading cold code into a running SBCL
4984 * Support for the CMU CL "scavenger hook" extension has been removed.
4985 (It was undocumented and unused in the CMU CL sources that SBCL was
4986 derived from, and stale in sbcl-0.6.1.)
4987 * Various errors in the cross-compiler type system were detected
4988 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4989 (enabling various consistency checks). Many of them were fixed,
4990 but some hard problems remain, so the compiler is back to
4991 running without *TYPE-SYSTEM-INITIALIZED* for now.
4992 * As part of the cross-compiler type system cleanup, I implemented
4993 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4994 * I have started adding UNCROSS calls throughout the type system
4995 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4996 on cross-compiler input files will be able to go away, and various
4998 * CONSTANTP now returns true for quoted forms (as explicitly required
5001 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5003 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5004 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5005 (MAKE-LIST -1). In the longer term, it ought to become true
5006 that declarations are assertions even at SAFETY 1. For now, it's not
5007 quite true even at SAFETY 3, but it's at least more nearly true..
5008 (Note that this change seems to increases the size of the system by
5009 O(5%) and to decrease the speed of the compiler by 20% or more.)
5010 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5011 for the problem of printing dozens of lines of distracting information
5012 about low-level system machinery as part of the top stack frame
5013 on entry to the debugger when an undefined function was called.
5014 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5015 is not reset to COMMON-LISP-USER.
5016 * Compilation of stuff related to dyncount.lisp has been made conditional
5017 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5018 smaller. The various dyncount-related symbols have been moved into
5019 a new "SB-DYNCOUNT" package.
5020 * tty-inspect.lisp has been renamed to inspect.lisp.
5021 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5022 feature has gone away. (When we eventually port to other flavors of
5023 libc and/or Unix, we'll try to make the differences between flavors
5024 invisible at the user level.)
5025 * Various other *FEATURES* tags, and/or their associated conditionals,
5026 have been removed if obsolescent, or given better documentation, or
5027 sometimes given more-mnemonic names.
5029 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5031 * tidied up "make.sh" script
5032 * tidied up system directory structure
5033 * better "clean.sh" behavior
5034 * added doc/FOR-CMUCL-DEVELOPERS
5035 * many many small tweaks to output format, e.g. removing possibly-confusing
5036 trailing #\. character in DESCRIBE-INSTANCE
5037 * (EQUALP #\A 'A) no longer signals an error.
5038 * new hashing code, including EQUALP hashing
5039 * tidied up Lisp initialization and toplevel
5040 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5041 * command line argument processing
5042 * added POSIX-GETENV function to deal with Unix-ish environment variables
5043 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5044 terminating SBCL on EOF
5045 * non-verbose GC by default
5046 * There is no more "sbcl" shell script; the sbcl file is now the C
5047 runtime executable (just like CMU CL).
5048 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5050 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5051 symbol and package information as Lisp native symbol and package objects
5052 instead of strings naming symbols and strings naming packages. This way,
5053 whenever packages are renamed (as in warm init), debug information is
5054 transformed along with everything else.
5055 * tweaked the optimization policy declarations which control the building
5056 of SBCL itself. Now, among other things, the system no longer saves
5057 source location debugging information. (This helps two problems at once
5058 by reducing SBCL size and by keeping SBCL from trying to look for its
5059 sources -- which may not exist -- when reporting errors.)
5060 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5061 debugging and testing purposes
5062 * cleaned up printing, making the printer call PRINT-OBJECT for
5063 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5064 methods, giving nearly-ANSI behavior
5065 * converted almost all special variables to use *FOO* naming convention
5066 * deleted PARSE-TIME functionality, since it can be done portably
5067 * moved some files out of cold init into warm init
5068 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5070 * regularized formatting of source files
5071 * added an install.sh script
5072 * fixed ridiculous memory usage of cross-compiler by making
5073 compiler/alloc.lisp not try to do pooling unless it can hook
5074 itself into the GC of the cross-compilation host. Now the system
5075 builds nicely on my old laptop.
5076 * added :SB-ALLOC in target-features.lisp-expr
5077 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5078 was not implemented)
5079 * re-did condition handling and note reporting in the compiler. Notes
5080 are no longer handled by signalling conditions. Style warnings
5081 and warnings are handled more correctly and reported in such a way
5082 that it's easy to find one or the other in your output (so that you
5083 can e.g. figure out which of many problems caused COMPILE-FILE to
5085 * changed the severity of several compiler warnings from full WARNING
5086 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5087 compiler note reporting so that it doesn't use the condition system
5088 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5089 COMPILE-FILE command)
5090 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5091 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5092 no longer supported, since their implementation was deeply intertwingled
5093 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5094 * removed bogus "support" for compiler macros named (SETF FOO), and
5095 removed the compiler macro for SETF INFO (but only after making a fool
5096 of myself on the cmucl-imp mailing list by posting a bogus patch for
5097 DEFINE-COMPILER-MACRO..)
5098 * Compiled files containing forms which have side effects on the Lisp
5099 reader (such as DEFPACKAGE forms) are now handled more correctly.
5100 (Compiler queuing of top level lambdas has been suppressed by setting
5101 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5102 * deleted various currently-unused source files, e.g. gengc.lisp. They
5103 may be added back at some point e.g. when porting to other architectures,
5104 but until they are it's distracting to distribute them and to try to
5106 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5107 there were so many of them they're just distractions, and UNCROSS is
5108 known to be able to handle the current sources
5109 * moved PROFILE functionality into TRACE, so that it will be clear
5110 how the wrapping and unwrapping of functions when you profile them
5111 interacts with the wrapping and unwrapping of functions when you
5112 trace them. (Actually, the functionality isn't there yet, but at least
5113 the interface specification is there. Hopefully, the functionality will
5114 arrive with some future maintenance release.)
5115 * removed host-oops.lisp
5116 * changed signature of QUIT function to allow UNIX-CODE argument
5117 * fixed READ-SEQUENCE bug
5118 * tweaked verbose GC output so that it looks more like the progress
5119 output that ANSI specifies for functions like LOAD
5120 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5121 * added <http://sbcl.sourceforge.com> to the banner information printed by