1 ;;;; -*- coding: utf-8; -*-
2 * bug fix: lexical type declarations are now correctly reported by
3 SB-CLTL2. (reported by Larry D'Anna)
4 * bug fix: STRING-TO-OCTETS did not handle :START properly when
5 using UTF-8 as external format. (thanks to Luis Oliveira)
7 changes in sbcl-1.0.23 relative to 1.0.22:
8 * enhancement: when disassembling method functions, disassembly
9 for the associated fast function is also produced.
10 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
12 * optimization: printing with *PRINT-PRETTY* true is now more
13 efficient as long as the object being printed doesn't require
14 special handling by the pretty printer.
15 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
16 now interact correctly with type declarations.
17 * partial bug fix: PCL detects infinite recursion during wrapper
18 validation. (thanks to Attila Lendvai)
19 * bug fix: #426; nested function calls are inlined properly.
20 Previously if FOO was an inline function, in calls of the form
21 (FOO (FOO ...)) the outer call was not inlined.
22 * bug fix: long long arguments passed in registers to alien
23 functions on PowerPC Linux and NetBSD are now handled correctly
24 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
26 changes in sbcl-1.0.22 relative to 1.0.21:
27 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
28 for the shared object in the current directory, but passes the native
29 namestring of the designated pathname to the operation system's shared
30 object loading function as-it.
31 * minor incompatible change: --disable-debugger toplevel option now takes
32 effect before processing of initialization files and --eval or --load
34 * new feature: new commandline argument: --script, which supports
35 shebang lines. See documentation for details. (based on work by
37 * new feature: SAVE-LISP-AND-DIE can save current values of
38 --dynamic-space-size and --control-stack-size in the executable core,
39 causing it to skip normal runtime option processing. See documentation
40 for details. (thanks to Zach Beane)
41 * enhancement: inoccous calls to EVAL or generic functions dispatching
42 on subclasses of eg. STREAM no longer cause compiler notes to appear.
43 * enhancement: the system no longer resignals errors from --load and
44 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
45 associated with the original error to be lost. (thanks to Ariel
47 * enhancement: :DONT-SAVE keyword argument has been added to
48 LOAD-SHARED-OBJECT for controlling interaction with
49 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
50 the effects of an earlier LOAD-SHARED-OBJECT call.
51 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
52 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
53 provided. (reported by Cedric St-Jean)
54 * bug fix: circularity handling in the reader did not treat raw
55 structure slots correctly. (reported by Cedric St-Jean)
56 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
57 file descriptors when there were none.
58 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
59 was broken. (reporteed by Willem Broekema)
60 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
61 pathnames without a directory.
62 * bug fix: COMPILE-FILE did not warn about undefined variable
63 references at toplevel, and LOAD of the resulting fasl did
65 * bug fix: functions with non-required arguments used to end up with
66 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
67 * bug fix: redefining a function with non-required arguments didn't
68 update the system's knowledge about its call signature properly.
69 * bug fix: fixed #431; incompatible alien record type redefinitions
70 are detected and handled. (reported by Neil Haven)
71 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
72 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
74 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
75 after alien stack frames.
76 * bug fix: strings passed to mk{s,d}temp lacked null termination
78 changes in sbcl-1.0.21 relative to 1.0.20:
79 * new feature: the compiler is able to track the effective type of a
80 generic function across method addition and removal.
81 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
82 appropriately typed structure slots without locking.
83 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
84 information like those gathered by TIME using a programming-friendly
86 * new feature: TIME reports time taken even if the form performs a
87 non-local transfer of control.
88 * enhancement: reduced conservativism on GENCGC platforms: on
89 average 45% less pages pinned (measured from SBCL self build).
90 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
91 mutate constant symbols or violate declaimed type of the symbol.
92 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
93 sleeping on the semaphore in heavily contested semaphores.
94 * bug fix: semaphores and condition variables were not interrupt
96 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
97 owned by other threads anymore.
98 * bug fix: FIND on lists called KEY outside the specified
99 subsequence. (reported by budden)
100 * bug fix: LOG doesn't use single-float intermediate results when
101 given mixed integer and double-float arguments, leading to better
102 precision. (reported by Bob Felts)
103 * bug fix: LOG with base zero returned values of inconsistent type.
105 changes in sbcl-1.0.20 relative to 1.0.19:
106 * minor incompatible change: OPTIMIZE qualities
107 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
108 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
109 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
110 * documentation: some slot access efficiency guidelines have been
111 added to the user manual.
112 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
113 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
114 as ASSOC and MEMEBER.
115 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
116 to more efficient EQ-comparison versions more often.
117 * optimization: enhanced derivation of DOLIST iteration variable type
119 * optimization: constant folding of simple (LIST ...) forms as DOLIST
121 * optimization: runtime lookup of function definitions can be
122 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
123 * optimization: compiler is able to derive the return type of
124 (AREF (THE STRING X) Y) as being CHARACTER.
125 * optimization: CLRHASH on empty hash-tables no longer does pointless
126 work. (thanks to Alec Berryman)
127 * optimization: REPLACE deftransforms don't punt when :START1 or
129 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
130 no longer emits a code deletion note. (reported by Andrew
132 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
133 functions with an odd number of &OPTIONAL arguments, a &REST
134 argument, and one or more &KEY arguments at the call site.
135 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
136 arguments appeared at call sites as well.
137 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
138 READ-LINE &co can no longer read from them afterwards. (reported
140 * bug fix: fixed #427: unused local aliens no longer cause compiler
141 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
143 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
144 alien-stack leakage. (reported by Andy Hefner)
145 * bug fix: PROGV signals an error when an attempt to violate declared
146 type of a variable or bind a constant is made.
147 * bug fix: SET signals an error when an attempt to violate declared
148 type of a variable is made.
149 * bug fix: restart computation during the execution of a restart
150 test function no longer causes infinite recursion. (reported by
152 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
153 signals an error. (thanks to Pierre Mai)
154 * bug fix: EXPORT left symbol unexported in conflict situations.
155 (thanks to Michael Weber)
156 * bug fix: correctly handle name conflicts involving CL:NIL.
157 (thanks to Michael Weber)
158 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
159 conflicts arising from USEing package with conflicting symbols
160 correctly. (thanks to Michael Weber)
162 changes in sbcl-1.0.19 relative to 1.0.18:
163 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
164 warnings that go otherwise unhandled will be muffled if they are
165 of the type that's the value of this variable.
166 * optimization: stack allocation is slightly more efficient on x86
168 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
169 where a raw slot always is initialized using the initform whose
170 type is not know sufficiently well a compile-time are now compiled
171 correctly. (reported by John Morrison)
172 * bug fix: compiler no longer makes erronous assumptions in the
173 presense of non-foldable SATISFIES types.
174 * bug fix: stack analysis missed cleanups of dynamic-extent
175 arguments in non-let-converted calls to local functions.
176 * improvements to the Windows port:
177 ** adjusted address spaces for building on both Win32 and
178 Win64. (thanks for John Connors)
179 * fixed some bugs revealed by Paul Dietz' test suite:
180 ** interval arithmetic during type derivation used inexact integer
181 to single-float coercions.
182 ** arithmetic operations involving large integers and single
183 floats give the same results in compiled and interpreted code.
184 ** deriving the result type of COERCE no longer signals an error
185 if the derived type of the second argument is a MEMBER type
186 containing invalid type specifiers.
187 ** ADJOIN with constant NIL as second argument works correctly.
188 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
190 changes in sbcl-1.0.18 relative to 1.0.17:
191 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
192 profiles only the current thread.
193 * minor incompatible change: changes to SYMBOL-VALUE of constants
194 defined with DEFCONSTANT now signal an error.
195 * enhancement: SB-SPROF now has support for wallclock profiling,
196 and is also able to profile specific threads. REPORT output
197 has also additional sorting options.
198 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
200 * optimization: structure allocation has been improved
201 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
202 ** out of line constructors are ~10% faster.
203 ** inline constructors are ~15% faster.
204 ** inline constructors are capable of dynamic extent allocation
205 (generally on x86 and x86-64, in some cases on other platforms
207 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
209 * optimization: file compiler is now able to coalesce non-circular
210 lists, non-base strings, and bit-vectors. Additionally, constants
211 are never referenced through SYMBOL-VALUE at runtime.
212 * optimization: code defining methods on PRINT-OBJECT (and other
213 generic functions in the COMMON-LISP package) now loads faster.
214 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
216 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
217 fasl is now deleted. (reported by Attila Lendvai)
218 * bug fix: READ-LINE always returned NIL for the last line in files.
219 (reported by Yoshinori Tahara)
220 * bug fix: more accurate disassembly annotations of foreign function
221 calls. (thanks to Andy Hefner)
222 * bug fix: trimming non-simple strings and non-string string
223 designators when the there is nothing to trim works properly.
224 (thanks to James Knight)
225 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
226 (thanks to Travis Cross)
227 * fixed some bugs revealed by Paul Dietz' test suite:
228 ** NIL is a valid function name (regression at 1.0.13.38)
229 ** FILL on lists was missing its return value (regression at 1.0.12.27)
230 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
231 fill pointers properly (regression at 1.0.12.23)
232 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
233 displacement indices properly (regression at 1.0.12.23)
235 changes in sbcl-1.0.17 relative to 1.0.16:
236 * temporary regression: user code can no longer allocate closure
237 variable storage on stack, due to bug 419 without explicitly
238 requesting it. Please consult sbcl-devel for advice if you need to
239 use this feature in the meanwhile.
240 * new feature: runtime argument --control-stack-size can be used to
241 adjust thread default control stack size.
242 * enhancement: improved TIME output
243 ** all times are reported using the measured accuracy (milliseconds
244 for real and GC times, microseconds for everything else.)
245 ** processor cycle counts on x86 and x86-64.
246 ** interpreted forms are counted for both evaluator modes.
247 ** number of lambdas converted by the compiler is reported.
248 ** CPU percentage report (computed from real and total run time.)
249 ** more comprehensive run time reporting, using a condenced format
250 ** interperted form, lambda, and page fault counts are omitted
252 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
254 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
255 * optimization: two argument forms of LAST are upto ~10% faster
256 in normal SPEED policies.
257 * optimization: NCONC no longer needs to heap cons its &REST list
258 in normal SPEED policies.
259 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
261 * bug fix: UNION and NUNION work with :TEST-NOT once more,
262 regression since 1.0.9.1. (thanks to Eric Marsden)
263 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
264 since 1.0.15.36. (thanks to Paul Khuong)
265 * bug fix: LAST when always returned the whole list when given a bignum
266 as the second argument.
267 * bug fix: dynamic extent allocation of nested lists and vectors
268 could leak to otherwise accessible parts.
269 * bug fix: invalid optimization of heap-allocated alien variable
271 * bug fix: fasl header checking is less vulnerable to different
272 platform word lengths.
273 * bug fix: more correct assembler syntax for GNU binutils
274 2.18.50.0.4 support. (thanks to Marijn Schouten)
275 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
277 * bug fix: the fopcompiler can handle LOCALLY forms (with no
278 declarations) successfully. (reported by Attila Lendvai)
280 changes in sbcl-1.0.16 relative to 1.0.15:
281 * minor incompatible change: revert the changes to sb-posix's error
282 signaling added in 1.0.14.
283 * minor incompatible change: change PROBE-FILE back to returning
284 NIL whenever we can't get a truename, as was the case before 1.0.14.
285 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
286 inherits from ERROR instead of just CONDITION.
287 * new feature: SB-INTROSPECT can provide source locations for instances
288 as well. (thanks to Tobian Ritterweiler)
289 * optimization: binding special variables now generates smaller code
290 on threaded platforms.
291 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
292 and cases where no :TEST is given but the compiler can infer that
293 the element to search is of type (OR FIXNUM (NOT NUMBER)).
294 * optimization: better LOGNOT on fixnums.
295 * optimization: modular arithmetic for a particular requested width
296 is implemented using a tagged representation unless a better
297 representation is available.
298 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
299 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
300 for wellformedness and returns a specialized vector. (reported by
301 Francois-Rene Rideau)
302 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
303 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
304 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
305 unscheduling timer at the same time as another timer fires.
306 * bug fix: don't reschedule timers for dead threads.
307 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
308 * bug fix: copying output from RUN-PROGRAM to a stream signalled
309 bogus errors if select() was interrupted.
310 * enhancement: add support for fcntl's struct flock to SB-POSIX.
312 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
313 * enhancement: cleaner backtraces for interactive interrupts, as
314 well as other cases where the interesting frames used to be
315 obscured by interrupt handling frames.
316 * enhancement: untracing a whole package using (UNTRACE "FOO") is
317 now supported, and tracing a whole package using (TRACE "FOO") now
318 traces SETF-functions as well.
319 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
320 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
321 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
322 * unreadably printed representation of hash-tables now includes
324 * bug fix: partially fixed #188: type propagation from assignments
325 is now more efficient.
326 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
327 'DOUBLE-FLOAT) are not flushable.
328 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
329 full address of the object, and none of the tag bits.
330 * bug fix: readably printing hash-tables now respects other printer
331 control variables. (reported by Cedric St-Jean)
332 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
333 keyword with MAKE-HASH-TABLE.
334 * bug fix: export SB-POSIX:MKSTEMP.
335 * bug fix: SORT was not interrupt safe.
336 * bug fix: XREF accounts for the last node of each basic-block as
338 * bug fix: MAKE-INSTANCE optimizations interacted badly with
339 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
340 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
341 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
342 create obsolete instances in the case of redefinition or
343 obsoletion of a superclass. (thanks to Andy Hefner)
344 * bug fix: Support for the Alpha architecture has been revived; it had
345 suffered somewhat from lack of maintenance since sbcl-1.0.
346 * improvements to the Windows port:
347 ** The system detects the codepage to use at startup, instead of
348 using the value from the saved core. (thanks to Kei Suzuki)
350 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
351 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
352 (see documentation for details.)
353 * revived support for OpenBSD (contributed by Josh Elsasser)
354 * partially fixed bug #108: ROOM no longer suffers from occasional
355 (AVER (SAP= CURRENT END)) failures.
356 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
357 bodies no longer cause a WARNING to be signalled. (reported by
359 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
360 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
361 the bug in TRUNCATE.) (reported by Michael Weber)
362 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
363 no samples. (reported by Andy Hefner)
364 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
365 no longer appear as (NIL ...) frames in backtraces.
366 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
367 is now actually exported from SB-EXT as documented. (reported by
369 * bug fix: sb-aclrepl now correctly understands how to inspect
370 single-floats on 64-bit platforms where single-floats are not boxed.
371 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
372 yet been finalized. (reported by Levente Meszaros)
373 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
374 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
375 SBCL-specific optimize qualities.
377 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
378 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
379 an executable in the search path, and does so in the child
380 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
381 been removed; it can be found in the CVS history, for somebody who
382 needs that search behavior (see the manual).
383 * minor incompatible change: compiler policy re. weakening type
384 checks has changed: now type checks are weakened only if SAFETY < 2
386 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
387 unparsing of directory pathnames as files. Analogously,
388 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
389 filename to parse into a directory pathname.
390 * enhancement: implicit generic function creation now signals a
391 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
392 users can bind handlers for and muffle around calls to LOAD.
393 * enhancement: RUN-PROGRAM allows unicode arguments and environments
394 to be used (using the default stream external format), and allows
395 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
396 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
397 strings and vectors whose element-type or simplicity is not fully
398 known at compile-time.
399 * optimization: STRING-TRIM and related functions no longer allocate
400 a new string when no trimming needs to be performed. These functions
401 are also faster than before when the input string has been declared
403 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
404 * optimization: READ-LINE is significantly faster for files containing
406 * optimization: non-open coded uses of character comparison operators
407 (e.g. char=) no longer cons when called with more than one parameter
408 on platforms supporting dynamic-extent allocation.
409 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
410 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
411 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
413 * bug fix: some sequence functions elided bounds checking when
415 * bug fix: too liberal weakening of union-type checks when SPEED >
417 * bug fix: more bogus fixnum declarations in ROOM implementation
420 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
421 * new feature: MAKE-HASH-TABLE now experimentally accepts a
422 :SYNCHRONIZED argument, which makes the hash-table safe for
423 concurrent accesses (but not iteration.) See also:
424 SB-EXT:WITH-LOCKED-HASH-TABLE, and
425 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
426 * optimization: CONCATENATE on strings is an order of magnitude faster
427 in code compiled with (> SPEED SPACE).
428 * optimization: SUBSEQ is ~50% faster on lists.
429 * bug fix: bug 417 fixed -- source location reporting is now more robust.
430 * bug fix: SUBSEQ on a list will now correctly signal an error if if
431 END is smaller then START.
432 * bug fix: SB-PROFILE will no longer report extra consing for nested
433 calls to profiled functions.
434 * bug fix: ROOM implementation had bogus fixnum declarations which
435 could cause type-errors when calling ROOM on large images.
436 * bug fix: if file compilation is aborted, the partial fasl is now
437 deleted, and COMPILE-FILE returns NIL as the primary value.
438 * bug fix: number of thread safety issues relating to SBCL's internal
439 hash-table usage have been fixed.
440 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
441 be returned from its body when the values were being returned
442 using unknown-values return convection and the W-P-O was wrapped
443 inside an UNWIND-PROTECT.
444 * bug fix: sb-posix should now compile again under Windows, enabling
447 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
448 * incompatible change: hash-table accessor functions are no longer
449 automatically protected by locks. Concurrent accesses on the same hash-table
450 from multiple threads can give inconsistent results or even corrupt the
451 hash-table completely. Multi-threaded applications should do their own
452 locking at the correct granularity. In the current implementation it is
453 still safe to have multiple readers access the same table, but it's not
454 guaranteed that this property will be maintained in future releases.
455 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
456 and will signal an error at runtime.
457 * enhancement: SB-THREAD package now exports a semaphore interface.
458 * enhancement: CONS can now stack-allocate on x86 and
459 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
461 * enhancement: nested lists can now be stack allocated on
462 platforms providing stack allocation support.
463 * enhancement: dynamic-extent support has been extended to support
464 cases where there are multiple possible sources for the stack
466 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
467 if the mutex is uncontested on Linux.
468 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
469 as specified by the CLHS. (thanks to Tobias Rittweiler)
470 * bug fix: NaN comparison now works on x86-64.
471 * bug fix: CLOSE :ABORT T on a stream with pending output now
473 * bug fix: instances of non-standard metaclasses using standard
474 instance structure protocol sometimes missed the slot type checks
476 * bug fix: known functions can be inlined in high-debug code.
478 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
479 * minor incompatible change: the MSI installer on Windows no longer
480 associates .lisp and .fasl files with the installed SBCL.
481 * minor incompatible change: :UNIX is no longer present in *FEATURES*
482 on Windows. (thanks to Luis Oliviera)
483 * new feature: SB-CLTL2 contrib module now implements
484 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
485 * optimization: scavenging weak pointers is now more efficient,
486 requiring O(1) instead of O(N) per weak pointer to identify
487 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
488 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
489 method bodies using constant slot-names when the first argument is
490 a specializer parameter for the method.
491 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
492 names, when the class of the instance is a direct instance
493 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
494 as slow as the constant slot-name case.)
495 * optimization: member type construction is now O(N) instead
497 * optimization: UNION and NUNION are now O(N+M) for large
498 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
500 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
501 now more readable in environments like Slime which display it.
502 (thanks to Tobias C. Rittweiler)
503 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
504 was able to infer the correct argument type for the object on which
505 the CAS operation was being performed.
506 * bug fix: copy propagation interfered with parallel assignment
507 semantics in local calls. (reported by Paul Khuong)
508 * bug fix: the signed modular fixnum shift compiled to wrong code on
509 x86 and x86-64. (spotted by a slight modification to some of
512 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
513 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
514 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
515 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
517 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
518 and generic function now print the object identity for anonymous
519 instances. (thanks to Attila Lendvai)
520 * enhancement: as an extension to MOP, SBCL now supports SETF of
521 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
522 (thanks to Attila Lendvai)
523 * enhancement: sb-cover has an alternate annotation mode which puts
524 the form annotations at the car of the form, instead of annotating
525 it completely. The benefit of this mode is that it shows better
526 which forms were instrumented by the compiler.
527 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
528 by printer control variables. (reported by Dan Corkill)
529 * bug fix: system leaked memory when delayed output was performed by
530 the OS in smaller chunks then expected. (thanks to David Smith)
531 * bug fix: system leaked memory when file streams were not closed
533 * bug fix: large objects written to slow streams that were modified
534 after the write could end up with the modified state written to
535 the underlying file descriptor.
536 * bug fix: multiple threads operating in parallel on the same stream
537 could cause buffer-overflows.
538 * bug fix: source location information is stored correctly
539 (broken since 1.0.6). This bug would generally show up as the
540 Slime debugger highlighting the wrong form when the sldb-show-source
542 * bug fix: Branch forms are again annotated as branches in the sb-cover
544 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
545 on x86-64. (reported by Gregory Vanuxem)
546 * bug fix: EQUALP could return wrong results for structures with raw
547 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
548 word). (reported by Vjacheslav Fyodorov)
549 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
552 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
553 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
554 atomic compare-and-swap operations on threaded platforms.
555 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
556 allows assigning a global minimum value to optimization qualities
557 (overriding proclamations and declarations).
558 * enhancement: closed over variables can be stack-allocated on x86
560 * performance bug fix: GETHASH and (SETF GETHASH) are once again
562 * optimization: slot definition lookup is now O(1). This speeds up
563 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
564 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
565 * optimization: ASSOC and MEMBER can now be open-coded for all
566 combinations of keyword arguments when second argument is constant
567 and SPEED >= SPACE. In other cases a specialized version is
569 * bug fix: using obsoleted structure instances with TYPEP and
570 generic functions now signals a sensible error.
571 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
572 (reported by Kristoffer Kvello)
573 * bug fix: backtrace construction is now more careful when making
574 lisp-objects from pointers on the stack, to avoid creating bogus
575 objects that can be seen by the GC.
576 * bug fix: defaulting of values in contexts expecting more than 7
577 variables now works on x86-64. (reported by Christopher Laux)
578 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
580 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
581 as the property-list of a symbol.
582 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
583 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
584 situations requested, are once again file-compileable. (reported
587 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
588 * MOP improvement: support for user-defined subclasses of
589 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
590 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
591 the expansion of DEFMETHOD, is responsible for generating a form
592 which creates a list of specializers when evaluated. Additional
593 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
594 debugging and introspective support.
595 * minor incompatible change: the (unsupported) spinlock interface
596 has changed: free spinlock now has the value NIL, and a held spinlock
597 has the owning thread as its value.
598 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
599 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
601 * enhancement: name of a socket-stream is now "a socket" instead of
603 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
604 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
605 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
606 for generic functions. (Thanks to Tobias C. Rittweiler)
607 * optimization: bignum printing speed has been improved by 20-40%
608 (depending on the bignum size.)
609 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
611 * bug fix: the cache used by the CLOS to store precomputed effective
612 methods, slot offsets, and constant return values is now thread and
614 * bug fix: generic function dispatch function updating is now thread
615 and interrupt safe (in the sense that the known issues have been
617 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
618 * bug fix: interning EQL-specializers is now thread and interrupt safe.
619 * bug fix: asdf systems with dependencies to the SB-POSIX or
620 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
621 * bug fix: interrupt safety of applicable method computation has been
624 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
625 * new contrib: sb-cover, an experimental code coverage tool, is included
627 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
628 significantly faster.
629 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
630 produce more efficient code on x86-64 if the compiler can determine
631 that (+ INDEX <constant>) does not require a bounds check and FOO
632 has an element type at least 8 bits wide.
633 * enhancement: a new, experimental synchronous timeout facility is
634 provided. Refer to SB-SYS:WITH-DEADLINE for details.
635 * enhancement: when a symbol name conflict error arises, the
636 conflicting symbols are always printed with a package prefix.
637 (thanks to Kevin Reid)
638 * enhancement: stepping is now once again supported on the SPARC and
639 MIPS platforms. (It is also now more likely to work on CheneyGC
641 * enhancement: sb-sprof can now also track and report accurate call
643 * bug fixes: the treatment of non-standard subclasses of
644 SB-MOP:SPECIALIZER is more correct.
645 * incompatible change: PURIFY no longer copies the data from the
646 dynamic space into the static and read-only spaces on platforms
647 that use the generational garbage collector
648 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
650 * bug fix: GC race condition occasionally resulting in crashes with
651 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
653 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
655 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
656 a STYLE-WARNING to be signalled (regression from 1.0.4.)
657 * bug fix: an asynchronous interrupt could previously leave the
658 system running with GC inhibited.
659 * bug fix: a DECLARE form evaluated at top-level now causes an error
660 rather than silently (or verbosely) returning NIL.
661 * bug fix: trying to dissassemble functions compiled at high DEBUG could
662 result in a error being signalled due to source form lookup errors.
663 (reported by Peter Graves)
665 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
666 * incompatible change: removed writer methods for host-ent-name,
667 host-ent-addresses -- changing the values did not update the DNS
669 * minor incompatible change: changed experimental JOIN-THREAD interface
670 * documentation: the manual now lists reader and writer methods
671 in class slot documentation sections. (thanks to Richard M Kreuter)
672 * documentation: unwinding from asyncronous events has been
673 documented as unsafe.
674 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
675 in multithreaded application code.
676 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
677 platforms. (thanks to James Anderson for the optimization hint)
678 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
680 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
681 variants no longer cons.
682 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
683 their NOT- variants no longer cons.
684 * optimization: EQUAL hash tables no longer use SXHASH for objects
685 of all data types, but instead use an EQL hash for types for which
686 EQUAL is the same as EQL.
687 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
688 are significantly faster.
689 * optimization: new STRING-OUTPUT-STREAM implementation is much
690 faster for large outputs and conses 30% less on average.
691 * enhancement: XREF information is now collected for references made
692 to global variables using SYMBOL-VALUE and a constant argument.
693 * enhancement: SIGINT now causes a specific condition
694 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
695 * bug fix: FILE-POSITION used to signal an error in some cases where
696 ANSI requires it to return NIL.
697 * bug fix: ADJUST-ARRAY is now interrupt-safe.
698 * bug fix: adding and removing fd-handlers is now interrupt-safe.
699 * bug fix: inlined calls to C now ensure 16byte stack alignment on
701 * bug fix: bad type declaration in the CLOS implementation has
702 been fixed. (reported by James Anderson)
703 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
704 (thanks to Lutz Euler)
705 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
706 required by the CLHS. (reported by Samium Gromoff)
707 * bug fix: dead unbound variable references now signal an error.
708 (reported by Marco Monteiro)
709 * bug fix: / with an unused value was being deleted in safe code.
710 (thanks to Marco Monteiro and Kevin Reid)
711 * bug fix: number of characters that can be written onto a single
712 line in a file is unlimited.
713 * bug fix: some GC deadlocks caused by asynchronous interrupts have
714 been fixed by inhibiting interrupts for when GC is disabled.
715 * bug fix: some interrupt safety issues with GETHASH, (SETF
716 GETHASH), CLRHASH and REMHASH have been fixed.
717 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
718 specifier no longer causes infinite recursion.
719 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
720 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
721 * bug fix: an &environment argument with the correct variable information
722 is passed to macros that are expanded during byte compilation.
723 (reported by Samium Gromoff)
724 * bug fix: a logic error was causing the hash function for bignums to
725 have excessive amounts of collisions. (reported by Faré Rideau)
726 * bug fix: modifying the contents of an array could change the return
727 value of SXHASH on that array, which is only allowed for strings
728 and bit vectors. (bug introduced in 0.9.16)
729 * improvement: the x86-64/darwin port now passes all tests (except
730 for the debugger tests) but should still be considered
731 experimental until this is fixed.
732 * improvement: a style-warning is signaled for CASE (etc) clauses with
733 duplicate keys. (thanks to Kevin Reid)
734 * improvement: macos/x86 and macos/x86-64 now use mach exception
735 handlers for dealing with illegal instructions (for trapping and
736 error handling) and memory protection violations (for GC).
738 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
739 * new platform: experimental support for x86-64/darwin (MacOS).
740 * incompatible change: the thread-safe (on most platforms) getaddrinfo
741 and getnameinfo sockets functions are used instead of gethostbyaddr
742 and gethostbyname, on platforms where the newer functions are available.
743 As a result, the ALIASES field of HOST-ENT will always be NIL on these
745 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
746 don't print the startup banner, but behave as if --noinform was passed
747 as a command line argument. (thanks to Kevin Reid)
748 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
749 * optimization: code using alien values with undeclared types is much faster.
750 * optimization: the compiler is now able to open code SEARCH in more cases.
751 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
752 * bug fix: using standardized COMMON-LISP special variables as loop
753 variables no longer signals bogus package lock violations. (reported
755 * bug fix: declaring local loop variables to be of a range-limited type
756 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
757 (reported by Andras Simon)
758 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
759 bugs remain on x86-64.)
760 * bug fix: the #= and ## reader macros now interact reasonably with
761 funcallable instances.
762 * bug fix: type-checks for function arguments were compiled using the
763 compiler policy settings of the wrong lexical environment (in 1.0.2
765 * bug fix: SHADOW accepts characters as string designators, as required
766 by the spec (thanks to Eric Marsden)
767 * bug fix: fixed GC safety issues when foreign functions are called with
768 non-base strings as arguments
769 * bug fix: more consistent error output for fatal-compiler-warnings, like
771 * bug fix: functions with proper names were showing up as NIL in
772 backtraces (reported by Edi Weitz, regression in 1.0.2)
774 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
775 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
776 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
777 produce more efficient code on the x86 if the compiler can determine
778 that (+ INDEX <constant>) does not require a bounds check and FOO
779 has an element type at least 8 bits wide.
780 * bug fix: references to undefined variables are handled the same way
781 in toplevel forms as in the normal compiler
782 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
784 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
785 (thanks to Jon Buller)
786 * improvement: faster compilation times for complex functions
787 * improvement: added readlink support to SB-POSIX (thanks to Richard
790 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
791 * improvement: experimental support for mach exception handling on
792 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
794 * improvement: support for GBK external format.
795 (thanks to Chun Tian (binghe))
796 * improvement: the debugger now displays variables that have been closed
797 over, in code compiled with (DEBUG 2) or higher
798 * improvement: support for executable cores on NetBSD (thanks to
800 * new feature: added a RESTART-FRAME debugger command
801 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
802 be used to provide an implementation for FILE-POSITION on Gray streams
803 (thanks to Eric Marsden)
804 * new feature: add syslog, openlog and closelog support to SB-POSIX
805 (thanks to Richard Kreuter)
806 * optimization: the function call overhead in code compiled with
807 a high DEBUG optimization setting is significantly
808 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
809 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
810 (thanks to Tony Martinez)
811 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
812 evaluated once (reported by Kevin Reid)
813 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
814 works again (thanks to Kilian Sprotte)
815 * bug fix: an error is signaled for tagbodies with duplicate tags
816 (thanks to Stephen Wilson)
817 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
818 * bug fix: Win32 port can now handle foreign code unwinding Lisp
819 stack frames from alien callbacks.
820 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
821 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
822 65k pages (thanks to David Woodhouse)
823 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
825 changes in sbcl-1.0.1 relative to sbcl-1.0:
826 * new platform: FreeBSD/x86-64, including support for threading.
827 * new feature: the compiler stores cross-referencing information
828 about function calls (who-calls), macroexpansion (who-macroexpands)
829 and special variables (who-binds, who-sets, who-references) for code
830 compiled with (< SPACE 3). This information is available through the
831 sb-introspect contrib.
832 * new feature: users may subclass SEQUENCE, and have instances of
833 these classes interoperate with standard Common Lisp functions if
834 a number of methods are defined. (This feature is experimental
835 and the interface subject to change based on feedback from SBCL
836 users and the general community)
837 * improvement: sb-sprof traces call stacks to an arbitrary depth on
838 x86/x86-64, rather than the previous fixed depth of 8
839 * improvement: another pthread back-end of mutex "pthread-futex".
840 * bug fix: non-ascii command-line arguments are processed correctly
841 (thanks to Yaroslav Kavenchuk)
842 * bug fix: non-required arguments were not passed correctly when a method
843 defined using DEFMETHOD was called from a mop-generated method using
844 CALL-NEXT-METHOD (reported by Pascal Costanza)
845 * bug fix: recursion is now permitted in accessors through
846 SLOT-UNBOUND. (reported by Pascal Costanza)
847 * bug fix: an error was signaled at startup if the HOME environment
848 variable was defined, but had an empty value (reported by Peter Van Eynde)
849 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
850 signal a PROGRAM-ERROR, not a BUG.
851 * bug fix: SB-POSIX:READDIR works when built with large file support.
852 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
854 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
855 for better automatic indentation support. (thanks to Matt Pillsbury)
856 * optimization: loading generic functions no longer takes O(n^2) time,
857 proportional to the amount of methods in the generic function
858 (reported by Todd Sabin and Jeremy Brown)
859 * optimization: the FIND and POSITION family of sequence functions
860 are significantly faster on arrays whose element types have been
862 * improvements to the Windows port:
863 ** Intermittent heap corruption problems have been fixed. (thanks
864 to Alastair Bridgewater)
865 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
867 ** Lisp is able to unwind foreign exception frames from alien
868 callbacks. (thanks to Alastair Bridgewater)
870 changes in sbcl-1.0 relative to sbcl-0.9.18:
871 * improvement: experimental support for threading on FreeBSD/x86.
872 (thanks to NIIMI Satoshi)
873 * improvement: runtime option --dynamic-space-size can be used
874 to set the size of the dynamic space reserved on startup.
875 * improvement: floating point modes in effect are now saved in
876 core, and restored on startup.
877 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
878 startup, not time since first call to GET-INTERNAL-REAL-TIME.
879 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
880 threads are not running after *SAVE-HOOKS* have run.
881 * improvement: writes to CLOS instance slots are type-checked in code
882 compiled with (SAFETY 3)
883 * improvement: floating-point exception handling on FreeBSD (thanks to
885 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
886 (thanks to Zach Beane)
887 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
889 * improvement: added support for the Shift-JIS external format.
890 (contributed by NIIMI Satoshi)
891 * improvement: callbacks are supported on Linux/PPC. (thanks to
893 * bug fix: compiler bug triggered by a (non-standard) VALUES
894 declaration in a LET* was fixed. (reported by Kaersten Poeck)
895 * bug fix: file compiler no longer confuses validated and already
896 dumped structurres. (reported by Kaersten Poeck)
897 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
898 fill-pointer signals a type-error as required. (thanks to
900 * bug fix: disassemly of funcallable instances works.
901 * bug fix: single stepping on PPC.
902 * bug fix: fix thread-safety problems in the type system (generally
903 manifesting as nonsensical errors like "STRING is a bad type specifier
904 for sequences" or "The value 1 is not of type FIXNUM")
905 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
906 (reported by Josip Gracin)
907 * bug fix: an error is signaled for attempts to displace arrays with
908 incompatible element types (thanks to Mario Mommer)
909 * bug fix: more correct handling of wide characters in debug info
910 (bug reported by Attila Lendvai and fixed by Juho Snellman)
911 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
912 and don't cause extra consing
913 * optimization: MAP and MAP-INTO are significantly faster on vectors
914 whose elements types have been declared.
915 * Improvements to SB-SPROF:
916 ** Support for allocation profiling
917 ** Reduced profiling overhead, especially for long profiling runs
918 * Improvements to the Windows port:
919 ** floating point exceptions are now reported correctly.
920 ** stack exhaustion detection works partially.
921 ** more accurate GET-INTERNAL-REAL-TIME.
922 ** better RUN-PROGRAM behaviour in case of errors in the child
924 ** PROBE-FILE now simplifies pathnames correctly.
925 ** DIRECTORY now works correctly with :WILD-INFERIORS.
926 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
927 1st 1970 (although time timezone data will be unreliable).
928 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
930 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
931 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
932 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
934 * bug fix: two potential GC deadlocks affecting threaded builds.
935 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
936 returns T (reported by Anton Kazennikov)
937 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
938 with non-variable places
939 * bug fix: remove a race condition in the setting of
940 funcallable-instance functions, this should make threaded CLOS
941 code more stable against memory faults.
942 * bug fix: corruption of specials when unbinding is interrupted by an
943 asynchronous unwind (reported by Hannu Koivisto)
944 * improvement: the debugger will now also display local variables that
945 are only used once, for code compiled with a DEBUG optimization quality
948 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
949 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
950 * incompatible change: External-format support for FFI calls. The
951 SB-ALIEN:C-STRING no longer implies an ASCII
952 external-format. Instead the string is subject to external-format
953 conversion. Additionally return values of type SB-ALIEN:C-STRING
954 are by default of type (SIMPLE-ARRAY CHARACTER), not
955 SIMPLE-BASE-STRING. For an alien type that behaves like the old
956 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
957 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
958 * incompatible change: SB-EXT package no longer contains the
959 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
960 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
962 * incompatible change: the single-stepper is no longer available
963 on Alpha, Mips and Sparc platforms.
964 * minor incompatible change: the direct superclasses of
965 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
966 not (STANDARD-OBJECT FUNCTION). This makes the
967 class-precedence-lists of GENERIC-FUNCTION and
968 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
970 * minor incompatible change: the default stream external format on
971 non-unicode SBCL is no longer detected from the locale, but is always
973 * new feature: Add a version of evaluator that uses an interpreter instead
974 of the compiler. EVAL still uses the compiler by default, to switch it
975 to use the interpreter, set the value of the variable
976 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
977 * minor incompatible change: the single-stepper REPL has been merged
978 with the normal debugger (see the "Stepping" heading of the debugger help
979 for more details). The debugger command STEP will no longer switch
980 to the single-stepper REPL.
981 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
982 for a type now works.
983 * bug fix: loading of default sysinit file works. (thanks to Leonid
985 * bug fix: better detection of circularities in the file-compiler.
986 (reported by Marco Monteiro)
987 * bug fix: the CL pathname functions now work with files that have
988 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
989 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
990 systems with Unix98 pty semantics.
991 * bug fix: ASDF-INSTALL will now work with bsd tar.
992 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
994 * bug fix: timers expiring in dead threads no longer cause a
995 type-error (reported by Paul "Nonny Mouse").
996 * bug fix: thanks to more lightweight single-stepper instrumentation,
997 code compiled with (DEBUG 3) will compile and execute significantly faster,
998 and will have more accurate type-inferencing than before
999 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1000 whose bindings are modified
1001 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1002 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1003 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1004 CL:SOFTWARE-VERSION with proper version
1006 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1007 * feature: implemented the READER-METHOD-CLASS and
1008 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1009 as specified by AMOP.
1010 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1012 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1013 (thanks to Lutz Euler)
1014 * optimization: hashing of general arrays and vectors has been
1015 improved. (reported by Any Fingerhut)
1016 * enhancement: SB-INTROSPECT is now able to find definitions of
1017 profiled functions. (thanks to Troels Henriksen)
1018 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1019 as well, allowing compiler-macros for SETF-functions to expand.
1020 * enhancement: step-instrumentation no longer wraps known
1021 single-value functions with multiple-value context, allowing
1022 better type inference.
1023 * fixed bug #337: use of MAKE-METHOD in method combination now works
1024 even in the presence of user-defined method classes. (reported by
1025 Bruno Haible and Pascal Costanza)
1026 * fixed bug #339(c): if there are applicable methods not part of any
1027 long-form method-combination group, call INVALID-METHOD-ERROR.
1028 (reported by Bruno Haible)
1029 * fixed bug #361: the :FUNCTION initarg in the protocol for
1030 initialization of methods can now be used to override
1031 internally-produced optimized functions. (reported by Bruno
1033 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1034 system-provided lambda expression no longer cause warnings about
1035 unbound #:|pv-table| symbols.
1036 * bug fix: improved the handling of type declarations and the
1037 detection of violations for keyword arguments with non-constant
1039 * bug fix: potentially erronous calls to PATHNAME and
1040 MERGE-PATHNAMES were being flushed in some cases.
1041 (reported by Richard Kreuter)
1042 * bug fix: compiled calls to TYPEP were mishandling obsolete
1043 instances. (reported by James Bielman and Attila Lendvai)
1044 * bug fix: format strings with ~> without matching ~< no longer
1045 trigger an AVER, but signal an understandable error instead.
1046 (reported by Antonio Martinez)
1047 * bug fix: specifying an output-file without a pathname-type for
1048 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1049 of input-file instead of "fasl". (reported by Robert Dodier)
1050 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1051 (reported by James Y Knight).
1052 * bug fix: compiler-macros-function did not consider the environment
1053 argument for shadowing by local functions.
1054 * bug fix: compiler-macros expansion was inhibited by local INLINE
1056 * bug fix: inline expansions of known functions were subject to
1057 step-instrumentation in high debug policies, leading to problems
1058 with type-inference.
1059 * bug fix: compiler failed to differentiate between different CONS
1060 types in some cases.
1061 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1062 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1063 * bug fix: #368: incorrect use of expressed vs. upgraded array
1065 * bug fix: #306a: more precise unions of array types.
1066 * thread-safety improvements:
1067 ** CONDITION-WAIT could return early on Linux, if the thread was
1068 interrupted and subsequently continued with SIGCONT.
1069 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1071 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1072 * added support for the ucs-2 external format. (contributed by Ivan
1074 * minor incompatible change: pretty printing of objects of type
1075 (cons symbol) is, in the default pprint-dispatch-table, now
1076 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1078 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1079 called later in a class's lifetime, possibly as late as when the
1080 first instance of the class is created. Previously,
1081 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1082 class became finalizeable.
1083 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1084 for multibyte external-format streams. (thanks to "vbzoli")
1085 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1086 with :ALLOCATION :CLASS inherited from superclasses of the
1088 * fixed bug: anonymous classes can now be created using the :NAME
1089 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1090 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1091 * fixed bug: core-files saved with :EXECUTABLE T can again be
1092 executed when SBCL_HOME isn't set. (reported by James Knight)
1093 * fixed bug: toplevel LOCALLY forms with declarations could
1094 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1095 * fixed bug: printing from several different threads using different
1096 values of *print-case* could cause invalid output, due to
1097 some internal special variables of the printer not being bound
1098 thread-locally (reported by Max Mikhanosha)
1099 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1100 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1101 (reported by Pascal Costanza)
1102 * fixed bug: FINALIZE-INHERITANCE is called from
1103 REINITIALIZE-INSTANCE on classes when the class has previously
1104 been finalized, as required by AMOP.
1105 * minor code generation optimizations:
1106 ** better register allocation in CLOS dispatching functions
1107 ** overflow detection when coercing signed bytes to fixnums on x86-64
1108 ** is now implemented with one IMUL instruction instead of three shifts
1109 ** more efficient bit-vector access on x86 and x86-64
1110 ** more efficient access to raw structure slots on x86 and x86-64
1111 * fixed some bugs revealed by Paul Dietz' test suite:
1112 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1113 return its argument.
1115 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1116 * feature: thread support on Solaris/x86, and experimental thread support
1118 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1120 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1121 default method for SLOT-UNBOUND.
1122 * minor incompatible change: prevent the user from specializing the
1123 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1124 somewhat counter to the intent of the protocol, I (CSR) think, and
1125 additionally it just doesn't work in SBCL as currently
1126 implemented, thanks to optimizations (that are always valid for
1127 the other three SLOT-VALUEish functions, but not for the setter).
1128 * minor incompatible change: the :SB-LDB feature is now enabled by
1129 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1130 the low-level debugger.
1131 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1132 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1133 down to an SB-SYS:FD-STREAM, allowing direct communication in
1134 more cases. Temporary files and pipes are still used as a fallback
1136 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1137 on the non-reentrant C functions localtime(3) and gmtime(3).
1138 * bug fix: global optimization policy was not visible in LOCALLY and
1140 * bug fix: class objects can be used as specializers in methods.
1141 (reported by Pascal Costanza)
1142 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1143 with a type error. (reported by blitz_ on #lisp)
1144 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1145 failed with a type error. (reported by Pascal Bourguignon)
1146 * bug fix: merging pathnames against defaults with :DIRECTORY
1147 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1148 (reported by James Y Knight)
1149 * bug fix: saving large (>2GB) cores on x86-64 now works
1150 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1151 constant index and a (simple-array (signed-byte 32)) array
1152 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1153 error (patch by Robert J. Macomber)
1154 * feature: TIME now displays the amount of run-time spent in GC
1155 * bug fix: The debugger now does a better job of respecting
1156 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1157 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1158 workaround for bug 403.)
1159 * bug fix: floating point exception handling now works on FreeBSD
1160 (thanks to NIIMI Satoshi)
1161 * fixed some bugs revealed by Paul Dietz' test suite:
1162 ** MISC.641: LET-conversion were not supposed to work in late
1164 * improvements to the Win32/x86 port:
1165 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1166 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1168 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1169 * new feature: source path information is generated for macro-expansion
1170 errors for use in IDE's like Slime (thanks to Helmut Eller)
1171 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1172 * bug fix: compiler does not loop forever on an invalid type in
1174 * improvement: compilation of most CLOS applications is significantly
1176 * optimization: added a limited bytecode compiler for simple toplevel
1177 forms, speeding up compilation and FASL loading
1178 * bug fix: the statistical profiler now properly distinguishes anonymous
1181 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1182 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1183 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1184 into subclasses of the finalized class. Now user calls to
1185 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1186 system return before any subclasses are finalized.
1187 * minor incompatible change: The reader no longer ignores errors
1188 regarding non-existent packages in #+ and #- feature tests.
1189 * new feature: command line options --no-sysinit, --no-userinit to
1190 inhibit loading the corresponding init files
1191 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1192 for sending data through UDP sockets (thanks to François-René Rideau)
1193 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1194 error is signalled instead (thanks to François-René Rideau)
1195 * minor incompatible change: Error signalling behaviour of lexical
1196 operations violating package locks has changed slightly. Refer to
1197 documentation on package locks for details.
1198 * enhancement: EVAL can process IF-expressions without resorting to the
1200 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1201 constant-folded properly due to missing out-of-line functions.
1202 (reported by Mika Pihlajamäki)
1203 * bug fix: LISTEN sometimes returned T even in cases where no data was
1204 immediately available from the stream
1205 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1206 were reversed. (reported by Levente Mészáros)
1207 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1208 when given a vector displaced to an adjustable array. (reported
1210 * bug fix: derivation of float boundaries from numbers outside the
1211 appropriate float range (reported by John Wiseman)
1212 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1214 * bug fix: precomputing cache entries for generic functions with
1215 some subclasses of specializers as yet invalid does not attempt to
1216 fill a cache line with a negative offset. (reported by Levente
1218 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1219 allow condition class objects as arguments to DOCUMENTATION and
1220 (SETF DOCUMENTATION); only find and set documentation for
1221 structure names for the STRUCTURE doc-type. (suggested by Gary
1223 * improvements to the Win32/x86 port:
1224 ** (user-homedir-pathname) and default initialization file
1225 locations now know about the user's "Documents and Settings"
1226 directory (thanks to Yaroslav Kavenchuk)
1227 ** run-program is implemented (thanks to Mike Thomas)
1228 ** sockets support (thanks to Timothy Ritchey)
1229 ** better backtrace support (thanks to Alastair Bridgewater)
1230 ** sb-grovel supported
1231 ** asdf-install and sb-posix work somewhat
1232 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1233 * improvements to the Solaris/x86 port:
1234 ** works on Solaris 11/Solaris Express
1235 ** floating-point exception handling support
1236 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1237 * fixed some bugs revealed by Paul Dietz' test suite:
1238 ** REMOVE-METHOD returns its generic function argument even when
1239 the method is not one of the generic functions' methods.
1240 ** objects declared as MEMBER types can be admissible arguments to
1241 structure accessors.
1242 ** printing characters should simply be printed by the FORMAT ~:C
1244 ** compiler failure when compiling functions with hairy constant
1245 defaults for optional parameters.
1246 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1247 function, which is already optimized.
1249 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1250 * new platform: experimental support for SBCL x86/Darwin, including
1251 MacOS X 10.4.5 on Intel.
1252 * new feature: Unicode character names are now known to the system
1253 (through CHAR-NAME and NAME-CHAR).
1254 * new feature: the filesystem location of SBCL's core file is
1255 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1256 * minor incompatible change: the contrib modules SB-POSIX and
1257 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1258 this change is to make it easier to distribute
1259 location-independent binaries.
1260 * bug fix: as implied by AMOP, standardized classes no longer have
1261 slots named by external symbols of public packages. (reported by
1263 * optimization: calling functions via constant symbols -- as in
1264 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1265 via the function object as in (FUNCALL #'FOO).
1266 * optimization: on x86 and x86-64, the calling convention for the
1267 case of dealing with an unknown number of values has been altered
1268 to be friendlier to the prediction heuristics implemented,
1269 particularly with reference to CALL and RET pairing. (thanks to
1270 Alastair Bridgewater)
1271 * optimization: on x86, the code for access to thread-local symbol
1272 values and binding/unbinding of thread-local symbols is smaller.
1273 (thanks to Alastair Bridgewater)
1274 * enhancement: CONSTANTP is now able to determine constantness of
1275 more complex forms, including calls to constant-foldable standardized
1276 functions and some special forms beyond QUOTE.
1277 * fixed bug: occasional GC crashes on Solaris/x86
1278 * optimization: x86-64 supports stack allocation of results of simple
1279 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1280 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1281 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1282 (thanks to James Knight)
1283 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1284 by the standard (thanks to Stephen Compall)
1286 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1287 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1288 be used for bundling the runtime and the core file into one
1289 executable binary. This feature is not currently supported on all SBCL
1290 platforms. (thanks to James Bielman and NIIMI Satoshi)
1291 * new feature: a generational or ephemeral garbage collector is now
1292 the default on the PowerPC platform (both Linux and Darwin). The
1293 old Cheney (stop and copy) collector is a build-time option.
1294 (thanks to Cyrus Harmon, after Raymond Toy)
1295 * minor incompatible change: the method by which SBCL finds its
1296 contributed modules has changed; it no longer relies on symbolic
1297 links from an $SBCL_HOME/systems directory, but searches directly
1298 in the subdirectories of $SBCL_HOME.
1299 * enhancement: the dynamic heap size on the Linux/PPC platform is
1300 markedly larger, even using the older Cheney garbage collector.
1301 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1302 certain complicated string types. (reported by Gary King)
1303 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1304 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1305 * fixed bug: DOCUMENTATION on structures no longer signals an
1306 error if no structure type of the right name exists. (reported by
1308 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1309 applicable non-standard primary or :AROUND method on
1310 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1311 no longer cause unbound variable errors. (reported by Kevin Reid)
1312 * fixed bug: metacircle resolution in cases where methods have slots
1313 added before the slots from STANDARD-METHOD. (reported by Jean
1315 * fixed bug: the Power PC assembly code for calling into lisp
1316 sometimes computed its return address wrongly (depending on the
1317 15th bit of the address, fixed at link time). This used to
1318 manifest itself in a segmentation violation while building PCL.
1319 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1320 many others over the years)
1321 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1322 the stack pointer (treated as a signed integer) is negative.
1323 (thanks to Peter van Eynde)
1325 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1326 * new platform: experimental support for the Windows operating
1327 system has been added. (thanks to Alastair Bridgewater)
1328 * fixed several bugs in and robustified the PPC FFI (including
1329 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1330 * bug fix: allow non-simple string symbol names (reported by Paul
1332 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1334 * bug fix: saving a core corrupted callbacks on x86/x86-64
1335 * bug fix: closed a loophole in metacircularity detection and
1336 grounding in the PCL implementation of CLOS.
1337 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1338 (reported by tomppa on #lisp)
1339 * bug fix: clear the direction flag on Lisp -> C transitions, as
1340 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1341 (reported by Andrej Grozin and Hendrik Maryns)
1342 * optimization: major improvements to GC efficiency on GENCGC platforms
1343 * optimization: faster implementation of EQUAL
1344 * optimization: emit more efficient opcodes for some common
1345 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1346 * optimization: several other minor code-generation improvements on x86-64
1347 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1349 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1350 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1351 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1352 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1353 by ANSI as a generic function, and for consistency (SETF
1354 GENERIC-FUNCTION-NAME) is treated likewise.
1355 * fixed bug #233.b: make constraint propagation notice when a variable
1356 value is changed after it is referenced but before it is used
1357 * fixed bug #296: no more arbitrary behaviour from filenames with
1358 odd characters as --load arguments. (reported by Adam Warner)
1359 * fixed bug #390: :CHARACTER-SET pathname components now work as
1360 expected. (reported by Tim Daly Jr)
1361 * fixed bug #391: complicated :TYPE intersections in slot
1362 definitions no longer cause an error in PCL internals.
1363 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1364 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1365 returning the number of octets which would be written to the
1366 file-stream. (thanks to Robert J. Macomber)
1367 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1368 arguments to the internal stream functions. (thanks to David
1370 * bug fix: stack exhaustion is now handled on the x86/SunOS
1371 platform. (thanks to Daisuke Homma)
1372 * bug fix: more accurate ROOM results on GENCGC platforms
1373 * optimization: improved type inference for arithmetic-for
1374 index variables in LOOP
1375 * optimization: faster floating-point SQRT on x86-64
1376 * fixed some bugs revealed by Paul Dietz' test suite:
1377 ** DOCUMENTATION returns NIL instead of "" for method combinations
1378 that don't have a docstring
1380 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1381 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1382 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1383 therefore are not customizeable by user code (as seems to be at
1384 least permitted and maybe required by AMOP). As a consolation,
1385 however, the SBCL implementation of these functions now calls
1386 REINITIALIZE-INSTANCE as specified by AMOP.
1387 * bug fix: slot-definition documentation is propagated as per ANSI
1388 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1389 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1390 argument, as do eslotd initialization methods. (from Pascal
1391 Costanza's "Closer" project)
1392 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1393 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1395 * bug fix: it is now possible to have more than one subclass of
1396 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1397 (reported by Bruno Haible, Pascal Costanza and others)
1398 * bug fix: the dependent update protocol now works for generic
1399 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1400 and Pascal Costanza)
1401 * bug fix: condition-class instances corresponding to
1402 DEFINE-CONDITION forms are now created eagerly. (reported by
1403 Kalle Olavi Niemitalo on comp.lang.lisp)
1404 * bug fix: floating point printing is more accurate in some
1405 circumstances. (thanks to Simon Alexander)
1406 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1407 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1408 * bug fix: callbacks on OS X now preserve stack-alignment.
1409 (thanks to Cyrus Harmon)
1410 * enhancement: the x86-64 disassembler is much better at
1411 disassembling SSE instructions. (thanks to Lutz Euler)
1412 * enhancement: CHECK-TYPE on a variable now allows type-inference
1413 (thanks to Vincent Arkesteijn)
1414 * optimization: improved performance of EUC-JP external format.
1415 (thanks to NIIMI Satoshi)
1416 * optimization: performance improvements to IO on file streams of
1417 :ELEMENT-TYPE CHARACTER
1418 * optimization: much faster memory allocation on x86-64
1419 * optimization: faster garbage collections (latency and throughput)
1421 * optimization: faster bignum subtraction and division on x86 and x86-64
1422 (thanks to Lutz Euler)
1423 * optimization: more accurate type-inference for loops that use a
1424 floating point index variable or a negative step.
1426 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1427 * bug fix: add a workaround to SBCL looping infinitely at startup on
1428 Linux kernels with apparently buggy implementations of personality().
1429 (thanks to Svein Ove Aas)
1430 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1431 (thanks to David Lichteblau)
1432 * bug fix: local bindings shadow global symbol macros for
1433 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1434 * optimization: non-open coded uses of numeric comparison operators
1435 (e.g. >) no longer cons when called with more than one parameter
1436 on platforms supporting dynamic-extent allocation.
1437 * enhancement: saving cores with foreign code loaded is now supported
1438 on MIPS/Linux in addition to the previously supported platforms.
1439 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1440 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1441 * bug fix: division by zero in sb-sprof when no samples were collected
1442 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1443 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1444 (thanks to Alexander Kjeldaas)
1445 * new feature: ignoring the return values of destructive functions
1446 like NREVERSE signals a compile-time style-warning.
1447 (thanks to Kevin Reid)
1448 * new feature: the alignment of alien structure fields can be
1449 explicitly specified. (thanks to Cyrus Harmon)
1450 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1451 and dump core on SIGQUIT
1453 ** incompatible change: threads do not inherit values of specials
1454 from their parents (see manual)
1455 ** bug fix: threads stacks belonging to dead threads are freed by the
1456 next exiting thread, no need to gc to collect thread stacks anymore
1457 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1458 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1459 ** bug fix: dynamic variable and thread start related gc lossage
1460 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1462 * fixed some bugs revealed by Paul Dietz' test suite:
1463 ** SUBTYPEP is slightly more accurate on heinously complicated
1464 CONS types where some of the members have uncertain (in the
1465 NIL, NIL sense) type relationships to each other.
1467 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1468 no longer purified unless :PURIFY T is explicitly specified.
1469 ** Non-purified cores are significantly smaller than before
1471 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1472 * new feature: timers based on Zach Beane's excellent timer package
1473 * added support for the following external formats: koi8-u,
1474 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1475 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1476 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1477 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1478 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1479 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1480 * incompatible change: a threaded SBCL will no longer revert to
1481 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1482 * bug fix: interrupts are disabled until startup is complete; no
1483 more sigsegvs when receiving a signal to soon
1484 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1486 * bug fix: add a workaround for the memory randomization features in
1487 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1488 workaround will only be in effect on systems with the proc filesystem
1490 * bug fix: printing objects of type HASH-TABLE signals a
1491 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1493 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1494 have been running for over 50 days. (reported by Gilbert Baumann)
1495 * bug fix: the logic for getting names of functions gets less
1496 confused when confronded with alternate-metaclass
1497 funcallable-instances. (reported by Cyrus Harmon)
1498 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1499 consistent, even on internal alternate-metaclass objects.
1500 * bug fix: flush closure information collected by physical
1501 environment analysis prepass before the main pass. (bug reported
1503 * bug fix: compiler pack phase does not modify a hashtable, which is
1504 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1505 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1506 may be expanded. (reported by Alexander <malishev> on c.l.l)
1507 * bug fix: no more style-warnings when compiling code using LAST
1508 under high-SPEED low-SPACE settings. (reported by David Wallin
1510 * fixed some incompatibilities between SBCL's MOP and the MOP
1512 ** the METAOBJECT class is now implemented;
1513 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1514 STANDARD-OBJECT, as required;
1515 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1516 now compatible (as required by VALIDATE-SUPERCLASS); there
1517 remains a consistency requirement over the presence or absence
1518 of the FUNCTION class in the superclass list of finalized
1519 classes; see the manual for more details;
1520 ** the :AROUND method for COMPUTE-SLOTS on
1521 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1522 requested slot ordering.
1524 ** bug fix: parent thread now can be gc'ed even with a live
1526 ** bug fix: binding a special with PROGV to no value is not
1527 the same as that symbol not having been bound (thanks to
1529 ** bug fix: binding specials is thread safe (thanks to
1531 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1532 and SB-SPROF do not die with 'no handler for signal XX in
1533 interrupt_handle_now(..)' anymore
1534 ** bug fix: WITH-TIMEOUT works with multiple running threads
1535 * fixed some bugs revealed by Paul Dietz' test suite:
1536 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1537 the :method-class keyword argument.
1539 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1540 * new port: the Solaris operating system on x86 processors is now
1541 mostly supported, though some rough edges in the environment
1542 remain. (thanks to Daisuke Homma)
1543 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1544 space for its heap. (thanks to Thiemo Seufer)
1545 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1546 closures. (thanks to Thiemo Seufer)
1547 * minor incompatible change: eof selects abort in the debugger.
1548 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1549 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1551 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1552 deprecated, and will go away in a future revision of SBCL.
1553 * minor incompatible change: GC-ON and GC-OFF are no longer
1554 implemented with a counter, it does not matter how many times gc
1555 is switched on or off
1556 * bug fix: discriminating functions for generic function classes
1557 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1558 make invalid assumptions about method precedence order. (reported
1560 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1561 platforms. (thanks to Thiemo Seufer)
1562 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1563 * bug fix: correct pathname printing with printer escaping is on.
1564 (thanks to Kevin Reid)
1565 * bug fix: complex VOP definitions in "user-space" no longer trigger
1566 package locks. (reported by Zach Beane)
1567 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1568 for funcallable-instance functions, and is no different from
1570 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1571 characters in the current readtable. (reported by Nicholas Neuss)
1572 * bug fix: SUBTYPEP on various CONS types returns more a more
1573 accurate acknowledgment of its certainty.
1574 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1575 lists in some cases. This partially fixes bug 384.
1576 * flush all standard streams before prompting in the REPL and the
1578 * bug fix: signal handling and triggering gc do not conflict
1579 directly anymore, in particular a high frequency sb-sprof does
1580 not prevent gc from running
1581 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1582 approximation for timezone and DST information between the
1583 universal time epoch and the smallest negative 32-bit time_t.
1584 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1585 year 1899 so long as the encoded time is non-negative.
1586 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1587 the correct answer for non-constant rotate amounts when performing
1588 an inline 32-bit rotation.
1590 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1591 there is only one thread in the session
1592 ** bug fix: memory leak for streams created in one thread and
1593 written to in another
1594 ** bug fix: lockup when compiled with gcc4
1595 ** bug fix: race that allows the gc to be triggered when gc is
1597 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1598 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1600 ** bug fix: finished threads are gc'ed properly
1601 * fixed some bugs revealed by Paul Dietz' test suite:
1602 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1603 has arguments with a different set of applicable methods from
1604 the orignal arguments.
1605 ** The type error thrown by MAP now has a correct expected-type
1607 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1608 name a compiled function.
1609 ** Three MISC tests where a large bignum was improperly coerced to
1610 a float that couldn't represent that bignum during type
1611 derivation were fixed.
1612 ** SUBTYPEP can now handle types involving the negation of a
1613 list-form FUNCTION type.
1614 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1615 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1616 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1618 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1619 * New feature: Experimental support for bivalent streams: streams
1620 opened with :element-type :default now allow character and binary
1621 (unsigned-byte 8) I/O
1622 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1623 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1625 * bug fix: no more highly sporadic "couldn't check whether ~S is
1626 readable" when reading a stream and an interrupt hits in the middle
1627 of a select system call
1628 * compiler better recognizes complex arrays (reported by Eduardo
1630 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1631 for variables on SBCL built with linkage-tables. (reported by Luis
1633 * various error reporting improvements.
1634 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1635 (Thanks to Thiemo Seufer)
1636 * bug fix: nested reader invokations work correctly
1637 * bug fix: it is possible to have simultaneous references to foreign
1638 code and foreign data with the same name.
1640 ** added x86-64 support
1641 ** incompatible change: the threading api now works with thread
1642 objects instead of thread ids
1643 ** bug fix: threads are protected from signals and interruption when
1644 starting up or going down
1645 ** bug fix: a race where an exiting thread could lose its stack to gc
1646 ** bug fix: don't halt on infinite error in threads if possible
1647 ** fixed numerous gc deadlocks introduced in the pthread merge
1648 ** bug fix: fixed thread safety issues in read and print
1649 ** bug fix: debugger doesn't hang on session lock if interrupted at
1650 an inappropriate moment
1651 ** bug fix: run-program is now thread safe(r)
1652 ** bug fix: inner with-recursive-lock no longer releases the mutex
1653 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1654 * fixed some bugs revealed by Paul Dietz' test suite:
1655 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1656 EXPECTED-TYPE slots filled.
1657 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1659 ** MAKE-CONDITION accepts classes as type-designators.
1660 ** COMPILE may never return NIL.
1661 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1662 range before calling Unix time functions
1664 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1665 * numerous signal handling fixes to increase stability
1666 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1667 * minor incompatible change: we now correctly canonize default
1668 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1669 as per the MOP, rather than the historical (INITARG INITFUNCTION
1670 INITFORM). (reported by Bruno Haible)
1671 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1672 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1673 for more information.
1674 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1675 pathname is a directory pathname.
1676 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1677 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1679 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1680 :METACLASS argument in addition to a class name. (reported by
1681 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1682 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1683 (reported by Bruno Haible; patch from Gerd Moellmann)
1684 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1686 * bug fix: more cleanups to the floating point exception handling on
1687 x86-64 (thanks to James Knight)
1688 * bug fix: the compiler does not try to constant fold calls of
1689 COERCE and BIT-* functions when they return freshly constructed
1690 objects. (reported by Nikodemus Siivola and Paul Dietz)
1691 * optimization: improved the MIPS versions of generic (in the
1692 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1693 * optimization: direct conversion of (unsigned-byte 32) to floats on
1694 the PowerPC platform.
1695 * optimization: structure instances with raw slots now use less
1696 memory, and probably show better memory locality. (thanks to
1698 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1699 are treated as requests for stack allocation on the x86-64,
1700 Alpha-32, PPC and SPARC platforms.
1701 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1702 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1703 (thanks to Tony Martinez)
1705 ** gcing a dead thread can no longer lead to lockups
1706 ** threads block signals until they are set up properly
1707 ** errno is no longer shared by threads
1708 ** interrupt-thread restores the eflags register on x86
1709 ** fixed some lockups due to gc/thread interaction
1710 * fixed some bugs revealed by Paul Dietz' test suite:
1711 ** invalid dotted lists no longer raise a read error when
1712 *READ-SUPPRESS* is T
1713 ** signal an error if a symbol that names a declaration is used
1714 as the name of a type, or vice versa
1715 ** allow using the (declare (typespec var*)) abbreviation for
1716 (declare (type typespec var*)) with all type specifiers
1717 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1718 ** always return NIL from PPRINT-POP when OBJECT is NIL
1719 ** don't signal errors when pretty-printing malformed LABELS,
1720 FLET or MACROLET forms
1721 ** declarations in a DOLIST body are also in scope for the
1723 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1725 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1728 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1729 * fixed cross-compiler leakages that prevented building a 32-bit
1730 target with a 64-bit host compiler.
1731 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1732 opened with :IF-EXISTS :OVERWRITE.
1733 * fixed bug 281: error for an invalid qualifier in a short-form method
1734 combination method is not signalled until the faulty method is called.
1735 * bug fix: iteration variable type inferrer failed to deal with open
1736 intervals. (reported by Alan Shields)
1737 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1738 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1739 or more runtime options were provided to the sbcl binary.
1740 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1741 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1742 * optimization: there's now a fast-path for fixnum arguments in the
1743 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1744 * optimization: the code generated on x86-64 is more compact thanks
1745 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1746 * bug fix: floating-point exception handling now partly works on
1747 x86-64. (Thanks to James Knight)
1748 * improvement to the MIPS backend from Thiemo Seufer: C-style
1749 64-bit long long arguments and return values to alien functions
1751 * fixed some bugs revealed by Paul Dietz' test suite:
1752 ** the type-error signalled from WARN has a filled-in DATUM slot.
1753 ** the type-error required when a stream is not associated with
1754 a file has the stream as its datum.
1755 ** type-errors on single-floats on x86-64 no longer have
1756 :INVALID-OBJECT as the datum
1757 ** the type-errors signalled for invalid function names now have
1758 a correct expected type
1759 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1760 for structures defined with a :TYPE. Documentation strings for
1761 typed structures are no longer immediately discarded
1762 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1763 broadcast streams and synonym streams. FILE-LENGTH now also works
1764 on broadcast streams.
1766 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1767 * incompatible change: the --noprogrammer option, deprecated since
1768 version 0.7.5, has been removed. Please use the equivalent
1769 --disable-debugger option instead.
1770 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1772 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1773 inoperational for a while now) has been completely removed.
1774 * null lexical environments are now printed as #<NULL-LEXENV>,
1775 significantly reducing the amount of clutter in typical backtraces.
1776 * documentation on weak pointers, finalization, and after GC hooks
1777 has been added to the manual.
1778 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1779 as other specialized array types, is much faster. SUBSEQ and
1780 COPY-SEQ on such arrays have also been sped up.
1781 * optimization: EQL is now more efficient when at least other argument
1782 is known to be of type (OR FIXNUM (NOT NUMBER)).
1783 * fixed bug: the runtime is now less vulnerable to changes in the
1784 size of the SBCL object on OS X, and virtual memory is reserved for
1785 all spaces that need to be at a fixed address.
1786 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1787 * fixed bug: finalizers and after GC hooks that cause consing are now
1789 * fixed bug: compiler error messages and summaries are now printed to
1790 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1791 * fixed inference of the upper bound of an iteration variable.
1792 (reported by Rajat Datta).
1793 * fixed bug 211e: calling local functions with duplicated constant
1794 keyword argument no longer causes a bogus style warning about an
1796 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1797 ftype declaration to be disregarded. (reported by Dave Roberts)
1798 * fixed bug 373: caused by erronous compilation of references to alien
1799 variables in the runtime on ppc/darwin.
1800 * fixed bug 376: CONJUGATE type deriver.
1801 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1802 (reported by Baughn on #lisp)
1803 * a cross-compiler bug on non-x86 platforms has been identified and
1804 fixed. (thanks to Bruno Haible)
1805 * improvements to the MIPS runtime code for increased stability.
1806 (thanks to Thiemo Seufer)
1807 * increased the maximimum compact environment size to allow
1808 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1809 * partial workaround for bug 135: don't name the function we're
1810 calling for hairy cases of EVAL, so as not to accumulate one environment
1811 entry per call to EVAL. (reported by Kevin Reid)
1812 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1813 * optimization: type testing for non-vector arrays should be faster.
1814 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1816 * bug fix: setting 31st element of a bit vector to zero did not work
1818 * bug fix: redefining a class definition which failed due to a
1819 previous accessor / function clash now works (but see BUGS entry
1820 #380 for more problems in this area). (thanks to Zach Beane)
1821 * the long form of DEFINE-METHOD-COMBINATION disables method group
1822 checking when given a single method group with pattern *. (thanks
1824 * on x86 compiler supports stack allocation of results of simple
1825 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1826 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1827 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1828 * fixed some bugs related to Unicode integration:
1829 ** the restarts for recovering from input and output encoding
1830 errors only appear when there is in fact such an error to
1832 * fixed some bugs revealed by Paul Dietz' test suite:
1833 ** MISC.549 and similar: late transformation of unsafe type
1834 assertions into derived types caused unexpected code
1836 ** SCALE-FLOAT type deriver is less wrong.
1837 ** type derivers for EXP, LOG and similar functions compute result
1838 types for complex arguments better.
1839 ** (MISC.563) CONJUGATE type deriver works for very restricted
1841 ** out-of-line type testers for character strings are available.
1842 ** EQUAL compiler transform understands specialness of objects
1844 ** accessing double-floats stored on the stack now works on x86-64.
1845 ** debugger internals could sometimes create invalid lispobjs,
1846 resulting in GC crashes.
1847 ** MISC.548: type check weakening can convert required type into
1849 ** initialization forms for bindings are not in scope of free special
1852 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1853 * incompatible change: thread support for non-NPTL systems has
1854 been removed - locking is buggy and unreliable. A threaded
1855 SBCL build will now warn at startup and refuse to create
1856 new threads, unless futex support is detected in the kernel
1857 * incompatible change: the top level REPL now has only an ABORT
1858 restart associated with it, not TOPLEVEL and ABORT as it used to.
1859 TOP and TOPLEVEL are now available as debugger commands for
1860 returning to the top level.
1861 * incompatible change: forms evaluated in the REPL now use the
1862 global optimization policy.
1863 * incompatible change: user- and system-initialization files are
1864 no longer processed with LOAD, but by READ and EVAL; hence the
1865 global optimization policy, startup package, readtable, etc,
1867 * The .fasl file format number has been incremented because of
1868 various incompatible changes.
1869 * internal entry point details and argument counts no longer appear
1870 in backtraces unless explicitly requested by setting
1871 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1872 * built-in and standard functions no longer have names like "top
1873 level local call to FOO".
1874 * fixed bug 32: functions defined in non-null lexical environments
1875 now have more legible printed representation
1876 * fixed bug 33: functions defined in non-null lexical environemnts
1877 are now more amenable to inspection by INSPECT.
1878 * workaround for bug 354: XEPs no longer appear in backtraces unless
1879 explicitly requested.
1880 * fixed bug: receiving the signal which results from attempting to
1881 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1882 SIGBUS on other *BSDs) on architectures where the C stack is also
1883 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1884 to debugger confusion or wild instability. (reported by Cheuksan
1886 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1887 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1888 (reported by Lutz Euler)
1889 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1890 compiled code. (reported by Paul Dietz)
1891 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1892 specialized parameter to a method to something that is not TYPEP
1893 the specializer is now possible.
1894 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1895 face of package deletion.
1896 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1897 pathname. (reported by David Tolpin on c.l.l)
1898 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1899 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1900 STRING-STREAM and FILE-STREAM, respectively.
1901 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1902 than previously; in addition, its test suite is now run on build.
1903 (thanks to Luke Gorrie)
1904 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1905 * optimization: added a immediate representation for single-floats
1907 * fixed some bugs revealed by Paul Dietz' test suite:
1908 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1909 ** attempting to create a package with a colliding nickname causes
1910 correctable errors to be signalled.
1911 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1912 ** several x86-64 backend bugs related to sign-extension of immediate
1915 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1916 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1917 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1919 * fixed loading of multiply forward-referenced layouts.
1920 (thanks to Cheuksan Wang)
1921 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1922 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1923 (see COMPILE-FILE documentation for details of :PRINT option)
1924 * fixed bugs 19 and 317: fixed-format floating point printing is
1925 more accurate. This also fixes a bug reported by Adam Warner
1926 related to the ~@F format directive.
1927 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1929 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1930 dispatch table if the from-char is a dispatch macro character.
1931 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1932 by the wordsize no longer ignore the last word. (reported by Lutz
1934 * fixed bug in type checking of dynamic-extent variables. (reported
1936 * optimization: sequence traversal functions use their freedom to
1937 coerce function designators to functions.
1938 * optimization: code with many calls to CLOS methods specialized on
1939 CLOS classes has had redundant type checks removed.
1940 * optimization: added declarations to speed up operations that access
1941 the internal character database (for example STRING-UPCASE)
1942 * optimization: comparison operations between floats and sufficiently small
1943 fixnums no longer create extra rationals
1944 * fixed some bugs related to Unicode integration:
1945 ** portions of multibyte characters at the end of buffers for
1946 character-based file input are correctly transferred to the
1947 start of the buffer at the next read.
1948 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1949 passing it through to OPEN.
1950 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1951 argument given to internal calls to OPEN.
1952 ** the built-in comment readers (introduced by character sequences
1953 ";" and "#|") are more forgiving to encoding errors; they will
1954 STYLE-WARN and then attempt to resync the stream at a character
1955 boundary. (thanks to Teemu Kalvas)
1956 * fixed some bugs in the x86-64 port:
1957 ** Negative short int return values from c-calls are sign-extended
1959 ** The stack is aligned to 16-bytes for c-calls, as required by
1960 the ABI. (thanks to Cheuksan Wang)
1961 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1962 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1963 for enums. (thanks to Vincent Arkesteijn)
1964 ** Multiple small optimizations and bugfixes for floating point
1966 * fixed some bugs revealed by Paul Dietz' test suite:
1967 ** Space, Tab, Linefeed, Return and Page have the invalid
1968 secondary constituent character trait.
1969 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1971 ** WITH-INPUT-FROM-STRING should only update the index place on
1973 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1974 works more reliably.
1975 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1976 with one-dimensional arrays with fill pointers.
1977 ** TYPE-OF failed on a complex with an integer realpart and a
1979 ** compiler failure during type inference for the code of form
1980 (IF (EQL X (THE ...)) ...) (MISC.535).
1982 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1983 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1984 platform. The port supports 61-bit fixnums, large memory spaces
1985 and reloading shared object files.
1986 * enhancement: saving cores with foreign code loaded is now
1987 supported on ppc/Darwin in addition to the previously supported
1989 * enhancement: the statistical profiler now walks deeper into the
1990 call stack for more meaningful call-graphs and accrued time
1991 reports (x86/x86-64 only). It also now reports time spent in
1993 * enhancement: it is now possible to trace most individual methods
1994 of a generic function in addition to tracing the generic function
1996 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1998 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1999 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2000 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2001 produces an error. (thanks to Vincent Arkesteijn)
2002 * bug fix: NAMESTRING on pathnames with :WILD components in their
2003 directories works correctly. (thanks to Artem V. Andreev)
2004 * fixed bug 125: compiler preserves identity of closures. (reported
2006 * bug fixed: functions with &REST arguments sometimes failed with
2007 "Undefined function" when compiled with (DEBUG 3). (reported by
2009 * bug fix: overflow during compiling of setting element of a bit
2010 vector with constant index and value. (reported by Timmy Douglas)
2011 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2012 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2013 * on x86 compiler partially supports stack allocation of dynamic-extent
2015 * GO and RETURN-FROM do not check the extent of their exit points
2016 when compiled with SAFETY 0.
2017 * fixed some bugs related to Unicode integration:
2018 ** encoding and decoding errors are now much more robustly
2019 handled; it should now be possible to recover even from invalid
2020 input or output to the terminal. (thanks to Teemu Kalvas)
2021 ** provided a first cut at implementing STRING-TO-OCTETS and
2022 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2023 ** altered the SB-MD5 contributed module slightly, changing the
2024 interface just enough for it to be supportable for builds where
2025 lisp characters are not eight bits.
2026 * fixed some bugs revealed by Paul Dietz' test suite:
2027 ** the FORMATTER-generated functions for ~V[ conditionals require
2028 the correct number of arguments.
2029 ** READ-FROM-STRING returns the mandated second value when applied
2030 to displaced strings.
2031 ** the #\Rubout and #\Backspace characters are treated as invalid
2032 constituent characters by the tokenizer.
2034 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2035 * new feature: reloading changed shared object files with
2036 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2037 * new feature: references to foreign variables and functions
2038 can now be compiled and loaded before the corresponding shared
2039 object file is loaded, as long as the foreign definitions are
2040 available at runtime.
2041 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2042 supported. (thanks to Dan Debertin)
2043 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2044 just keywords. (thanks to Vincent Arkesteijn)
2045 * fixed bug #331: structure-class instances corresponding to
2046 DEFSTRUCT forms are now created eagerly.
2047 * fixed bug #345: backtraces from calls to undefined functions work
2048 on x86 as well. Related bug #61 is now also partially fixed on x86
2049 (backtraces from throws to unknown catch tags.)
2050 * bug fix: lambda-list parsing is now stricter vrt. order and number
2051 of lambda-list keywords.
2052 * bug fix: as specified by AMOP, an error is signalled if a
2053 class-option appears multiple times in a DEFCLASS form. (reported
2055 * bug fix: RANDOM can be compiled when the compiler derives the type
2056 of its numeric argument as a disjoint set of small integers.
2057 (reported by Paul Dietz)
2058 * bug fix: starting a core saved with shared objects loaded when
2059 those objects are not available no longer causes threaded SBCL to
2060 hang. (reported by Sean Ross)
2061 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2063 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2064 argument, which must always be NIL. (reported by Kalle Niemitalo)
2065 * bug fix: printing 1.0d+23 no longer results in an error.
2066 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2067 * bug fix: structure slot setters preserve evaluation order. (thanks
2069 * fixed some bugs related to Unicode integration:
2070 ** RUN-PROGRAM can allow its child to take input from a Lisp
2071 stream. (reported by Stefan Scholl)
2072 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2073 locales. (reported by Ken Causey)
2074 * fixed some bugs revealed by Paul Dietz' test suite:
2075 ** INCF, DECF and REMF evaluate their place form as specified in
2077 ** FORMATTER expands ~{ ~} iteration directives with V or #
2078 parameters correctly.
2079 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2080 consequent uses no arguments correctly.
2081 ** the system has a partial understanding of the (COMPLEX RATIO)
2083 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2086 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2087 * new feature: a build-time option (controlled by the :SB-UNICODE
2088 keyword feature, enabled by default) for building the system with
2089 support for the entire 21-bit character space defined by the
2091 * new feature: the system now has rudimentary external-format
2092 support; the primary user-visible change at this time is that
2093 characters with the high bit set (such as lower-case-e-acute) will
2094 print correctly to a terminal in a UTF-8 environment.
2095 * minor incompatible change: BASE-CHAR no longer names a class;
2096 however, CHARACTER continues to do so, as required by ANSI.
2097 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2098 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2099 has been moved to the SB-EXT package (temporarily re-exported from
2101 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2102 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2104 * fixed bug #351: better error-handlind and reporting for malformed
2106 * fixed bug #350: bignum-printing is now more memory-efficient,
2107 allowing printing of very large bignums, eg. (expt 2 10000000).
2108 (reported by Bruno Haible)
2109 * fixed bug #302: better primitive-type selection for intersection
2111 * fixed bug #308: non-graphic characters now all have names, as
2112 required. (reported by Bruno Haible)
2113 * bug fix: redefining a class with different superclasses now correctly
2114 removes it from the direct-subclasses of its previous superclasses.
2115 (reported by David Morse)
2116 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2117 new value now works. (reported by Bruno Haible)
2118 * bug fix: correct canonicalization of multiple non-standard slot
2119 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2120 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2121 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2122 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2123 now exists, an signals an error.
2124 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2125 during macroexpansion for non-iteration variables as well. (reported
2126 by Bruno Haible for CMUCL)
2127 * bug fix: Cyclic structures and unprintable objects in compiler
2128 messages no longer cause errors. (reported by Bruno Haible)
2129 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2130 and READ-FROM-STRING all now return a primary value of NIL if
2131 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2132 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2133 (reported by Bruno Haible for CMUCL)
2134 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2135 do not designate a condition. (reported by Bruno Haible for
2137 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2138 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2139 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2140 permitting resolution in favour of any of the conflicting symbols.
2141 (reported by Bruno Haible for CMUCL)
2142 * bug fix: EQUAL compiler optimizations is less aggressive on
2143 strings which can potentially compare true despite having distinct
2144 specialized array element types.
2145 * bug fix: unit enumerations can be defined without dividing by
2146 zero. (thanks to Vincent Arkesteijn)
2147 * FORMAT compile-time argument count checking has been enhanced.
2148 (report from Bruno Haible for CMUCL)
2149 * a partial workaround for the bug 262: the compiler does not try to
2150 inline-expand a local function doing RETURN-FROM from a deleted
2151 BLOCK. (thanks to Peter Denno for the bug report and to David
2152 Wragg for the simple test case)
2153 * fixed some bugs revealed by Paul Dietz' test suite:
2154 ** RENAME-PACKAGE allows all package designators as new package
2156 ** constraint propagation and lambda variable substitution are
2157 more cautious in dealing with partially deleted code.
2158 ** compiler sometimes lost reoptimization passes.
2159 ** CERROR, given a condition as condition designator, passes its
2160 remaining arguments to the continue format control without
2162 ** Case-altering FORMAT directives work correctly on non-ASCII
2164 ** The REAL type specifier handles bounds outside the floating
2165 point ranges without signalling FLOATING-POINT-OVERFLOW.
2166 ** Functions with IR1-transformations can create intercomponent
2167 references to global functions.
2168 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2170 ** FORMAT ~R treats a nil value for its first parameter correctly.
2172 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2173 * enhancement: saving cores with foreign code loaded is now
2174 supported on x86/NetBSD and sparc/Linux in addition to the previously
2175 supported platforms.
2176 * bug fix: on some platforms repeated installations caused multiple
2177 copies of HTML documentation to be installed -- should not happen
2178 any more. (reported by Stefan Scholl)
2179 * bug fix: parsing self-recursive alien record types multiple times
2180 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2181 original patch by Helmut Eller for CMUCL)
2182 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2183 (thanks to Richard Kreuter)
2184 * bug fix: defining classes whose accessors are methods on existing
2185 generic functions in other (locked) packages no longer signals
2186 bogus package lock violations. (reported by François-René Rideau)
2187 * bug fix: special variables as DEFMETHOD parameters no longer have
2188 associated bogus type declarations. (reported by David Wragg and
2190 * bug fix: read-write consistency on streams of element-type
2191 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2192 * bug fix: redefiniton of the only method of a generic function with
2193 no DEFGENERIC no longer emits a full WARNING. In addition,
2194 redefinition of generic functions with no DEFGENERIC to an
2195 incompatible lambda list now signals an error. (thanks to Zach
2197 * bug fix: DEFGENERIC now works even when there's a function of the
2198 same name in an enclosing lexical environment. (thanks to Zach
2200 * fixed compiler failure, caused by instrumenting code during
2201 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2202 * optimization: added loop analysis and improved register allocation
2203 * fixed some bugs revealed by Paul Dietz' test suite:
2204 ** POSITION on displaced vectors with non-zero displacement
2205 returns the right answer.
2206 ** (SIMPLE-STRING) is a valid type specifier for sequence
2208 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2210 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2211 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2213 ** The justification version of the FORMAT ~< directive treats
2214 non-zero minpad parameter correctly.
2216 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2217 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2218 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2219 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2220 the supported interface.
2221 * new feature: Single-stepping of code compiled with DEBUG 2 or
2222 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2223 * new feature: saving cores with foreign code loaded is now
2224 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2225 Timothy Moore's work for CMUCL)
2226 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2227 parameters to * instead of NIL if no initform is supplied.
2228 (reported by Johan Bockgaard)
2229 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2230 correctly when FUNCALL appears as the car of the form. Note:
2231 despite this FUNCALL forms are not currently subject to
2232 compiler-macro expansion. (port of Raymond Toy's fix for the
2233 same from CMUCL, reported by Johan Bockgaard)
2234 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2235 (thanks for Teemu Kalvas)
2236 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2237 case instead of terminating the iteration (thanks for Julian
2238 Squires, Sean Champ and Raymond Toy)
2239 * bug fix: incorrect expansion of defgeneric that caused a style
2240 warning. (thanks for Zach Beane)
2241 * bug fix: slot accessor effective method computation works properly
2242 for classes with multiple non-standard applicable methods on
2243 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2244 * on x86 compiler supports stack allocation of results of LIST and
2245 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2246 CMUCL implementation by Gerd Moellmann)
2247 * fixed some bugs revealed by Paul Dietz' test suite:
2248 ** FORMAT strings with both the ~<~:;~> form of the justification
2249 directive and pretty-printing directives cause an error.
2251 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2252 * incompatible change: the internal functions
2253 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2254 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2255 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2256 instead of the old functions.
2257 * new feature: on platforms where "dladdr" is available foreign
2258 function names now appear in backtraces. (based on Helmut Eller's
2260 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2261 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2263 * documentation: ASDF manual is now distributed with SBCL.
2264 * bug fix: SBCL can now load its contributed modules with REQUIRE
2265 even if the system-provided entries have been removed from
2266 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2268 * bug fix: backtraces involving undefined functions or assembly
2269 routines are more informative. (thanks to Brian Downing)
2270 * bug fix: mutually referent alien structures now work correctly.
2271 (reported by Rick Taube)
2272 * bug fix: structures defined by WITH-ALIEN can be referred to
2273 within other definitions in the same WITH-ALIEN.
2274 * bug fix: division operators (MOD, TRUNCATE and the like) with
2275 constant zero divisors and integer dividends no longer generate
2277 * bug fix: provide default methods for INPUT-STREAM-P and
2278 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2279 * bug fix: improve the detection and resolution of MOP metacycles
2280 (where effective method computation depends on the generic function
2281 having its effective method computed). (reported by Bruno Haible)
2282 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2283 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2284 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2285 * bug fix: pathnames with relative directory components can be
2286 represented relative to default pathnames.
2287 * optimization: in taking the GCD of bignums, reduce the two bignums
2288 to approximately the same size (using Euclid's algorithm) before
2289 applying the more sophisticated binary GCD. (thanks to Juho
2291 * optimization: COUNT on bitvectors now operates word-at-a-time.
2292 * optimization: ASH with a positive, but not necessarily constant,
2293 (leftwards) shift, when in a modular context, is compiled to a
2295 * fixed some bugs revealed by Paul Dietz' test suite:
2296 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2297 if the corresponding argument is NIL.
2298 ** FORMAT directives accepting a minpad parameter treat negative
2300 ** literal commas as character parameters to FORMAT directives are
2302 ** literal spaces directly after ~<Newline> directives within a
2303 format-logical-block (~:< ~@:>) do not induce :FILL-style
2304 conditional newlines.
2305 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2306 as specified: it no longer includes conditional newlines).
2307 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2309 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2310 :PER-LINE-PREFIX argument does not evaluate to a string.
2312 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2313 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2314 the manual for details; for now, package locks can be disabled by
2315 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2316 if no major problems are found then it is likely that they will be
2317 compiled in unconditionally.
2318 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2319 unsupported operators on all platforms. To load a shared library
2320 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2321 object file, link it into a shared library outside of SBCL and
2322 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2324 * bug workaround: ROOM T and the GC were not working together
2325 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2326 reporting facility aren't preserved. That reporting has been
2327 disabled, so now until and unless someone figures out how to make it
2328 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2329 * minor incompatible change: as threatened around sbcl-0.8.0, the
2330 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2331 The symbols are also exported from SB-PCL for backwards
2332 compatibility, but more so than before SB-PCL should be treated as
2333 an implementation-internal package.
2334 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2336 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2337 It is known as of this release not to work on the Alpha, however.
2338 * fixed bug #167: errors signalled due to illegal syntax in method
2339 bodies are now more legible.
2340 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2341 designators and can hence be used with TYPEP.
2342 * fixed bug #333: CHECK-TYPE now ensures that the type error
2343 signalled, if any, has the right object to be accessed by
2344 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2345 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2346 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2348 * fixed bug #334: programmatic addition of slots using specialized
2349 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2350 and :ALLOCATION :CLASS slots.
2351 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2352 float exponents. (rereported by Peter Seibel)
2353 * fixed a bug: #\Space (and other whitespace characters) are no
2354 longer considered to be macro characters in standard syntax by
2355 GET-MACRO-CHARACTER.
2356 * fixed bug: initialization of condition class metaobjects no longer
2357 causes an instance of the condition to be created. (reported by
2359 * fixed bug: it is now possible to have slots such that
2360 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2361 description is neither :INSTANCE nor :CLASS.
2362 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2363 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2364 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2365 non-standard SLOT-DEFINITION-ALLOCATION values.
2366 * fixed bugs: various race conditions handling exiting threads.
2367 CL-PPCRE's thread torture test now passes.
2368 * fixed arguably-a-bug: GC time proportional to number of threads in
2369 system even when most of them are idle
2370 * optimization: improved performance of BIT and SBIT on bit-vectors.
2371 * .fasl file incompatibility: The fasl file version number has
2372 been incremented because of changes associated with package locks.
2374 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2375 * minor incompatible change: the system no longer provides
2376 optimization hints (conditions which are TYPEP
2377 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2378 mode; these hints are still emitted when higher SPEED optimization
2380 * new contrib module: a sampling profiler (profiling by statistical
2381 sampling, rather than by instrumenting functions) is available as
2382 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2383 * the behaviour of the standard function ED is now customizeable by
2384 third parties through a hook variable: see ED's documentation
2385 string for information on the protocol.
2386 * the compiler no longer emits efficiency notes for (FUNCALL X)
2387 when the type of X is uncertain under default optimization
2389 * fixed bug 276: mutating a binding of a specialized parameter to a
2390 method to something that is not TYPEP the specializer is now
2392 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2393 exhibit the required behaviour on the x86 platform. (thanks to
2394 Peter van Eynde, Eric Marsden and Bruno Haible)
2395 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2396 even for difficult arguments. (reported by Peter Graves)
2397 * fixed bug 141a: the backquote printer now descends quoted
2399 * fixed another bug in backquote printing: no more destructive
2400 modification of the form's list structure. (reported by Brian
2402 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2403 move between its address being taken and the call to
2404 interrupt_thread, fixing a crashing race condition.
2405 * the SB-POSIX contrib implementation has been adjusted so that it
2406 no longer exhibits ridiculously poor performance when constructing
2407 instances corresponding to C structs.
2409 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2410 * minor incompatible change: the sb-grovel contrib now treats C
2411 structures as alien (in the sense of SB-ALIEN) objects rather than
2412 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2413 has implications for memory management of client code
2414 (sb-grovel-returned objects must now be manually managed) and for
2415 type safety (alien objects now have full types).
2416 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2417 used to control emission of compiler diagnostics, rather than the
2418 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2419 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2420 quality should be considered deprecated.
2421 * install.sh now installs the user manual as well
2422 * (not quite a new documentable feature, but worth considering in
2423 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2424 beginnings of a semantically meaningful condition hierarchy is
2425 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2427 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2428 designator as the defaults argument.
2429 * fixed bug: Displaced arrays whose displaced-to array has become
2430 too small now cause ARRAY-DIMENSION to signal an error, providing
2431 for safer bounds-checking. (reported by Bruno Haible)
2432 * fixed bug: DEFCLASS slot definitions with identical :READER and
2433 :WRITER names now signal a reasonable error. (reported by Thomas
2435 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2437 * fixed bug: Invalid slot specification errors now print correctly.
2438 (thanks to Zach Beane)
2439 * fixed bug 320: Shared to local slot value transfers in class
2440 redefinitions now happen corrently from superclasses as
2441 well. (reported by Bruno Haible)
2442 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2444 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2445 improper lists correctly. (reported by Bruno Haible)
2446 * fixed bug 313: source-transform for <fun-name> was erroneously
2447 applied to a call of a value of a variable with name <fun-name>.
2448 (reported by Antonio Menezes Leitao)
2449 * fixed bug 307: The obsolete instance protocol ensures that
2450 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2451 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2452 remove dead unknown-values globs from the middle of the stack.
2453 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2454 easier to resume long computations after using *BREAK-ON-SIGNALS*
2455 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2456 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2457 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2458 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2459 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2460 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2462 * fixed a (fairly theoretical) bug in string printing: if
2463 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2464 does not have array-element-type equal to the most general string
2466 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2467 function name. (thanks to Nikodemus Siivola)
2468 * fixed bug: compile-time format string checker failed on
2469 non-closed ~{. (reported by Thomas F Burdick)
2470 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2471 don't warn on legal loop code involving more than one aggregate
2472 boolean. (thanks to Nikodemus Siivola)
2473 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2474 in keyword position to MAKE-PACKAGE was not being reported as
2476 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2477 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2479 * fixed bugs 280 and 312: the checking for multiple definitions in a
2480 file is less likely to become confused by uses of inline
2482 * fixed bug: the #S reader macro performs the keyword coercion
2483 specified for slot names. (reported by Kalle Niemitalo)
2484 * fixed bug: lambda lists may contain symbols whose names start with
2485 & but are not lambda-list-keywords; their occurrence triggers a
2487 * fixed bug 321: define-method-combination argument lambda lists do
2488 not cause invalid code to be generated when &optional or &aux
2489 variables have default values. (reported by Bruno Haible)
2490 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2491 have CLOS classes; this appears also to have fixed reported
2492 instability in stack exhaustion detection.
2493 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2494 MUFFLE-WARNING when no associated restart is present is now
2496 * optimization: rearranged the expansion of various defining macros
2497 so that each expands into only one top-level form in a
2498 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2500 * optimization: used a previously little-used slot in symbols to
2501 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2503 * fixed some bugs revealed by Paul Dietz' test suite:
2504 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2506 ** Fixed an optimization bug related to inheritance of initforms
2507 from local to shared slots.
2508 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2509 ** CAST optimizer forgot to flush argument derived type.
2510 ** print/read consistency on floats is now orders of magnitude
2511 more likely. (thanks also to Bruno Haible for a similar report
2513 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2514 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2515 for objects of type REAL. Make it so.
2516 ** SXHASH is specified (infelicitously) to respect similarity,
2517 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2518 it so. (thanks to Markus Ziegler)
2519 ** on the Alpha, the compiler succeeds in compiling functions
2520 returning a known number of arguments greater than 63.
2521 ** fixed handling of invalid NIL arguments in keyword position
2523 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2525 ** provided a readably-printable representation for RANDOM-STATE
2527 ** ensured that pathnames, where they have a namestring, always
2528 print using #P"..." syntax.
2530 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2531 * Support for the forthcoming 2.0 version of the NetBSD kernel
2532 running on x86 hardware has been added. (thanks to Perry
2533 E. Metzger most immediately, and others for their past work)
2534 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2535 current release version; SBCL's OpenBSD support had been broken
2536 since about the time of OpenBSD's switch to ELF binary format.)
2537 * [placeholder for DX summary]
2538 ** user code with &REST lists declared dynamic-extent, under high
2539 speed or space and low safety and debug optimization policy.
2540 * The manual has been converted to Texinfo format and the debugger
2541 chapter from the cmucl manual has been added.
2542 * A facility has been added to extract documentation strings from
2543 sbcl and store them as Texinfo-formatted snippets for inclusion in
2544 the manual (via Texinfo's @include directive)
2545 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2546 the test case to Dave Roberts)
2547 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2548 pointers. (reported by Sean Ross)
2549 * bug fix: PROFILE output is printed nicely even for large numerical
2550 values. (thanks to Zach Beane)
2551 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2552 greater than 32 handle EOF correctly.
2553 * bug fix: on X86 an immediate argument of the IMUL instruction is
2554 correctly printed in disassembly. (thanks to Lutz Euler)
2555 * bug fix: class slots in redefined classes preserve their old
2556 values. (thanks to Bruno Haible and Nikodemus Siivola)
2557 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2558 succeeds. (reported by Marco Baringer)
2559 * fixed some bugs revealed by Paul Dietz' test suite:
2560 ** READ-SEQUENCE now works on ECHO-STREAMs.
2561 ** RATIONALIZE works more according to its specification. (thanks
2564 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2565 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2566 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2567 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2568 should matter to you only if you rebind the printer control
2569 variables and then find you want different bindings in the
2570 debugger than in the ordinary execution of your program.)
2571 * The runtime build system has been tweaked to support building
2572 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2573 assembler and linker. (thanks to Nikodemus Siivola)
2574 * Unbound, undefined, undeclared variables now trigger full
2575 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2576 more often programmer error than deliberate exploitation of undefined
2578 * optimization: the hash algorithm for strings has changed to one
2579 that is less vulnerable to spurious collisions. (thanks to Juho
2581 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2582 needless bounds checking. (thanks to Juho Snellman)
2583 * optimization: implemented multiplication as a modular
2584 (UNSIGNED-BYTE 32) operation on the PPC backend.
2585 * fixed some bugs revealed by Paul Dietz' test suite:
2586 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2588 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2589 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2590 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2592 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2594 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2596 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2597 their output stream on EOF from read.
2598 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2599 have been read to end-of-file.
2600 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2602 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2603 description of determination of which consecutive characters
2605 ** Printing the "Space" character with escaping on now yields "#\\ ",
2606 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2607 ** Reading floating-point numbers with *READ-BASE* set to a number
2608 less than 10 works correctly.
2609 ** Reading floating-point numbers with *READ-BASE* set to a number
2610 more than 10 works correctly.
2611 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2612 the readtable currently in effect.
2614 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2615 * minor incompatible change: parsing of namestrings on a physical
2616 (Unix) host has changed; numbers after the final #\. in a
2617 namestring are no longer interpreted as a version field. This is
2618 intented to be largely invisible to the user, except that the
2619 meaning of the namestring "*.*.*" has changed: it now refers to a
2620 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2621 should usually be replaced by
2622 (make-pathname :name :wild :type :wild :version :wild)
2623 with the added benefit that this is more likely to be portable.
2624 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2625 option to OPEN now signals an error if the file being opened
2626 exists; this may have an impact on existing code.
2627 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2628 signals received, on the PowerPC platforms (both Linux and
2629 Darwin). (thanks to Pierre Mai for pointing out the location of
2631 * several fixes on OS X: The system now builds and runs cleanly on
2632 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2633 (thanks to Brian Mastenbrook)
2634 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2635 second-resolution: integer multiples of 1/3600 between -24 and 24.
2636 (thanks to Vincent Arkesteijn)
2637 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2638 type when called with 1 argument; PEEK-CHAR checked type of
2639 PEEK-TYPE only after having read first character from a
2640 stream. (reported by Peter Graves)
2641 * bug fix: the garbage collector now has much better locality
2642 behaviour, and in particular no longer treats all memory as being
2643 exhausted when many small objects point to each other in a deeply
2645 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2646 recognized as being TYPEP their class.
2647 * bug fix: the PUSHNEW documentation string has been corrected.
2648 (thanks to Vincent Arkesteijn)
2649 * bug fix: defaulting of the value for the last of an atypically
2650 large number of multiple values being bound was not being
2651 performed correctly on the Alpha or PPC platforms
2652 * optimization: implemented multiplication as a modular
2653 (UNSIGNED-BYTE 32) operation on the x86 backend.
2654 * optimization: SEARCH on simple-base-strings can now be open-coded.
2655 (see also contrib/compiler-extras.lisp for inspiration for
2656 teaching the compiler about the Boyer-Moore algorithm).
2657 * value, returned by MAX (and MIN) called with several EQUALP, but
2658 not EQL, arguments now does not depend on compiler settings.
2659 * fixed some bugs revealed by Paul Dietz' test suite:
2660 ** in stack analysis liveness information is propagated from
2661 non-local entry points.
2662 ** pathwise CAST removing failed when the CAST node did not start
2664 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2665 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2667 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2668 host is already defined.
2669 ** RENAME-FILE works on streams instead of signalling an internal
2671 ** PEEK-CHAR uses the current readtable when determining whether
2672 or not a character is whitespace.
2673 ** MERGE-PATHNAMES handles the case when the pathname does not
2674 specify a name while the default-pathname specifies a version.
2675 ** Pathnames now stand a better chance of respecting print/read
2677 ** Attempting to use standardized file system operators with a
2678 pathname with invalid :DIRECTORY components signals a
2680 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2682 ** DIRECTORY on logical pathnames is more correct.
2683 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2684 signal a TYPE-ERROR if their argument is not a stream.
2685 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2686 designator argument does not designate a stream.
2687 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2688 examining the synonym.
2689 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2691 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2692 element-type larger than ([UN]SIGNED-BYTE 32).
2694 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2695 * When built with the :SB-FUTEX feature, threaded builds now take
2696 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2697 for faster/more reliable mutex and condition variable support.
2698 * Incompatible change (but one you probably shouldn't have been using
2699 anyway): the interface and code for arbitrating between multiple
2700 threads in the same user session has been redesigned.
2701 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2702 error when called without an explicit environment argument.
2703 (thanks to Nikodemus Siivola)
2704 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2705 confused about their position. (thanks to Adam Warner and Gerd
2707 * bug fix: Pretty printing backquoted forms with unquotations in the
2708 argument list position of various code constructs such as LAMBDA
2709 now works correctly. (reported by Paul Dietz)
2710 * bug fix: Pretty printing unquotations no longer loses all
2711 stream position information.
2712 * optimization: performance of string output streams is now less
2713 poor for multiple small sequence writes.
2714 * optimization: performance of CSUBTYPEP in the presence of complex
2715 expressions involving CONS and NOT many times has been improved.
2716 (reported by Paul Dietz)
2717 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2719 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2720 streams for the REPL, for the debugger, and so on. (thanks to
2722 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2723 * fixed some bugs revealed by Paul Dietz' test suite:
2724 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2725 is ignored if it is too small, rather than propagating through
2726 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2727 ** extremely complex negations of CONS types were not being
2728 sufficiently canonicalized, leading to inconsistencies in
2730 ** VALUES tranformer lost derived type.
2732 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2733 * fixed a bootstrapping bug: the build process no longer assumes
2734 that the various BOOLE-related constants have the same value in
2735 host and target lisps. (noted by Paul Dietz' test suite on an
2736 SBCL binary built from CLISP)
2737 * The system can now be dynamically linked on the MIPS platform,
2738 which enables dynamic loading of foreign code from Lisp. (thanks
2739 to Ralf Baechle for discussions on the MIPS ABI)
2740 * The system now records debugging information for its own source
2741 files in a filesystem-position-independent manner, relative to
2742 the "SYS" logical host.
2743 * fixed a compiler bug: MV-LET convertion did not check references
2744 to the "max args" entry point. (reported by Brian Downing)
2745 * tweaked disassembly notes to be less confident about proclaiming
2746 some instruction as an LRA. (thanks to Brian Downing)
2747 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2748 now each have their own history, command character, and other
2749 characteristics. (thanks to David Lichteblau)
2750 * fixed some bugs revealed by Paul Dietz' test suite:
2751 ** compiler failure in compiling LOGAND expressions including a
2753 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2754 shift greater than 32.
2755 ** FLUSH-DEST did not mark blocks for type check regeneration.
2756 ** HANDLER-CASE failed to accept declarations in handler clauses
2757 in some circumstances.
2759 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2760 * New code in contrib/sb-introspect (still probably not entirely
2761 stable yet) provides some support for smart Lisp development
2762 environments like SLIME.
2763 * The conditions signalled for errors occurring when loading .fasl
2764 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2765 in a way which should help ASDF recover gracefully.
2766 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2767 stuff has been cleaned up. If you code contrib/ stuff, this might
2768 affect you, and you can look at contrib/README, contrib/STANDARDS,
2769 and/or the 0.8.4.27 diff to check.
2770 * In full calls the compiler now does not generate checks for declared
2771 argument types for all arguments.
2772 * various threading fixes
2773 ** and some experimental patches which didn't make it into
2774 the main tree for this release, but which are shipped in
2775 contrib/experimental-thread.patch as a possible fix for some
2776 failures (deadlock, spinning...) in GC-intensive multithreaded
2778 * fixed PPC build problem (source code incompatibility of different
2779 library versions): added offsetof() hackery which attempts to divine
2780 where glibc maintainers put uc_mcontext today
2781 * fixed bug 282: compiler does not trust type assertions while passing
2782 arguments to a full call.
2783 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2784 &OPTIONAL VALUES type parameter.
2785 * fix bug 214: algorithm for noting rejected templates is now more
2786 similar to that of template seletion. (also reported by rydis on
2788 * fixed bug 141b: printing backquoted information readably and prettily
2789 inserts a space where necessary.
2790 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2791 (CHARACTER 10) are now reported as errors, rather than propagated
2792 as unknown types. (reported by piso on #lisp)
2793 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2794 with values NIL and :ERROR. (thanks to Milan Zamazal)
2795 * fixed bug 191c: CLOS now does proper keyword argument checking as
2796 described in CLHS 7.6.5 and 7.6.5.1.
2797 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2798 longer raise an error; further, using a list as a for-as-arithmetic
2799 counter now raises a meaningful error.
2800 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2801 now understood by sequence creation functions such as MAKE-SEQUENCE
2803 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2804 a STRING-INPUT-STREAM.
2805 * compiler enhancement: SIGNUM is now better able to derive the type
2807 * type declarations inside WITH-SLOTS are checked. (reported by
2809 * fixed some bugs revealed by Paul Dietz' test suite:
2810 ** incorrect optimization of TRUNCATE for a positive first
2811 argument and negative second.
2812 ** compiler failure in let-convertion during flushing dead code.
2813 ** compiler failure while deriving type of TRUNCATE on an
2814 interval, containing 0.
2815 ** ASH of a negative bignum by a negative bignum count now returns
2817 ** intersection of CONS types now canonicalizes properly, fixing
2818 inconsistencies in SUBTYPEP.
2820 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2821 * incompatible change: The --disable-debugger command line
2822 option now clobbers the debugger at a more fundamental
2823 level, by redefining #'INVOKE-DEBUGGER instead of by
2824 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2825 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2826 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2827 enter the debugger REPL and then suffer recursive errors
2828 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2829 while under the new system, BREAK in --disable-debugger mode
2830 terminates the system just as an unhandled error would.
2831 * fixed compiler performance when processing loops with a step >1;
2832 * bug fix: DOCUMENTATION now retrieves generic function
2833 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2834 support has been systematized, and now supports the methods
2835 specified by ANSI, along with a default method and a method for
2836 slot documentation. (reported by Nathan Froyd)
2837 * bug fix: effective methods associated with a generic function are
2838 no longer cached over a change of that generic function's method
2839 combination. (reported by Andreas Fuchs)
2840 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2841 names a non-existent directory. (thanks to Andreas Fuchs)
2842 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2843 the PPC platform, lead to stack corruption; this has been fixed.
2844 (reported by Rainer Joswig)
2845 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2846 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2847 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2848 arithmetic in general, is now much more reliable on the PPC
2850 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2851 platform now returns the right answer.
2852 * optimization: restored some effective method precomputation in
2853 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2854 precomputation is now tunable.
2855 * optimization: compiler-internal data structure use has been
2856 reviewed, and changes have been made that should improve the
2857 performance of the compiler by about 20%.
2858 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2859 simple-base-strings and simple-bit-vectors is improved.
2860 * optimization: the optimization of 32-bit logical and arithmetic
2861 functions introduced in version 0.8.3 on the x86 has been
2862 implemented on the mips, ppc and sparc platforms; an
2863 implementation of the same facility, but for 64-bit arithmetic,
2864 has been added for the alpha.
2865 * microoptimization: the compiler is better able to make use of the
2866 x86 LEA instruction for multiplication by constants.
2867 * bug fix: in some situations compiler did not report usage of
2868 generic arithmetic in (SPEED 3) policy.
2869 * bug 145b fix: compiler used wrong type specifier while converting
2870 MEMBER-types to numeric.
2871 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2873 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2874 index argument now works properly on the Alpha platform.
2875 * bug fix: floating point exception treatment on the Alpha platform
2877 * bug fix: FILE-POSITION works much better on string input and
2878 output streams. (thanks to Nikodemus Siivola)
2879 * bug fix: many threading/garbage collection symptoms sorted.
2880 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2881 might be pseudo-atomic.
2882 * internal change: Stopping for GC is now done with signals not
2883 ptrace. GC is now done in whichever thread wanted it, instead of
2885 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2887 * bug fix: result form in DO is not contained in the implicit
2889 * incompatible change: ICR structure is changed; the value part of
2890 CONTINUATION is now called LVAR; corresponding functions are
2891 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2892 * added type deriver for ISQRT (thanks to Robert E. Brown).
2893 * bug fix: better support for loading from the command line when an
2894 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2896 * fixed some bugs revealed by Paul Dietz' test suite:
2897 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2898 ** ROUND and FROUND now give the right answer when given very
2899 small float arguments.
2900 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2902 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2903 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2904 ** LCM should return a non-negative integer.
2905 ** PARSE-INTEGER returned the index of a terminator instead of the
2906 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2907 ** PARSE-INTEGER returned an incorrect index being applied to a
2909 ** LCM with two arguments of 0 returns 0 rather than signalling
2911 ** unsigned addition of a 32-bit constant with the high bit set no
2912 longer causes an internal compiler error on the x86.
2913 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2914 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2915 with negative last argument.
2916 ** byte specifiers with zero size and position no longer cause
2917 an error during type derivation.
2918 ** bignum multiplication on the Alpha platform now returns the
2920 * porting: The system now builds on SuSE AMD64, although it still
2921 generates a 32-bit binary.
2922 * .fasl file incompatibility: The fasl file version number has
2923 been incremented (because of the changes to internal compiler
2924 data structures referred to above).
2926 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2927 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2928 more accurately, on the Darwin kernel running on PowerPC hardware.
2929 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2930 * Compiler code deletion notes now signal a condition of type
2931 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2932 an associated MUFFLE-WARNING restart.
2933 * The compiler now performs limited argument count validation of
2934 constant format strings in FORMAT, and where appropriate in ERROR,
2935 CERROR and WARN. (thanks to Gerd Moellmann)
2936 * New ASDF-INSTALL contrib can be used for automatic download and
2937 installation of third-party Lisp code from CCLAN or other sites
2939 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2940 which forces another thread to execute a function supplied by the
2942 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2943 now accept and act upon their :ELEMENT-TYPE keyword argument.
2944 (reported by Martin Atzmueller, Edi Weitz)
2945 * bug fix: FILE-POSITION now accepts position designators up to
2946 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2947 is the greater. (thanks to Patrik Nordebo)
2948 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2949 Kalle Olavi Niemitalo)
2950 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2951 printable. (reported by Eric Marsden)
2952 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2953 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2954 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2955 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2956 (thanks to Stig Sandoe)
2957 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2958 UNREAD-CHAR now correctly includes the unread character in the
2959 target sequence. (thanks to Gerd Moellmann)
2960 * bug fix in threaded builds: the system can now be suspended and
2961 resumed by shell job control with minimal disruption.
2962 * bug fixes in times and timezones >2038AD
2963 * better handling of "where is GNU make?" problem in build scripts
2964 (thanks to Nikodemus Siivola)
2965 * new optimization: inside a named function any reference to a
2966 function with the same name is considered to be a self-reference;
2967 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2968 optimization quality.
2969 * new optimization on x86: logical functions and + now have
2970 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2971 used when the result is truncated to 32 bits.
2972 * VALUES declaration is partially enabled.
2973 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2974 * fixed some bugs revealed by Paul Dietz' test suite:
2975 ** The system now obeys the constraint imposed by
2976 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2977 types form a lattice under type intersection.
2978 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2979 ** ASSOC now ignores NIL elements in an alist.
2980 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2981 and (1+ MOST-POSITIVE-FIXNUM) answers.
2982 ** The addition of a method with invalid qualifiers to a generic
2983 function does not cause an error to be signalled immediately;
2984 a warning is signalled, and the error is generated only on
2985 calling the generic function.
2986 * changed .fasl file version number, in part to add type codes for
2987 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2988 obscure ANSI requirements
2990 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2991 * fixed bug 148: failure to inline-expand a local function left
2992 garbage, confusing the compiler.
2993 * fixed bugs 3cd: structure slot readers perform type check if the
2994 slot can have an invalid value (i.e. it is either not initialized
2995 or can be written with a less specific slot writer).
2996 * bug fix: the compiler now traps array references to elements off
2997 the end of an array; previously, the bounds checking in some
2998 circumstances could go off-by-one.
2999 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3001 * type declarations for array element types now obey the description
3002 on the CLHS page "Declaration TYPE", as per discussions on
3003 sbcl-help around 2003-05-08. This means that a declaration
3004 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3005 declaration, all references to BAR will be asserted or assumed
3006 (with THE, so dependent on compiler policy) to involve objects of
3007 type FOO. Note that no such declaration is implied in
3008 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3009 * declared types of functions from the "Conditions"
3010 chapter. (reported by Paul Dietz)
3011 * bug fix: CERROR accepts a function as its first argument.
3012 * bug fix: NTH an NTHCDR accept a bignum as index
3013 arguments. (reported by Adam Warner)
3014 * optimization: character compare routines now optimize comparing
3015 against a constant character. (reported by Gilbert Baumann)
3016 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3017 argument now works properly on the MIPS platform.
3018 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3019 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3020 declared to have a functional type, does not cause an error
3022 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3023 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3024 * fixed bug: if last continuation of a deleted block has a
3025 destination, this destination should be deleted too. (reported by
3027 * fixed a bug in the bootstrap process: the host compiler's values
3028 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3029 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3030 test case from Patrik Nordebo)
3031 * improved the ability of the disassembler on the PPC platform to
3032 provide helpful disassembly notes.
3033 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3034 the class in more cases than previously.
3035 * bug fix: FILE-POSITION now understands :START and :END for
3036 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3037 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3038 8) data. (thanks to David Lichteblau)
3039 * bug fix: it is possible to add a method to a generic function
3040 without lambda list.
3041 * bug fix: reader failed to signal END-OF-FILE inside an
3042 object representation. (reported by Nikodemus Siivola)
3043 * fixed some bugs revealed by Paul Dietz' test suite:
3044 ** LAST and [N]BUTLAST should accept a bignum.
3045 ** condition slot accessors are methods.
3046 ** (VECTOR NIL) is a subtype of STRING.
3048 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3049 * minor incompatible change: some nonsensical specialized lambda
3050 lists (used in DEFMETHOD) which were previously ignored now signal
3052 * minor incompatible change: the system is now aware of the types of
3053 variables in the COMMON-LISP package, and will signal errors for
3054 most violations of these type constraints (where previously they
3055 were silently accepted).
3056 * minor incompatible change: COMPILE-FILE now uses the freedom
3057 afforded (ANSI 3.2.2.3) to use derived function types for
3058 functions defined in the same file. This also permits the system
3059 to warn on static type mismatches and function
3060 redefinition. (Currently it does not work with high DEBUG level.)
3061 * minor incompatible change: VALUES declaration is disabled.
3062 * When issuing notes, the compiler now signals a condition of type
3063 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3064 restart for use in user handlers. It is expected that the
3065 COMPILER-NOTE condition will eventually become a condition
3066 supertype to a hierarchy of note types, which will then be
3067 handleable in a similar fashion. However, at the moment, no such
3068 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3069 but it's an implementation detail, not a classification for the
3071 * Changes in type checking closed the following bugs:
3072 ** type checking of unused values (192b, 194d, 203);
3073 ** template selection based on unsafe type assertions (192c, 236);
3074 ** type checking in branches (194bc).
3075 * A short form of VALUES type specifier has ANSI meaning (it has
3076 increased the number of situations when SBCL cannot perform type
3078 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3079 keywords or constants is permissible.
3080 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3081 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3082 classes. (thanks to Antonio Martinez)
3083 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3084 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3085 argument to a pretty-print stream. Instead, it leaves any such
3086 operations to DESCRIBE-OBJECT methods.
3087 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3089 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3090 lambda lists. (thanks to David Lichteblau)
3091 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3092 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3093 (thanks to Gerd Moellmann)
3094 * fixed bug 246: increased compilation speed of long
3095 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3097 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3099 * DEFCONSTANT now throws a condition of type
3100 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3101 constant to a non-EQL value; CONTINUE and ABORT restarts
3102 respectively change and preserve the value.
3103 * fixed bug 63: The code walker, part of the implementation of CLOS,
3104 is now better at handling symbol macros.
3105 * bug fix: There is no longer an internal implementation type named
3106 CL:LENGTH. (reported by Raymond Toy)
3107 * bug fix: In macro-like defining macros/special operators the
3108 implicit block does not enclose the lambda list.
3109 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3110 enclosed in parentheses) are not suitable as type specifiers, and
3111 their use properly signals an error now.
3112 * bug fix: An argument count mismatch for a type specifier in code
3113 being compiled no longer causes an unhandled error at compile
3114 time, but signals a compile-time warning.
3115 * fixed simple vector readable printing
3116 * bug fix: DESCRIBE takes more care over whether the class
3117 precedence list slot of a class is bound before accessing it.
3118 (reported by Markus Krummenacker)
3119 * bug fix: FORMATTER can successfully compile pretty-printer format
3120 strings which use variants of the ~* directive inside.
3121 * bug fix: SEARCH now applies its TEST predicate to the elements of
3122 the arguments in the correct order. (thanks to Wolfhard Buss)
3123 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3124 in type inference. (thanks to Robert E. Brown)
3125 * bug fix: Reading in symbols with an explicit package name of ""
3126 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3127 * fixed some bugs revealed by Paul Dietz' test suite:
3128 ** NIL is now allowed as a structure slot name.
3129 ** Arbitrary numbers, not just REALs, are allowed in certain
3130 circumstances in LOOP for-as-arithmetic clauses.
3131 ** Multiple class redefinitions before slot access no longer
3132 causes a type error.
3133 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3134 association between the name and a class.
3135 ** Generic functions with non-standard method-combination and over
3136 five methods all of which return constants no longer return NIL
3137 after the first few invocations. (thanks to Gerd Moellmann)
3138 ** CALL-NEXT-METHOD with no arguments now passes the original
3139 values of the arguments, even in the presence of assignment.
3140 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3142 ** DISASSEMBLE works with closures and funcallable instances.
3143 ** ADD-METHOD now returns the generic function, not the new method.
3144 ** FIND-METHOD signals an error if the lengths of the specializers
3145 is incompatible with the generic function, even if the ERRORP
3147 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3148 which its argument is a member.
3149 ** DEFCLASS only redefines the class named by its class-name
3150 argument if that name is the proper name of the class;
3151 otherwise, it creates a new class.
3152 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3153 of the UNBOUND-SLOT condition to the name of the slot.
3154 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3155 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3156 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3157 SLOT-MAKUNBOUND in the specified fashion.
3159 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3160 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3161 cross-compilation host. As a consequence, we can now bootstrap our
3162 way up to SBCL starting with a bare gcc toolchain and human-readable
3163 source code (first the source to CLISP, then the source to SBCL).
3164 * A contributed module containing a partial implementation of the
3165 simple-streams interface has been included. (thanks to Rudi
3167 * A contributed module implementing the RFC1321 Message Digest
3168 Algorithm, known as MD5, has been included.
3169 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3170 no longer has any effect, as the code controlled by this feature
3171 has been deleted. (As far as we know, no-one has ever built using
3172 this feature, and its semantics were confused in any case).
3173 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3174 work on conditions (as required by the ANSI specification),
3175 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3176 expected behaviour on conditions. Users should note, however,
3177 that such behaviour is not required by the ANSI specification,
3178 and so use of this behaviour may render their code unportable.
3179 * fixed some bugs revealed by Paul Dietz' test suite:
3180 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3182 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3183 ** SXHASH on condition objects no longer returns NIL.
3184 ** :ALLOCATION :CLASS slots are better treated; their values are
3185 updated on class redefinition, and initforms inherited from
3186 superclasses are applied.
3187 ** REMOVE-METHOD returns its generic function argument even when
3188 no method was removed.
3189 ** SHARED-INITIALIZE now initializes the values of the requested
3190 slots, including those with :ALLOCATION :CLASS.
3191 ** ALLOCATE-INSTANCE now works on structure classes defined via
3192 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3194 ** SLOT-EXISTS-P now works on conditions, as well as structures
3196 ** MAKE-LOAD-FORM now has the required methods on
3197 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3198 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3199 keyword, and now implements the SLOT-NAMES argument.
3200 ** methods with &OPTIONAL arguments no longer allow too many
3201 arguments to be passed in the call without error.
3202 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3203 option is consistent with the required arguments of the generic
3204 function lambda list.
3205 * bug fix: REQUIRE accepts a string designator. (Thanks to
3207 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3208 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3209 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3210 * bug fix: The compiler now checks for duplicated variables in macro
3212 * bug fix: SETQ on globals returns the correct value.
3213 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3214 not a COMPILER-ERROR (followed by some other strange error on
3215 choosing the CONTINUE restart).
3216 * bug fix: make.sh and friends are now more consistent in the way that
3217 they look for GNU "make".
3219 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3220 * experimental native threads support (on x86 Linux >=2.4 only).
3221 This is not compiled in by default: you need to add :SB-THREAD to
3222 the target features. See the "Beyond ANSI" chapter of the manual
3224 * fix for longstanding nonANSIism: The old distinction between
3225 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3226 The return value from CL:FIND-CLASS is now a CLOS class, and
3227 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3228 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3229 * An interface to the MetaObject Protocol, as described in Kiczales,
3230 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3231 MIT Press, 1991, is available from the SB-MOP package.
3232 * incompatible change: the SB-PCL package should now be considered
3233 a private implementation detail, and no longer a semi-private MOP
3235 * minor incompatible change: due to rearrangement for threads, the
3236 control stack and binding stack are now allocated at arbitrary
3237 addresses instead of being hardcoded per-port. Users affected by
3238 this probably have to be doing advanced things with shared
3239 libraries, and will know who they are.
3240 * minor incompatible change: Previously, all --eval forms used were
3241 processed with READ before any of them were processed with EVAL.
3242 Now each --eval form is processed with both READ and EVAL before
3243 the next --eval form is processed. (Thus package operations like
3244 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3245 work as the user might reasonably expect.)
3246 * minor incompatible change: *STANDARD-INPUT* is now only an
3247 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3249 * minor incompatible change: Y-OR-N-P is now character-oriented,
3250 not line oriented. Also, YES-OR-NO-P now works without errors.
3251 (thanks to Antonio Martinez)
3252 * sb-aclrepl module improvements: an integrated inspector, added
3253 repl features, and a bug fix to :trace command.
3254 * Known functions, which cannot be open coded by the backend, are
3255 considered to be able to check types of their arguments. (fixing
3256 a bug report by Nathan J. Froyd)
3257 * fixed a bug in computing method discriminating functions: It is
3258 now possible to define methods specialized on classes which have
3259 forward-referenced superclasses. (thanks to Gerd Moellmann)
3260 * fixed evaluation order in optional entries (reported by Gilbert
3262 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3263 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3264 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3265 required-by-AMOP NAME argument, as well as CLASS and
3266 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3267 * fixed bug 20: DEFMETHOD can define methods using names that are
3268 not the proper names of classes to designate class specializers.
3269 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3270 Unix file descriptors, instead of blowing up. (thanks to Antonio
3272 * Garbage collection refactoring: user-visible change is that a
3273 call to the GC function during WITHOUT-GCING will not do garbage
3274 collection until the end of the WITHOUT-GCING. If you were doing
3275 this you were probably losing anyway.
3276 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3277 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3278 TYPEP the latter but not the former.
3279 * The compiler issues a full WARNING for calls to undefined functions
3280 with names from the CL package.
3281 * MAP-INTO for a vector destination is open coded. (reported by
3282 Brian Downing on c.l.l)
3283 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3284 documentation string.
3285 * fixed some bugs revealed by Paul Dietz' test suite:
3286 ** COPY-ALIST now signals an error if its argument is a dotted
3288 ** Condition slots are now accessed more correctly in the presence
3289 of multiple initargs for a given slot.
3290 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3291 exclude restarts of the same name associated with a different
3293 ** DEFCLASS of forward-referenced classes with another
3294 forward-referenced class in the superclasses list no longer
3296 ** Condition slots are now initialized once each, not multiple
3297 times. (thanks to Gerd Moellmann)
3298 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3300 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3301 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3302 arguments contain duplicated elements.
3303 ** RESTART-CASE understands local macros.
3304 ** RESTART-CASE associates exactly its own restarts with a condition.
3305 ** ENDP in safe mode checks its argument to be of type LIST.
3306 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3307 in question is unbound.
3308 ** Optimized MAKE-INSTANCE functions no longer cause internal
3309 assertion failures in the presence of duplicate initargs.
3310 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3311 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3312 propagated even in the ctor optimized implementation of
3314 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3316 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3317 * a better implementation of SXHASH on (simple) bit vectors,
3318 measured both in execution speed and in distribution of results
3319 over the positive fixnums, has been installed. Likewise, a better
3320 implementation of EQUAL for simple bit vectors is now available.
3321 * fixed CEILING optimization for a divisor of form 2^k.
3322 * fixed bug 240 (emitting extra style warnings "using the lexical
3323 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3324 by Antonio Martinez)
3325 * fixed SXHASH, giving different results for NIL depending on type
3326 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3327 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3328 variable. (found by Rolf Wester)
3329 * SBCL does not ignore type declarations for special
3330 variables. (reported by rif on c.l.l 2003-03-05)
3331 * some bug fixes in contrib/sb-aclrepl/
3332 * fixed some bugs revealed by Paul Dietz' test suite:
3333 ** a bug in the CONS type specifier, whereby the CAR and CDR
3334 types got intertwined, has been fixed;
3335 ** the type system is now able to reason about the interaction
3336 between INTEGER and RATIO types more completely;
3337 ** APPEND, [N]REVERSE and NRECONC check that those their
3338 arguments, which must be proper lists, are really so;
3339 ** An array specialized to be unable to hold elements has been
3340 implemented, as required -- yes, really -- by ANSI;
3341 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3342 on malformed property lists;
3344 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3345 * incompatible packaging change: in line with Unix convention,
3346 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3347 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3348 * REQUIRE and PROVIDE are now optionally capable of doing something
3349 useful. See the documentation string for REQUIRE.
3350 * infrastructure for a managed SBCL contrib system: contributed
3351 modules in this release include:
3352 ** the ASDF system definition facility;
3353 ** an interface to the BSD Sockets API;
3354 ** an ACL-like convenience interface to the repl;
3355 (thanks to Kevin Rosenberg)
3356 ** an implementation of ROTATE-BYTE, with efficient implementation
3358 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3359 gives a closer approximation to the right answer than previously.
3360 (thanks to Raymond Toy)
3361 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3362 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3363 optional environment arguments, as required by ANSI.
3364 * fixed bugs in other functions taking environment objects, allowing
3365 calls with an explicit NIL environment argument to be compiled
3367 * fixed bug 228: primary return values from
3368 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3369 COMPILE or FUNCTION.
3370 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3371 :TYPE will no longer signal errors on innocuous objects.
3372 * fixed bug 231b: SETQ is better at respecting type declarations in
3373 the lexical environment.
3374 * fixed a bug in DEFCLASS: classes named by symbols with no or
3375 unprintable packages can now be defined.
3376 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3377 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3378 * fixed compiler failure related to checking types of functions.
3379 (reported by Robert E. Brown)
3380 * the compiler is now much more consistent in its error-checking
3381 treatment of bounding index arguments to sequence functions: in
3382 (SAFETY 3) code, errors will be signalled in almost all cases if
3383 invalid sequence bounding indices are passed to functions defined
3384 by ANSI to operate on sequences.
3385 * fixed a bug in the build procedure: documentation of SBCL-specific
3386 packages is now preserved and available in the final Lisp image.
3387 * lifted FDEFINITION lookup out of loops in the implementation of
3388 many list operations. (thanks to Robert E. Brown)
3389 * fixed a bug in the reader: the #n# reader macro now works for
3390 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3391 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3392 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3393 E. Brown and Tony Martinez respectively)
3394 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3395 SLOT-BOUNDP: in particular, it is now able to optimize them much
3396 better, and is now not vulnerable to having packages renamed.
3397 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3398 always called when a slot is not present in an instance. (thanks
3400 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3401 longer signal an error on unreading a character following EOF on
3402 the previous constituent stream. (thanks to Tony Martinez)
3403 * fixed some bugs revealed by Paul Dietz' test suite:
3404 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3405 not just nonnegative fixnums;
3406 ** the logical bit-array operators such as BIT-AND now accept an
3407 explicit NIL for their "opt-arg" argument (to indicate a
3408 freshly-consed result bit-array);
3409 ** ELT now signals an error on an invalid sequence index in safe
3411 ** the type system is now cleverer about negations of numeric
3412 types, and consequently understands the BIGNUM and RATIO types
3414 ** the type system is now cleverer about the interaction between
3415 INTEGER and RATIO types: while bugs still remain, many more
3416 cases are accurately computed;
3417 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3418 if it is in the last clause;
3419 ** CONSTANTLY now correctly returns a side-effect-free function in
3421 ** DECLARE is no longer treated as a special-operator; in
3422 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3423 * incremented fasl file version number due to the change in the
3424 DEFSTRUCT-SLOT-DESCRIPTION structure.
3426 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3427 * minor incompatible change: code processed by the "interpreter" or
3428 EVAL now has a compilation optimization policy of (DEBUG 2)
3429 (changed from (DEBUG 1)) to improve debuggability of interactive
3430 development, and to allow the use of the debug RETURN command in
3432 * an experimental implementation of the RETURN command for the
3433 debugger has been included. (thanks to Frederik Kuivinen)
3434 * fixed bug 62: constraints were not propagated into a loop.
3435 * fixed bug in embedded calls of SORT (reported and investigated by
3437 * fixed some bugs revealed by Paul Dietz' test suite:
3438 ** printing and reading of arrays with some dimensions having
3439 length 0 (thanks to Gerd Moellmann);
3440 ** BOA constructor with &AUX argument without a default value does
3441 not cause a type error;
3442 ** CONSTANTP now returns true for all self-evaluating objects.
3444 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3445 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3446 accessors that are related by inheritance, as specified in the
3447 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3448 Valtteri Vuorikoski)
3449 * The compiler is now able to inline functions that were defined in
3450 a complex lexical environment (e.g. inside a MACROLET).
3451 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3453 * Support for the upcoming FreeBSD-5.0 release has been included.
3454 (thanks to Dag-Erling Smorgrav)
3455 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3456 effect when it is not in a toplevel context.
3457 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3458 stand a better chance of being correct. (thanks to Gerd
3460 * fixed bug in COERCE, which now signals an error on coercing a
3461 rational to a bounded real type which excludes the expected
3463 * The compiler is now able to derive types more accurately from the
3464 COERCE and COMPILE functions.
3465 * fixed bug 223: functional binding is considered to be constant
3466 only for symbols in the CL package.
3467 * fixed bug 231: SETQ did not check the type of a variable being set
3468 (reported by Robert E. Brown)
3469 * A new optimization for MAKE-INSTANCE has been included, fixing
3470 various bugs (including relating to :ALLOCATION :CLASS slots and
3471 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3473 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3474 ** As required by ANSI, LOOP now disallows anonymous collection
3475 clauses such as COLLECT I in conjunction with aggregate boolean
3476 clauses such as THEREIS (= I 1);
3477 ** LOOP now signals an error when any variable is reused in the
3478 same loop (including the potentially useful construct analogous
3479 to WITH A = 1 WITH A = (1+ A);
3480 ** IT is only a special loop symbol within the first clause of a
3481 conditional loop clause;
3482 ** LOOP with a typed iteration variable over a hashtable now
3483 signals a type error iff it should.
3484 * fixed some other bugs revealed by Paul Dietz' test suite:
3485 ** FILE-STREAM now names the class previously known as FD-STREAM;
3486 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3487 argument) no longer signals an error;
3488 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3489 of the slot symbol, rather than using the current package
3490 ((:CONC-NAME "") continues to intern the slot's name in the
3492 * incremented fasl file version number, because of the incompatible
3493 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3494 of the new implementation of DEFINE-COMPILER-MACRO.
3496 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3497 * Support for building SBCL for MIPS platforms running in
3498 little-endian mode has now been checked in, and basic
3499 functionality on said platforms verified.
3500 * minor incompatible change: PCL now records the pathname of a file
3501 in which methods and the like are defined, rather than its
3503 * minor incompatible change: TRUENAME now considers the truename of
3504 a file naming a directory to be the pathname with :DIRECTORY
3505 component indicating that directory.
3506 * minor incompatible change: a NAMED clause in the extended form of
3507 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3508 reason for the previous behaviour is unclear.
3509 * more systematization and improvement of CLOS and MOP conformance
3510 in PCL (thanks to Gerd Moellman and Pierre Mai):
3511 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3513 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3514 primary methods with no specializers;
3515 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3517 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3518 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3519 FORWARD-REFERENCED-CLASSes; error reporting on
3520 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3522 ** SXHASH on CLOS instances now uses a slot internal to the
3523 instance to return different numbers on distinct instances,
3524 while preserving the same return value through invocations of
3526 ** DEFMETHOD signals errors when methods with longer incongruent
3527 lambda lists are added to generic functions;
3528 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3529 CLASS, as specified in AMOP;
3530 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3531 on the order returned by the primary method for classes of
3532 class STANDARD-CLASS;
3533 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3534 * fixed some bugs shown by Paul Dietz' test suite:
3535 ** DOLIST puts its body in TAGBODY;
3536 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3538 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3539 value producing form;
3540 ** if more variables are given to PROGV than values, extra
3541 variables are bound and made to have no value;
3542 ** NSUBSTITUTE on list arguments gets the right answer with
3544 ** ELT signals an error of type TYPE-ERROR when the index argument
3545 is not a valid sequence index;
3546 ** LOOP signals (at macroexpansion time) an error of type
3547 PROGRAM-ERROR when duplicate variable names are found;
3548 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3549 ** FUNCALL of special-operators now cause an error of type
3551 ** PSETQ now works as required in the presence of side-effecting
3552 symbol-macro places;
3553 ** NCONC accepts any object as its last argument;
3554 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3556 ** loop-for-as-package does not require a package to be explicitely
3558 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3560 * fixed bug 166: compiler preserves "there is a way to go"
3561 invariant when deleting code.
3562 * fixed bug 172: macro lambda lists with required arguments after
3563 &REST arguments now cause an error to be signalled. (thanks to
3565 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3566 bug. (thanks to Matthew Danish)
3567 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3569 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3570 arguments are lexically rebound. (thanks to Gerd Moellmann and
3572 * fixed bug 194: error messages are now more informative when there
3573 is no primary method applicable in a call to a generic
3574 function. (thanks to Gerd Moellmann)
3575 * fixed bug in command line argument checking (thanks to Julian
3577 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3578 (thanks to Matthew Danish)
3579 * incremented fasl file version number, because of the
3580 SXHASH-related changes in the layout of CLOS data structures
3582 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3583 * minor incompatible change: The runtime (the Unix executable named
3584 "sbcl") is now much pickier about the .core files it will load.
3585 Essentially it now requires .core files to descend from the same
3586 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3587 as the runtime does. (The intent is to prevent the crashes which
3588 can occur, and which can even be reported as mysterious failures,
3589 when people patch the sources or change the build parameters
3590 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3591 sbcl and .core files.)
3592 * fixed bug: VALUES-LIST is no longer optimized away.
3593 * fixed bug 142: The FFI conversion of C string values to Lisp
3594 string values no longer conses excessively. (thanks to Nathan
3595 Froyd porting Raymond Toy's fix to CMU CL)
3596 * began to systematize and improve MOP conformance in PCL (thanks to
3597 Nathan Froyd, Gerd Moellman and Pierre Mai):
3598 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3600 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3601 ** FINALIZE-INHERITANCE is now called on class finalization;
3602 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3603 argument precedence order.
3604 * fixed bug 202: The compiler no longer fails on functions whose
3605 derived types contradict their declared type.
3606 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3607 so it can be non-toplevel.
3608 * The fasl file version number has changed (because of the new
3609 implementation of DEFMACRO).
3610 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3611 safe code, that any length requirement by their type-specifier
3612 argument is valid. The exceptions to this are described in bug
3614 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3615 functions now check the types of their inputs as required by ANSI.
3616 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3617 introduced symbol is DECLAREd to be SPECIAL.
3618 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3619 * fixed bug: :COUNT argument to sequence functions may be negative
3620 * fixed bug: body of DO-SYMBOLS may contain declarations
3621 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3622 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3623 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3624 symbol macro only once
3625 * fixed printing of call frame when argument list is unavailable
3626 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3627 * compiler no longer signals WARNING on unknown keyword
3630 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3631 * A beta-quality port to the mips architecture running Linux,
3632 based on the old CMUCL backend, has been made. It has been tested
3633 on a big-endian kernel, and works sufficiently well to be able to
3634 rebuild itself; it has not been tested in little-endian mode.
3635 * fixed an inconsistency between gencgc.c and purify.c which made
3636 dumping/loading .core files unreliable
3637 * fixed bug 120a: The compiler now deals correctly with IFs where
3638 the consequent is the same as the alternative, instead of
3639 misderiving the return type. (thanks to Alexey Dejneka)
3640 * fixed bug 113: Logical pathnames are now dumpable (the logical
3641 host is resolved at load-time, throwing an error if it is not
3643 * fixed bug 174: FORMAT's error message is slightly clearer when a
3644 non-printing character is used in a format directive.
3645 * fixed several bugs in compiler checking of type declarations, i.e.
3646 violations of the Python "declarations are assertions" principle
3647 (thanks to Alexey Dejneka)
3648 * fixed several bugs in PCL's error checking (thanks to Gerd
3650 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3652 * fixed bug in compilation of functions as first class values
3653 (thanks to Antonio Martinez-Shotton)
3654 * The compiler's handling TYPE-ERRORs which it can prove will
3655 inevitably happen at runtime has been cleaned up and corrected
3656 in several ways. (thanks to Alexey Dejneka)
3657 * improved argument type checking for various basic arithmetic
3658 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3659 lavished on them in the past that they can be compiled in many
3660 ways in different special cases
3661 * fixed bug 181: compiler checks validity of user supplied type
3663 * cleaned up code flushing in optimization: Function calls which
3664 should signal errors for safety purposes (e.g. which ANSI says
3665 should signal errors when their arguments are of incorrect type)
3666 are no longer optimized away.
3667 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3668 * incremented fasl file version number, because changes in the
3669 implementation of sequence functions like COERCE caused
3670 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3671 in old inline expansions) to become undefined. (Actually these
3672 changes were later undone, so we might very well be binary
3673 compatible with 0.7.7 after all, but leaving the version number
3674 incremented seemed like the simplest and most conservative
3677 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3678 * An alpha-quality port to the parisc architecture running Linux,
3679 based on the old CMUCL backend, has been made. This, even more so
3680 than the other backends, should be considered still a work in
3681 progress; known problems include that the Linux kernel in 64-bit
3682 mode does not propagate the correct sigcontext structure to
3683 userspace, and consequently SBCL on a parisc64 kernel will not
3685 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3686 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3687 suppressed.) Also now that the compiler is looking at declarations
3688 in the environment, it checks optimization declarations as well,
3689 and suppresses inlining when (> DEBUG SPEED).
3690 * More fixes have been made to treatment of floating point exception
3691 treatment and other Unix signals. In particular, floating point
3692 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3693 * The detection and handling of control stack exhaustion (infinite
3694 or very deeply nested recursion) has changed. Stack exhaustion
3695 detection is now done by write-protecting pages at the OS level
3696 and applies at all optimization settings; when found, a
3697 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3698 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3699 be caught using IGNORE-ERRORS.
3700 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3701 correctly when transforming recursive LABELS functions to LETs.
3702 (thanks to Alexey Dejneka)
3703 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3704 that are names of constants or global variables.
3705 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3706 alien routines with docstrings.
3707 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3708 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3710 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3711 (thanks to Eric Marsden)
3712 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3713 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3714 * Bug fix: The SPARC backend can now compile functions involving
3715 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3716 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3717 to a functions expecting a PATHNAME-DESIGNATOR.
3718 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3719 lambda lists. (thanks to Alexey Dejneka)
3720 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3721 functions so that EXPORTing the name of the class doesn't cause
3722 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3723 to Antonio Martinez for reporting this)
3724 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3725 constant-folded without failing an assertion. (thanks to Einar
3726 Floystad Dorum for reporting this)
3727 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3728 types are now dealt with more correctly by the compiler.
3729 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3730 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3731 behaviour analogous to RENAME-FILE. This puts its behaviour more
3732 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3734 * The fasl file version number has changed again. (because of the
3735 bug fix involving the names of PCL MAKE-INSTANCE functions)
3737 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3738 * bug fix: Floating point exceptions are treated much more
3739 consistently on the x86/Linux and PPC/Linux platforms.
3740 * Array initialization with :INITIAL-ELEMENT is now much faster for
3741 cases when the compiler cannot open code the array creation, but
3742 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3743 array accesses have also seen a speed increase.
3744 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3745 specified. (This was at the root of some bad interactions between
3746 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3748 * bug fix: Internal error arguments for undefined functions are now
3749 computed correctly on the PPC/Linux platform.
3750 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3751 Raymond Toy's patch for CMU CL)
3752 * Support for the Solaris 9 operating environment has been included
3753 (thanks to Daniel Merritt)
3754 * A very ugly but hopefully complete draft of the missing FFI chapter
3755 of the manual has been created by reformatting the corresponding
3756 CMU CL manual chapter into (currently very ugly and incoherent)
3757 DocBook and bringing it up to date for SBCL behavior. Thus, the
3758 manual is now essentially complete, at least by my extreme
3759 once-and-only-once standards, whereby it's acceptable to refer to
3760 the doc strings of SB-EXT functions as the primary documentation.
3761 * The fasl file version number has changed again, due to cleanup of
3762 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3763 is no longer a static symbol.)
3765 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3766 * SBCL now builds with OpenMCL (version 0.12) as the
3767 cross-compilation host; also, more progress has been made toward
3768 bootstrapping under CLISP.
3769 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3771 * bug 158 fixed: The compiler can now deal with integer loop
3772 increments different from 1; fixing this turned out also to fix
3774 * bug 169 fixed: no more bogus warnings about using lexical bindings
3775 despite the presence of perfectly good SPECIAL declarations (thanks
3776 to David Lichteblau)
3777 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3778 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3779 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3781 * bug 180 fixed: Method combination specifications no longer ignore
3782 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3783 * bug fix: Structure type predicate functions now check their argument
3784 count as they should.
3785 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3786 correctly. (thanks to Pierre Mai)
3787 * minor incompatible change: The --noprogrammer option is deprecated
3788 in favor of the new --disable-debugger option, which is very similar.
3789 (The major difference is that it takes effect at a slightly different
3790 time at startup, causing handling of errors in --sysinit and
3791 --userinit files will be affected differently.) The
3792 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3793 been added to allow this functionality to be controlled from ordinary
3794 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3795 maintainers, who might want to run non-interactive scripts to
3796 build SBCL cores which will later be used interactively.)
3797 * minor incompatible change: The LOAD function no longer, when given
3798 a wild pathname to load, loads all files matching that pathname.
3799 Instead, an error of type FILE-ERROR is signalled.
3801 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3802 * bug 147 fixed: The compiler preserves its block link/count
3803 invariants more correctly now so that it doesn't crash. (thanks
3805 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3807 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3808 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3809 no-value-for-this-character as NIL (as specified by ANSI).
3810 * HOST-NAMESTRING on physical pathnames now returns a string that is
3811 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3812 (thanks to Christophe Rhodes)
3813 * The Alpha port handles icache flushing more correctly. (thanks to
3815 * More progress has been made toward bootstrapping under CLISP. (thanks
3816 to Christophe Rhodes)
3817 * The fasl file format has changed again, because dynamic loading
3818 on OpenBSD (which has non-ELF object files) motivated some cleanups
3819 in the way that foreign symbols are transformed and passed around.
3820 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3821 is no longer treated as whitespace by the reader, but instead as
3822 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3823 |A
\7fB|, instead of A as it used to.
3825 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3826 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3827 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3828 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3829 this as of 0.7.1.45, but was left out of the previous news section
3830 (thanks to Dan Barlow)
3831 * SBCL now runs on the Solaris operating system on SPARC architectures
3832 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3833 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3834 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3835 * SPARC backend cleanups, allowing builds of cores optimized for V8
3836 and V9 SPARCS, and also emission of code targeted to a particular
3837 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3839 * SBCL is closer to bootstrapping under CLISP, thanks to various
3840 fixes by Christophe Rhodes.
3841 * The fasl file format has changed again, to allow the compiler's
3842 INFO database to support symbol macros.
3843 * The user manual (in doc/) is formatted into HTML more nicely.
3844 (thanks to coreythomas)
3845 * The system is smarter about SUBTYPEP relationships, especially
3846 those involving NOT types (including types such as ATOM which are
3847 represented internally using NOT types). Thus SUBTYPEP is less
3848 likely to return (VALUES NIL NIL) in general, and in particular
3849 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3851 * The fasl file format has changed again, because the internal
3852 representation of types now includes a new slot to support the new
3853 SUBTYPEP-of-NOT-types logic.
3854 * (not a change in the main branch of SBCL, but a related prototype
3855 which can hopefully be merged into the main branch of SBCL in the
3856 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3857 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3858 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3859 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3860 dispatch macro character combinations. (thanks to Alexey Dejneka)
3861 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3862 that superficially look like logical namestrings correctly.
3863 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3865 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3866 default supplied-p) for &optional and &key arguments. (thanks to
3869 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3870 * incompatible change: The compiler is now less aggressive about
3871 tail call optimization, doing it only when (> SPACE DEBUG) or
3872 (> SPEED DEBUG). (This is an incompatible change because there are
3873 programs which relied on the old CMU-CL-style behavior to optimize
3874 away their unbounded recursion which will now die of stack overflow.)
3875 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3876 for non-GENCGC systems has been increased to 20M (since that
3877 seems much closer to the likely performance optimum for modern
3878 systems than the old 4M value was)
3879 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3880 and *DEBUG-PRINT-LEVEL*
3881 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3882 of CMU CL's support for SPARC, and various endianness and other
3883 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3884 * new syntactic sugar for the Unix command line: --load foo.bar is now
3885 an alternate notation for --eval '(load "foo.bar")'.
3887 ** The system now detects stack overflow and handles it gracefully,
3888 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3889 optimization settings. (This is a good thing in general, and
3890 its introduction in this version should be particularly timely
3891 for anyone whose code fails because of suppression of tail
3893 ** The system now hunts for the C variable "environ" in a more
3894 devious way, to avoid segfaults when the C library version differs
3895 between compile time and run time. (thanks to Christophe Rhodes)
3896 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3897 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3898 ** The compiler no longer issues bogus style warnings for undefined
3899 classes in the same source file as the DEFCLASSes which defined
3900 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3902 ** fixes in CONDITION class precedence list for undefined function
3903 errors (thanks to Alexei Dejneka)
3904 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3905 correctly. (thanks to Dan Barlow)
3906 ** portability fixes aiming at bootstrapping under CLISP (thanks
3907 to Dave McDonald and Christophe Rhodes)
3908 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3909 ** fixes in type translation and and type inference (thanks to
3911 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3912 ** various fixes in the new ports (thanks to Dan Barlow)
3913 * several changes related to debugging:
3914 ** suppression of tail recursion, as noted above
3915 ** stack overflow detection, as noted above
3916 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3917 is now the default. (For some time encapsulation has been more
3918 reliable than the breakpoint-based :ENCAPSULATE NIL
3919 implementation, at least on X86 systems; and I just noticed that
3920 encapsulation also seems closer to the spirit of the ANSI
3923 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3925 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3926 up properly again. (There was a packaging bug in 0.7.0 which
3927 left their definitions in SB-SYS::LOAD-FOREIGN and
3928 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3929 vital for most things which interface to C-level interfaces,
3930 like extensions working with sockets or databases or
3931 Perl-compatible regexes or whatever, and the need to fix
3932 this bug is the main reason that 0.7.1 was released so
3934 ** DEFGENERIC is now choosier about the methods it redefines, so that
3935 reLOADing a previously-LOADed file containing DEFGENERICs does
3936 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3937 cycle now works as it should. (thanks to Alexey Dejneka)
3938 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3939 fixed by Christophe Rhodes. (He actually submitted this patch
3940 months ago, and I delayed until after 0.7.0.)
3941 ** Bug 111 (internal compiler confusion about runtime checks on
3942 FUNCTION types) was fixed by Alexey Dejneka.
3943 * Some internal cleanups (getting rid of variables which aren't
3944 needed now that the byte interpreter is gone) caused the fasl
3945 file format number to change again.
3947 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3948 * major incompatible change: The default fasl file extension, i.e. the
3949 default extension for files produced by COMPILE-FILE, has changed
3950 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3952 ** There are many changes in the implementation of the compiler.
3953 SBCL is now essentially a compiler-only implementation of ANSI
3954 Common Lisp. EVAL still "interprets" a few special cases, but
3955 almost all the interesting cases are handled by creating
3956 a LAMBDA expression, calling COMPILE on it, then calling
3957 FUNCALL on the result.
3958 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3959 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3960 Since the code is newer, there might still be some new bugs
3961 (though not as many as before Martin Atzmueller's fixes:-). But
3962 the new code is substantially simpler and clearer, and hopefully
3963 any remaining bugs will be simpler, less fundamental, and more
3964 fixable then the bugs in the old code.
3965 ** The revised compiler is still a little unsteady on its feet.
3967 *** The debugging information it produces (particularly the names
3968 of FUNCTION objects) is sometimes much less useful than what
3969 the old compiler produced.
3970 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3971 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3972 within a MACROLET) has been temporarily weakened.
3973 ** There are new compiler optimizations for various functions:
3974 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3975 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3976 *** the math functions TRUNCATE, FLOOR, and CEILING
3977 *** the function-of-all-trades COERCE
3978 Mostly these should be transparent, but there's one
3979 potentially-annoying problem (bug 117): when the compiler
3980 inline-expands a function and does type analysis on the result,
3981 it can create control paths which have type mismatches, and
3982 when it can't prove that those control paths aren't taken,
3983 it will issue WARNINGs about the type mismatches. This is
3984 a particular problem in practice for the new sequence functions.
3985 It's not clear how this should be fixed, and for now, a
3986 workaround is given in the entry for 117 in the BUGS file.
3987 ** (Because of the interaction between the two previous items --
3988 occasional inlining problems and new inline expansions -- some
3989 of the new sequence function optimizations won't really kick in
3990 completely until debugging information, and then inlining, are
3991 straightened out in some future version.)
3992 * minor incompatible changes:
3993 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3994 DIRECTORY no longer implicitly promotes NIL slots of its
3995 pathname argument to :WILD. In particular, when you ask for the
3996 contents of a directory (which you used to be able to do without
3997 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3998 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3999 ** changes in behavior that ANSI explicitly defines to be
4000 implementation dependent:
4001 *** The new compiler-only implementation still conforms with ANSI,
4002 but acts a little different than before. Besides the obvious
4003 changes in performance tradeoffs (that the cost per form passed
4004 to EVAL has gone up, and the cost per form executed by EVAL
4005 has gone down), the behavior of the system changes a little
4006 because there are no longer any interpreted function objects.
4007 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4008 e.g. doing COMPILE on the output of interactive DEFUN is
4010 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4012 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4013 has changed from (:CL) to NIL.
4014 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4015 e.g. #\Newline and #\Tab, don't have names specified in the
4016 ANSI Common Lisp standard, is now based on their ASCII symbolic
4017 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4018 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4019 are no longer used for output.
4020 ** changes in internal implementation constants:
4021 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4022 4 million. (If your application spends a lot of time GCing and
4023 you have a lot of RAM, you might want to experiment with
4024 increasing it even more.)
4025 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4026 However, almost all old code should still continue to work without
4027 immediate update, as SB-C-CALL is now a (deprecated) nickname
4029 ** Old operator names in the style DEF-FOO are now deprecated in
4030 favor of new corresponding names DEFINE-FOO, for consistency with
4031 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4032 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4033 internal symbols, but a few supported extensions like
4034 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4035 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4036 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4037 etc. as you get deeper into recursive calls to the debugger
4038 command loop, instead of the old "5]", "5]]", "5]]]"
4039 sequence. (I was motivated to do this when squabbles between
4040 ILISP and SBCL left me very deeply nested in the debugger. In the
4041 short term, this change will probably provoke more ILISP/SBCL
4042 squabbles, but hopefully it will be an improvement in the long run.)
4043 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4044 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4045 function, since declaiming return types involving aliens is
4046 (1) annoyingly messy to do by hand and (2) vital to efficient
4047 compilation of code which calls such functions.
4048 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4049 longer reexported by the SB-EXT package. They're solely useful
4050 for alien code, so it seems more logical that you should get
4051 them from the SB-ALIEN package, not in SB-EXT.
4052 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4053 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4054 features. Instead, the code that they used to control is always
4055 built into the system.
4056 * many other bug fixes
4057 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4058 advantage of the new EVAL-WHEN stuff and to clean them up in
4059 general, and they are now more ANSI-compliant in a number of
4060 ways. Martin Atzmueller is responsible for a lot of this.
4061 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4063 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4064 *** correct ERROR type for various file operations
4065 *** some fixes for Lisp streams
4066 *** DEFMETHOD syntax checking
4067 *** changing old weird representation of debug information as
4068 strings (which, among their other deficiencies, don't transform
4069 correctly when you rename packages, and don't change their
4070 print representation when you change things like *PACKAGE*
4071 and *PRINT-LENGTH*) to symbols and lists of symbols
4072 He also made several improvements and fixed several bugs in DESCRIBE.
4073 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4075 *** misbehavior of WRITE-STRING/WRITE-LINE
4076 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4077 and several other LOOP problems as well
4078 *** DIRECTORY when similar filenames are present
4079 *** DEFGENERIC with :METHOD options
4080 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4081 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4082 *** argument ordering in FIND with :TEST option
4083 *** mishandled package designator argument in APROPOS-LIST
4084 *** various problems in the backquote readmacro
4085 *** a bug in APROPOS
4086 *** probably some others that I'm not describing very well here,
4087 since the CVS log documents them by reference to sbcl-devel
4088 messages, and the SourceForge archives aren't working well.:-(
4089 ** Dan Barlow improved the Alpha port (and is making progress on the
4090 PPC port, for those of you who think different).
4091 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4092 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4093 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4094 continues to work on the SPARC port (for those of you in a position
4095 to look down upon our little PC-compatible boxes from a great height).
4096 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4097 and uses that copy, rather than the current dynamic value, when
4098 it's trying to decide whether to truncate output. Thus e.g.
4099 (let ((*print-lines* 50))
4100 (pprint-logical-block (stream nil)
4102 (let ((*print-lines* 8))
4103 (print (aref possiblybigthings i) stream)))))
4104 should now truncate the logical block only at 50 lines, instead of
4105 often truncating it at 8 lines, as it did before.
4106 * The doc/cmucl/ directory, containing old CMU CL documentation
4107 from the time of the fork, is no longer part of the base system.
4108 SourceForge has shut down its anonymous FTP service, and with it
4109 my original plan for distributing the old CMU CL documentation
4110 there. For now, if you need these files you can download an old
4111 SBCL source release and extract them from it.
4112 * The fasl file version number changed again, for dozens of reasons,
4113 some of which are apparent above.
4115 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4116 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4117 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4118 patch, so that DISASSEMBLE output is much nicer.
4119 * The code in the SB-PROFILE package now seems reasonably stable.
4120 I still haven't decided what the final interface should look like
4121 (I'd like PROFILE to interact cleanly with TRACE, since both
4122 facilities use function encapsulation) but if you have a need
4123 for profiling now, you can probably use it successfully with
4124 the current CMU-CL-style interface.
4125 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4126 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4127 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4128 link now returns the dangling link itself, and for similar
4129 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4130 link itself. (In these cases the old code signalled an error and
4131 looped endlessly, respectively.) Thus, DIRECTORY now works even
4132 in the presence of dangling and cyclic symbolic links.
4133 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4134 is now a supported extension again, since the consensus on
4135 sbcl-devel was that it can be useful for ordinary development
4136 work, not just for debugging SBCL itself.
4137 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4138 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4139 that currently-defined functions might be redefined later with
4140 different return types.
4141 * Hash tables can be printed readably, as inspired by CMU CL code
4142 of Eric Marsden and SBCL code of Martin Atzmueller.
4143 * better error handling in CLOS method combination, thanks to
4144 Martin Atzmueller porting Pierre Mai's CMU CL patches
4145 * more overflow fixes for >16Mbyte I/O buffers
4146 * A bug in READ has been fixed, so that now a single Ctrl-D
4147 character suffices to cause end-of-file on character streams.
4148 In particular, now you only need one Ctrl-D at the command
4149 line (not two) to exit SBCL.
4150 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4152 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4153 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4154 does the right thing.
4155 * The implementation of some type tests, especially for CONDITION
4156 types, is now tidier and maybe faster, due to CMU CL code
4157 originally by Douglas Crosher, ported by Martin Atzmueller.
4158 * Some math functions have been fixed, and there are new
4159 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4160 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4161 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4162 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4163 likely to be built into sbcl-0.7.0.)
4164 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4165 CL, ported to SBCL by Martin Atzmueller)
4166 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4167 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4168 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4169 * There's a new slam.sh hack to shorten the edit/compile/debug
4170 cycle for low-level changes to SBCL itself, and a new
4171 :SB-AFTER-XC-CORE target feature to control the generation of
4172 the after-xc.core file needed by slam.sh.
4173 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4174 COMPILE-FILE is no longer supported, so that now every function
4175 gets an entry point, so that block compilation looks a little
4176 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4177 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4178 returns the number of bytes consed since the system started,
4179 rather than the number consed since the first time the function
4180 was called. (The new definition parallels ANSI functions like
4181 CL:GET-INTERNAL-RUN-TIME.)
4182 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4183 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4184 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4185 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4186 * Fasl file version numbers are now independent of the target CPU,
4187 since historically most system changes which required version
4188 number changes have affected all CPUs equally. Similarly,
4189 the byte fasl file version is now equal to the ordinary
4192 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4193 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4194 is no longer recognized. I apologize for this, because it was
4195 listed in SB-EXT as a supported extension, but I found that
4196 its existing behavior was poorly specified, as well as incorrectly
4197 specified, and it looked like too much of a mess to straighten it
4198 out. I have enough on my hands trying to get ANSI stuff to work..
4199 * many patches ported from CMU CL by Martin Atzmueller, with
4200 half a dozen bug fixes in pretty-printing and the debugger, and
4201 half a dozen others elsewhere
4202 * fixed bug 13: Floating point infinities are now supported again.
4203 They might still be a little bit flaky, but thanks to bug reports
4204 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4205 as flaky as they were.
4206 * The --noprogrammer command line option is now supported. (Its
4207 behavior is slightly different in detail from what the old man
4208 page claimed it would do, but it's still appropriate under the
4209 same circumstances that the man page talks about.)
4210 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4211 are now supported, and enabled by default. Thus, the compiler can
4212 handle many floating point and complex operations much less
4213 inefficiently. (Thus e.g. you can implement a complex FFT
4215 * The compiler now detects type mismatches between DECLAIM FTYPE
4216 and DEFUN better, and implements CHECK-TYPE more correctly, and
4217 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4218 thanks to patches from Martin Atzmueller.
4219 * various fixes to make the cross-compiler more portable to
4220 ANSI-conforming-but-different cross-compilation hosts (notably
4221 Lispworks for Windows, following bug reports from Arthur Lemmens)
4222 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4223 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4224 fixed thanks to Pierre Mai's CMU CL patches.
4225 * fixes to keep the system from overflowing internal counters when
4226 it tries to use i/o buffers larger than 16M bytes
4227 * fixed bug 45a: Various internal functions required to support
4228 complex special functions have been merged from CMU CL sources.
4229 (When I was first setting up SBCL, I misunderstood a compile-time
4230 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4231 * improved support for type intersection and union, fixing bug 12
4232 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4233 more obscure bugs as well
4234 * some steps toward byte-compiling non-performance-critical
4235 parts of the system, courtesy of patches from Martin Atzmueller
4236 * Christophe Rhodes has made some debian packages of sbcl at
4237 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4238 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4239 stable, but are nonetheless usable. When he's ready, I'd be happy
4240 to add them to the SourceForge "File Releases" section. (And if
4241 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4242 * new fasl file format version number (because of changes in
4243 internal representation of (OR ..) types to accommodate the new
4244 support for (AND ..) types, among other things)
4246 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4247 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4249 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4251 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4252 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4253 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4254 * DESCRIBE now works on structure objects again.
4255 * Most function call argument type mismatches are now handled as
4256 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4257 know whether the function will be redefined before the call is
4258 executed. (The compiler could flag local calls with full WARNINGs,
4259 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4260 it doesn't keep track of enough information to know whether calls
4261 are local in this sense.)
4262 * Compiler output is now more verbose, with messages truncated
4263 later than before. (There should be some supported way for users
4264 to override the default verbosity, but I haven't decided how to
4265 provide it yet, so this behavior is still controlled by the internal
4266 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4267 src/compiler/ir1util.lisp.)
4268 * Fasl file format version numbers have increased again, because
4269 support for the Gray streams extension changes the layout of the
4270 system's STREAM objects.
4271 * The Gray subclassable streams extension now works, thanks to a
4272 patch from Martin Atzmueller.
4273 * The full LOAD-FOREIGN extension (not just the primitive
4274 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4275 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4276 but like most other programs, it defaults to copying the Unix
4277 environment from the original process instead of starting the
4278 new process in an empty environment.
4279 * Extensions which manipulate the Unix environment now support
4280 an :ENVIRONMENT keyword option which doesn't smash case or
4281 do other bad things. The CMU-CL-style :ENV option is retained
4282 for porting convenience.
4283 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4284 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4286 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4288 * A patch from Martin Atzmueller seems to have solved the SIGINT
4289 problem, and as far as we know, signal-handling now works cleanly.
4290 (If you find any new bugs, please report them!)
4291 * The system no longer defaults Lisp source file names to types
4292 ".l", ".cl", or ".lsp", but only to ".lisp".
4293 * The compiler no longer uses special default file extensions for
4294 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4295 seems to expect a single default extension for all compiled code,
4296 and there's no compelling reason to try to stretch the standard
4297 to allow two different extensions.) Instead, byte-compiled files
4298 default to the same extension as native-compiled files.
4299 * Fasl file format version numbers have increased again, because
4300 a rearrangement of internal implementation packages made some
4301 dumped symbols in old fasl files unreadable in new cores.
4302 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4303 many fewer weird special cases.
4304 * Bug #17 (differing COMPILE-FILE behavior between logical and
4305 physical pathnames) has been fixed, and some related misbehavior too,
4306 thanks to a patch from Martin Atzmueller.
4307 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4308 by Tim Moore, ported to SBCL by Martin Atzmueller.
4309 * Martin Atzmueller fixed several filesystem-related problems,
4310 including bug #36, in part by porting CMU CL patches, which were
4311 written in part by Paul Werkowski.
4312 * More compiler warnings in src/runtime/ are gone, thanks to
4313 more patches from Martin Atzmueller.
4314 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4317 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4319 * DESCRIBE now works on CONDITION objects.
4320 * The debugger now handles errors which arise when trying to print
4321 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4323 * The build system now uses an additional file, customize-target-features.lisp,
4324 to allow local modifications to the target *FEATURES* list. (The point of
4325 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4326 debugging features enabled, without having to worry about propagating it
4327 into everyone's system when I do a "cvs update".) When no
4328 customize-target-features.lisp file exists, the target *FEATURES* list
4329 should be constructed the same way as before.
4330 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4331 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4332 warn about behavior which is undefined under the ANSI spec. Note: This
4333 is specified by ANSI, but it's not very popular with programmers.
4334 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4335 macro in the SBCL sources for an example of a workaround which you
4336 might use to make portable ANSI-standard code which does what you want.
4337 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4338 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4339 when it's used as a non-toplevel form. (This is required in order
4340 to implement the DEFCONSTANT-EQX macro.)
4341 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4342 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4343 funny things, due to bugs in the implementation of EVAL-WHEN
4344 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4345 non-toplevel DEFCONSTANTs are uncommon.)
4346 * The core file version number and fasl file version number have been
4347 incremented, because the old noncompliant DEFCONSTANT behavior involved
4348 calling functions which no longer exist, and because I also took the
4349 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4350 * fixed bug 1 (error handling before read-eval-print loop starts), and
4351 redid debugger restarts and related debugger commands somewhat while
4353 ** The QUIT debugger command is gone, since it did something
4354 rather different than the SB-EXT:QUIT command, and since it never
4355 worked properly outside the main toplevel read/eval/print loop.
4356 Invoking the new TOPLEVEL restart provides the same functionality.
4357 ** The GO debugger command is also gone, since you can just invoke
4358 the CONTINUE restart directly instead.
4359 ** The TOP debugger command is also gone, since it's redundant with the
4360 FRAME 0 command, and since it interfered with abbreviations for the
4362 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4364 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4365 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4366 some porting work by Martin Atzmueller)
4367 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4368 with a boatload of patches which he ported to SBCL. Now that those
4370 ** The system tries to make sure that its low-priority messages
4371 are prefixed by semicolons, to help people who like to use
4372 syntax highlighting in their ILISP buffer. (This patch
4373 was originally due to Raymond Toy.)
4374 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4375 patches originally written by Raymond Toy.
4376 ** The compiler understands coercion between single-value and
4377 multiple-VALUES type expressions better, getting rid of some very
4378 weird behavior, thanks to patches originally by Robert MacLachlan
4379 and Douglas Crosher.
4380 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4381 lambda lists, thanks to a patch originally by Pierre Mai.
4382 ** The system no longer bogusly warns about "abbreviated type
4384 ** The compiler gets less confused by inlining and RETURN-FROM,
4385 thanks to some patches originally by Tim Moore.
4386 ** The system no longer hangs when dumping circular lists to fasl
4387 files, thanks to a patch originally from Douglas Crosher.
4388 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4389 undefined function error.
4390 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4391 in structure slot initforms) for now, documented workaround instead:-|
4392 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4393 * fixed bug 5: added stubs for various Gray stream functions called
4394 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4395 installed, at least appropriate type errors are generated
4396 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4397 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4398 consistently in DEFMETHOD forms.
4399 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4400 it doesn't seem to affect SBCL after all
4401 * The C runtime system now builds with better optimization and many
4402 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4404 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4406 * The system is now under CVS at SourceForge (instead of the
4407 CVS repository on my home machine).
4408 * The new signal handling code has been tweaked to treat register
4409 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4410 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4411 downstream, e.g. in debug-int.lisp, has implicit dependencies
4412 on the unsignedness of integer representation of machine words,
4413 and that caused the system to bomb out with infinite regress
4414 when trying to recover from type errors involving signed values,
4415 e.g. (BUTLAST '(1 2 3) -1).)
4416 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4417 (This was one of the bugs Peter Van Eynde reported back in July.)
4418 * The system now uses code inspired by Colin Walters' O(N)
4419 implementation of MAP (from the cmucl-imp@cons.org mailing
4420 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4421 inline the MAP operation, and there is more than one
4422 sequence argument to the MAP call (so that it can't just
4423 do ETYPECASE once and for all based on the type of the
4424 single sequence argument). (The old non-inline implementation
4425 of the general M-argument sequence-of-length-N case required
4426 O(M*N*N) time when any of the sequence arguments were LISTs.)
4427 * The QUIT :UNIX-CODE keyword argument has been renamed to
4428 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4429 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4431 * Raymond Wiker's patches to port dynamic loading from Linux to
4432 FreeBSD have been added.
4433 * The BUGS file is now more nearly up to date, thanks in large part
4434 to Martin Atzmueller's review of it.
4435 * The debugger now flushes standard output streams before it begins
4436 its output ("debugger invoked" and so forth).
4437 * The core version number and fasl file version number have both
4438 been incremented, because of incompatible changes in the layout
4440 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4441 used to not be called for a saved Lisp image.)
4442 * Martin Atzmueller's version of a patch to fix a compiler crash,
4443 as posted on sbcl-devel 13 September 2000, has been installed.
4444 * Instead of installing Martin Atzmueller's patch for the
4445 compiler transform for SUBSEQ, I deleted the compiler transform,
4446 and transforms for some similar consing operations.
4447 * A bug in signal handling which kept TRACE from working on OpenBSD
4449 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4450 away by constant folding
4451 * The system now defines its address space constants in one place
4452 (in the Lisp sources), and propagates them automatically elsewhere
4453 (through GENESIS and the sbcl.h file). Therefore, patching the
4454 address map is less unnecessarily tedious and error-prone. The
4455 Lisp names of address space constants have also been systematized.
4456 * CVS tags like dollar-Header-dollar have been removed from
4457 the sources, because they have never saved me trouble and
4458 they've been source of trouble working with patches and other
4459 diff-related operations.
4460 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4461 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4463 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4465 * The system has been ported to OpenBSD.
4466 * The system now compiles with a simple "sh make.sh" on the systems
4467 that it's supported on. I.e., now you no longer need to tweak
4468 text in the target-features.lisp-expr and symlinks in src/runtime/
4469 by hand, the make.sh takes care of it for you.
4470 * The system is no longer so grossly inefficient when compiling code
4471 involving vectors implemented as general (not simple) vectors (VECTOR T),
4472 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4473 which dares to use the various sequence functions on non-simple
4474 vectors, takes less of a performance hit.
4475 * There is now a primitive type predicate VECTOR-T-P
4476 to test for the (VECTOR T) type, so that e.g.
4477 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4478 can now be compiled with some semblance of efficiency. (The old code
4479 turned the type declaration into a full call to %TYPEP at runtime!)
4480 * AREF on (VECTOR T) is still not fast, since it's still compiled
4481 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4482 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4483 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4484 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4485 without making full calls to SUBTYPEP at runtime.
4486 (Some analogous efficiency issues for non-simple vectors specialized to
4487 element types other than T, or for non-simple multidimensional arrays,
4488 have not been addressed. They could almost certainly be handled the
4489 same way if anyone is motivated to do so.)
4490 * The changes in array handling break binary compatibility, so
4491 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4492 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4493 instead of (VALUES T T).
4494 * By following the instructions that Dan Barlow posted to sbcl-devel
4495 on 2 July 2000, I was able to enable primitive dynamic object
4496 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4497 functionality is not implemented (since it calls ld to resolve
4498 library references automatically, requiring RUN-PROGRAM for its
4499 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4500 try to resolve library references) is now supported.
4501 * The system now flushes the standard output streams when it terminates,
4502 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4503 them at several other probably-convenient times, e.g. in each pass of
4504 the toplevel read-eval-print loop, and after evaluating a form given
4505 as an "--eval" command-line option. (These changes were motivated by a
4506 discussion of stream flushing issues on cmucl-imp in August 2000.)
4507 * The source transform for TYPEP of array types no longer assumes
4508 that an array whose element type is a not-yet-defined type
4509 is implemented as an array of T, but instead punts, so that the
4510 type will be interpreted at runtime.
4511 * There is now some support for cross-compiling in make.sh: each of
4512 the phases of make.sh has its own script. (This should be transparent
4513 to people doing ordinary, non-cross-compile builds.)
4514 * Since my laptop doesn't have hundreds of megabytes of memory like
4515 my desktop machine, I became more motivated to do some items on
4516 my to-do list in order to reduce the size of the system a little:
4517 ** Arrange for various needed-only-at-cold-init things to be
4518 uninterned after cold init. To support this, those things have
4519 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4520 symbols with such names are now uninterned after cold init).
4521 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4522 fasl files for cold load.
4523 ** Remove the old compiler structure pooling code (which used to
4524 be conditional on the target feature :SB-ALLOC) completely.
4525 ** Redo the representation of some data in cold init to be more compact.
4526 (I also looked into supporting byte compiled code at bootstrap time,
4527 which would probably reduce the size of the system a lot, but that
4528 looked too complicated, so I punted for now.)
4529 * The maximum signal nesting depth in the src/runtime/ support code has
4530 been reduced from 4096 to 256. (I don't know any reason for the very
4531 large old value. If the new smaller value turns out to break something,
4532 I'll probably just bump it back up.)
4533 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4535 * Many, many bugs reported by Peter Van Eynde have been added to
4536 the BUGS list; some have even been fixed.
4537 * While enabling dynamic object file loading, I tried to make the
4538 code easier to understand, renaming various functions and variables
4539 with less ambiguous names, and changing some function calling
4540 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4541 * While trying to figure out how to do the OpenBSD port, I tried to
4542 clean up some of the code in src/runtime/. In particular, I dropped
4543 support for non-POSIX signal handling, added various comments,
4544 tweaked the code to reduce the number of compilation warnings, and
4545 renamed some files to increase consistency.
4546 * To support the new automatic configuration functionality in make.sh,
4547 the source file target-features.lisp-expr has been replaced with the
4548 source file base-target-features.lisp-expr and the machine-generated
4549 file local-target-features.lisp-expr.
4550 * fixed a stupid quoting error in make.sh so that using CMU CL
4551 "lisp -batch" as cross-compilation host works again
4553 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4555 * DESCRIBE no longer tries to call itself recursively to describe
4556 bound/fbound values, so that it no longer fails on symbols which are
4557 bound to themselves (like keywords, T, and NIL).
4558 * DESCRIBE now works on generic functions.
4559 * The printer now prints less-screwed-up representations of closures
4560 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4561 * A private symbol is used instead of the :EMPTY keyword previously
4562 used to mark empty slots in hash tables. Thus
4563 (DEFVAR *HT* (MAKE-HASH-TABLE))
4564 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4565 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4566 now does what ANSI says that it should. (You can still get
4567 similar noncompliant behavior if bang on the hash table
4568 implementation with all the symbols you get back from
4569 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4570 This breaks binary compatibility, since tests for equality to
4571 :EMPTY are wired into things like the macroexpansion of
4572 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4574 * There's now a minimal placeholder implementation for CL:STEP,
4575 as required by ANSI.
4576 * An obscure bug in the interaction of the normal compiler, the byte
4577 compiler, inlining, and structure predicates has been patched
4578 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4579 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4581 * Missing ordinary arguments in a macro call are now detected even
4582 when the macro lambda list contains &KEY or &REST.
4583 * The debugger no longer complains about encountering the top of the
4584 stack when you type "FRAME 0" to explicitly instruct it to go to
4585 the top of the stack. And it now prints the frame you request even
4586 if it's the current frame (instead of saying "You are here.").
4587 * As specified by ANSI, the system now always prints keywords
4588 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4590 * The default initial SIZE of HASH-TABLEs is now smaller.
4591 * Type information from CLOS class dispatch is now propagated
4592 into DEFMETHOD bodies, so that e.g.
4593 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4595 is now basically equivalent to
4596 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4597 (DECLARE (TYPE SINGLE-FLOAT X))
4599 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4600 operation, without having to do run-time type dispatch.
4601 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4602 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4603 * Since I don't seem to be making big file reorganizations very often
4604 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4605 is overflowing my ability to conveniently back them up), I've finally
4606 checked the system into CVS. (The CVS repository is on my home system,
4607 not at SourceForge -- putting it on SourceForge might come later.)
4608 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4609 high-level GC-NOTIFY-FOO functions send their output. (There's
4610 still very little control of where low-level verbose GC functions
4611 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4612 less than it used to -- the GC-NOTIFY-FOO functions are now under
4613 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4614 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4615 in only one place in the source code, and propagates it automatically
4616 everywhere that it's needed. Thus e.g. when I bump the version from
4617 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4618 * The C source files now include boilerplate legalese and documentation
4619 at the head of each file (just as the Lisp source files already did).
4620 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4621 to his page will be replaced with a link to his new CLiki service.
4623 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4625 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4626 * The build process now looks for GNU make under the default name "gmake",
4627 instead of "make" as it used to. If GNU make is not available as "gmake"
4628 on your system, you can change this default behavior by setting the
4629 GNUMAKE environment variable.
4630 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4631 can build without error under CMU CL.
4633 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4635 * There is now a partial SBCL user manual (with some new text and some
4636 text cribbed from the CMU CL manual).
4637 * The beginnings of a profiler have been added (starting with the
4638 CMU CL profiler and simplifying and cleaning up). Eventually the
4639 main interface should be through the TRACE macro, but for now,
4640 it's still accessed through vaguely CMU-CL-style functions and macros
4641 exported from the package SB-PROFILE.
4642 * Some problems left over from porting CMU CL to the new
4643 cross-compilation bootstrap process have been cleaned up:
4644 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4645 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4646 part when getting rid of the compiler *BACKEND* stuff.)
4647 ** The value of *NULL-TYPE* was screwed up, because it was
4648 being initialized before the type system knew the final
4649 definition of the 'NULL type. This screwed up several key
4650 optimizations in the compiler, causing inefficiency in all sorts
4651 of places. (I found it because I wanted to understand why
4652 GET-INTERNAL-RUN-TIME was consing.)
4653 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4654 PROCLAIM FTYPE information. Unfortunately this broke binary
4655 compatibility again, since now the forms output by DEFGENERIC
4656 to refer to functions which didn't exist in 0.6.3.
4657 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4658 can use the new (as of 0.6.3) transform for SOME into MAP into
4660 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4661 values to (INTEGER 0 1000000), so that the time code will no longer
4662 occasionally get blown up by Linux returning 1000000 microseconds
4663 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4664 its output conform to the ANSI spec. (Alas, this makes its output
4665 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4666 * A full call to MAP NIL with a single sequence argument no longer conses.
4667 * fixes to problems pointed out by Martin Atzmueller:
4668 * The manual page no longer talks about multiprocessing as though
4669 it were currently supported.
4670 * The ILISP support patches have been removed from the distribution,
4671 because as of version 5.10.1, ILISP now supports SBCL without us
4672 having to maintain patches.
4673 * added a modified version of Raymond Toy's recent CMU CL patch for
4674 EQUALP comparison of HASH-TABLE
4676 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4678 * The system still can't cross-compile itself with
4679 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4680 entails), but at least it can compile more of itself that way
4681 than it used to be able to, and various buglets which were uncovered
4682 by trying to cross-compile itself that way have now been fixed.
4683 * This release breaks binary compatibility again. This time
4684 at least I've incremented the FASL file format version to 2, so that the
4685 problem can be detected reliably instead of just causing weird errors.
4686 * various new style warnings:
4687 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4688 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4689 ** using the lexical binding of a variable named in the *FOO* style
4690 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4691 as specified by ANSI.
4692 * *RANDOM-STATE* is no longer automatically initialized from
4693 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4694 default behavior of the system is to repeat its behavior every time
4695 it's run. If you'd like to change this behavior, you can always
4696 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4697 old convention there was no comparably easy way to get the system to
4698 repeat its behavior every time it was run.
4699 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4700 FTYPE declarations has been removed, in favor of their ANSI
4701 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4702 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4703 the types of their sequence arguments can be determined at compile time.
4704 This is done through a new open code expansion for MAP which eliminates
4705 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4706 argument types can be determined at compile time.
4707 * The optimizer now transforms COERCE into an identity operation when it
4708 can prove that the coerced object is already of the correct type. (This
4709 can be a win for machine generated code, including the output of other
4710 optimization transforms, such as the MAP transform above.)
4711 * Credit information has been moved from source file headers into CREDITS.
4712 * Source file headers have been made more standard.
4713 * The CASE macro now compiles without complaining even when it has
4716 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4718 * (Note that the way that the PCL macroexpansions were rewritten
4719 to accommodate the change in DEFGENERIC below breaks binary
4720 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4721 not run under sbcl-0.6.2. Once we get out of alpha releases,
4722 i.e. hit release 1.0.0, we'll probably try to maintain binary
4723 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4724 and sbcl-1.4.4. Until then, however, it might be fairly common
4725 for maintenance releases to break binary compatibility.)
4726 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4728 * The reporting of unhandled signals has been changed to print some
4729 explanatory text as well as the report form. (Previously only
4730 the report form was printed.)
4731 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4732 it defines, so that the compiler no longer issues undefined function
4733 warnings for compiled-but-not-yet-loaded generic functions.
4734 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4735 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4736 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4737 section of the ANSI spec are supported.
4738 * The "" nickname for the "KEYWORD" package has been removed.
4739 The reader still handles symbol tokens which begin with a package marker
4740 as keywords, but it doesn't expose its mechanism for doing so in the
4741 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4742 * The system now issues STYLE-WARNINGs for contradictory TYPE
4743 proclamations. (Warnings for contradictory FTYPE proclamations would
4744 be nice too, but those can't be done usefully unless the type system
4745 is made smarter about FUNCTION types.)
4746 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4747 systematized, so that "*target-*.lisp is supposed to exist only on the
4748 target and imply that there's a related file which exists on the
4749 host, and *host-*.lisp is supposed to exist only on the host and imply
4750 that there's a related file which exists on the target. This involves a
4751 lot of renaming. Hopefully the acute confusion caused by the renaming
4752 will be justified by the reduction in chronic confusion..
4753 ** runtime-type.lisp -> early-target-type.lisp
4754 ** target-type.lisp -> late-target-type.lisp
4755 ** early-host-format.lisp -> early-format.lisp
4756 ** late-host-format.lisp -> late-format.lisp
4757 ** host-error.lisp -> misc-error.lisp
4758 ** early-error.lisp -> early-target-error.lisp
4759 ** late-error.lisp -> late-target-error.lisp
4760 ** host-defboot.lisp -> early-defboot.lisp
4761 ** code/misc.lisp -> code/target-misc.lisp
4762 ** code/host-misc.lisp -> code/misc.lisp
4763 ** code/numbers.lisp -> code/target-numbers.lisp
4764 ** code/early-numbers.lisp -> numbers.lisp
4765 ** early-host-type.lisp -> early-type.lisp
4766 ** late-host-type.lisp -> late-type.lisp
4767 ** host-typep.lisp -> typep.lisp
4768 ** load.lisp -> target-load.lisp
4769 ** host-load.lisp -> load.lisp
4770 ** host-disassem.lisp -> disassem.lisp
4771 ** host-insts.lisp -> insts.lisp
4772 ** byte-comp.lisp -> target-byte-comp.lisp
4773 ** host-byte-comp.lisp -> byte-comp.lisp
4774 ** host-signal.lisp -> signal.lisp
4775 ** host-defstruct.lisp -> defstruct.lisp
4776 ** late-target-type.lisp -> deftypes-for-target.lisp
4777 Furthermore, several other previously target-only files foo.lisp (e.g.
4778 hash-table.lisp and random.lisp) have been split into a target-and-host
4779 foo.lisp file and a target-only target-foo.lisp file, with their key type
4780 definitions in the target-and-host part, so that the cross-compiler will
4781 know more about target types.
4782 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4783 gone away. In their place are various *BACKEND-FOO* variables
4784 corresponding to the slots of the old structure.
4785 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4786 into the target SBCL has been fixed.
4787 * The chill.lisp system for loading cold code into a running SBCL
4789 * Support for the CMU CL "scavenger hook" extension has been removed.
4790 (It was undocumented and unused in the CMU CL sources that SBCL was
4791 derived from, and stale in sbcl-0.6.1.)
4792 * Various errors in the cross-compiler type system were detected
4793 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4794 (enabling various consistency checks). Many of them were fixed,
4795 but some hard problems remain, so the compiler is back to
4796 running without *TYPE-SYSTEM-INITIALIZED* for now.
4797 * As part of the cross-compiler type system cleanup, I implemented
4798 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4799 * I have started adding UNCROSS calls throughout the type system
4800 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4801 on cross-compiler input files will be able to go away, and various
4803 * CONSTANTP now returns true for quoted forms (as explicitly required
4806 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4808 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4809 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4810 (MAKE-LIST -1). In the longer term, it ought to become true
4811 that declarations are assertions even at SAFETY 1. For now, it's not
4812 quite true even at SAFETY 3, but it's at least more nearly true..
4813 (Note that this change seems to increases the size of the system by
4814 O(5%) and to decrease the speed of the compiler by 20% or more.)
4815 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4816 for the problem of printing dozens of lines of distracting information
4817 about low-level system machinery as part of the top stack frame
4818 on entry to the debugger when an undefined function was called.
4819 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4820 is not reset to COMMON-LISP-USER.
4821 * Compilation of stuff related to dyncount.lisp has been made conditional
4822 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4823 smaller. The various dyncount-related symbols have been moved into
4824 a new "SB-DYNCOUNT" package.
4825 * tty-inspect.lisp has been renamed to inspect.lisp.
4826 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4827 feature has gone away. (When we eventually port to other flavors of
4828 libc and/or Unix, we'll try to make the differences between flavors
4829 invisible at the user level.)
4830 * Various other *FEATURES* tags, and/or their associated conditionals,
4831 have been removed if obsolescent, or given better documentation, or
4832 sometimes given more-mnemonic names.
4834 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4836 * tidied up "make.sh" script
4837 * tidied up system directory structure
4838 * better "clean.sh" behavior
4839 * added doc/FOR-CMUCL-DEVELOPERS
4840 * many many small tweaks to output format, e.g. removing possibly-confusing
4841 trailing #\. character in DESCRIBE-INSTANCE
4842 * (EQUALP #\A 'A) no longer signals an error.
4843 * new hashing code, including EQUALP hashing
4844 * tidied up Lisp initialization and toplevel
4845 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4846 * command line argument processing
4847 * added POSIX-GETENV function to deal with Unix-ish environment variables
4848 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4849 terminating SBCL on EOF
4850 * non-verbose GC by default
4851 * There is no more "sbcl" shell script; the sbcl file is now the C
4852 runtime executable (just like CMU CL).
4853 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4855 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4856 symbol and package information as Lisp native symbol and package objects
4857 instead of strings naming symbols and strings naming packages. This way,
4858 whenever packages are renamed (as in warm init), debug information is
4859 transformed along with everything else.
4860 * tweaked the optimization policy declarations which control the building
4861 of SBCL itself. Now, among other things, the system no longer saves
4862 source location debugging information. (This helps two problems at once
4863 by reducing SBCL size and by keeping SBCL from trying to look for its
4864 sources -- which may not exist -- when reporting errors.)
4865 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4866 debugging and testing purposes
4867 * cleaned up printing, making the printer call PRINT-OBJECT for
4868 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4869 methods, giving nearly-ANSI behavior
4870 * converted almost all special variables to use *FOO* naming convention
4871 * deleted PARSE-TIME functionality, since it can be done portably
4872 * moved some files out of cold init into warm init
4873 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4875 * regularized formatting of source files
4876 * added an install.sh script
4877 * fixed ridiculous memory usage of cross-compiler by making
4878 compiler/alloc.lisp not try to do pooling unless it can hook
4879 itself into the GC of the cross-compilation host. Now the system
4880 builds nicely on my old laptop.
4881 * added :SB-ALLOC in target-features.lisp-expr
4882 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4883 was not implemented)
4884 * re-did condition handling and note reporting in the compiler. Notes
4885 are no longer handled by signalling conditions. Style warnings
4886 and warnings are handled more correctly and reported in such a way
4887 that it's easy to find one or the other in your output (so that you
4888 can e.g. figure out which of many problems caused COMPILE-FILE to
4890 * changed the severity of several compiler warnings from full WARNING
4891 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4892 compiler note reporting so that it doesn't use the condition system
4893 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4894 COMPILE-FILE command)
4895 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4896 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4897 no longer supported, since their implementation was deeply intertwingled
4898 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4899 * removed bogus "support" for compiler macros named (SETF FOO), and
4900 removed the compiler macro for SETF INFO (but only after making a fool
4901 of myself on the cmucl-imp mailing list by posting a bogus patch for
4902 DEFINE-COMPILER-MACRO..)
4903 * Compiled files containing forms which have side effects on the Lisp
4904 reader (such as DEFPACKAGE forms) are now handled more correctly.
4905 (Compiler queuing of top level lambdas has been suppressed by setting
4906 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4907 * deleted various currently-unused source files, e.g. gengc.lisp. They
4908 may be added back at some point e.g. when porting to other architectures,
4909 but until they are it's distracting to distribute them and to try to
4911 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4912 there were so many of them they're just distractions, and UNCROSS is
4913 known to be able to handle the current sources
4914 * moved PROFILE functionality into TRACE, so that it will be clear
4915 how the wrapping and unwrapping of functions when you profile them
4916 interacts with the wrapping and unwrapping of functions when you
4917 trace them. (Actually, the functionality isn't there yet, but at least
4918 the interface specification is there. Hopefully, the functionality will
4919 arrive with some future maintenance release.)
4920 * removed host-oops.lisp
4921 * changed signature of QUIT function to allow UNIX-CODE argument
4922 * fixed READ-SEQUENCE bug
4923 * tweaked verbose GC output so that it looks more like the progress
4924 output that ANSI specifies for functions like LOAD
4925 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4926 * added <http://sbcl.sourceforge.com> to the banner information printed by