1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.22 relative to 1.0.21:
3 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
4 for the shared object in the current directory, but passes the native
5 namestring of the designated pathname to the operation system's shared
6 object loading function as-it.
7 * minor incompatible change: --disable-debugger toplevel option now takes
8 effect before processing of initialization files and --eval or --load
10 * new feature: new commandline argument: --script, which supports
11 shebang lines. See documentation for details. (based on work by
13 * new feature: SAVE-LISP-AND-DIE can save current values of
14 --dynamic-space-size and --control-stack-size in the executable core,
15 causing it to skip normal runtime option processing. See documentation
16 for details. (thanks to Zach Beane)
17 * enhancement: inoccous calls to EVAL or generic functions dispatching
18 on subclasses of eg. STREAM no longer cause compiler notes to appear.
19 * enhancement: the system no longer resignals errors from --load and
20 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
21 associated with the original error to be lost. (thanks to Ariel
23 * enhancement: :DONT-SAVE keyword argument has been added to
24 LOAD-SHARED-OBJECT for controlling interaction with
26 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
27 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
28 provided. (reported by Cedric St-Jean)
29 * bug fix: circularity handling in the reader did not treat raw
30 structure slots correctly. (reported by Cedric St-Jean)
31 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
32 file descriptors when there were none.
33 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
34 was broken. (reporteed by Willem Broekema)
35 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
36 pathnames without a directory.
37 * bug fix: COMPILE-FILE did not warn about undefined variable
38 references at toplevel, and LOAD of the resulting fasl did
41 changes in sbcl-1.0.21 relative to 1.0.20:
42 * new feature: the compiler is able to track the effective type of a
43 generic function across method addition and removal.
44 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
45 appropriately typed structure slots without locking.
46 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
47 information like those gathered by TIME using a programming-friendly
49 * new feature: TIME reports time taken even if the form performs a
50 non-local transfer of control.
51 * enhancement: reduced conservativism on GENCGC platforms: on
52 average 45% less pages pinned (measured from SBCL self build).
53 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
54 mutate constant symbols or violate declaimed type of the symbol.
55 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
56 sleeping on the semaphore in heavily contested semaphores.
57 * bug fix: semaphores and condition variables were not interrupt
59 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
60 owned by other threads anymore.
61 * bug fix: FIND on lists called KEY outside the specified
62 subsequence. (reported by budden)
63 * bug fix: LOG doesn't use single-float intermediate results when
64 given mixed integer and double-float arguments, leading to better
65 precision. (reported by Bob Felts)
66 * bug fix: LOG with base zero returned values of inconsistent type.
68 changes in sbcl-1.0.20 relative to 1.0.19:
69 * minor incompatible change: OPTIMIZE qualities
70 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
71 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
72 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
73 * documentation: some slot access efficiency guidelines have been
74 added to the user manual.
75 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
76 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
78 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
79 to more efficient EQ-comparison versions more often.
80 * optimization: enhanced derivation of DOLIST iteration variable type
82 * optimization: constant folding of simple (LIST ...) forms as DOLIST
84 * optimization: runtime lookup of function definitions can be
85 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
86 * optimization: compiler is able to derive the return type of
87 (AREF (THE STRING X) Y) as being CHARACTER.
88 * optimization: CLRHASH on empty hash-tables no longer does pointless
89 work. (thanks to Alec Berryman)
90 * optimization: REPLACE deftransforms don't punt when :START1 or
92 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
93 no longer emits a code deletion note. (reported by Andrew
95 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
96 functions with an odd number of &OPTIONAL arguments, a &REST
97 argument, and one or more &KEY arguments at the call site.
98 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
99 arguments appeared at call sites as well.
100 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
101 READ-LINE &co can no longer read from them afterwards. (reported
103 * bug fix: fixed #427: unused local aliens no longer cause compiler
104 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
106 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
107 alien-stack leakage. (reported by Andy Hefner)
108 * bug fix: PROGV signals an error when an attempt to violate declared
109 type of a variable or bind a constant is made.
110 * bug fix: SET signals an error when an attempt to violate declared
111 type of a variable is made.
112 * bug fix: restart computation during the execution of a restart
113 test function no longer causes infinite recursion. (reported by
115 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
116 signals an error. (thanks to Pierre Mai)
117 * bug fix: EXPORT left symbol unexported in conflict situations.
118 (thanks to Michael Weber)
119 * bug fix: correctly handle name conflicts involving CL:NIL.
120 (thanks to Michael Weber)
121 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
122 conflicts arising from USEing package with conflicting symbols
123 correctly. (thanks to Michael Weber)
125 changes in sbcl-1.0.19 relative to 1.0.18:
126 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
127 warnings that go otherwise unhandled will be muffled if they are
128 of the type that's the value of this variable.
129 * optimization: stack allocation is slightly more efficient on x86
131 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
132 where a raw slot always is initialized using the initform whose
133 type is not know sufficiently well a compile-time are now compiled
134 correctly. (reported by John Morrison)
135 * bug fix: compiler no longer makes erronous assumptions in the
136 presense of non-foldable SATISFIES types.
137 * bug fix: stack analysis missed cleanups of dynamic-extent
138 arguments in non-let-converted calls to local functions.
139 * improvements to the Windows port:
140 ** adjusted address spaces for building on both Win32 and
141 Win64. (thanks for John Connors)
142 * fixed some bugs revealed by Paul Dietz' test suite:
143 ** interval arithmetic during type derivation used inexact integer
144 to single-float coercions.
145 ** arithmetic operations involving large integers and single
146 floats give the same results in compiled and interpreted code.
147 ** deriving the result type of COERCE no longer signals an error
148 if the derived type of the second argument is a MEMBER type
149 containing invalid type specifiers.
150 ** ADJOIN with constant NIL as second argument works correctly.
151 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
153 changes in sbcl-1.0.18 relative to 1.0.17:
154 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
155 profiles only the current thread.
156 * minor incompatible change: changes to SYMBOL-VALUE of constants
157 defined with DEFCONSTANT now signal an error.
158 * enhancement: SB-SPROF now has support for wallclock profiling,
159 and is also able to profile specific threads. REPORT output
160 has also additional sorting options.
161 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
163 * optimization: structure allocation has been improved
164 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
165 ** out of line constructors are ~10% faster.
166 ** inline constructors are ~15% faster.
167 ** inline constructors are capable of dynamic extent allocation
168 (generally on x86 and x86-64, in some cases on other platforms
170 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
172 * optimization: file compiler is now able to coalesce non-circular
173 lists, non-base strings, and bit-vectors. Additionally, constants
174 are never referenced through SYMBOL-VALUE at runtime.
175 * optimization: code defining methods on PRINT-OBJECT (and other
176 generic functions in the COMMON-LISP package) now loads faster.
177 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
179 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
180 fasl is now deleted. (reported by Attila Lendvai)
181 * bug fix: READ-LINE always returned NIL for the last line in files.
182 (reported by Yoshinori Tahara)
183 * bug fix: more accurate disassembly annotations of foreign function
184 calls. (thanks to Andy Hefner)
185 * bug fix: trimming non-simple strings and non-string string
186 designators when the there is nothing to trim works properly.
187 (thanks to James Knight)
188 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
189 (thanks to Travis Cross)
190 * fixed some bugs revealed by Paul Dietz' test suite:
191 ** NIL is a valid function name (regression at 1.0.13.38)
192 ** FILL on lists was missing its return value (regression at 1.0.12.27)
193 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
194 fill pointers properly (regression at 1.0.12.23)
195 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
196 displacement indices properly (regression at 1.0.12.23)
198 changes in sbcl-1.0.17 relative to 1.0.16:
199 * temporary regression: user code can no longer allocate closure
200 variable storage on stack, due to bug 419 without explicitly
201 requesting it. Please consult sbcl-devel for advice if you need to
202 use this feature in the meanwhile.
203 * new feature: runtime argument --control-stack-size can be used to
204 adjust thread default control stack size.
205 * enhancement: improved TIME output
206 ** all times are reported using the measured accuracy (milliseconds
207 for real and GC times, microseconds for everything else.)
208 ** processor cycle counts on x86 and x86-64.
209 ** interpreted forms are counted for both evaluator modes.
210 ** number of lambdas converted by the compiler is reported.
211 ** CPU percentage report (computed from real and total run time.)
212 ** more comprehensive run time reporting, using a condenced format
213 ** interperted form, lambda, and page fault counts are omitted
215 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
217 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
218 * optimization: two argument forms of LAST are upto ~10% faster
219 in normal SPEED policies.
220 * optimization: NCONC no longer needs to heap cons its &REST list
221 in normal SPEED policies.
222 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
224 * bug fix: UNION and NUNION work with :TEST-NOT once more,
225 regression since 1.0.9.1. (thanks to Eric Marsden)
226 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
227 since 1.0.15.36. (thanks to Paul Khuong)
228 * bug fix: LAST when always returned the whole list when given a bignum
229 as the second argument.
230 * bug fix: dynamic extent allocation of nested lists and vectors
231 could leak to otherwise accessible parts.
232 * bug fix: invalid optimization of heap-allocated alien variable
234 * bug fix: fasl header checking is less vulnerable to different
235 platform word lengths.
236 * bug fix: more correct assembler syntax for GNU binutils
237 2.18.50.0.4 support. (thanks to Marijn Schouten)
238 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
240 * bug fix: the fopcompiler can handle LOCALLY forms (with no
241 declarations) successfully. (reported by Attila Lendvai)
243 changes in sbcl-1.0.16 relative to 1.0.15:
244 * minor incompatible change: revert the changes to sb-posix's error
245 signaling added in 1.0.14.
246 * minor incompatible change: change PROBE-FILE back to returning
247 NIL whenever we can't get a truename, as was the case before 1.0.14.
248 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
249 inherits from ERROR instead of just CONDITION.
250 * new feature: SB-INTROSPECT can provide source locations for instances
251 as well. (thanks to Tobian Ritterweiler)
252 * optimization: binding special variables now generates smaller code
253 on threaded platforms.
254 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
255 and cases where no :TEST is given but the compiler can infer that
256 the element to search is of type (OR FIXNUM (NOT NUMBER)).
257 * optimization: better LOGNOT on fixnums.
258 * optimization: modular arithmetic for a particular requested width
259 is implemented using a tagged representation unless a better
260 representation is available.
261 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
262 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
263 for wellformedness and returns a specialized vector. (reported by
264 Francois-Rene Rideau)
265 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
266 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
267 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
268 unscheduling timer at the same time as another timer fires.
269 * bug fix: don't reschedule timers for dead threads.
270 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
271 * bug fix: copying output from RUN-PROGRAM to a stream signalled
272 bogus errors if select() was interrupted.
273 * enhancement: add support for fcntl's struct flock to SB-POSIX.
275 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
276 * enhancement: cleaner backtraces for interactive interrupts, as
277 well as other cases where the interesting frames used to be
278 obscured by interrupt handling frames.
279 * enhancement: untracing a whole package using (UNTRACE "FOO") is
280 now supported, and tracing a whole package using (TRACE "FOO") now
281 traces SETF-functions as well.
282 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
283 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
284 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
285 * unreadably printed representation of hash-tables now includes
287 * bug fix: partially fixed #188: type propagation from assignments
288 is now more efficient.
289 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
290 'DOUBLE-FLOAT) are not flushable.
291 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
292 full address of the object, and none of the tag bits.
293 * bug fix: readably printing hash-tables now respects other printer
294 control variables. (reported by Cedric St-Jean)
295 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
296 keyword with MAKE-HASH-TABLE.
297 * bug fix: export SB-POSIX:MKSTEMP.
298 * bug fix: SORT was not interrupt safe.
299 * bug fix: XREF accounts for the last node of each basic-block as
301 * bug fix: MAKE-INSTANCE optimizations interacted badly with
302 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
303 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
304 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
305 create obsolete instances in the case of redefinition or
306 obsoletion of a superclass. (thanks to Andy Hefner)
307 * bug fix: Support for the Alpha architecture has been revived; it had
308 suffered somewhat from lack of maintenance since sbcl-1.0.
309 * improvements to the Windows port:
310 ** The system detects the codepage to use at startup, instead of
311 using the value from the saved core. (thanks to Kei Suzuki)
313 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
314 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
315 (see documentation for details.)
316 * revived support for OpenBSD (contributed by Josh Elsasser)
317 * partially fixed bug #108: ROOM no longer suffers from occasional
318 (AVER (SAP= CURRENT END)) failures.
319 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
320 bodies no longer cause a WARNING to be signalled. (reported by
322 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
323 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
324 the bug in TRUNCATE.) (reported by Michael Weber)
325 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
326 no samples. (reported by Andy Hefner)
327 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
328 no longer appear as (NIL ...) frames in backtraces.
329 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
330 is now actually exported from SB-EXT as documented. (reported by
332 * bug fix: sb-aclrepl now correctly understands how to inspect
333 single-floats on 64-bit platforms where single-floats are not boxed.
334 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
335 yet been finalized. (reported by Levente Meszaros)
336 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
337 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
338 SBCL-specific optimize qualities.
340 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
341 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
342 an executable in the search path, and does so in the child
343 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
344 been removed; it can be found in the CVS history, for somebody who
345 needs that search behavior (see the manual).
346 * minor incompatible change: compiler policy re. weakening type
347 checks has changed: now type checks are weakened only if SAFETY < 2
349 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
350 unparsing of directory pathnames as files. Analogously,
351 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
352 filename to parse into a directory pathname.
353 * enhancement: implicit generic function creation now signals a
354 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
355 users can bind handlers for and muffle around calls to LOAD.
356 * enhancement: RUN-PROGRAM allows unicode arguments and environments
357 to be used (using the default stream external format), and allows
358 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
359 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
360 strings and vectors whose element-type or simplicity is not fully
361 known at compile-time.
362 * optimization: STRING-TRIM and related functions no longer allocate
363 a new string when no trimming needs to be performed. These functions
364 are also faster than before when the input string has been declared
366 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
367 * optimization: READ-LINE is significantly faster for files containing
369 * optimization: non-open coded uses of character comparison operators
370 (e.g. char=) no longer cons when called with more than one parameter
371 on platforms supporting dynamic-extent allocation.
372 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
373 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
374 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
376 * bug fix: some sequence functions elided bounds checking when
378 * bug fix: too liberal weakening of union-type checks when SPEED >
380 * bug fix: more bogus fixnum declarations in ROOM implementation
383 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
384 * new feature: MAKE-HASH-TABLE now experimentally accepts a
385 :SYNCHRONIZED argument, which makes the hash-table safe for
386 concurrent accesses (but not iteration.) See also:
387 SB-EXT:WITH-LOCKED-HASH-TABLE, and
388 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
389 * optimization: CONCATENATE on strings is an order of magnitude faster
390 in code compiled with (> SPEED SPACE).
391 * optimization: SUBSEQ is ~50% faster on lists.
392 * bug fix: bug 417 fixed -- source location reporting is now more robust.
393 * bug fix: SUBSEQ on a list will now correctly signal an error if if
394 END is smaller then START.
395 * bug fix: SB-PROFILE will no longer report extra consing for nested
396 calls to profiled functions.
397 * bug fix: ROOM implementation had bogus fixnum declarations which
398 could cause type-errors when calling ROOM on large images.
399 * bug fix: if file compilation is aborted, the partial fasl is now
400 deleted, and COMPILE-FILE returns NIL as the primary value.
401 * bug fix: number of thread safety issues relating to SBCL's internal
402 hash-table usage have been fixed.
403 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
404 be returned from its body when the values were being returned
405 using unknown-values return convection and the W-P-O was wrapped
406 inside an UNWIND-PROTECT.
407 * bug fix: sb-posix should now compile again under Windows, enabling
410 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
411 * incompatible change: hash-table accessor functions are no longer
412 automatically protected by locks. Concurrent accesses on the same hash-table
413 from multiple threads can give inconsistent results or even corrupt the
414 hash-table completely. Multi-threaded applications should do their own
415 locking at the correct granularity. In the current implementation it is
416 still safe to have multiple readers access the same table, but it's not
417 guaranteed that this property will be maintained in future releases.
418 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
419 and will signal an error at runtime.
420 * enhancement: SB-THREAD package now exports a semaphore interface.
421 * enhancement: CONS can now stack-allocate on x86 and
422 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
424 * enhancement: nested lists can now be stack allocated on
425 platforms providing stack allocation support.
426 * enhancement: dynamic-extent support has been extended to support
427 cases where there are multiple possible sources for the stack
429 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
430 if the mutex is uncontested on Linux.
431 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
432 as specified by the CLHS. (thanks to Tobias Rittweiler)
433 * bug fix: NaN comparison now works on x86-64.
434 * bug fix: CLOSE :ABORT T on a stream with pending output now
436 * bug fix: instances of non-standard metaclasses using standard
437 instance structure protocol sometimes missed the slot type checks
439 * bug fix: known functions can be inlined in high-debug code.
441 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
442 * minor incompatible change: the MSI installer on Windows no longer
443 associates .lisp and .fasl files with the installed SBCL.
444 * minor incompatible change: :UNIX is no longer present in *FEATURES*
445 on Windows. (thanks to Luis Oliviera)
446 * new feature: SB-CLTL2 contrib module now implements
447 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
448 * optimization: scavenging weak pointers is now more efficient,
449 requiring O(1) instead of O(N) per weak pointer to identify
450 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
451 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
452 method bodies using constant slot-names when the first argument is
453 a specializer parameter for the method.
454 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
455 names, when the class of the instance is a direct instance
456 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
457 as slow as the constant slot-name case.)
458 * optimization: member type construction is now O(N) instead
460 * optimization: UNION and NUNION are now O(N+M) for large
461 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
463 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
464 now more readable in environments like Slime which display it.
465 (thanks to Tobias C. Rittweiler)
466 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
467 was able to infer the correct argument type for the object on which
468 the CAS operation was being performed.
469 * bug fix: copy propagation interfered with parallel assignment
470 semantics in local calls. (reported by Paul Khuong)
471 * bug fix: the signed modular fixnum shift compiled to wrong code on
472 x86 and x86-64. (spotted by a slight modification to some of
475 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
476 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
477 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
478 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
480 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
481 and generic function now print the object identity for anonymous
482 instances. (thanks to Attila Lendvai)
483 * enhancement: as an extension to MOP, SBCL now supports SETF of
484 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
485 (thanks to Attila Lendvai)
486 * enhancement: sb-cover has an alternate annotation mode which puts
487 the form annotations at the car of the form, instead of annotating
488 it completely. The benefit of this mode is that it shows better
489 which forms were instrumented by the compiler.
490 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
491 by printer control variables. (reported by Dan Corkill)
492 * bug fix: system leaked memory when delayed output was performed by
493 the OS in smaller chunks then expected. (thanks to David Smith)
494 * bug fix: system leaked memory when file streams were not closed
496 * bug fix: large objects written to slow streams that were modified
497 after the write could end up with the modified state written to
498 the underlying file descriptor.
499 * bug fix: multiple threads operating in parallel on the same stream
500 could cause buffer-overflows.
501 * bug fix: source location information is stored correctly
502 (broken since 1.0.6). This bug would generally show up as the
503 Slime debugger highlighting the wrong form when the sldb-show-source
505 * bug fix: Branch forms are again annotated as branches in the sb-cover
507 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
508 on x86-64. (reported by Gregory Vanuxem)
509 * bug fix: EQUALP could return wrong results for structures with raw
510 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
511 word). (reported by Vjacheslav Fyodorov)
512 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
515 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
516 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
517 atomic compare-and-swap operations on threaded platforms.
518 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
519 allows assigning a global minimum value to optimization qualities
520 (overriding proclamations and declarations).
521 * enhancement: closed over variables can be stack-allocated on x86
523 * performance bug fix: GETHASH and (SETF GETHASH) are once again
525 * optimization: slot definition lookup is now O(1). This speeds up
526 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
527 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
528 * optimization: ASSOC and MEMBER can now be open-coded for all
529 combinations of keyword arguments when second argument is constant
530 and SPEED >= SPACE. In other cases a specialized version is
532 * bug fix: using obsoleted structure instances with TYPEP and
533 generic functions now signals a sensible error.
534 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
535 (reported by Kristoffer Kvello)
536 * bug fix: backtrace construction is now more careful when making
537 lisp-objects from pointers on the stack, to avoid creating bogus
538 objects that can be seen by the GC.
539 * bug fix: defaulting of values in contexts expecting more than 7
540 variables now works on x86-64. (reported by Christopher Laux)
541 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
543 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
544 as the property-list of a symbol.
545 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
546 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
547 situations requested, are once again file-compileable. (reported
550 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
551 * MOP improvement: support for user-defined subclasses of
552 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
553 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
554 the expansion of DEFMETHOD, is responsible for generating a form
555 which creates a list of specializers when evaluated. Additional
556 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
557 debugging and introspective support.
558 * minor incompatible change: the (unsupported) spinlock interface
559 has changed: free spinlock now has the value NIL, and a held spinlock
560 has the owning thread as its value.
561 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
562 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
564 * enhancement: name of a socket-stream is now "a socket" instead of
566 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
567 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
568 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
569 for generic functions. (Thanks to Tobias C. Rittweiler)
570 * optimization: bignum printing speed has been improved by 20-40%
571 (depending on the bignum size.)
572 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
574 * bug fix: the cache used by the CLOS to store precomputed effective
575 methods, slot offsets, and constant return values is now thread and
577 * bug fix: generic function dispatch function updating is now thread
578 and interrupt safe (in the sense that the known issues have been
580 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
581 * bug fix: interning EQL-specializers is now thread and interrupt safe.
582 * bug fix: asdf systems with dependencies to the SB-POSIX or
583 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
584 * bug fix: interrupt safety of applicable method computation has been
587 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
588 * new contrib: sb-cover, an experimental code coverage tool, is included
590 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
591 significantly faster.
592 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
593 produce more efficient code on x86-64 if the compiler can determine
594 that (+ INDEX <constant>) does not require a bounds check and FOO
595 has an element type at least 8 bits wide.
596 * enhancement: a new, experimental synchronous timeout facility is
597 provided. Refer to SB-SYS:WITH-DEADLINE for details.
598 * enhancement: when a symbol name conflict error arises, the
599 conflicting symbols are always printed with a package prefix.
600 (thanks to Kevin Reid)
601 * enhancement: stepping is now once again supported on the SPARC and
602 MIPS platforms. (It is also now more likely to work on CheneyGC
604 * enhancement: sb-sprof can now also track and report accurate call
606 * bug fixes: the treatment of non-standard subclasses of
607 SB-MOP:SPECIALIZER is more correct.
608 * incompatible change: PURIFY no longer copies the data from the
609 dynamic space into the static and read-only spaces on platforms
610 that use the generational garbage collector
611 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
613 * bug fix: GC race condition occasionally resulting in crashes with
614 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
616 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
618 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
619 a STYLE-WARNING to be signalled (regression from 1.0.4.)
620 * bug fix: an asynchronous interrupt could previously leave the
621 system running with GC inhibited.
622 * bug fix: a DECLARE form evaluated at top-level now causes an error
623 rather than silently (or verbosely) returning NIL.
624 * bug fix: trying to dissassemble functions compiled at high DEBUG could
625 result in a error being signalled due to source form lookup errors.
626 (reported by Peter Graves)
628 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
629 * incompatible change: removed writer methods for host-ent-name,
630 host-ent-addresses -- changing the values did not update the DNS
632 * minor incompatible change: changed experimental JOIN-THREAD interface
633 * documentation: the manual now lists reader and writer methods
634 in class slot documentation sections. (thanks to Richard M Kreuter)
635 * documentation: unwinding from asyncronous events has been
636 documented as unsafe.
637 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
638 in multithreaded application code.
639 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
640 platforms. (thanks to James Anderson for the optimization hint)
641 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
643 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
644 variants no longer cons.
645 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
646 their NOT- variants no longer cons.
647 * optimization: EQUAL hash tables no longer use SXHASH for objects
648 of all data types, but instead use an EQL hash for types for which
649 EQUAL is the same as EQL.
650 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
651 are significantly faster.
652 * optimization: new STRING-OUTPUT-STREAM implementation is much
653 faster for large outputs and conses 30% less on average.
654 * enhancement: XREF information is now collected for references made
655 to global variables using SYMBOL-VALUE and a constant argument.
656 * enhancement: SIGINT now causes a specific condition
657 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
658 * bug fix: FILE-POSITION used to signal an error in some cases where
659 ANSI requires it to return NIL.
660 * bug fix: ADJUST-ARRAY is now interrupt-safe.
661 * bug fix: adding and removing fd-handlers is now interrupt-safe.
662 * bug fix: inlined calls to C now ensure 16byte stack alignment on
664 * bug fix: bad type declaration in the CLOS implementation has
665 been fixed. (reported by James Anderson)
666 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
667 (thanks to Lutz Euler)
668 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
669 required by the CLHS. (reported by Samium Gromoff)
670 * bug fix: dead unbound variable references now signal an error.
671 (reported by Marco Monteiro)
672 * bug fix: / with an unused value was being deleted in safe code.
673 (thanks to Marco Monteiro and Kevin Reid)
674 * bug fix: number of characters that can be written onto a single
675 line in a file is unlimited.
676 * bug fix: some GC deadlocks caused by asynchronous interrupts have
677 been fixed by inhibiting interrupts for when GC is disabled.
678 * bug fix: some interrupt safety issues with GETHASH, (SETF
679 GETHASH), CLRHASH and REMHASH have been fixed.
680 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
681 specifier no longer causes infinite recursion.
682 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
683 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
684 * bug fix: an &environment argument with the correct variable information
685 is passed to macros that are expanded during byte compilation.
686 (reported by Samium Gromoff)
687 * bug fix: a logic error was causing the hash function for bignums to
688 have excessive amounts of collisions. (reported by Faré Rideau)
689 * bug fix: modifying the contents of an array could change the return
690 value of SXHASH on that array, which is only allowed for strings
691 and bit vectors. (bug introduced in 0.9.16)
692 * improvement: the x86-64/darwin port now passes all tests (except
693 for the debugger tests) but should still be considered
694 experimental until this is fixed.
695 * improvement: a style-warning is signaled for CASE (etc) clauses with
696 duplicate keys. (thanks to Kevin Reid)
697 * improvement: macos/x86 and macos/x86-64 now use mach exception
698 handlers for dealing with illegal instructions (for trapping and
699 error handling) and memory protection violations (for GC).
701 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
702 * new platform: experimental support for x86-64/darwin (MacOS).
703 * incompatible change: the thread-safe (on most platforms) getaddrinfo
704 and getnameinfo sockets functions are used instead of gethostbyaddr
705 and gethostbyname, on platforms where the newer functions are available.
706 As a result, the ALIASES field of HOST-ENT will always be NIL on these
708 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
709 don't print the startup banner, but behave as if --noinform was passed
710 as a command line argument. (thanks to Kevin Reid)
711 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
712 * optimization: code using alien values with undeclared types is much faster.
713 * optimization: the compiler is now able to open code SEARCH in more cases.
714 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
715 * bug fix: using standardized COMMON-LISP special variables as loop
716 variables no longer signals bogus package lock violations. (reported
718 * bug fix: declaring local loop variables to be of a range-limited type
719 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
720 (reported by Andras Simon)
721 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
722 bugs remain on x86-64.)
723 * bug fix: the #= and ## reader macros now interact reasonably with
724 funcallable instances.
725 * bug fix: type-checks for function arguments were compiled using the
726 compiler policy settings of the wrong lexical environment (in 1.0.2
728 * bug fix: SHADOW accepts characters as string designators, as required
729 by the spec (thanks to Eric Marsden)
730 * bug fix: fixed GC safety issues when foreign functions are called with
731 non-base strings as arguments
732 * bug fix: more consistent error output for fatal-compiler-warnings, like
734 * bug fix: functions with proper names were showing up as NIL in
735 backtraces (reported by Edi Weitz, regression in 1.0.2)
737 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
738 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
739 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
740 produce more efficient code on the x86 if the compiler can determine
741 that (+ INDEX <constant>) does not require a bounds check and FOO
742 has an element type at least 8 bits wide.
743 * bug fix: references to undefined variables are handled the same way
744 in toplevel forms as in the normal compiler
745 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
747 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
748 (thanks to Jon Buller)
749 * improvement: faster compilation times for complex functions
750 * improvement: added readlink support to SB-POSIX (thanks to Richard
753 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
754 * improvement: experimental support for mach exception handling on
755 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
757 * improvement: support for GBK external format.
758 (thanks to Chun Tian (binghe))
759 * improvement: the debugger now displays variables that have been closed
760 over, in code compiled with (DEBUG 2) or higher
761 * improvement: support for executable cores on NetBSD (thanks to
763 * new feature: added a RESTART-FRAME debugger command
764 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
765 be used to provide an implementation for FILE-POSITION on Gray streams
766 (thanks to Eric Marsden)
767 * new feature: add syslog, openlog and closelog support to SB-POSIX
768 (thanks to Richard Kreuter)
769 * optimization: the function call overhead in code compiled with
770 a high DEBUG optimization setting is significantly
771 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
772 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
773 (thanks to Tony Martinez)
774 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
775 evaluated once (reported by Kevin Reid)
776 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
777 works again (thanks to Kilian Sprotte)
778 * bug fix: an error is signaled for tagbodies with duplicate tags
779 (thanks to Stephen Wilson)
780 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
781 * bug fix: Win32 port can now handle foreign code unwinding Lisp
782 stack frames from alien callbacks.
783 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
784 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
785 65k pages (thanks to David Woodhouse)
786 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
788 changes in sbcl-1.0.1 relative to sbcl-1.0:
789 * new platform: FreeBSD/x86-64, including support for threading.
790 * new feature: the compiler stores cross-referencing information
791 about function calls (who-calls), macroexpansion (who-macroexpands)
792 and special variables (who-binds, who-sets, who-references) for code
793 compiled with (< SPACE 3). This information is available through the
794 sb-introspect contrib.
795 * new feature: users may subclass SEQUENCE, and have instances of
796 these classes interoperate with standard Common Lisp functions if
797 a number of methods are defined. (This feature is experimental
798 and the interface subject to change based on feedback from SBCL
799 users and the general community)
800 * improvement: sb-sprof traces call stacks to an arbitrary depth on
801 x86/x86-64, rather than the previous fixed depth of 8
802 * improvement: another pthread back-end of mutex "pthread-futex".
803 * bug fix: non-ascii command-line arguments are processed correctly
804 (thanks to Yaroslav Kavenchuk)
805 * bug fix: non-required arguments were not passed correctly when a method
806 defined using DEFMETHOD was called from a mop-generated method using
807 CALL-NEXT-METHOD (reported by Pascal Costanza)
808 * bug fix: recursion is now permitted in accessors through
809 SLOT-UNBOUND. (reported by Pascal Costanza)
810 * bug fix: an error was signaled at startup if the HOME environment
811 variable was defined, but had an empty value (reported by Peter Van Eynde)
812 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
813 signal a PROGRAM-ERROR, not a BUG.
814 * bug fix: SB-POSIX:READDIR works when built with large file support.
815 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
817 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
818 for better automatic indentation support. (thanks to Matt Pillsbury)
819 * optimization: loading generic functions no longer takes O(n^2) time,
820 proportional to the amount of methods in the generic function
821 (reported by Todd Sabin and Jeremy Brown)
822 * optimization: the FIND and POSITION family of sequence functions
823 are significantly faster on arrays whose element types have been
825 * improvements to the Windows port:
826 ** Intermittent heap corruption problems have been fixed. (thanks
827 to Alastair Bridgewater)
828 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
830 ** Lisp is able to unwind foreign exception frames from alien
831 callbacks. (thanks to Alastair Bridgewater)
833 changes in sbcl-1.0 relative to sbcl-0.9.18:
834 * improvement: experimental support for threading on FreeBSD/x86.
835 (thanks to NIIMI Satoshi)
836 * improvement: runtime option --dynamic-space-size can be used
837 to set the size of the dynamic space reserved on startup.
838 * improvement: floating point modes in effect are now saved in
839 core, and restored on startup.
840 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
841 startup, not time since first call to GET-INTERNAL-REAL-TIME.
842 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
843 threads are not running after *SAVE-HOOKS* have run.
844 * improvement: writes to CLOS instance slots are type-checked in code
845 compiled with (SAFETY 3)
846 * improvement: floating-point exception handling on FreeBSD (thanks to
848 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
849 (thanks to Zach Beane)
850 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
852 * improvement: added support for the Shift-JIS external format.
853 (contributed by NIIMI Satoshi)
854 * improvement: callbacks are supported on Linux/PPC. (thanks to
856 * bug fix: compiler bug triggered by a (non-standard) VALUES
857 declaration in a LET* was fixed. (reported by Kaersten Poeck)
858 * bug fix: file compiler no longer confuses validated and already
859 dumped structurres. (reported by Kaersten Poeck)
860 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
861 fill-pointer signals a type-error as required. (thanks to
863 * bug fix: disassemly of funcallable instances works.
864 * bug fix: single stepping on PPC.
865 * bug fix: fix thread-safety problems in the type system (generally
866 manifesting as nonsensical errors like "STRING is a bad type specifier
867 for sequences" or "The value 1 is not of type FIXNUM")
868 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
869 (reported by Josip Gracin)
870 * bug fix: an error is signaled for attempts to displace arrays with
871 incompatible element types (thanks to Mario Mommer)
872 * bug fix: more correct handling of wide characters in debug info
873 (bug reported by Attila Lendvai and fixed by Juho Snellman)
874 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
875 and don't cause extra consing
876 * optimization: MAP and MAP-INTO are significantly faster on vectors
877 whose elements types have been declared.
878 * Improvements to SB-SPROF:
879 ** Support for allocation profiling
880 ** Reduced profiling overhead, especially for long profiling runs
881 * Improvements to the Windows port:
882 ** floating point exceptions are now reported correctly.
883 ** stack exhaustion detection works partially.
884 ** more accurate GET-INTERNAL-REAL-TIME.
885 ** better RUN-PROGRAM behaviour in case of errors in the child
887 ** PROBE-FILE now simplifies pathnames correctly.
888 ** DIRECTORY now works correctly with :WILD-INFERIORS.
889 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
890 1st 1970 (although time timezone data will be unreliable).
891 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
893 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
894 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
895 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
897 * bug fix: two potential GC deadlocks affecting threaded builds.
898 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
899 returns T (reported by Anton Kazennikov)
900 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
901 with non-variable places
902 * bug fix: remove a race condition in the setting of
903 funcallable-instance functions, this should make threaded CLOS
904 code more stable against memory faults.
905 * bug fix: corruption of specials when unbinding is interrupted by an
906 asynchronous unwind (reported by Hannu Koivisto)
907 * improvement: the debugger will now also display local variables that
908 are only used once, for code compiled with a DEBUG optimization quality
911 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
912 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
913 * incompatible change: External-format support for FFI calls. The
914 SB-ALIEN:C-STRING no longer implies an ASCII
915 external-format. Instead the string is subject to external-format
916 conversion. Additionally return values of type SB-ALIEN:C-STRING
917 are by default of type (SIMPLE-ARRAY CHARACTER), not
918 SIMPLE-BASE-STRING. For an alien type that behaves like the old
919 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
920 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
921 * incompatible change: SB-EXT package no longer contains the
922 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
923 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
925 * incompatible change: the single-stepper is no longer available
926 on Alpha, Mips and Sparc platforms.
927 * minor incompatible change: the direct superclasses of
928 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
929 not (STANDARD-OBJECT FUNCTION). This makes the
930 class-precedence-lists of GENERIC-FUNCTION and
931 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
933 * minor incompatible change: the default stream external format on
934 non-unicode SBCL is no longer detected from the locale, but is always
936 * new feature: Add a version of evaluator that uses an interpreter instead
937 of the compiler. EVAL still uses the compiler by default, to switch it
938 to use the interpreter, set the value of the variable
939 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
940 * minor incompatible change: the single-stepper REPL has been merged
941 with the normal debugger (see the "Stepping" heading of the debugger help
942 for more details). The debugger command STEP will no longer switch
943 to the single-stepper REPL.
944 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
945 for a type now works.
946 * bug fix: loading of default sysinit file works. (thanks to Leonid
948 * bug fix: better detection of circularities in the file-compiler.
949 (reported by Marco Monteiro)
950 * bug fix: the CL pathname functions now work with files that have
951 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
952 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
953 systems with Unix98 pty semantics.
954 * bug fix: ASDF-INSTALL will now work with bsd tar.
955 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
957 * bug fix: timers expiring in dead threads no longer cause a
958 type-error (reported by Paul "Nonny Mouse").
959 * bug fix: thanks to more lightweight single-stepper instrumentation,
960 code compiled with (DEBUG 3) will compile and execute significantly faster,
961 and will have more accurate type-inferencing than before
962 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
963 whose bindings are modified
964 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
965 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
966 ** replace dummy implementations of CL:MACHINE-INSTANCE and
967 CL:SOFTWARE-VERSION with proper version
969 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
970 * feature: implemented the READER-METHOD-CLASS and
971 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
972 as specified by AMOP.
973 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
975 * optimization: faster LOGCOUNT implementation on x86 and x86-64
976 (thanks to Lutz Euler)
977 * optimization: hashing of general arrays and vectors has been
978 improved. (reported by Any Fingerhut)
979 * enhancement: SB-INTROSPECT is now able to find definitions of
980 profiled functions. (thanks to Troels Henriksen)
981 * enhancement: compiler-macro expansion applies now to FUNCALL forms
982 as well, allowing compiler-macros for SETF-functions to expand.
983 * enhancement: step-instrumentation no longer wraps known
984 single-value functions with multiple-value context, allowing
985 better type inference.
986 * fixed bug #337: use of MAKE-METHOD in method combination now works
987 even in the presence of user-defined method classes. (reported by
988 Bruno Haible and Pascal Costanza)
989 * fixed bug #339(c): if there are applicable methods not part of any
990 long-form method-combination group, call INVALID-METHOD-ERROR.
991 (reported by Bruno Haible)
992 * fixed bug #361: the :FUNCTION initarg in the protocol for
993 initialization of methods can now be used to override
994 internally-produced optimized functions. (reported by Bruno
996 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
997 system-provided lambda expression no longer cause warnings about
998 unbound #:|pv-table| symbols.
999 * bug fix: improved the handling of type declarations and the
1000 detection of violations for keyword arguments with non-constant
1002 * bug fix: potentially erronous calls to PATHNAME and
1003 MERGE-PATHNAMES were being flushed in some cases.
1004 (reported by Richard Kreuter)
1005 * bug fix: compiled calls to TYPEP were mishandling obsolete
1006 instances. (reported by James Bielman and Attila Lendvai)
1007 * bug fix: format strings with ~> without matching ~< no longer
1008 trigger an AVER, but signal an understandable error instead.
1009 (reported by Antonio Martinez)
1010 * bug fix: specifying an output-file without a pathname-type for
1011 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1012 of input-file instead of "fasl". (reported by Robert Dodier)
1013 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1014 (reported by James Y Knight).
1015 * bug fix: compiler-macros-function did not consider the environment
1016 argument for shadowing by local functions.
1017 * bug fix: compiler-macros expansion was inhibited by local INLINE
1019 * bug fix: inline expansions of known functions were subject to
1020 step-instrumentation in high debug policies, leading to problems
1021 with type-inference.
1022 * bug fix: compiler failed to differentiate between different CONS
1023 types in some cases.
1024 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1025 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1026 * bug fix: #368: incorrect use of expressed vs. upgraded array
1028 * bug fix: #306a: more precise unions of array types.
1029 * thread-safety improvements:
1030 ** CONDITION-WAIT could return early on Linux, if the thread was
1031 interrupted and subsequently continued with SIGCONT.
1032 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1034 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1035 * added support for the ucs-2 external format. (contributed by Ivan
1037 * minor incompatible change: pretty printing of objects of type
1038 (cons symbol) is, in the default pprint-dispatch-table, now
1039 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1041 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1042 called later in a class's lifetime, possibly as late as when the
1043 first instance of the class is created. Previously,
1044 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1045 class became finalizeable.
1046 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1047 for multibyte external-format streams. (thanks to "vbzoli")
1048 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1049 with :ALLOCATION :CLASS inherited from superclasses of the
1051 * fixed bug: anonymous classes can now be created using the :NAME
1052 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1053 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1054 * fixed bug: core-files saved with :EXECUTABLE T can again be
1055 executed when SBCL_HOME isn't set. (reported by James Knight)
1056 * fixed bug: toplevel LOCALLY forms with declarations could
1057 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1058 * fixed bug: printing from several different threads using different
1059 values of *print-case* could cause invalid output, due to
1060 some internal special variables of the printer not being bound
1061 thread-locally (reported by Max Mikhanosha)
1062 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1063 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1064 (reported by Pascal Costanza)
1065 * fixed bug: FINALIZE-INHERITANCE is called from
1066 REINITIALIZE-INSTANCE on classes when the class has previously
1067 been finalized, as required by AMOP.
1068 * minor code generation optimizations:
1069 ** better register allocation in CLOS dispatching functions
1070 ** overflow detection when coercing signed bytes to fixnums on x86-64
1071 ** is now implemented with one IMUL instruction instead of three shifts
1072 ** more efficient bit-vector access on x86 and x86-64
1073 ** more efficient access to raw structure slots on x86 and x86-64
1074 * fixed some bugs revealed by Paul Dietz' test suite:
1075 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1076 return its argument.
1078 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1079 * feature: thread support on Solaris/x86, and experimental thread support
1081 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1083 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1084 default method for SLOT-UNBOUND.
1085 * minor incompatible change: prevent the user from specializing the
1086 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1087 somewhat counter to the intent of the protocol, I (CSR) think, and
1088 additionally it just doesn't work in SBCL as currently
1089 implemented, thanks to optimizations (that are always valid for
1090 the other three SLOT-VALUEish functions, but not for the setter).
1091 * minor incompatible change: the :SB-LDB feature is now enabled by
1092 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1093 the low-level debugger.
1094 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1095 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1096 down to an SB-SYS:FD-STREAM, allowing direct communication in
1097 more cases. Temporary files and pipes are still used as a fallback
1099 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1100 on the non-reentrant C functions localtime(3) and gmtime(3).
1101 * bug fix: global optimization policy was not visible in LOCALLY and
1103 * bug fix: class objects can be used as specializers in methods.
1104 (reported by Pascal Costanza)
1105 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1106 with a type error. (reported by blitz_ on #lisp)
1107 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1108 failed with a type error. (reported by Pascal Bourguignon)
1109 * bug fix: merging pathnames against defaults with :DIRECTORY
1110 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1111 (reported by James Y Knight)
1112 * bug fix: saving large (>2GB) cores on x86-64 now works
1113 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1114 constant index and a (simple-array (signed-byte 32)) array
1115 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1116 error (patch by Robert J. Macomber)
1117 * feature: TIME now displays the amount of run-time spent in GC
1118 * bug fix: The debugger now does a better job of respecting
1119 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1120 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1121 workaround for bug 403.)
1122 * bug fix: floating point exception handling now works on FreeBSD
1123 (thanks to NIIMI Satoshi)
1124 * fixed some bugs revealed by Paul Dietz' test suite:
1125 ** MISC.641: LET-conversion were not supposed to work in late
1127 * improvements to the Win32/x86 port:
1128 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1129 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1131 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1132 * new feature: source path information is generated for macro-expansion
1133 errors for use in IDE's like Slime (thanks to Helmut Eller)
1134 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1135 * bug fix: compiler does not loop forever on an invalid type in
1137 * improvement: compilation of most CLOS applications is significantly
1139 * optimization: added a limited bytecode compiler for simple toplevel
1140 forms, speeding up compilation and FASL loading
1141 * bug fix: the statistical profiler now properly distinguishes anonymous
1144 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1145 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1146 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1147 into subclasses of the finalized class. Now user calls to
1148 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1149 system return before any subclasses are finalized.
1150 * minor incompatible change: The reader no longer ignores errors
1151 regarding non-existent packages in #+ and #- feature tests.
1152 * new feature: command line options --no-sysinit, --no-userinit to
1153 inhibit loading the corresponding init files
1154 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1155 for sending data through UDP sockets (thanks to François-René Rideau)
1156 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1157 error is signalled instead (thanks to François-René Rideau)
1158 * minor incompatible change: Error signalling behaviour of lexical
1159 operations violating package locks has changed slightly. Refer to
1160 documentation on package locks for details.
1161 * enhancement: EVAL can process IF-expressions without resorting to the
1163 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1164 constant-folded properly due to missing out-of-line functions.
1165 (reported by Mika Pihlajamäki)
1166 * bug fix: LISTEN sometimes returned T even in cases where no data was
1167 immediately available from the stream
1168 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1169 were reversed. (reported by Levente Mészáros)
1170 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1171 when given a vector displaced to an adjustable array. (reported
1173 * bug fix: derivation of float boundaries from numbers outside the
1174 appropriate float range (reported by John Wiseman)
1175 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1177 * bug fix: precomputing cache entries for generic functions with
1178 some subclasses of specializers as yet invalid does not attempt to
1179 fill a cache line with a negative offset. (reported by Levente
1181 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1182 allow condition class objects as arguments to DOCUMENTATION and
1183 (SETF DOCUMENTATION); only find and set documentation for
1184 structure names for the STRUCTURE doc-type. (suggested by Gary
1186 * improvements to the Win32/x86 port:
1187 ** (user-homedir-pathname) and default initialization file
1188 locations now know about the user's "Documents and Settings"
1189 directory (thanks to Yaroslav Kavenchuk)
1190 ** run-program is implemented (thanks to Mike Thomas)
1191 ** sockets support (thanks to Timothy Ritchey)
1192 ** better backtrace support (thanks to Alastair Bridgewater)
1193 ** sb-grovel supported
1194 ** asdf-install and sb-posix work somewhat
1195 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1196 * improvements to the Solaris/x86 port:
1197 ** works on Solaris 11/Solaris Express
1198 ** floating-point exception handling support
1199 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1200 * fixed some bugs revealed by Paul Dietz' test suite:
1201 ** REMOVE-METHOD returns its generic function argument even when
1202 the method is not one of the generic functions' methods.
1203 ** objects declared as MEMBER types can be admissible arguments to
1204 structure accessors.
1205 ** printing characters should simply be printed by the FORMAT ~:C
1207 ** compiler failure when compiling functions with hairy constant
1208 defaults for optional parameters.
1209 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1210 function, which is already optimized.
1212 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1213 * new platform: experimental support for SBCL x86/Darwin, including
1214 MacOS X 10.4.5 on Intel.
1215 * new feature: Unicode character names are now known to the system
1216 (through CHAR-NAME and NAME-CHAR).
1217 * new feature: the filesystem location of SBCL's core file is
1218 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1219 * minor incompatible change: the contrib modules SB-POSIX and
1220 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1221 this change is to make it easier to distribute
1222 location-independent binaries.
1223 * bug fix: as implied by AMOP, standardized classes no longer have
1224 slots named by external symbols of public packages. (reported by
1226 * optimization: calling functions via constant symbols -- as in
1227 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1228 via the function object as in (FUNCALL #'FOO).
1229 * optimization: on x86 and x86-64, the calling convention for the
1230 case of dealing with an unknown number of values has been altered
1231 to be friendlier to the prediction heuristics implemented,
1232 particularly with reference to CALL and RET pairing. (thanks to
1233 Alastair Bridgewater)
1234 * optimization: on x86, the code for access to thread-local symbol
1235 values and binding/unbinding of thread-local symbols is smaller.
1236 (thanks to Alastair Bridgewater)
1237 * enhancement: CONSTANTP is now able to determine constantness of
1238 more complex forms, including calls to constant-foldable standardized
1239 functions and some special forms beyond QUOTE.
1240 * fixed bug: occasional GC crashes on Solaris/x86
1241 * optimization: x86-64 supports stack allocation of results of simple
1242 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1243 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1244 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1245 (thanks to James Knight)
1246 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1247 by the standard (thanks to Stephen Compall)
1249 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1250 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1251 be used for bundling the runtime and the core file into one
1252 executable binary. This feature is not currently supported on all SBCL
1253 platforms. (thanks to James Bielman and NIIMI Satoshi)
1254 * new feature: a generational or ephemeral garbage collector is now
1255 the default on the PowerPC platform (both Linux and Darwin). The
1256 old Cheney (stop and copy) collector is a build-time option.
1257 (thanks to Cyrus Harmon, after Raymond Toy)
1258 * minor incompatible change: the method by which SBCL finds its
1259 contributed modules has changed; it no longer relies on symbolic
1260 links from an $SBCL_HOME/systems directory, but searches directly
1261 in the subdirectories of $SBCL_HOME.
1262 * enhancement: the dynamic heap size on the Linux/PPC platform is
1263 markedly larger, even using the older Cheney garbage collector.
1264 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1265 certain complicated string types. (reported by Gary King)
1266 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1267 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1268 * fixed bug: DOCUMENTATION on structures no longer signals an
1269 error if no structure type of the right name exists. (reported by
1271 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1272 applicable non-standard primary or :AROUND method on
1273 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1274 no longer cause unbound variable errors. (reported by Kevin Reid)
1275 * fixed bug: metacircle resolution in cases where methods have slots
1276 added before the slots from STANDARD-METHOD. (reported by Jean
1278 * fixed bug: the Power PC assembly code for calling into lisp
1279 sometimes computed its return address wrongly (depending on the
1280 15th bit of the address, fixed at link time). This used to
1281 manifest itself in a segmentation violation while building PCL.
1282 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1283 many others over the years)
1284 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1285 the stack pointer (treated as a signed integer) is negative.
1286 (thanks to Peter van Eynde)
1288 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1289 * new platform: experimental support for the Windows operating
1290 system has been added. (thanks to Alastair Bridgewater)
1291 * fixed several bugs in and robustified the PPC FFI (including
1292 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1293 * bug fix: allow non-simple string symbol names (reported by Paul
1295 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1297 * bug fix: saving a core corrupted callbacks on x86/x86-64
1298 * bug fix: closed a loophole in metacircularity detection and
1299 grounding in the PCL implementation of CLOS.
1300 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1301 (reported by tomppa on #lisp)
1302 * bug fix: clear the direction flag on Lisp -> C transitions, as
1303 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1304 (reported by Andrej Grozin and Hendrik Maryns)
1305 * optimization: major improvements to GC efficiency on GENCGC platforms
1306 * optimization: faster implementation of EQUAL
1307 * optimization: emit more efficient opcodes for some common
1308 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1309 * optimization: several other minor code-generation improvements on x86-64
1310 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1312 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1313 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1314 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1315 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1316 by ANSI as a generic function, and for consistency (SETF
1317 GENERIC-FUNCTION-NAME) is treated likewise.
1318 * fixed bug #233.b: make constraint propagation notice when a variable
1319 value is changed after it is referenced but before it is used
1320 * fixed bug #296: no more arbitrary behaviour from filenames with
1321 odd characters as --load arguments. (reported by Adam Warner)
1322 * fixed bug #390: :CHARACTER-SET pathname components now work as
1323 expected. (reported by Tim Daly Jr)
1324 * fixed bug #391: complicated :TYPE intersections in slot
1325 definitions no longer cause an error in PCL internals.
1326 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1327 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1328 returning the number of octets which would be written to the
1329 file-stream. (thanks to Robert J. Macomber)
1330 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1331 arguments to the internal stream functions. (thanks to David
1333 * bug fix: stack exhaustion is now handled on the x86/SunOS
1334 platform. (thanks to Daisuke Homma)
1335 * bug fix: more accurate ROOM results on GENCGC platforms
1336 * optimization: improved type inference for arithmetic-for
1337 index variables in LOOP
1338 * optimization: faster floating-point SQRT on x86-64
1339 * fixed some bugs revealed by Paul Dietz' test suite:
1340 ** DOCUMENTATION returns NIL instead of "" for method combinations
1341 that don't have a docstring
1343 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1344 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1345 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1346 therefore are not customizeable by user code (as seems to be at
1347 least permitted and maybe required by AMOP). As a consolation,
1348 however, the SBCL implementation of these functions now calls
1349 REINITIALIZE-INSTANCE as specified by AMOP.
1350 * bug fix: slot-definition documentation is propagated as per ANSI
1351 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1352 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1353 argument, as do eslotd initialization methods. (from Pascal
1354 Costanza's "Closer" project)
1355 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1356 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1358 * bug fix: it is now possible to have more than one subclass of
1359 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1360 (reported by Bruno Haible, Pascal Costanza and others)
1361 * bug fix: the dependent update protocol now works for generic
1362 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1363 and Pascal Costanza)
1364 * bug fix: condition-class instances corresponding to
1365 DEFINE-CONDITION forms are now created eagerly. (reported by
1366 Kalle Olavi Niemitalo on comp.lang.lisp)
1367 * bug fix: floating point printing is more accurate in some
1368 circumstances. (thanks to Simon Alexander)
1369 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1370 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1371 * bug fix: callbacks on OS X now preserve stack-alignment.
1372 (thanks to Cyrus Harmon)
1373 * enhancement: the x86-64 disassembler is much better at
1374 disassembling SSE instructions. (thanks to Lutz Euler)
1375 * enhancement: CHECK-TYPE on a variable now allows type-inference
1376 (thanks to Vincent Arkesteijn)
1377 * optimization: improved performance of EUC-JP external format.
1378 (thanks to NIIMI Satoshi)
1379 * optimization: performance improvements to IO on file streams of
1380 :ELEMENT-TYPE CHARACTER
1381 * optimization: much faster memory allocation on x86-64
1382 * optimization: faster garbage collections (latency and throughput)
1384 * optimization: faster bignum subtraction and division on x86 and x86-64
1385 (thanks to Lutz Euler)
1386 * optimization: more accurate type-inference for loops that use a
1387 floating point index variable or a negative step.
1389 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1390 * bug fix: add a workaround to SBCL looping infinitely at startup on
1391 Linux kernels with apparently buggy implementations of personality().
1392 (thanks to Svein Ove Aas)
1393 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1394 (thanks to David Lichteblau)
1395 * bug fix: local bindings shadow global symbol macros for
1396 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1397 * optimization: non-open coded uses of numeric comparison operators
1398 (e.g. >) no longer cons when called with more than one parameter
1399 on platforms supporting dynamic-extent allocation.
1400 * enhancement: saving cores with foreign code loaded is now supported
1401 on MIPS/Linux in addition to the previously supported platforms.
1402 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1403 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1404 * bug fix: division by zero in sb-sprof when no samples were collected
1405 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1406 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1407 (thanks to Alexander Kjeldaas)
1408 * new feature: ignoring the return values of destructive functions
1409 like NREVERSE signals a compile-time style-warning.
1410 (thanks to Kevin Reid)
1411 * new feature: the alignment of alien structure fields can be
1412 explicitly specified. (thanks to Cyrus Harmon)
1413 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1414 and dump core on SIGQUIT
1416 ** incompatible change: threads do not inherit values of specials
1417 from their parents (see manual)
1418 ** bug fix: threads stacks belonging to dead threads are freed by the
1419 next exiting thread, no need to gc to collect thread stacks anymore
1420 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1421 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1422 ** bug fix: dynamic variable and thread start related gc lossage
1423 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1425 * fixed some bugs revealed by Paul Dietz' test suite:
1426 ** SUBTYPEP is slightly more accurate on heinously complicated
1427 CONS types where some of the members have uncertain (in the
1428 NIL, NIL sense) type relationships to each other.
1430 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1431 no longer purified unless :PURIFY T is explicitly specified.
1432 ** Non-purified cores are significantly smaller than before
1434 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1435 * new feature: timers based on Zach Beane's excellent timer package
1436 * added support for the following external formats: koi8-u,
1437 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1438 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1439 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1440 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1441 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1442 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1443 * incompatible change: a threaded SBCL will no longer revert to
1444 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1445 * bug fix: interrupts are disabled until startup is complete; no
1446 more sigsegvs when receiving a signal to soon
1447 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1449 * bug fix: add a workaround for the memory randomization features in
1450 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1451 workaround will only be in effect on systems with the proc filesystem
1453 * bug fix: printing objects of type HASH-TABLE signals a
1454 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1456 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1457 have been running for over 50 days. (reported by Gilbert Baumann)
1458 * bug fix: the logic for getting names of functions gets less
1459 confused when confronded with alternate-metaclass
1460 funcallable-instances. (reported by Cyrus Harmon)
1461 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1462 consistent, even on internal alternate-metaclass objects.
1463 * bug fix: flush closure information collected by physical
1464 environment analysis prepass before the main pass. (bug reported
1466 * bug fix: compiler pack phase does not modify a hashtable, which is
1467 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1468 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1469 may be expanded. (reported by Alexander <malishev> on c.l.l)
1470 * bug fix: no more style-warnings when compiling code using LAST
1471 under high-SPEED low-SPACE settings. (reported by David Wallin
1473 * fixed some incompatibilities between SBCL's MOP and the MOP
1475 ** the METAOBJECT class is now implemented;
1476 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1477 STANDARD-OBJECT, as required;
1478 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1479 now compatible (as required by VALIDATE-SUPERCLASS); there
1480 remains a consistency requirement over the presence or absence
1481 of the FUNCTION class in the superclass list of finalized
1482 classes; see the manual for more details;
1483 ** the :AROUND method for COMPUTE-SLOTS on
1484 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1485 requested slot ordering.
1487 ** bug fix: parent thread now can be gc'ed even with a live
1489 ** bug fix: binding a special with PROGV to no value is not
1490 the same as that symbol not having been bound (thanks to
1492 ** bug fix: binding specials is thread safe (thanks to
1494 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1495 and SB-SPROF do not die with 'no handler for signal XX in
1496 interrupt_handle_now(..)' anymore
1497 ** bug fix: WITH-TIMEOUT works with multiple running threads
1498 * fixed some bugs revealed by Paul Dietz' test suite:
1499 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1500 the :method-class keyword argument.
1502 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1503 * new port: the Solaris operating system on x86 processors is now
1504 mostly supported, though some rough edges in the environment
1505 remain. (thanks to Daisuke Homma)
1506 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1507 space for its heap. (thanks to Thiemo Seufer)
1508 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1509 closures. (thanks to Thiemo Seufer)
1510 * minor incompatible change: eof selects abort in the debugger.
1511 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1512 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1514 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1515 deprecated, and will go away in a future revision of SBCL.
1516 * minor incompatible change: GC-ON and GC-OFF are no longer
1517 implemented with a counter, it does not matter how many times gc
1518 is switched on or off
1519 * bug fix: discriminating functions for generic function classes
1520 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1521 make invalid assumptions about method precedence order. (reported
1523 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1524 platforms. (thanks to Thiemo Seufer)
1525 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1526 * bug fix: correct pathname printing with printer escaping is on.
1527 (thanks to Kevin Reid)
1528 * bug fix: complex VOP definitions in "user-space" no longer trigger
1529 package locks. (reported by Zach Beane)
1530 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1531 for funcallable-instance functions, and is no different from
1533 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1534 characters in the current readtable. (reported by Nicholas Neuss)
1535 * bug fix: SUBTYPEP on various CONS types returns more a more
1536 accurate acknowledgment of its certainty.
1537 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1538 lists in some cases. This partially fixes bug 384.
1539 * flush all standard streams before prompting in the REPL and the
1541 * bug fix: signal handling and triggering gc do not conflict
1542 directly anymore, in particular a high frequency sb-sprof does
1543 not prevent gc from running
1544 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1545 approximation for timezone and DST information between the
1546 universal time epoch and the smallest negative 32-bit time_t.
1547 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1548 year 1899 so long as the encoded time is non-negative.
1549 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1550 the correct answer for non-constant rotate amounts when performing
1551 an inline 32-bit rotation.
1553 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1554 there is only one thread in the session
1555 ** bug fix: memory leak for streams created in one thread and
1556 written to in another
1557 ** bug fix: lockup when compiled with gcc4
1558 ** bug fix: race that allows the gc to be triggered when gc is
1560 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1561 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1563 ** bug fix: finished threads are gc'ed properly
1564 * fixed some bugs revealed by Paul Dietz' test suite:
1565 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1566 has arguments with a different set of applicable methods from
1567 the orignal arguments.
1568 ** The type error thrown by MAP now has a correct expected-type
1570 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1571 name a compiled function.
1572 ** Three MISC tests where a large bignum was improperly coerced to
1573 a float that couldn't represent that bignum during type
1574 derivation were fixed.
1575 ** SUBTYPEP can now handle types involving the negation of a
1576 list-form FUNCTION type.
1577 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1578 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1579 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1581 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1582 * New feature: Experimental support for bivalent streams: streams
1583 opened with :element-type :default now allow character and binary
1584 (unsigned-byte 8) I/O
1585 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1586 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1588 * bug fix: no more highly sporadic "couldn't check whether ~S is
1589 readable" when reading a stream and an interrupt hits in the middle
1590 of a select system call
1591 * compiler better recognizes complex arrays (reported by Eduardo
1593 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1594 for variables on SBCL built with linkage-tables. (reported by Luis
1596 * various error reporting improvements.
1597 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1598 (Thanks to Thiemo Seufer)
1599 * bug fix: nested reader invokations work correctly
1600 * bug fix: it is possible to have simultaneous references to foreign
1601 code and foreign data with the same name.
1603 ** added x86-64 support
1604 ** incompatible change: the threading api now works with thread
1605 objects instead of thread ids
1606 ** bug fix: threads are protected from signals and interruption when
1607 starting up or going down
1608 ** bug fix: a race where an exiting thread could lose its stack to gc
1609 ** bug fix: don't halt on infinite error in threads if possible
1610 ** fixed numerous gc deadlocks introduced in the pthread merge
1611 ** bug fix: fixed thread safety issues in read and print
1612 ** bug fix: debugger doesn't hang on session lock if interrupted at
1613 an inappropriate moment
1614 ** bug fix: run-program is now thread safe(r)
1615 ** bug fix: inner with-recursive-lock no longer releases the mutex
1616 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1617 * fixed some bugs revealed by Paul Dietz' test suite:
1618 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1619 EXPECTED-TYPE slots filled.
1620 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1622 ** MAKE-CONDITION accepts classes as type-designators.
1623 ** COMPILE may never return NIL.
1624 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1625 range before calling Unix time functions
1627 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1628 * numerous signal handling fixes to increase stability
1629 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1630 * minor incompatible change: we now correctly canonize default
1631 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1632 as per the MOP, rather than the historical (INITARG INITFUNCTION
1633 INITFORM). (reported by Bruno Haible)
1634 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1635 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1636 for more information.
1637 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1638 pathname is a directory pathname.
1639 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1640 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1642 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1643 :METACLASS argument in addition to a class name. (reported by
1644 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1645 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1646 (reported by Bruno Haible; patch from Gerd Moellmann)
1647 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1649 * bug fix: more cleanups to the floating point exception handling on
1650 x86-64 (thanks to James Knight)
1651 * bug fix: the compiler does not try to constant fold calls of
1652 COERCE and BIT-* functions when they return freshly constructed
1653 objects. (reported by Nikodemus Siivola and Paul Dietz)
1654 * optimization: improved the MIPS versions of generic (in the
1655 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1656 * optimization: direct conversion of (unsigned-byte 32) to floats on
1657 the PowerPC platform.
1658 * optimization: structure instances with raw slots now use less
1659 memory, and probably show better memory locality. (thanks to
1661 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1662 are treated as requests for stack allocation on the x86-64,
1663 Alpha-32, PPC and SPARC platforms.
1664 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1665 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1666 (thanks to Tony Martinez)
1668 ** gcing a dead thread can no longer lead to lockups
1669 ** threads block signals until they are set up properly
1670 ** errno is no longer shared by threads
1671 ** interrupt-thread restores the eflags register on x86
1672 ** fixed some lockups due to gc/thread interaction
1673 * fixed some bugs revealed by Paul Dietz' test suite:
1674 ** invalid dotted lists no longer raise a read error when
1675 *READ-SUPPRESS* is T
1676 ** signal an error if a symbol that names a declaration is used
1677 as the name of a type, or vice versa
1678 ** allow using the (declare (typespec var*)) abbreviation for
1679 (declare (type typespec var*)) with all type specifiers
1680 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1681 ** always return NIL from PPRINT-POP when OBJECT is NIL
1682 ** don't signal errors when pretty-printing malformed LABELS,
1683 FLET or MACROLET forms
1684 ** declarations in a DOLIST body are also in scope for the
1686 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1688 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1691 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1692 * fixed cross-compiler leakages that prevented building a 32-bit
1693 target with a 64-bit host compiler.
1694 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1695 opened with :IF-EXISTS :OVERWRITE.
1696 * fixed bug 281: error for an invalid qualifier in a short-form method
1697 combination method is not signalled until the faulty method is called.
1698 * bug fix: iteration variable type inferrer failed to deal with open
1699 intervals. (reported by Alan Shields)
1700 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1701 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1702 or more runtime options were provided to the sbcl binary.
1703 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1704 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1705 * optimization: there's now a fast-path for fixnum arguments in the
1706 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1707 * optimization: the code generated on x86-64 is more compact thanks
1708 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1709 * bug fix: floating-point exception handling now partly works on
1710 x86-64. (Thanks to James Knight)
1711 * improvement to the MIPS backend from Thiemo Seufer: C-style
1712 64-bit long long arguments and return values to alien functions
1714 * fixed some bugs revealed by Paul Dietz' test suite:
1715 ** the type-error signalled from WARN has a filled-in DATUM slot.
1716 ** the type-error required when a stream is not associated with
1717 a file has the stream as its datum.
1718 ** type-errors on single-floats on x86-64 no longer have
1719 :INVALID-OBJECT as the datum
1720 ** the type-errors signalled for invalid function names now have
1721 a correct expected type
1722 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1723 for structures defined with a :TYPE. Documentation strings for
1724 typed structures are no longer immediately discarded
1725 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1726 broadcast streams and synonym streams. FILE-LENGTH now also works
1727 on broadcast streams.
1729 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1730 * incompatible change: the --noprogrammer option, deprecated since
1731 version 0.7.5, has been removed. Please use the equivalent
1732 --disable-debugger option instead.
1733 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1735 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1736 inoperational for a while now) has been completely removed.
1737 * null lexical environments are now printed as #<NULL-LEXENV>,
1738 significantly reducing the amount of clutter in typical backtraces.
1739 * documentation on weak pointers, finalization, and after GC hooks
1740 has been added to the manual.
1741 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1742 as other specialized array types, is much faster. SUBSEQ and
1743 COPY-SEQ on such arrays have also been sped up.
1744 * optimization: EQL is now more efficient when at least other argument
1745 is known to be of type (OR FIXNUM (NOT NUMBER)).
1746 * fixed bug: the runtime is now less vulnerable to changes in the
1747 size of the SBCL object on OS X, and virtual memory is reserved for
1748 all spaces that need to be at a fixed address.
1749 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1750 * fixed bug: finalizers and after GC hooks that cause consing are now
1752 * fixed bug: compiler error messages and summaries are now printed to
1753 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1754 * fixed inference of the upper bound of an iteration variable.
1755 (reported by Rajat Datta).
1756 * fixed bug 211e: calling local functions with duplicated constant
1757 keyword argument no longer causes a bogus style warning about an
1759 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1760 ftype declaration to be disregarded. (reported by Dave Roberts)
1761 * fixed bug 373: caused by erronous compilation of references to alien
1762 variables in the runtime on ppc/darwin.
1763 * fixed bug 376: CONJUGATE type deriver.
1764 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1765 (reported by Baughn on #lisp)
1766 * a cross-compiler bug on non-x86 platforms has been identified and
1767 fixed. (thanks to Bruno Haible)
1768 * improvements to the MIPS runtime code for increased stability.
1769 (thanks to Thiemo Seufer)
1770 * increased the maximimum compact environment size to allow
1771 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1772 * partial workaround for bug 135: don't name the function we're
1773 calling for hairy cases of EVAL, so as not to accumulate one environment
1774 entry per call to EVAL. (reported by Kevin Reid)
1775 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1776 * optimization: type testing for non-vector arrays should be faster.
1777 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1779 * bug fix: setting 31st element of a bit vector to zero did not work
1781 * bug fix: redefining a class definition which failed due to a
1782 previous accessor / function clash now works (but see BUGS entry
1783 #380 for more problems in this area). (thanks to Zach Beane)
1784 * the long form of DEFINE-METHOD-COMBINATION disables method group
1785 checking when given a single method group with pattern *. (thanks
1787 * on x86 compiler supports stack allocation of results of simple
1788 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1789 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1790 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1791 * fixed some bugs related to Unicode integration:
1792 ** the restarts for recovering from input and output encoding
1793 errors only appear when there is in fact such an error to
1795 * fixed some bugs revealed by Paul Dietz' test suite:
1796 ** MISC.549 and similar: late transformation of unsafe type
1797 assertions into derived types caused unexpected code
1799 ** SCALE-FLOAT type deriver is less wrong.
1800 ** type derivers for EXP, LOG and similar functions compute result
1801 types for complex arguments better.
1802 ** (MISC.563) CONJUGATE type deriver works for very restricted
1804 ** out-of-line type testers for character strings are available.
1805 ** EQUAL compiler transform understands specialness of objects
1807 ** accessing double-floats stored on the stack now works on x86-64.
1808 ** debugger internals could sometimes create invalid lispobjs,
1809 resulting in GC crashes.
1810 ** MISC.548: type check weakening can convert required type into
1812 ** initialization forms for bindings are not in scope of free special
1815 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1816 * incompatible change: thread support for non-NPTL systems has
1817 been removed - locking is buggy and unreliable. A threaded
1818 SBCL build will now warn at startup and refuse to create
1819 new threads, unless futex support is detected in the kernel
1820 * incompatible change: the top level REPL now has only an ABORT
1821 restart associated with it, not TOPLEVEL and ABORT as it used to.
1822 TOP and TOPLEVEL are now available as debugger commands for
1823 returning to the top level.
1824 * incompatible change: forms evaluated in the REPL now use the
1825 global optimization policy.
1826 * incompatible change: user- and system-initialization files are
1827 no longer processed with LOAD, but by READ and EVAL; hence the
1828 global optimization policy, startup package, readtable, etc,
1830 * The .fasl file format number has been incremented because of
1831 various incompatible changes.
1832 * internal entry point details and argument counts no longer appear
1833 in backtraces unless explicitly requested by setting
1834 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1835 * built-in and standard functions no longer have names like "top
1836 level local call to FOO".
1837 * fixed bug 32: functions defined in non-null lexical environments
1838 now have more legible printed representation
1839 * fixed bug 33: functions defined in non-null lexical environemnts
1840 are now more amenable to inspection by INSPECT.
1841 * workaround for bug 354: XEPs no longer appear in backtraces unless
1842 explicitly requested.
1843 * fixed bug: receiving the signal which results from attempting to
1844 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1845 SIGBUS on other *BSDs) on architectures where the C stack is also
1846 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1847 to debugger confusion or wild instability. (reported by Cheuksan
1849 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1850 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1851 (reported by Lutz Euler)
1852 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1853 compiled code. (reported by Paul Dietz)
1854 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1855 specialized parameter to a method to something that is not TYPEP
1856 the specializer is now possible.
1857 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1858 face of package deletion.
1859 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1860 pathname. (reported by David Tolpin on c.l.l)
1861 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1862 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1863 STRING-STREAM and FILE-STREAM, respectively.
1864 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1865 than previously; in addition, its test suite is now run on build.
1866 (thanks to Luke Gorrie)
1867 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1868 * optimization: added a immediate representation for single-floats
1870 * fixed some bugs revealed by Paul Dietz' test suite:
1871 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1872 ** attempting to create a package with a colliding nickname causes
1873 correctable errors to be signalled.
1874 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1875 ** several x86-64 backend bugs related to sign-extension of immediate
1878 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1879 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1880 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1882 * fixed loading of multiply forward-referenced layouts.
1883 (thanks to Cheuksan Wang)
1884 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1885 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1886 (see COMPILE-FILE documentation for details of :PRINT option)
1887 * fixed bugs 19 and 317: fixed-format floating point printing is
1888 more accurate. This also fixes a bug reported by Adam Warner
1889 related to the ~@F format directive.
1890 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1892 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1893 dispatch table if the from-char is a dispatch macro character.
1894 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1895 by the wordsize no longer ignore the last word. (reported by Lutz
1897 * fixed bug in type checking of dynamic-extent variables. (reported
1899 * optimization: sequence traversal functions use their freedom to
1900 coerce function designators to functions.
1901 * optimization: code with many calls to CLOS methods specialized on
1902 CLOS classes has had redundant type checks removed.
1903 * optimization: added declarations to speed up operations that access
1904 the internal character database (for example STRING-UPCASE)
1905 * optimization: comparison operations between floats and sufficiently small
1906 fixnums no longer create extra rationals
1907 * fixed some bugs related to Unicode integration:
1908 ** portions of multibyte characters at the end of buffers for
1909 character-based file input are correctly transferred to the
1910 start of the buffer at the next read.
1911 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1912 passing it through to OPEN.
1913 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1914 argument given to internal calls to OPEN.
1915 ** the built-in comment readers (introduced by character sequences
1916 ";" and "#|") are more forgiving to encoding errors; they will
1917 STYLE-WARN and then attempt to resync the stream at a character
1918 boundary. (thanks to Teemu Kalvas)
1919 * fixed some bugs in the x86-64 port:
1920 ** Negative short int return values from c-calls are sign-extended
1922 ** The stack is aligned to 16-bytes for c-calls, as required by
1923 the ABI. (thanks to Cheuksan Wang)
1924 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1925 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1926 for enums. (thanks to Vincent Arkesteijn)
1927 ** Multiple small optimizations and bugfixes for floating point
1929 * fixed some bugs revealed by Paul Dietz' test suite:
1930 ** Space, Tab, Linefeed, Return and Page have the invalid
1931 secondary constituent character trait.
1932 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1934 ** WITH-INPUT-FROM-STRING should only update the index place on
1936 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1937 works more reliably.
1938 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1939 with one-dimensional arrays with fill pointers.
1940 ** TYPE-OF failed on a complex with an integer realpart and a
1942 ** compiler failure during type inference for the code of form
1943 (IF (EQL X (THE ...)) ...) (MISC.535).
1945 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1946 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1947 platform. The port supports 61-bit fixnums, large memory spaces
1948 and reloading shared object files.
1949 * enhancement: saving cores with foreign code loaded is now
1950 supported on ppc/Darwin in addition to the previously supported
1952 * enhancement: the statistical profiler now walks deeper into the
1953 call stack for more meaningful call-graphs and accrued time
1954 reports (x86/x86-64 only). It also now reports time spent in
1956 * enhancement: it is now possible to trace most individual methods
1957 of a generic function in addition to tracing the generic function
1959 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1961 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1962 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1963 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1964 produces an error. (thanks to Vincent Arkesteijn)
1965 * bug fix: NAMESTRING on pathnames with :WILD components in their
1966 directories works correctly. (thanks to Artem V. Andreev)
1967 * fixed bug 125: compiler preserves identity of closures. (reported
1969 * bug fixed: functions with &REST arguments sometimes failed with
1970 "Undefined function" when compiled with (DEBUG 3). (reported by
1972 * bug fix: overflow during compiling of setting element of a bit
1973 vector with constant index and value. (reported by Timmy Douglas)
1974 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1975 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1976 * on x86 compiler partially supports stack allocation of dynamic-extent
1978 * GO and RETURN-FROM do not check the extent of their exit points
1979 when compiled with SAFETY 0.
1980 * fixed some bugs related to Unicode integration:
1981 ** encoding and decoding errors are now much more robustly
1982 handled; it should now be possible to recover even from invalid
1983 input or output to the terminal. (thanks to Teemu Kalvas)
1984 ** provided a first cut at implementing STRING-TO-OCTETS and
1985 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1986 ** altered the SB-MD5 contributed module slightly, changing the
1987 interface just enough for it to be supportable for builds where
1988 lisp characters are not eight bits.
1989 * fixed some bugs revealed by Paul Dietz' test suite:
1990 ** the FORMATTER-generated functions for ~V[ conditionals require
1991 the correct number of arguments.
1992 ** READ-FROM-STRING returns the mandated second value when applied
1993 to displaced strings.
1994 ** the #\Rubout and #\Backspace characters are treated as invalid
1995 constituent characters by the tokenizer.
1997 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1998 * new feature: reloading changed shared object files with
1999 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2000 * new feature: references to foreign variables and functions
2001 can now be compiled and loaded before the corresponding shared
2002 object file is loaded, as long as the foreign definitions are
2003 available at runtime.
2004 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2005 supported. (thanks to Dan Debertin)
2006 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2007 just keywords. (thanks to Vincent Arkesteijn)
2008 * fixed bug #331: structure-class instances corresponding to
2009 DEFSTRUCT forms are now created eagerly.
2010 * fixed bug #345: backtraces from calls to undefined functions work
2011 on x86 as well. Related bug #61 is now also partially fixed on x86
2012 (backtraces from throws to unknown catch tags.)
2013 * bug fix: lambda-list parsing is now stricter vrt. order and number
2014 of lambda-list keywords.
2015 * bug fix: as specified by AMOP, an error is signalled if a
2016 class-option appears multiple times in a DEFCLASS form. (reported
2018 * bug fix: RANDOM can be compiled when the compiler derives the type
2019 of its numeric argument as a disjoint set of small integers.
2020 (reported by Paul Dietz)
2021 * bug fix: starting a core saved with shared objects loaded when
2022 those objects are not available no longer causes threaded SBCL to
2023 hang. (reported by Sean Ross)
2024 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2026 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2027 argument, which must always be NIL. (reported by Kalle Niemitalo)
2028 * bug fix: printing 1.0d+23 no longer results in an error.
2029 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2030 * bug fix: structure slot setters preserve evaluation order. (thanks
2032 * fixed some bugs related to Unicode integration:
2033 ** RUN-PROGRAM can allow its child to take input from a Lisp
2034 stream. (reported by Stefan Scholl)
2035 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2036 locales. (reported by Ken Causey)
2037 * fixed some bugs revealed by Paul Dietz' test suite:
2038 ** INCF, DECF and REMF evaluate their place form as specified in
2040 ** FORMATTER expands ~{ ~} iteration directives with V or #
2041 parameters correctly.
2042 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2043 consequent uses no arguments correctly.
2044 ** the system has a partial understanding of the (COMPLEX RATIO)
2046 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2049 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2050 * new feature: a build-time option (controlled by the :SB-UNICODE
2051 keyword feature, enabled by default) for building the system with
2052 support for the entire 21-bit character space defined by the
2054 * new feature: the system now has rudimentary external-format
2055 support; the primary user-visible change at this time is that
2056 characters with the high bit set (such as lower-case-e-acute) will
2057 print correctly to a terminal in a UTF-8 environment.
2058 * minor incompatible change: BASE-CHAR no longer names a class;
2059 however, CHARACTER continues to do so, as required by ANSI.
2060 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2061 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2062 has been moved to the SB-EXT package (temporarily re-exported from
2064 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2065 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2067 * fixed bug #351: better error-handlind and reporting for malformed
2069 * fixed bug #350: bignum-printing is now more memory-efficient,
2070 allowing printing of very large bignums, eg. (expt 2 10000000).
2071 (reported by Bruno Haible)
2072 * fixed bug #302: better primitive-type selection for intersection
2074 * fixed bug #308: non-graphic characters now all have names, as
2075 required. (reported by Bruno Haible)
2076 * bug fix: redefining a class with different superclasses now correctly
2077 removes it from the direct-subclasses of its previous superclasses.
2078 (reported by David Morse)
2079 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2080 new value now works. (reported by Bruno Haible)
2081 * bug fix: correct canonicalization of multiple non-standard slot
2082 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2083 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2084 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2085 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2086 now exists, an signals an error.
2087 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2088 during macroexpansion for non-iteration variables as well. (reported
2089 by Bruno Haible for CMUCL)
2090 * bug fix: Cyclic structures and unprintable objects in compiler
2091 messages no longer cause errors. (reported by Bruno Haible)
2092 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2093 and READ-FROM-STRING all now return a primary value of NIL if
2094 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2095 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2096 (reported by Bruno Haible for CMUCL)
2097 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2098 do not designate a condition. (reported by Bruno Haible for
2100 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2101 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2102 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2103 permitting resolution in favour of any of the conflicting symbols.
2104 (reported by Bruno Haible for CMUCL)
2105 * bug fix: EQUAL compiler optimizations is less aggressive on
2106 strings which can potentially compare true despite having distinct
2107 specialized array element types.
2108 * bug fix: unit enumerations can be defined without dividing by
2109 zero. (thanks to Vincent Arkesteijn)
2110 * FORMAT compile-time argument count checking has been enhanced.
2111 (report from Bruno Haible for CMUCL)
2112 * a partial workaround for the bug 262: the compiler does not try to
2113 inline-expand a local function doing RETURN-FROM from a deleted
2114 BLOCK. (thanks to Peter Denno for the bug report and to David
2115 Wragg for the simple test case)
2116 * fixed some bugs revealed by Paul Dietz' test suite:
2117 ** RENAME-PACKAGE allows all package designators as new package
2119 ** constraint propagation and lambda variable substitution are
2120 more cautious in dealing with partially deleted code.
2121 ** compiler sometimes lost reoptimization passes.
2122 ** CERROR, given a condition as condition designator, passes its
2123 remaining arguments to the continue format control without
2125 ** Case-altering FORMAT directives work correctly on non-ASCII
2127 ** The REAL type specifier handles bounds outside the floating
2128 point ranges without signalling FLOATING-POINT-OVERFLOW.
2129 ** Functions with IR1-transformations can create intercomponent
2130 references to global functions.
2131 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2133 ** FORMAT ~R treats a nil value for its first parameter correctly.
2135 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2136 * enhancement: saving cores with foreign code loaded is now
2137 supported on x86/NetBSD and sparc/Linux in addition to the previously
2138 supported platforms.
2139 * bug fix: on some platforms repeated installations caused multiple
2140 copies of HTML documentation to be installed -- should not happen
2141 any more. (reported by Stefan Scholl)
2142 * bug fix: parsing self-recursive alien record types multiple times
2143 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2144 original patch by Helmut Eller for CMUCL)
2145 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2146 (thanks to Richard Kreuter)
2147 * bug fix: defining classes whose accessors are methods on existing
2148 generic functions in other (locked) packages no longer signals
2149 bogus package lock violations. (reported by François-René Rideau)
2150 * bug fix: special variables as DEFMETHOD parameters no longer have
2151 associated bogus type declarations. (reported by David Wragg and
2153 * bug fix: read-write consistency on streams of element-type
2154 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2155 * bug fix: redefiniton of the only method of a generic function with
2156 no DEFGENERIC no longer emits a full WARNING. In addition,
2157 redefinition of generic functions with no DEFGENERIC to an
2158 incompatible lambda list now signals an error. (thanks to Zach
2160 * bug fix: DEFGENERIC now works even when there's a function of the
2161 same name in an enclosing lexical environment. (thanks to Zach
2163 * fixed compiler failure, caused by instrumenting code during
2164 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2165 * optimization: added loop analysis and improved register allocation
2166 * fixed some bugs revealed by Paul Dietz' test suite:
2167 ** POSITION on displaced vectors with non-zero displacement
2168 returns the right answer.
2169 ** (SIMPLE-STRING) is a valid type specifier for sequence
2171 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2173 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2174 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2176 ** The justification version of the FORMAT ~< directive treats
2177 non-zero minpad parameter correctly.
2179 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2180 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2181 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2182 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2183 the supported interface.
2184 * new feature: Single-stepping of code compiled with DEBUG 2 or
2185 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2186 * new feature: saving cores with foreign code loaded is now
2187 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2188 Timothy Moore's work for CMUCL)
2189 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2190 parameters to * instead of NIL if no initform is supplied.
2191 (reported by Johan Bockgaard)
2192 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2193 correctly when FUNCALL appears as the car of the form. Note:
2194 despite this FUNCALL forms are not currently subject to
2195 compiler-macro expansion. (port of Raymond Toy's fix for the
2196 same from CMUCL, reported by Johan Bockgaard)
2197 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2198 (thanks for Teemu Kalvas)
2199 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2200 case instead of terminating the iteration (thanks for Julian
2201 Squires, Sean Champ and Raymond Toy)
2202 * bug fix: incorrect expansion of defgeneric that caused a style
2203 warning. (thanks for Zach Beane)
2204 * bug fix: slot accessor effective method computation works properly
2205 for classes with multiple non-standard applicable methods on
2206 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2207 * on x86 compiler supports stack allocation of results of LIST and
2208 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2209 CMUCL implementation by Gerd Moellmann)
2210 * fixed some bugs revealed by Paul Dietz' test suite:
2211 ** FORMAT strings with both the ~<~:;~> form of the justification
2212 directive and pretty-printing directives cause an error.
2214 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2215 * incompatible change: the internal functions
2216 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2217 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2218 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2219 instead of the old functions.
2220 * new feature: on platforms where "dladdr" is available foreign
2221 function names now appear in backtraces. (based on Helmut Eller's
2223 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2224 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2226 * documentation: ASDF manual is now distributed with SBCL.
2227 * bug fix: SBCL can now load its contributed modules with REQUIRE
2228 even if the system-provided entries have been removed from
2229 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2231 * bug fix: backtraces involving undefined functions or assembly
2232 routines are more informative. (thanks to Brian Downing)
2233 * bug fix: mutually referent alien structures now work correctly.
2234 (reported by Rick Taube)
2235 * bug fix: structures defined by WITH-ALIEN can be referred to
2236 within other definitions in the same WITH-ALIEN.
2237 * bug fix: division operators (MOD, TRUNCATE and the like) with
2238 constant zero divisors and integer dividends no longer generate
2240 * bug fix: provide default methods for INPUT-STREAM-P and
2241 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2242 * bug fix: improve the detection and resolution of MOP metacycles
2243 (where effective method computation depends on the generic function
2244 having its effective method computed). (reported by Bruno Haible)
2245 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2246 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2247 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2248 * bug fix: pathnames with relative directory components can be
2249 represented relative to default pathnames.
2250 * optimization: in taking the GCD of bignums, reduce the two bignums
2251 to approximately the same size (using Euclid's algorithm) before
2252 applying the more sophisticated binary GCD. (thanks to Juho
2254 * optimization: COUNT on bitvectors now operates word-at-a-time.
2255 * optimization: ASH with a positive, but not necessarily constant,
2256 (leftwards) shift, when in a modular context, is compiled to a
2258 * fixed some bugs revealed by Paul Dietz' test suite:
2259 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2260 if the corresponding argument is NIL.
2261 ** FORMAT directives accepting a minpad parameter treat negative
2263 ** literal commas as character parameters to FORMAT directives are
2265 ** literal spaces directly after ~<Newline> directives within a
2266 format-logical-block (~:< ~@:>) do not induce :FILL-style
2267 conditional newlines.
2268 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2269 as specified: it no longer includes conditional newlines).
2270 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2272 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2273 :PER-LINE-PREFIX argument does not evaluate to a string.
2275 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2276 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2277 the manual for details; for now, package locks can be disabled by
2278 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2279 if no major problems are found then it is likely that they will be
2280 compiled in unconditionally.
2281 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2282 unsupported operators on all platforms. To load a shared library
2283 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2284 object file, link it into a shared library outside of SBCL and
2285 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2287 * bug workaround: ROOM T and the GC were not working together
2288 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2289 reporting facility aren't preserved. That reporting has been
2290 disabled, so now until and unless someone figures out how to make it
2291 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2292 * minor incompatible change: as threatened around sbcl-0.8.0, the
2293 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2294 The symbols are also exported from SB-PCL for backwards
2295 compatibility, but more so than before SB-PCL should be treated as
2296 an implementation-internal package.
2297 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2299 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2300 It is known as of this release not to work on the Alpha, however.
2301 * fixed bug #167: errors signalled due to illegal syntax in method
2302 bodies are now more legible.
2303 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2304 designators and can hence be used with TYPEP.
2305 * fixed bug #333: CHECK-TYPE now ensures that the type error
2306 signalled, if any, has the right object to be accessed by
2307 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2308 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2309 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2311 * fixed bug #334: programmatic addition of slots using specialized
2312 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2313 and :ALLOCATION :CLASS slots.
2314 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2315 float exponents. (rereported by Peter Seibel)
2316 * fixed a bug: #\Space (and other whitespace characters) are no
2317 longer considered to be macro characters in standard syntax by
2318 GET-MACRO-CHARACTER.
2319 * fixed bug: initialization of condition class metaobjects no longer
2320 causes an instance of the condition to be created. (reported by
2322 * fixed bug: it is now possible to have slots such that
2323 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2324 description is neither :INSTANCE nor :CLASS.
2325 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2326 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2327 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2328 non-standard SLOT-DEFINITION-ALLOCATION values.
2329 * fixed bugs: various race conditions handling exiting threads.
2330 CL-PPCRE's thread torture test now passes.
2331 * fixed arguably-a-bug: GC time proportional to number of threads in
2332 system even when most of them are idle
2333 * optimization: improved performance of BIT and SBIT on bit-vectors.
2334 * .fasl file incompatibility: The fasl file version number has
2335 been incremented because of changes associated with package locks.
2337 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2338 * minor incompatible change: the system no longer provides
2339 optimization hints (conditions which are TYPEP
2340 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2341 mode; these hints are still emitted when higher SPEED optimization
2343 * new contrib module: a sampling profiler (profiling by statistical
2344 sampling, rather than by instrumenting functions) is available as
2345 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2346 * the behaviour of the standard function ED is now customizeable by
2347 third parties through a hook variable: see ED's documentation
2348 string for information on the protocol.
2349 * the compiler no longer emits efficiency notes for (FUNCALL X)
2350 when the type of X is uncertain under default optimization
2352 * fixed bug 276: mutating a binding of a specialized parameter to a
2353 method to something that is not TYPEP the specializer is now
2355 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2356 exhibit the required behaviour on the x86 platform. (thanks to
2357 Peter van Eynde, Eric Marsden and Bruno Haible)
2358 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2359 even for difficult arguments. (reported by Peter Graves)
2360 * fixed bug 141a: the backquote printer now descends quoted
2362 * fixed another bug in backquote printing: no more destructive
2363 modification of the form's list structure. (reported by Brian
2365 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2366 move between its address being taken and the call to
2367 interrupt_thread, fixing a crashing race condition.
2368 * the SB-POSIX contrib implementation has been adjusted so that it
2369 no longer exhibits ridiculously poor performance when constructing
2370 instances corresponding to C structs.
2372 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2373 * minor incompatible change: the sb-grovel contrib now treats C
2374 structures as alien (in the sense of SB-ALIEN) objects rather than
2375 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2376 has implications for memory management of client code
2377 (sb-grovel-returned objects must now be manually managed) and for
2378 type safety (alien objects now have full types).
2379 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2380 used to control emission of compiler diagnostics, rather than the
2381 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2382 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2383 quality should be considered deprecated.
2384 * install.sh now installs the user manual as well
2385 * (not quite a new documentable feature, but worth considering in
2386 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2387 beginnings of a semantically meaningful condition hierarchy is
2388 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2390 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2391 designator as the defaults argument.
2392 * fixed bug: Displaced arrays whose displaced-to array has become
2393 too small now cause ARRAY-DIMENSION to signal an error, providing
2394 for safer bounds-checking. (reported by Bruno Haible)
2395 * fixed bug: DEFCLASS slot definitions with identical :READER and
2396 :WRITER names now signal a reasonable error. (reported by Thomas
2398 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2400 * fixed bug: Invalid slot specification errors now print correctly.
2401 (thanks to Zach Beane)
2402 * fixed bug 320: Shared to local slot value transfers in class
2403 redefinitions now happen corrently from superclasses as
2404 well. (reported by Bruno Haible)
2405 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2407 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2408 improper lists correctly. (reported by Bruno Haible)
2409 * fixed bug 313: source-transform for <fun-name> was erroneously
2410 applied to a call of a value of a variable with name <fun-name>.
2411 (reported by Antonio Menezes Leitao)
2412 * fixed bug 307: The obsolete instance protocol ensures that
2413 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2414 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2415 remove dead unknown-values globs from the middle of the stack.
2416 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2417 easier to resume long computations after using *BREAK-ON-SIGNALS*
2418 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2419 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2420 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2421 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2422 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2423 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2425 * fixed a (fairly theoretical) bug in string printing: if
2426 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2427 does not have array-element-type equal to the most general string
2429 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2430 function name. (thanks to Nikodemus Siivola)
2431 * fixed bug: compile-time format string checker failed on
2432 non-closed ~{. (reported by Thomas F Burdick)
2433 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2434 don't warn on legal loop code involving more than one aggregate
2435 boolean. (thanks to Nikodemus Siivola)
2436 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2437 in keyword position to MAKE-PACKAGE was not being reported as
2439 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2440 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2442 * fixed bugs 280 and 312: the checking for multiple definitions in a
2443 file is less likely to become confused by uses of inline
2445 * fixed bug: the #S reader macro performs the keyword coercion
2446 specified for slot names. (reported by Kalle Niemitalo)
2447 * fixed bug: lambda lists may contain symbols whose names start with
2448 & but are not lambda-list-keywords; their occurrence triggers a
2450 * fixed bug 321: define-method-combination argument lambda lists do
2451 not cause invalid code to be generated when &optional or &aux
2452 variables have default values. (reported by Bruno Haible)
2453 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2454 have CLOS classes; this appears also to have fixed reported
2455 instability in stack exhaustion detection.
2456 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2457 MUFFLE-WARNING when no associated restart is present is now
2459 * optimization: rearranged the expansion of various defining macros
2460 so that each expands into only one top-level form in a
2461 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2463 * optimization: used a previously little-used slot in symbols to
2464 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2466 * fixed some bugs revealed by Paul Dietz' test suite:
2467 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2469 ** Fixed an optimization bug related to inheritance of initforms
2470 from local to shared slots.
2471 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2472 ** CAST optimizer forgot to flush argument derived type.
2473 ** print/read consistency on floats is now orders of magnitude
2474 more likely. (thanks also to Bruno Haible for a similar report
2476 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2477 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2478 for objects of type REAL. Make it so.
2479 ** SXHASH is specified (infelicitously) to respect similarity,
2480 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2481 it so. (thanks to Markus Ziegler)
2482 ** on the Alpha, the compiler succeeds in compiling functions
2483 returning a known number of arguments greater than 63.
2484 ** fixed handling of invalid NIL arguments in keyword position
2486 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2488 ** provided a readably-printable representation for RANDOM-STATE
2490 ** ensured that pathnames, where they have a namestring, always
2491 print using #P"..." syntax.
2493 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2494 * Support for the forthcoming 2.0 version of the NetBSD kernel
2495 running on x86 hardware has been added. (thanks to Perry
2496 E. Metzger most immediately, and others for their past work)
2497 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2498 current release version; SBCL's OpenBSD support had been broken
2499 since about the time of OpenBSD's switch to ELF binary format.)
2500 * [placeholder for DX summary]
2501 ** user code with &REST lists declared dynamic-extent, under high
2502 speed or space and low safety and debug optimization policy.
2503 * The manual has been converted to Texinfo format and the debugger
2504 chapter from the cmucl manual has been added.
2505 * A facility has been added to extract documentation strings from
2506 sbcl and store them as Texinfo-formatted snippets for inclusion in
2507 the manual (via Texinfo's @include directive)
2508 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2509 the test case to Dave Roberts)
2510 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2511 pointers. (reported by Sean Ross)
2512 * bug fix: PROFILE output is printed nicely even for large numerical
2513 values. (thanks to Zach Beane)
2514 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2515 greater than 32 handle EOF correctly.
2516 * bug fix: on X86 an immediate argument of the IMUL instruction is
2517 correctly printed in disassembly. (thanks to Lutz Euler)
2518 * bug fix: class slots in redefined classes preserve their old
2519 values. (thanks to Bruno Haible and Nikodemus Siivola)
2520 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2521 succeeds. (reported by Marco Baringer)
2522 * fixed some bugs revealed by Paul Dietz' test suite:
2523 ** READ-SEQUENCE now works on ECHO-STREAMs.
2524 ** RATIONALIZE works more according to its specification. (thanks
2527 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2528 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2529 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2530 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2531 should matter to you only if you rebind the printer control
2532 variables and then find you want different bindings in the
2533 debugger than in the ordinary execution of your program.)
2534 * The runtime build system has been tweaked to support building
2535 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2536 assembler and linker. (thanks to Nikodemus Siivola)
2537 * Unbound, undefined, undeclared variables now trigger full
2538 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2539 more often programmer error than deliberate exploitation of undefined
2541 * optimization: the hash algorithm for strings has changed to one
2542 that is less vulnerable to spurious collisions. (thanks to Juho
2544 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2545 needless bounds checking. (thanks to Juho Snellman)
2546 * optimization: implemented multiplication as a modular
2547 (UNSIGNED-BYTE 32) operation on the PPC backend.
2548 * fixed some bugs revealed by Paul Dietz' test suite:
2549 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2551 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2552 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2553 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2555 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2557 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2559 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2560 their output stream on EOF from read.
2561 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2562 have been read to end-of-file.
2563 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2565 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2566 description of determination of which consecutive characters
2568 ** Printing the "Space" character with escaping on now yields "#\\ ",
2569 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2570 ** Reading floating-point numbers with *READ-BASE* set to a number
2571 less than 10 works correctly.
2572 ** Reading floating-point numbers with *READ-BASE* set to a number
2573 more than 10 works correctly.
2574 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2575 the readtable currently in effect.
2577 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2578 * minor incompatible change: parsing of namestrings on a physical
2579 (Unix) host has changed; numbers after the final #\. in a
2580 namestring are no longer interpreted as a version field. This is
2581 intented to be largely invisible to the user, except that the
2582 meaning of the namestring "*.*.*" has changed: it now refers to a
2583 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2584 should usually be replaced by
2585 (make-pathname :name :wild :type :wild :version :wild)
2586 with the added benefit that this is more likely to be portable.
2587 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2588 option to OPEN now signals an error if the file being opened
2589 exists; this may have an impact on existing code.
2590 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2591 signals received, on the PowerPC platforms (both Linux and
2592 Darwin). (thanks to Pierre Mai for pointing out the location of
2594 * several fixes on OS X: The system now builds and runs cleanly on
2595 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2596 (thanks to Brian Mastenbrook)
2597 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2598 second-resolution: integer multiples of 1/3600 between -24 and 24.
2599 (thanks to Vincent Arkesteijn)
2600 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2601 type when called with 1 argument; PEEK-CHAR checked type of
2602 PEEK-TYPE only after having read first character from a
2603 stream. (reported by Peter Graves)
2604 * bug fix: the garbage collector now has much better locality
2605 behaviour, and in particular no longer treats all memory as being
2606 exhausted when many small objects point to each other in a deeply
2608 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2609 recognized as being TYPEP their class.
2610 * bug fix: the PUSHNEW documentation string has been corrected.
2611 (thanks to Vincent Arkesteijn)
2612 * bug fix: defaulting of the value for the last of an atypically
2613 large number of multiple values being bound was not being
2614 performed correctly on the Alpha or PPC platforms
2615 * optimization: implemented multiplication as a modular
2616 (UNSIGNED-BYTE 32) operation on the x86 backend.
2617 * optimization: SEARCH on simple-base-strings can now be open-coded.
2618 (see also contrib/compiler-extras.lisp for inspiration for
2619 teaching the compiler about the Boyer-Moore algorithm).
2620 * value, returned by MAX (and MIN) called with several EQUALP, but
2621 not EQL, arguments now does not depend on compiler settings.
2622 * fixed some bugs revealed by Paul Dietz' test suite:
2623 ** in stack analysis liveness information is propagated from
2624 non-local entry points.
2625 ** pathwise CAST removing failed when the CAST node did not start
2627 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2628 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2630 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2631 host is already defined.
2632 ** RENAME-FILE works on streams instead of signalling an internal
2634 ** PEEK-CHAR uses the current readtable when determining whether
2635 or not a character is whitespace.
2636 ** MERGE-PATHNAMES handles the case when the pathname does not
2637 specify a name while the default-pathname specifies a version.
2638 ** Pathnames now stand a better chance of respecting print/read
2640 ** Attempting to use standardized file system operators with a
2641 pathname with invalid :DIRECTORY components signals a
2643 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2645 ** DIRECTORY on logical pathnames is more correct.
2646 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2647 signal a TYPE-ERROR if their argument is not a stream.
2648 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2649 designator argument does not designate a stream.
2650 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2651 examining the synonym.
2652 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2654 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2655 element-type larger than ([UN]SIGNED-BYTE 32).
2657 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2658 * When built with the :SB-FUTEX feature, threaded builds now take
2659 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2660 for faster/more reliable mutex and condition variable support.
2661 * Incompatible change (but one you probably shouldn't have been using
2662 anyway): the interface and code for arbitrating between multiple
2663 threads in the same user session has been redesigned.
2664 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2665 error when called without an explicit environment argument.
2666 (thanks to Nikodemus Siivola)
2667 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2668 confused about their position. (thanks to Adam Warner and Gerd
2670 * bug fix: Pretty printing backquoted forms with unquotations in the
2671 argument list position of various code constructs such as LAMBDA
2672 now works correctly. (reported by Paul Dietz)
2673 * bug fix: Pretty printing unquotations no longer loses all
2674 stream position information.
2675 * optimization: performance of string output streams is now less
2676 poor for multiple small sequence writes.
2677 * optimization: performance of CSUBTYPEP in the presence of complex
2678 expressions involving CONS and NOT many times has been improved.
2679 (reported by Paul Dietz)
2680 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2682 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2683 streams for the REPL, for the debugger, and so on. (thanks to
2685 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2686 * fixed some bugs revealed by Paul Dietz' test suite:
2687 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2688 is ignored if it is too small, rather than propagating through
2689 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2690 ** extremely complex negations of CONS types were not being
2691 sufficiently canonicalized, leading to inconsistencies in
2693 ** VALUES tranformer lost derived type.
2695 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2696 * fixed a bootstrapping bug: the build process no longer assumes
2697 that the various BOOLE-related constants have the same value in
2698 host and target lisps. (noted by Paul Dietz' test suite on an
2699 SBCL binary built from CLISP)
2700 * The system can now be dynamically linked on the MIPS platform,
2701 which enables dynamic loading of foreign code from Lisp. (thanks
2702 to Ralf Baechle for discussions on the MIPS ABI)
2703 * The system now records debugging information for its own source
2704 files in a filesystem-position-independent manner, relative to
2705 the "SYS" logical host.
2706 * fixed a compiler bug: MV-LET convertion did not check references
2707 to the "max args" entry point. (reported by Brian Downing)
2708 * tweaked disassembly notes to be less confident about proclaiming
2709 some instruction as an LRA. (thanks to Brian Downing)
2710 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2711 now each have their own history, command character, and other
2712 characteristics. (thanks to David Lichteblau)
2713 * fixed some bugs revealed by Paul Dietz' test suite:
2714 ** compiler failure in compiling LOGAND expressions including a
2716 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2717 shift greater than 32.
2718 ** FLUSH-DEST did not mark blocks for type check regeneration.
2719 ** HANDLER-CASE failed to accept declarations in handler clauses
2720 in some circumstances.
2722 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2723 * New code in contrib/sb-introspect (still probably not entirely
2724 stable yet) provides some support for smart Lisp development
2725 environments like SLIME.
2726 * The conditions signalled for errors occurring when loading .fasl
2727 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2728 in a way which should help ASDF recover gracefully.
2729 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2730 stuff has been cleaned up. If you code contrib/ stuff, this might
2731 affect you, and you can look at contrib/README, contrib/STANDARDS,
2732 and/or the 0.8.4.27 diff to check.
2733 * In full calls the compiler now does not generate checks for declared
2734 argument types for all arguments.
2735 * various threading fixes
2736 ** and some experimental patches which didn't make it into
2737 the main tree for this release, but which are shipped in
2738 contrib/experimental-thread.patch as a possible fix for some
2739 failures (deadlock, spinning...) in GC-intensive multithreaded
2741 * fixed PPC build problem (source code incompatibility of different
2742 library versions): added offsetof() hackery which attempts to divine
2743 where glibc maintainers put uc_mcontext today
2744 * fixed bug 282: compiler does not trust type assertions while passing
2745 arguments to a full call.
2746 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2747 &OPTIONAL VALUES type parameter.
2748 * fix bug 214: algorithm for noting rejected templates is now more
2749 similar to that of template seletion. (also reported by rydis on
2751 * fixed bug 141b: printing backquoted information readably and prettily
2752 inserts a space where necessary.
2753 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2754 (CHARACTER 10) are now reported as errors, rather than propagated
2755 as unknown types. (reported by piso on #lisp)
2756 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2757 with values NIL and :ERROR. (thanks to Milan Zamazal)
2758 * fixed bug 191c: CLOS now does proper keyword argument checking as
2759 described in CLHS 7.6.5 and 7.6.5.1.
2760 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2761 longer raise an error; further, using a list as a for-as-arithmetic
2762 counter now raises a meaningful error.
2763 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2764 now understood by sequence creation functions such as MAKE-SEQUENCE
2766 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2767 a STRING-INPUT-STREAM.
2768 * compiler enhancement: SIGNUM is now better able to derive the type
2770 * type declarations inside WITH-SLOTS are checked. (reported by
2772 * fixed some bugs revealed by Paul Dietz' test suite:
2773 ** incorrect optimization of TRUNCATE for a positive first
2774 argument and negative second.
2775 ** compiler failure in let-convertion during flushing dead code.
2776 ** compiler failure while deriving type of TRUNCATE on an
2777 interval, containing 0.
2778 ** ASH of a negative bignum by a negative bignum count now returns
2780 ** intersection of CONS types now canonicalizes properly, fixing
2781 inconsistencies in SUBTYPEP.
2783 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2784 * incompatible change: The --disable-debugger command line
2785 option now clobbers the debugger at a more fundamental
2786 level, by redefining #'INVOKE-DEBUGGER instead of by
2787 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2788 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2789 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2790 enter the debugger REPL and then suffer recursive errors
2791 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2792 while under the new system, BREAK in --disable-debugger mode
2793 terminates the system just as an unhandled error would.
2794 * fixed compiler performance when processing loops with a step >1;
2795 * bug fix: DOCUMENTATION now retrieves generic function
2796 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2797 support has been systematized, and now supports the methods
2798 specified by ANSI, along with a default method and a method for
2799 slot documentation. (reported by Nathan Froyd)
2800 * bug fix: effective methods associated with a generic function are
2801 no longer cached over a change of that generic function's method
2802 combination. (reported by Andreas Fuchs)
2803 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2804 names a non-existent directory. (thanks to Andreas Fuchs)
2805 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2806 the PPC platform, lead to stack corruption; this has been fixed.
2807 (reported by Rainer Joswig)
2808 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2809 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2810 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2811 arithmetic in general, is now much more reliable on the PPC
2813 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2814 platform now returns the right answer.
2815 * optimization: restored some effective method precomputation in
2816 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2817 precomputation is now tunable.
2818 * optimization: compiler-internal data structure use has been
2819 reviewed, and changes have been made that should improve the
2820 performance of the compiler by about 20%.
2821 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2822 simple-base-strings and simple-bit-vectors is improved.
2823 * optimization: the optimization of 32-bit logical and arithmetic
2824 functions introduced in version 0.8.3 on the x86 has been
2825 implemented on the mips, ppc and sparc platforms; an
2826 implementation of the same facility, but for 64-bit arithmetic,
2827 has been added for the alpha.
2828 * microoptimization: the compiler is better able to make use of the
2829 x86 LEA instruction for multiplication by constants.
2830 * bug fix: in some situations compiler did not report usage of
2831 generic arithmetic in (SPEED 3) policy.
2832 * bug 145b fix: compiler used wrong type specifier while converting
2833 MEMBER-types to numeric.
2834 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2836 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2837 index argument now works properly on the Alpha platform.
2838 * bug fix: floating point exception treatment on the Alpha platform
2840 * bug fix: FILE-POSITION works much better on string input and
2841 output streams. (thanks to Nikodemus Siivola)
2842 * bug fix: many threading/garbage collection symptoms sorted.
2843 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2844 might be pseudo-atomic.
2845 * internal change: Stopping for GC is now done with signals not
2846 ptrace. GC is now done in whichever thread wanted it, instead of
2848 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2850 * bug fix: result form in DO is not contained in the implicit
2852 * incompatible change: ICR structure is changed; the value part of
2853 CONTINUATION is now called LVAR; corresponding functions are
2854 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2855 * added type deriver for ISQRT (thanks to Robert E. Brown).
2856 * bug fix: better support for loading from the command line when an
2857 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2859 * fixed some bugs revealed by Paul Dietz' test suite:
2860 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2861 ** ROUND and FROUND now give the right answer when given very
2862 small float arguments.
2863 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2865 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2866 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2867 ** LCM should return a non-negative integer.
2868 ** PARSE-INTEGER returned the index of a terminator instead of the
2869 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2870 ** PARSE-INTEGER returned an incorrect index being applied to a
2872 ** LCM with two arguments of 0 returns 0 rather than signalling
2874 ** unsigned addition of a 32-bit constant with the high bit set no
2875 longer causes an internal compiler error on the x86.
2876 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2877 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2878 with negative last argument.
2879 ** byte specifiers with zero size and position no longer cause
2880 an error during type derivation.
2881 ** bignum multiplication on the Alpha platform now returns the
2883 * porting: The system now builds on SuSE AMD64, although it still
2884 generates a 32-bit binary.
2885 * .fasl file incompatibility: The fasl file version number has
2886 been incremented (because of the changes to internal compiler
2887 data structures referred to above).
2889 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2890 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2891 more accurately, on the Darwin kernel running on PowerPC hardware.
2892 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2893 * Compiler code deletion notes now signal a condition of type
2894 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2895 an associated MUFFLE-WARNING restart.
2896 * The compiler now performs limited argument count validation of
2897 constant format strings in FORMAT, and where appropriate in ERROR,
2898 CERROR and WARN. (thanks to Gerd Moellmann)
2899 * New ASDF-INSTALL contrib can be used for automatic download and
2900 installation of third-party Lisp code from CCLAN or other sites
2902 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2903 which forces another thread to execute a function supplied by the
2905 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2906 now accept and act upon their :ELEMENT-TYPE keyword argument.
2907 (reported by Martin Atzmueller, Edi Weitz)
2908 * bug fix: FILE-POSITION now accepts position designators up to
2909 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2910 is the greater. (thanks to Patrik Nordebo)
2911 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2912 Kalle Olavi Niemitalo)
2913 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2914 printable. (reported by Eric Marsden)
2915 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2916 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2917 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2918 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2919 (thanks to Stig Sandoe)
2920 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2921 UNREAD-CHAR now correctly includes the unread character in the
2922 target sequence. (thanks to Gerd Moellmann)
2923 * bug fix in threaded builds: the system can now be suspended and
2924 resumed by shell job control with minimal disruption.
2925 * bug fixes in times and timezones >2038AD
2926 * better handling of "where is GNU make?" problem in build scripts
2927 (thanks to Nikodemus Siivola)
2928 * new optimization: inside a named function any reference to a
2929 function with the same name is considered to be a self-reference;
2930 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2931 optimization quality.
2932 * new optimization on x86: logical functions and + now have
2933 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2934 used when the result is truncated to 32 bits.
2935 * VALUES declaration is partially enabled.
2936 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2937 * fixed some bugs revealed by Paul Dietz' test suite:
2938 ** The system now obeys the constraint imposed by
2939 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2940 types form a lattice under type intersection.
2941 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2942 ** ASSOC now ignores NIL elements in an alist.
2943 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2944 and (1+ MOST-POSITIVE-FIXNUM) answers.
2945 ** The addition of a method with invalid qualifiers to a generic
2946 function does not cause an error to be signalled immediately;
2947 a warning is signalled, and the error is generated only on
2948 calling the generic function.
2949 * changed .fasl file version number, in part to add type codes for
2950 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2951 obscure ANSI requirements
2953 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2954 * fixed bug 148: failure to inline-expand a local function left
2955 garbage, confusing the compiler.
2956 * fixed bugs 3cd: structure slot readers perform type check if the
2957 slot can have an invalid value (i.e. it is either not initialized
2958 or can be written with a less specific slot writer).
2959 * bug fix: the compiler now traps array references to elements off
2960 the end of an array; previously, the bounds checking in some
2961 circumstances could go off-by-one.
2962 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2964 * type declarations for array element types now obey the description
2965 on the CLHS page "Declaration TYPE", as per discussions on
2966 sbcl-help around 2003-05-08. This means that a declaration
2967 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2968 declaration, all references to BAR will be asserted or assumed
2969 (with THE, so dependent on compiler policy) to involve objects of
2970 type FOO. Note that no such declaration is implied in
2971 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2972 * declared types of functions from the "Conditions"
2973 chapter. (reported by Paul Dietz)
2974 * bug fix: CERROR accepts a function as its first argument.
2975 * bug fix: NTH an NTHCDR accept a bignum as index
2976 arguments. (reported by Adam Warner)
2977 * optimization: character compare routines now optimize comparing
2978 against a constant character. (reported by Gilbert Baumann)
2979 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2980 argument now works properly on the MIPS platform.
2981 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2982 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2983 declared to have a functional type, does not cause an error
2985 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2986 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2987 * fixed bug: if last continuation of a deleted block has a
2988 destination, this destination should be deleted too. (reported by
2990 * fixed a bug in the bootstrap process: the host compiler's values
2991 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2992 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2993 test case from Patrik Nordebo)
2994 * improved the ability of the disassembler on the PPC platform to
2995 provide helpful disassembly notes.
2996 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2997 the class in more cases than previously.
2998 * bug fix: FILE-POSITION now understands :START and :END for
2999 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3000 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3001 8) data. (thanks to David Lichteblau)
3002 * bug fix: it is possible to add a method to a generic function
3003 without lambda list.
3004 * bug fix: reader failed to signal END-OF-FILE inside an
3005 object representation. (reported by Nikodemus Siivola)
3006 * fixed some bugs revealed by Paul Dietz' test suite:
3007 ** LAST and [N]BUTLAST should accept a bignum.
3008 ** condition slot accessors are methods.
3009 ** (VECTOR NIL) is a subtype of STRING.
3011 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3012 * minor incompatible change: some nonsensical specialized lambda
3013 lists (used in DEFMETHOD) which were previously ignored now signal
3015 * minor incompatible change: the system is now aware of the types of
3016 variables in the COMMON-LISP package, and will signal errors for
3017 most violations of these type constraints (where previously they
3018 were silently accepted).
3019 * minor incompatible change: COMPILE-FILE now uses the freedom
3020 afforded (ANSI 3.2.2.3) to use derived function types for
3021 functions defined in the same file. This also permits the system
3022 to warn on static type mismatches and function
3023 redefinition. (Currently it does not work with high DEBUG level.)
3024 * minor incompatible change: VALUES declaration is disabled.
3025 * When issuing notes, the compiler now signals a condition of type
3026 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3027 restart for use in user handlers. It is expected that the
3028 COMPILER-NOTE condition will eventually become a condition
3029 supertype to a hierarchy of note types, which will then be
3030 handleable in a similar fashion. However, at the moment, no such
3031 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3032 but it's an implementation detail, not a classification for the
3034 * Changes in type checking closed the following bugs:
3035 ** type checking of unused values (192b, 194d, 203);
3036 ** template selection based on unsafe type assertions (192c, 236);
3037 ** type checking in branches (194bc).
3038 * A short form of VALUES type specifier has ANSI meaning (it has
3039 increased the number of situations when SBCL cannot perform type
3041 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3042 keywords or constants is permissible.
3043 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3044 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3045 classes. (thanks to Antonio Martinez)
3046 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3047 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3048 argument to a pretty-print stream. Instead, it leaves any such
3049 operations to DESCRIBE-OBJECT methods.
3050 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3052 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3053 lambda lists. (thanks to David Lichteblau)
3054 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3055 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3056 (thanks to Gerd Moellmann)
3057 * fixed bug 246: increased compilation speed of long
3058 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3060 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3062 * DEFCONSTANT now throws a condition of type
3063 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3064 constant to a non-EQL value; CONTINUE and ABORT restarts
3065 respectively change and preserve the value.
3066 * fixed bug 63: The code walker, part of the implementation of CLOS,
3067 is now better at handling symbol macros.
3068 * bug fix: There is no longer an internal implementation type named
3069 CL:LENGTH. (reported by Raymond Toy)
3070 * bug fix: In macro-like defining macros/special operators the
3071 implicit block does not enclose the lambda list.
3072 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3073 enclosed in parentheses) are not suitable as type specifiers, and
3074 their use properly signals an error now.
3075 * bug fix: An argument count mismatch for a type specifier in code
3076 being compiled no longer causes an unhandled error at compile
3077 time, but signals a compile-time warning.
3078 * fixed simple vector readable printing
3079 * bug fix: DESCRIBE takes more care over whether the class
3080 precedence list slot of a class is bound before accessing it.
3081 (reported by Markus Krummenacker)
3082 * bug fix: FORMATTER can successfully compile pretty-printer format
3083 strings which use variants of the ~* directive inside.
3084 * bug fix: SEARCH now applies its TEST predicate to the elements of
3085 the arguments in the correct order. (thanks to Wolfhard Buss)
3086 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3087 in type inference. (thanks to Robert E. Brown)
3088 * bug fix: Reading in symbols with an explicit package name of ""
3089 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3090 * fixed some bugs revealed by Paul Dietz' test suite:
3091 ** NIL is now allowed as a structure slot name.
3092 ** Arbitrary numbers, not just REALs, are allowed in certain
3093 circumstances in LOOP for-as-arithmetic clauses.
3094 ** Multiple class redefinitions before slot access no longer
3095 causes a type error.
3096 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3097 association between the name and a class.
3098 ** Generic functions with non-standard method-combination and over
3099 five methods all of which return constants no longer return NIL
3100 after the first few invocations. (thanks to Gerd Moellmann)
3101 ** CALL-NEXT-METHOD with no arguments now passes the original
3102 values of the arguments, even in the presence of assignment.
3103 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3105 ** DISASSEMBLE works with closures and funcallable instances.
3106 ** ADD-METHOD now returns the generic function, not the new method.
3107 ** FIND-METHOD signals an error if the lengths of the specializers
3108 is incompatible with the generic function, even if the ERRORP
3110 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3111 which its argument is a member.
3112 ** DEFCLASS only redefines the class named by its class-name
3113 argument if that name is the proper name of the class;
3114 otherwise, it creates a new class.
3115 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3116 of the UNBOUND-SLOT condition to the name of the slot.
3117 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3118 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3119 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3120 SLOT-MAKUNBOUND in the specified fashion.
3122 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3123 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3124 cross-compilation host. As a consequence, we can now bootstrap our
3125 way up to SBCL starting with a bare gcc toolchain and human-readable
3126 source code (first the source to CLISP, then the source to SBCL).
3127 * A contributed module containing a partial implementation of the
3128 simple-streams interface has been included. (thanks to Rudi
3130 * A contributed module implementing the RFC1321 Message Digest
3131 Algorithm, known as MD5, has been included.
3132 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3133 no longer has any effect, as the code controlled by this feature
3134 has been deleted. (As far as we know, no-one has ever built using
3135 this feature, and its semantics were confused in any case).
3136 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3137 work on conditions (as required by the ANSI specification),
3138 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3139 expected behaviour on conditions. Users should note, however,
3140 that such behaviour is not required by the ANSI specification,
3141 and so use of this behaviour may render their code unportable.
3142 * fixed some bugs revealed by Paul Dietz' test suite:
3143 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3145 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3146 ** SXHASH on condition objects no longer returns NIL.
3147 ** :ALLOCATION :CLASS slots are better treated; their values are
3148 updated on class redefinition, and initforms inherited from
3149 superclasses are applied.
3150 ** REMOVE-METHOD returns its generic function argument even when
3151 no method was removed.
3152 ** SHARED-INITIALIZE now initializes the values of the requested
3153 slots, including those with :ALLOCATION :CLASS.
3154 ** ALLOCATE-INSTANCE now works on structure classes defined via
3155 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3157 ** SLOT-EXISTS-P now works on conditions, as well as structures
3159 ** MAKE-LOAD-FORM now has the required methods on
3160 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3161 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3162 keyword, and now implements the SLOT-NAMES argument.
3163 ** methods with &OPTIONAL arguments no longer allow too many
3164 arguments to be passed in the call without error.
3165 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3166 option is consistent with the required arguments of the generic
3167 function lambda list.
3168 * bug fix: REQUIRE accepts a string designator. (Thanks to
3170 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3171 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3172 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3173 * bug fix: The compiler now checks for duplicated variables in macro
3175 * bug fix: SETQ on globals returns the correct value.
3176 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3177 not a COMPILER-ERROR (followed by some other strange error on
3178 choosing the CONTINUE restart).
3179 * bug fix: make.sh and friends are now more consistent in the way that
3180 they look for GNU "make".
3182 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3183 * experimental native threads support (on x86 Linux >=2.4 only).
3184 This is not compiled in by default: you need to add :SB-THREAD to
3185 the target features. See the "Beyond ANSI" chapter of the manual
3187 * fix for longstanding nonANSIism: The old distinction between
3188 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3189 The return value from CL:FIND-CLASS is now a CLOS class, and
3190 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3191 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3192 * An interface to the MetaObject Protocol, as described in Kiczales,
3193 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3194 MIT Press, 1991, is available from the SB-MOP package.
3195 * incompatible change: the SB-PCL package should now be considered
3196 a private implementation detail, and no longer a semi-private MOP
3198 * minor incompatible change: due to rearrangement for threads, the
3199 control stack and binding stack are now allocated at arbitrary
3200 addresses instead of being hardcoded per-port. Users affected by
3201 this probably have to be doing advanced things with shared
3202 libraries, and will know who they are.
3203 * minor incompatible change: Previously, all --eval forms used were
3204 processed with READ before any of them were processed with EVAL.
3205 Now each --eval form is processed with both READ and EVAL before
3206 the next --eval form is processed. (Thus package operations like
3207 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3208 work as the user might reasonably expect.)
3209 * minor incompatible change: *STANDARD-INPUT* is now only an
3210 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3212 * minor incompatible change: Y-OR-N-P is now character-oriented,
3213 not line oriented. Also, YES-OR-NO-P now works without errors.
3214 (thanks to Antonio Martinez)
3215 * sb-aclrepl module improvements: an integrated inspector, added
3216 repl features, and a bug fix to :trace command.
3217 * Known functions, which cannot be open coded by the backend, are
3218 considered to be able to check types of their arguments. (fixing
3219 a bug report by Nathan J. Froyd)
3220 * fixed a bug in computing method discriminating functions: It is
3221 now possible to define methods specialized on classes which have
3222 forward-referenced superclasses. (thanks to Gerd Moellmann)
3223 * fixed evaluation order in optional entries (reported by Gilbert
3225 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3226 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3227 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3228 required-by-AMOP NAME argument, as well as CLASS and
3229 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3230 * fixed bug 20: DEFMETHOD can define methods using names that are
3231 not the proper names of classes to designate class specializers.
3232 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3233 Unix file descriptors, instead of blowing up. (thanks to Antonio
3235 * Garbage collection refactoring: user-visible change is that a
3236 call to the GC function during WITHOUT-GCING will not do garbage
3237 collection until the end of the WITHOUT-GCING. If you were doing
3238 this you were probably losing anyway.
3239 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3240 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3241 TYPEP the latter but not the former.
3242 * The compiler issues a full WARNING for calls to undefined functions
3243 with names from the CL package.
3244 * MAP-INTO for a vector destination is open coded. (reported by
3245 Brian Downing on c.l.l)
3246 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3247 documentation string.
3248 * fixed some bugs revealed by Paul Dietz' test suite:
3249 ** COPY-ALIST now signals an error if its argument is a dotted
3251 ** Condition slots are now accessed more correctly in the presence
3252 of multiple initargs for a given slot.
3253 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3254 exclude restarts of the same name associated with a different
3256 ** DEFCLASS of forward-referenced classes with another
3257 forward-referenced class in the superclasses list no longer
3259 ** Condition slots are now initialized once each, not multiple
3260 times. (thanks to Gerd Moellmann)
3261 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3263 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3264 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3265 arguments contain duplicated elements.
3266 ** RESTART-CASE understands local macros.
3267 ** RESTART-CASE associates exactly its own restarts with a condition.
3268 ** ENDP in safe mode checks its argument to be of type LIST.
3269 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3270 in question is unbound.
3271 ** Optimized MAKE-INSTANCE functions no longer cause internal
3272 assertion failures in the presence of duplicate initargs.
3273 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3274 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3275 propagated even in the ctor optimized implementation of
3277 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3279 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3280 * a better implementation of SXHASH on (simple) bit vectors,
3281 measured both in execution speed and in distribution of results
3282 over the positive fixnums, has been installed. Likewise, a better
3283 implementation of EQUAL for simple bit vectors is now available.
3284 * fixed CEILING optimization for a divisor of form 2^k.
3285 * fixed bug 240 (emitting extra style warnings "using the lexical
3286 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3287 by Antonio Martinez)
3288 * fixed SXHASH, giving different results for NIL depending on type
3289 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3290 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3291 variable. (found by Rolf Wester)
3292 * SBCL does not ignore type declarations for special
3293 variables. (reported by rif on c.l.l 2003-03-05)
3294 * some bug fixes in contrib/sb-aclrepl/
3295 * fixed some bugs revealed by Paul Dietz' test suite:
3296 ** a bug in the CONS type specifier, whereby the CAR and CDR
3297 types got intertwined, has been fixed;
3298 ** the type system is now able to reason about the interaction
3299 between INTEGER and RATIO types more completely;
3300 ** APPEND, [N]REVERSE and NRECONC check that those their
3301 arguments, which must be proper lists, are really so;
3302 ** An array specialized to be unable to hold elements has been
3303 implemented, as required -- yes, really -- by ANSI;
3304 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3305 on malformed property lists;
3307 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3308 * incompatible packaging change: in line with Unix convention,
3309 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3310 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3311 * REQUIRE and PROVIDE are now optionally capable of doing something
3312 useful. See the documentation string for REQUIRE.
3313 * infrastructure for a managed SBCL contrib system: contributed
3314 modules in this release include:
3315 ** the ASDF system definition facility;
3316 ** an interface to the BSD Sockets API;
3317 ** an ACL-like convenience interface to the repl;
3318 (thanks to Kevin Rosenberg)
3319 ** an implementation of ROTATE-BYTE, with efficient implementation
3321 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3322 gives a closer approximation to the right answer than previously.
3323 (thanks to Raymond Toy)
3324 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3325 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3326 optional environment arguments, as required by ANSI.
3327 * fixed bugs in other functions taking environment objects, allowing
3328 calls with an explicit NIL environment argument to be compiled
3330 * fixed bug 228: primary return values from
3331 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3332 COMPILE or FUNCTION.
3333 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3334 :TYPE will no longer signal errors on innocuous objects.
3335 * fixed bug 231b: SETQ is better at respecting type declarations in
3336 the lexical environment.
3337 * fixed a bug in DEFCLASS: classes named by symbols with no or
3338 unprintable packages can now be defined.
3339 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3340 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3341 * fixed compiler failure related to checking types of functions.
3342 (reported by Robert E. Brown)
3343 * the compiler is now much more consistent in its error-checking
3344 treatment of bounding index arguments to sequence functions: in
3345 (SAFETY 3) code, errors will be signalled in almost all cases if
3346 invalid sequence bounding indices are passed to functions defined
3347 by ANSI to operate on sequences.
3348 * fixed a bug in the build procedure: documentation of SBCL-specific
3349 packages is now preserved and available in the final Lisp image.
3350 * lifted FDEFINITION lookup out of loops in the implementation of
3351 many list operations. (thanks to Robert E. Brown)
3352 * fixed a bug in the reader: the #n# reader macro now works for
3353 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3354 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3355 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3356 E. Brown and Tony Martinez respectively)
3357 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3358 SLOT-BOUNDP: in particular, it is now able to optimize them much
3359 better, and is now not vulnerable to having packages renamed.
3360 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3361 always called when a slot is not present in an instance. (thanks
3363 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3364 longer signal an error on unreading a character following EOF on
3365 the previous constituent stream. (thanks to Tony Martinez)
3366 * fixed some bugs revealed by Paul Dietz' test suite:
3367 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3368 not just nonnegative fixnums;
3369 ** the logical bit-array operators such as BIT-AND now accept an
3370 explicit NIL for their "opt-arg" argument (to indicate a
3371 freshly-consed result bit-array);
3372 ** ELT now signals an error on an invalid sequence index in safe
3374 ** the type system is now cleverer about negations of numeric
3375 types, and consequently understands the BIGNUM and RATIO types
3377 ** the type system is now cleverer about the interaction between
3378 INTEGER and RATIO types: while bugs still remain, many more
3379 cases are accurately computed;
3380 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3381 if it is in the last clause;
3382 ** CONSTANTLY now correctly returns a side-effect-free function in
3384 ** DECLARE is no longer treated as a special-operator; in
3385 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3386 * incremented fasl file version number due to the change in the
3387 DEFSTRUCT-SLOT-DESCRIPTION structure.
3389 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3390 * minor incompatible change: code processed by the "interpreter" or
3391 EVAL now has a compilation optimization policy of (DEBUG 2)
3392 (changed from (DEBUG 1)) to improve debuggability of interactive
3393 development, and to allow the use of the debug RETURN command in
3395 * an experimental implementation of the RETURN command for the
3396 debugger has been included. (thanks to Frederik Kuivinen)
3397 * fixed bug 62: constraints were not propagated into a loop.
3398 * fixed bug in embedded calls of SORT (reported and investigated by
3400 * fixed some bugs revealed by Paul Dietz' test suite:
3401 ** printing and reading of arrays with some dimensions having
3402 length 0 (thanks to Gerd Moellmann);
3403 ** BOA constructor with &AUX argument without a default value does
3404 not cause a type error;
3405 ** CONSTANTP now returns true for all self-evaluating objects.
3407 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3408 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3409 accessors that are related by inheritance, as specified in the
3410 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3411 Valtteri Vuorikoski)
3412 * The compiler is now able to inline functions that were defined in
3413 a complex lexical environment (e.g. inside a MACROLET).
3414 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3416 * Support for the upcoming FreeBSD-5.0 release has been included.
3417 (thanks to Dag-Erling Smorgrav)
3418 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3419 effect when it is not in a toplevel context.
3420 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3421 stand a better chance of being correct. (thanks to Gerd
3423 * fixed bug in COERCE, which now signals an error on coercing a
3424 rational to a bounded real type which excludes the expected
3426 * The compiler is now able to derive types more accurately from the
3427 COERCE and COMPILE functions.
3428 * fixed bug 223: functional binding is considered to be constant
3429 only for symbols in the CL package.
3430 * fixed bug 231: SETQ did not check the type of a variable being set
3431 (reported by Robert E. Brown)
3432 * A new optimization for MAKE-INSTANCE has been included, fixing
3433 various bugs (including relating to :ALLOCATION :CLASS slots and
3434 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3436 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3437 ** As required by ANSI, LOOP now disallows anonymous collection
3438 clauses such as COLLECT I in conjunction with aggregate boolean
3439 clauses such as THEREIS (= I 1);
3440 ** LOOP now signals an error when any variable is reused in the
3441 same loop (including the potentially useful construct analogous
3442 to WITH A = 1 WITH A = (1+ A);
3443 ** IT is only a special loop symbol within the first clause of a
3444 conditional loop clause;
3445 ** LOOP with a typed iteration variable over a hashtable now
3446 signals a type error iff it should.
3447 * fixed some other bugs revealed by Paul Dietz' test suite:
3448 ** FILE-STREAM now names the class previously known as FD-STREAM;
3449 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3450 argument) no longer signals an error;
3451 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3452 of the slot symbol, rather than using the current package
3453 ((:CONC-NAME "") continues to intern the slot's name in the
3455 * incremented fasl file version number, because of the incompatible
3456 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3457 of the new implementation of DEFINE-COMPILER-MACRO.
3459 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3460 * Support for building SBCL for MIPS platforms running in
3461 little-endian mode has now been checked in, and basic
3462 functionality on said platforms verified.
3463 * minor incompatible change: PCL now records the pathname of a file
3464 in which methods and the like are defined, rather than its
3466 * minor incompatible change: TRUENAME now considers the truename of
3467 a file naming a directory to be the pathname with :DIRECTORY
3468 component indicating that directory.
3469 * minor incompatible change: a NAMED clause in the extended form of
3470 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3471 reason for the previous behaviour is unclear.
3472 * more systematization and improvement of CLOS and MOP conformance
3473 in PCL (thanks to Gerd Moellman and Pierre Mai):
3474 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3476 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3477 primary methods with no specializers;
3478 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3480 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3481 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3482 FORWARD-REFERENCED-CLASSes; error reporting on
3483 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3485 ** SXHASH on CLOS instances now uses a slot internal to the
3486 instance to return different numbers on distinct instances,
3487 while preserving the same return value through invocations of
3489 ** DEFMETHOD signals errors when methods with longer incongruent
3490 lambda lists are added to generic functions;
3491 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3492 CLASS, as specified in AMOP;
3493 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3494 on the order returned by the primary method for classes of
3495 class STANDARD-CLASS;
3496 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3497 * fixed some bugs shown by Paul Dietz' test suite:
3498 ** DOLIST puts its body in TAGBODY;
3499 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3501 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3502 value producing form;
3503 ** if more variables are given to PROGV than values, extra
3504 variables are bound and made to have no value;
3505 ** NSUBSTITUTE on list arguments gets the right answer with
3507 ** ELT signals an error of type TYPE-ERROR when the index argument
3508 is not a valid sequence index;
3509 ** LOOP signals (at macroexpansion time) an error of type
3510 PROGRAM-ERROR when duplicate variable names are found;
3511 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3512 ** FUNCALL of special-operators now cause an error of type
3514 ** PSETQ now works as required in the presence of side-effecting
3515 symbol-macro places;
3516 ** NCONC accepts any object as its last argument;
3517 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3519 ** loop-for-as-package does not require a package to be explicitely
3521 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3523 * fixed bug 166: compiler preserves "there is a way to go"
3524 invariant when deleting code.
3525 * fixed bug 172: macro lambda lists with required arguments after
3526 &REST arguments now cause an error to be signalled. (thanks to
3528 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3529 bug. (thanks to Matthew Danish)
3530 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3532 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3533 arguments are lexically rebound. (thanks to Gerd Moellmann and
3535 * fixed bug 194: error messages are now more informative when there
3536 is no primary method applicable in a call to a generic
3537 function. (thanks to Gerd Moellmann)
3538 * fixed bug in command line argument checking (thanks to Julian
3540 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3541 (thanks to Matthew Danish)
3542 * incremented fasl file version number, because of the
3543 SXHASH-related changes in the layout of CLOS data structures
3545 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3546 * minor incompatible change: The runtime (the Unix executable named
3547 "sbcl") is now much pickier about the .core files it will load.
3548 Essentially it now requires .core files to descend from the same
3549 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3550 as the runtime does. (The intent is to prevent the crashes which
3551 can occur, and which can even be reported as mysterious failures,
3552 when people patch the sources or change the build parameters
3553 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3554 sbcl and .core files.)
3555 * fixed bug: VALUES-LIST is no longer optimized away.
3556 * fixed bug 142: The FFI conversion of C string values to Lisp
3557 string values no longer conses excessively. (thanks to Nathan
3558 Froyd porting Raymond Toy's fix to CMU CL)
3559 * began to systematize and improve MOP conformance in PCL (thanks to
3560 Nathan Froyd, Gerd Moellman and Pierre Mai):
3561 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3563 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3564 ** FINALIZE-INHERITANCE is now called on class finalization;
3565 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3566 argument precedence order.
3567 * fixed bug 202: The compiler no longer fails on functions whose
3568 derived types contradict their declared type.
3569 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3570 so it can be non-toplevel.
3571 * The fasl file version number has changed (because of the new
3572 implementation of DEFMACRO).
3573 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3574 safe code, that any length requirement by their type-specifier
3575 argument is valid. The exceptions to this are described in bug
3577 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3578 functions now check the types of their inputs as required by ANSI.
3579 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3580 introduced symbol is DECLAREd to be SPECIAL.
3581 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3582 * fixed bug: :COUNT argument to sequence functions may be negative
3583 * fixed bug: body of DO-SYMBOLS may contain declarations
3584 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3585 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3586 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3587 symbol macro only once
3588 * fixed printing of call frame when argument list is unavailable
3589 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3590 * compiler no longer signals WARNING on unknown keyword
3593 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3594 * A beta-quality port to the mips architecture running Linux,
3595 based on the old CMUCL backend, has been made. It has been tested
3596 on a big-endian kernel, and works sufficiently well to be able to
3597 rebuild itself; it has not been tested in little-endian mode.
3598 * fixed an inconsistency between gencgc.c and purify.c which made
3599 dumping/loading .core files unreliable
3600 * fixed bug 120a: The compiler now deals correctly with IFs where
3601 the consequent is the same as the alternative, instead of
3602 misderiving the return type. (thanks to Alexey Dejneka)
3603 * fixed bug 113: Logical pathnames are now dumpable (the logical
3604 host is resolved at load-time, throwing an error if it is not
3606 * fixed bug 174: FORMAT's error message is slightly clearer when a
3607 non-printing character is used in a format directive.
3608 * fixed several bugs in compiler checking of type declarations, i.e.
3609 violations of the Python "declarations are assertions" principle
3610 (thanks to Alexey Dejneka)
3611 * fixed several bugs in PCL's error checking (thanks to Gerd
3613 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3615 * fixed bug in compilation of functions as first class values
3616 (thanks to Antonio Martinez-Shotton)
3617 * The compiler's handling TYPE-ERRORs which it can prove will
3618 inevitably happen at runtime has been cleaned up and corrected
3619 in several ways. (thanks to Alexey Dejneka)
3620 * improved argument type checking for various basic arithmetic
3621 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3622 lavished on them in the past that they can be compiled in many
3623 ways in different special cases
3624 * fixed bug 181: compiler checks validity of user supplied type
3626 * cleaned up code flushing in optimization: Function calls which
3627 should signal errors for safety purposes (e.g. which ANSI says
3628 should signal errors when their arguments are of incorrect type)
3629 are no longer optimized away.
3630 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3631 * incremented fasl file version number, because changes in the
3632 implementation of sequence functions like COERCE caused
3633 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3634 in old inline expansions) to become undefined. (Actually these
3635 changes were later undone, so we might very well be binary
3636 compatible with 0.7.7 after all, but leaving the version number
3637 incremented seemed like the simplest and most conservative
3640 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3641 * An alpha-quality port to the parisc architecture running Linux,
3642 based on the old CMUCL backend, has been made. This, even more so
3643 than the other backends, should be considered still a work in
3644 progress; known problems include that the Linux kernel in 64-bit
3645 mode does not propagate the correct sigcontext structure to
3646 userspace, and consequently SBCL on a parisc64 kernel will not
3648 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3649 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3650 suppressed.) Also now that the compiler is looking at declarations
3651 in the environment, it checks optimization declarations as well,
3652 and suppresses inlining when (> DEBUG SPEED).
3653 * More fixes have been made to treatment of floating point exception
3654 treatment and other Unix signals. In particular, floating point
3655 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3656 * The detection and handling of control stack exhaustion (infinite
3657 or very deeply nested recursion) has changed. Stack exhaustion
3658 detection is now done by write-protecting pages at the OS level
3659 and applies at all optimization settings; when found, a
3660 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3661 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3662 be caught using IGNORE-ERRORS.
3663 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3664 correctly when transforming recursive LABELS functions to LETs.
3665 (thanks to Alexey Dejneka)
3666 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3667 that are names of constants or global variables.
3668 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3669 alien routines with docstrings.
3670 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3671 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3673 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3674 (thanks to Eric Marsden)
3675 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3676 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3677 * Bug fix: The SPARC backend can now compile functions involving
3678 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3679 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3680 to a functions expecting a PATHNAME-DESIGNATOR.
3681 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3682 lambda lists. (thanks to Alexey Dejneka)
3683 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3684 functions so that EXPORTing the name of the class doesn't cause
3685 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3686 to Antonio Martinez for reporting this)
3687 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3688 constant-folded without failing an assertion. (thanks to Einar
3689 Floystad Dorum for reporting this)
3690 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3691 types are now dealt with more correctly by the compiler.
3692 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3693 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3694 behaviour analogous to RENAME-FILE. This puts its behaviour more
3695 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3697 * The fasl file version number has changed again. (because of the
3698 bug fix involving the names of PCL MAKE-INSTANCE functions)
3700 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3701 * bug fix: Floating point exceptions are treated much more
3702 consistently on the x86/Linux and PPC/Linux platforms.
3703 * Array initialization with :INITIAL-ELEMENT is now much faster for
3704 cases when the compiler cannot open code the array creation, but
3705 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3706 array accesses have also seen a speed increase.
3707 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3708 specified. (This was at the root of some bad interactions between
3709 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3711 * bug fix: Internal error arguments for undefined functions are now
3712 computed correctly on the PPC/Linux platform.
3713 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3714 Raymond Toy's patch for CMU CL)
3715 * Support for the Solaris 9 operating environment has been included
3716 (thanks to Daniel Merritt)
3717 * A very ugly but hopefully complete draft of the missing FFI chapter
3718 of the manual has been created by reformatting the corresponding
3719 CMU CL manual chapter into (currently very ugly and incoherent)
3720 DocBook and bringing it up to date for SBCL behavior. Thus, the
3721 manual is now essentially complete, at least by my extreme
3722 once-and-only-once standards, whereby it's acceptable to refer to
3723 the doc strings of SB-EXT functions as the primary documentation.
3724 * The fasl file version number has changed again, due to cleanup of
3725 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3726 is no longer a static symbol.)
3728 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3729 * SBCL now builds with OpenMCL (version 0.12) as the
3730 cross-compilation host; also, more progress has been made toward
3731 bootstrapping under CLISP.
3732 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3734 * bug 158 fixed: The compiler can now deal with integer loop
3735 increments different from 1; fixing this turned out also to fix
3737 * bug 169 fixed: no more bogus warnings about using lexical bindings
3738 despite the presence of perfectly good SPECIAL declarations (thanks
3739 to David Lichteblau)
3740 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3741 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3742 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3744 * bug 180 fixed: Method combination specifications no longer ignore
3745 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3746 * bug fix: Structure type predicate functions now check their argument
3747 count as they should.
3748 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3749 correctly. (thanks to Pierre Mai)
3750 * minor incompatible change: The --noprogrammer option is deprecated
3751 in favor of the new --disable-debugger option, which is very similar.
3752 (The major difference is that it takes effect at a slightly different
3753 time at startup, causing handling of errors in --sysinit and
3754 --userinit files will be affected differently.) The
3755 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3756 been added to allow this functionality to be controlled from ordinary
3757 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3758 maintainers, who might want to run non-interactive scripts to
3759 build SBCL cores which will later be used interactively.)
3760 * minor incompatible change: The LOAD function no longer, when given
3761 a wild pathname to load, loads all files matching that pathname.
3762 Instead, an error of type FILE-ERROR is signalled.
3764 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3765 * bug 147 fixed: The compiler preserves its block link/count
3766 invariants more correctly now so that it doesn't crash. (thanks
3768 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3770 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3771 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3772 no-value-for-this-character as NIL (as specified by ANSI).
3773 * HOST-NAMESTRING on physical pathnames now returns a string that is
3774 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3775 (thanks to Christophe Rhodes)
3776 * The Alpha port handles icache flushing more correctly. (thanks to
3778 * More progress has been made toward bootstrapping under CLISP. (thanks
3779 to Christophe Rhodes)
3780 * The fasl file format has changed again, because dynamic loading
3781 on OpenBSD (which has non-ELF object files) motivated some cleanups
3782 in the way that foreign symbols are transformed and passed around.
3783 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3784 is no longer treated as whitespace by the reader, but instead as
3785 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3786 |A
\7fB|, instead of A as it used to.
3788 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3789 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3790 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3791 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3792 this as of 0.7.1.45, but was left out of the previous news section
3793 (thanks to Dan Barlow)
3794 * SBCL now runs on the Solaris operating system on SPARC architectures
3795 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3796 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3797 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3798 * SPARC backend cleanups, allowing builds of cores optimized for V8
3799 and V9 SPARCS, and also emission of code targeted to a particular
3800 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3802 * SBCL is closer to bootstrapping under CLISP, thanks to various
3803 fixes by Christophe Rhodes.
3804 * The fasl file format has changed again, to allow the compiler's
3805 INFO database to support symbol macros.
3806 * The user manual (in doc/) is formatted into HTML more nicely.
3807 (thanks to coreythomas)
3808 * The system is smarter about SUBTYPEP relationships, especially
3809 those involving NOT types (including types such as ATOM which are
3810 represented internally using NOT types). Thus SUBTYPEP is less
3811 likely to return (VALUES NIL NIL) in general, and in particular
3812 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3814 * The fasl file format has changed again, because the internal
3815 representation of types now includes a new slot to support the new
3816 SUBTYPEP-of-NOT-types logic.
3817 * (not a change in the main branch of SBCL, but a related prototype
3818 which can hopefully be merged into the main branch of SBCL in the
3819 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3820 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3821 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3822 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3823 dispatch macro character combinations. (thanks to Alexey Dejneka)
3824 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3825 that superficially look like logical namestrings correctly.
3826 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3828 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3829 default supplied-p) for &optional and &key arguments. (thanks to
3832 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3833 * incompatible change: The compiler is now less aggressive about
3834 tail call optimization, doing it only when (> SPACE DEBUG) or
3835 (> SPEED DEBUG). (This is an incompatible change because there are
3836 programs which relied on the old CMU-CL-style behavior to optimize
3837 away their unbounded recursion which will now die of stack overflow.)
3838 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3839 for non-GENCGC systems has been increased to 20M (since that
3840 seems much closer to the likely performance optimum for modern
3841 systems than the old 4M value was)
3842 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3843 and *DEBUG-PRINT-LEVEL*
3844 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3845 of CMU CL's support for SPARC, and various endianness and other
3846 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3847 * new syntactic sugar for the Unix command line: --load foo.bar is now
3848 an alternate notation for --eval '(load "foo.bar")'.
3850 ** The system now detects stack overflow and handles it gracefully,
3851 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3852 optimization settings. (This is a good thing in general, and
3853 its introduction in this version should be particularly timely
3854 for anyone whose code fails because of suppression of tail
3856 ** The system now hunts for the C variable "environ" in a more
3857 devious way, to avoid segfaults when the C library version differs
3858 between compile time and run time. (thanks to Christophe Rhodes)
3859 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3860 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3861 ** The compiler no longer issues bogus style warnings for undefined
3862 classes in the same source file as the DEFCLASSes which defined
3863 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3865 ** fixes in CONDITION class precedence list for undefined function
3866 errors (thanks to Alexei Dejneka)
3867 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3868 correctly. (thanks to Dan Barlow)
3869 ** portability fixes aiming at bootstrapping under CLISP (thanks
3870 to Dave McDonald and Christophe Rhodes)
3871 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3872 ** fixes in type translation and and type inference (thanks to
3874 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3875 ** various fixes in the new ports (thanks to Dan Barlow)
3876 * several changes related to debugging:
3877 ** suppression of tail recursion, as noted above
3878 ** stack overflow detection, as noted above
3879 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3880 is now the default. (For some time encapsulation has been more
3881 reliable than the breakpoint-based :ENCAPSULATE NIL
3882 implementation, at least on X86 systems; and I just noticed that
3883 encapsulation also seems closer to the spirit of the ANSI
3886 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3888 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3889 up properly again. (There was a packaging bug in 0.7.0 which
3890 left their definitions in SB-SYS::LOAD-FOREIGN and
3891 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3892 vital for most things which interface to C-level interfaces,
3893 like extensions working with sockets or databases or
3894 Perl-compatible regexes or whatever, and the need to fix
3895 this bug is the main reason that 0.7.1 was released so
3897 ** DEFGENERIC is now choosier about the methods it redefines, so that
3898 reLOADing a previously-LOADed file containing DEFGENERICs does
3899 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3900 cycle now works as it should. (thanks to Alexey Dejneka)
3901 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3902 fixed by Christophe Rhodes. (He actually submitted this patch
3903 months ago, and I delayed until after 0.7.0.)
3904 ** Bug 111 (internal compiler confusion about runtime checks on
3905 FUNCTION types) was fixed by Alexey Dejneka.
3906 * Some internal cleanups (getting rid of variables which aren't
3907 needed now that the byte interpreter is gone) caused the fasl
3908 file format number to change again.
3910 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3911 * major incompatible change: The default fasl file extension, i.e. the
3912 default extension for files produced by COMPILE-FILE, has changed
3913 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3915 ** There are many changes in the implementation of the compiler.
3916 SBCL is now essentially a compiler-only implementation of ANSI
3917 Common Lisp. EVAL still "interprets" a few special cases, but
3918 almost all the interesting cases are handled by creating
3919 a LAMBDA expression, calling COMPILE on it, then calling
3920 FUNCALL on the result.
3921 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3922 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3923 Since the code is newer, there might still be some new bugs
3924 (though not as many as before Martin Atzmueller's fixes:-). But
3925 the new code is substantially simpler and clearer, and hopefully
3926 any remaining bugs will be simpler, less fundamental, and more
3927 fixable then the bugs in the old code.
3928 ** The revised compiler is still a little unsteady on its feet.
3930 *** The debugging information it produces (particularly the names
3931 of FUNCTION objects) is sometimes much less useful than what
3932 the old compiler produced.
3933 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3934 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3935 within a MACROLET) has been temporarily weakened.
3936 ** There are new compiler optimizations for various functions:
3937 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3938 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3939 *** the math functions TRUNCATE, FLOOR, and CEILING
3940 *** the function-of-all-trades COERCE
3941 Mostly these should be transparent, but there's one
3942 potentially-annoying problem (bug 117): when the compiler
3943 inline-expands a function and does type analysis on the result,
3944 it can create control paths which have type mismatches, and
3945 when it can't prove that those control paths aren't taken,
3946 it will issue WARNINGs about the type mismatches. This is
3947 a particular problem in practice for the new sequence functions.
3948 It's not clear how this should be fixed, and for now, a
3949 workaround is given in the entry for 117 in the BUGS file.
3950 ** (Because of the interaction between the two previous items --
3951 occasional inlining problems and new inline expansions -- some
3952 of the new sequence function optimizations won't really kick in
3953 completely until debugging information, and then inlining, are
3954 straightened out in some future version.)
3955 * minor incompatible changes:
3956 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3957 DIRECTORY no longer implicitly promotes NIL slots of its
3958 pathname argument to :WILD. In particular, when you ask for the
3959 contents of a directory (which you used to be able to do without
3960 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3961 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3962 ** changes in behavior that ANSI explicitly defines to be
3963 implementation dependent:
3964 *** The new compiler-only implementation still conforms with ANSI,
3965 but acts a little different than before. Besides the obvious
3966 changes in performance tradeoffs (that the cost per form passed
3967 to EVAL has gone up, and the cost per form executed by EVAL
3968 has gone down), the behavior of the system changes a little
3969 because there are no longer any interpreted function objects.
3970 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3971 e.g. doing COMPILE on the output of interactive DEFUN is
3973 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3975 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3976 has changed from (:CL) to NIL.
3977 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3978 e.g. #\Newline and #\Tab, don't have names specified in the
3979 ANSI Common Lisp standard, is now based on their ASCII symbolic
3980 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3981 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3982 are no longer used for output.
3983 ** changes in internal implementation constants:
3984 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3985 4 million. (If your application spends a lot of time GCing and
3986 you have a lot of RAM, you might want to experiment with
3987 increasing it even more.)
3988 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3989 However, almost all old code should still continue to work without
3990 immediate update, as SB-C-CALL is now a (deprecated) nickname
3992 ** Old operator names in the style DEF-FOO are now deprecated in
3993 favor of new corresponding names DEFINE-FOO, for consistency with
3994 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3995 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3996 internal symbols, but a few supported extensions like
3997 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3998 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3999 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4000 etc. as you get deeper into recursive calls to the debugger
4001 command loop, instead of the old "5]", "5]]", "5]]]"
4002 sequence. (I was motivated to do this when squabbles between
4003 ILISP and SBCL left me very deeply nested in the debugger. In the
4004 short term, this change will probably provoke more ILISP/SBCL
4005 squabbles, but hopefully it will be an improvement in the long run.)
4006 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4007 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4008 function, since declaiming return types involving aliens is
4009 (1) annoyingly messy to do by hand and (2) vital to efficient
4010 compilation of code which calls such functions.
4011 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4012 longer reexported by the SB-EXT package. They're solely useful
4013 for alien code, so it seems more logical that you should get
4014 them from the SB-ALIEN package, not in SB-EXT.
4015 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4016 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4017 features. Instead, the code that they used to control is always
4018 built into the system.
4019 * many other bug fixes
4020 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4021 advantage of the new EVAL-WHEN stuff and to clean them up in
4022 general, and they are now more ANSI-compliant in a number of
4023 ways. Martin Atzmueller is responsible for a lot of this.
4024 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4026 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4027 *** correct ERROR type for various file operations
4028 *** some fixes for Lisp streams
4029 *** DEFMETHOD syntax checking
4030 *** changing old weird representation of debug information as
4031 strings (which, among their other deficiencies, don't transform
4032 correctly when you rename packages, and don't change their
4033 print representation when you change things like *PACKAGE*
4034 and *PRINT-LENGTH*) to symbols and lists of symbols
4035 He also made several improvements and fixed several bugs in DESCRIBE.
4036 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4038 *** misbehavior of WRITE-STRING/WRITE-LINE
4039 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4040 and several other LOOP problems as well
4041 *** DIRECTORY when similar filenames are present
4042 *** DEFGENERIC with :METHOD options
4043 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4044 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4045 *** argument ordering in FIND with :TEST option
4046 *** mishandled package designator argument in APROPOS-LIST
4047 *** various problems in the backquote readmacro
4048 *** a bug in APROPOS
4049 *** probably some others that I'm not describing very well here,
4050 since the CVS log documents them by reference to sbcl-devel
4051 messages, and the SourceForge archives aren't working well.:-(
4052 ** Dan Barlow improved the Alpha port (and is making progress on the
4053 PPC port, for those of you who think different).
4054 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4055 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4056 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4057 continues to work on the SPARC port (for those of you in a position
4058 to look down upon our little PC-compatible boxes from a great height).
4059 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4060 and uses that copy, rather than the current dynamic value, when
4061 it's trying to decide whether to truncate output. Thus e.g.
4062 (let ((*print-lines* 50))
4063 (pprint-logical-block (stream nil)
4065 (let ((*print-lines* 8))
4066 (print (aref possiblybigthings i) stream)))))
4067 should now truncate the logical block only at 50 lines, instead of
4068 often truncating it at 8 lines, as it did before.
4069 * The doc/cmucl/ directory, containing old CMU CL documentation
4070 from the time of the fork, is no longer part of the base system.
4071 SourceForge has shut down its anonymous FTP service, and with it
4072 my original plan for distributing the old CMU CL documentation
4073 there. For now, if you need these files you can download an old
4074 SBCL source release and extract them from it.
4075 * The fasl file version number changed again, for dozens of reasons,
4076 some of which are apparent above.
4078 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4079 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4080 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4081 patch, so that DISASSEMBLE output is much nicer.
4082 * The code in the SB-PROFILE package now seems reasonably stable.
4083 I still haven't decided what the final interface should look like
4084 (I'd like PROFILE to interact cleanly with TRACE, since both
4085 facilities use function encapsulation) but if you have a need
4086 for profiling now, you can probably use it successfully with
4087 the current CMU-CL-style interface.
4088 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4089 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4090 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4091 link now returns the dangling link itself, and for similar
4092 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4093 link itself. (In these cases the old code signalled an error and
4094 looped endlessly, respectively.) Thus, DIRECTORY now works even
4095 in the presence of dangling and cyclic symbolic links.
4096 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4097 is now a supported extension again, since the consensus on
4098 sbcl-devel was that it can be useful for ordinary development
4099 work, not just for debugging SBCL itself.
4100 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4101 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4102 that currently-defined functions might be redefined later with
4103 different return types.
4104 * Hash tables can be printed readably, as inspired by CMU CL code
4105 of Eric Marsden and SBCL code of Martin Atzmueller.
4106 * better error handling in CLOS method combination, thanks to
4107 Martin Atzmueller porting Pierre Mai's CMU CL patches
4108 * more overflow fixes for >16Mbyte I/O buffers
4109 * A bug in READ has been fixed, so that now a single Ctrl-D
4110 character suffices to cause end-of-file on character streams.
4111 In particular, now you only need one Ctrl-D at the command
4112 line (not two) to exit SBCL.
4113 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4115 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4116 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4117 does the right thing.
4118 * The implementation of some type tests, especially for CONDITION
4119 types, is now tidier and maybe faster, due to CMU CL code
4120 originally by Douglas Crosher, ported by Martin Atzmueller.
4121 * Some math functions have been fixed, and there are new
4122 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4123 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4124 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4125 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4126 likely to be built into sbcl-0.7.0.)
4127 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4128 CL, ported to SBCL by Martin Atzmueller)
4129 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4130 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4131 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4132 * There's a new slam.sh hack to shorten the edit/compile/debug
4133 cycle for low-level changes to SBCL itself, and a new
4134 :SB-AFTER-XC-CORE target feature to control the generation of
4135 the after-xc.core file needed by slam.sh.
4136 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4137 COMPILE-FILE is no longer supported, so that now every function
4138 gets an entry point, so that block compilation looks a little
4139 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4140 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4141 returns the number of bytes consed since the system started,
4142 rather than the number consed since the first time the function
4143 was called. (The new definition parallels ANSI functions like
4144 CL:GET-INTERNAL-RUN-TIME.)
4145 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4146 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4147 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4148 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4149 * Fasl file version numbers are now independent of the target CPU,
4150 since historically most system changes which required version
4151 number changes have affected all CPUs equally. Similarly,
4152 the byte fasl file version is now equal to the ordinary
4155 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4156 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4157 is no longer recognized. I apologize for this, because it was
4158 listed in SB-EXT as a supported extension, but I found that
4159 its existing behavior was poorly specified, as well as incorrectly
4160 specified, and it looked like too much of a mess to straighten it
4161 out. I have enough on my hands trying to get ANSI stuff to work..
4162 * many patches ported from CMU CL by Martin Atzmueller, with
4163 half a dozen bug fixes in pretty-printing and the debugger, and
4164 half a dozen others elsewhere
4165 * fixed bug 13: Floating point infinities are now supported again.
4166 They might still be a little bit flaky, but thanks to bug reports
4167 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4168 as flaky as they were.
4169 * The --noprogrammer command line option is now supported. (Its
4170 behavior is slightly different in detail from what the old man
4171 page claimed it would do, but it's still appropriate under the
4172 same circumstances that the man page talks about.)
4173 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4174 are now supported, and enabled by default. Thus, the compiler can
4175 handle many floating point and complex operations much less
4176 inefficiently. (Thus e.g. you can implement a complex FFT
4178 * The compiler now detects type mismatches between DECLAIM FTYPE
4179 and DEFUN better, and implements CHECK-TYPE more correctly, and
4180 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4181 thanks to patches from Martin Atzmueller.
4182 * various fixes to make the cross-compiler more portable to
4183 ANSI-conforming-but-different cross-compilation hosts (notably
4184 Lispworks for Windows, following bug reports from Arthur Lemmens)
4185 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4186 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4187 fixed thanks to Pierre Mai's CMU CL patches.
4188 * fixes to keep the system from overflowing internal counters when
4189 it tries to use i/o buffers larger than 16M bytes
4190 * fixed bug 45a: Various internal functions required to support
4191 complex special functions have been merged from CMU CL sources.
4192 (When I was first setting up SBCL, I misunderstood a compile-time
4193 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4194 * improved support for type intersection and union, fixing bug 12
4195 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4196 more obscure bugs as well
4197 * some steps toward byte-compiling non-performance-critical
4198 parts of the system, courtesy of patches from Martin Atzmueller
4199 * Christophe Rhodes has made some debian packages of sbcl at
4200 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4201 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4202 stable, but are nonetheless usable. When he's ready, I'd be happy
4203 to add them to the SourceForge "File Releases" section. (And if
4204 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4205 * new fasl file format version number (because of changes in
4206 internal representation of (OR ..) types to accommodate the new
4207 support for (AND ..) types, among other things)
4209 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4210 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4212 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4214 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4215 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4216 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4217 * DESCRIBE now works on structure objects again.
4218 * Most function call argument type mismatches are now handled as
4219 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4220 know whether the function will be redefined before the call is
4221 executed. (The compiler could flag local calls with full WARNINGs,
4222 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4223 it doesn't keep track of enough information to know whether calls
4224 are local in this sense.)
4225 * Compiler output is now more verbose, with messages truncated
4226 later than before. (There should be some supported way for users
4227 to override the default verbosity, but I haven't decided how to
4228 provide it yet, so this behavior is still controlled by the internal
4229 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4230 src/compiler/ir1util.lisp.)
4231 * Fasl file format version numbers have increased again, because
4232 support for the Gray streams extension changes the layout of the
4233 system's STREAM objects.
4234 * The Gray subclassable streams extension now works, thanks to a
4235 patch from Martin Atzmueller.
4236 * The full LOAD-FOREIGN extension (not just the primitive
4237 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4238 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4239 but like most other programs, it defaults to copying the Unix
4240 environment from the original process instead of starting the
4241 new process in an empty environment.
4242 * Extensions which manipulate the Unix environment now support
4243 an :ENVIRONMENT keyword option which doesn't smash case or
4244 do other bad things. The CMU-CL-style :ENV option is retained
4245 for porting convenience.
4246 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4247 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4249 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4251 * A patch from Martin Atzmueller seems to have solved the SIGINT
4252 problem, and as far as we know, signal-handling now works cleanly.
4253 (If you find any new bugs, please report them!)
4254 * The system no longer defaults Lisp source file names to types
4255 ".l", ".cl", or ".lsp", but only to ".lisp".
4256 * The compiler no longer uses special default file extensions for
4257 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4258 seems to expect a single default extension for all compiled code,
4259 and there's no compelling reason to try to stretch the standard
4260 to allow two different extensions.) Instead, byte-compiled files
4261 default to the same extension as native-compiled files.
4262 * Fasl file format version numbers have increased again, because
4263 a rearrangement of internal implementation packages made some
4264 dumped symbols in old fasl files unreadable in new cores.
4265 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4266 many fewer weird special cases.
4267 * Bug #17 (differing COMPILE-FILE behavior between logical and
4268 physical pathnames) has been fixed, and some related misbehavior too,
4269 thanks to a patch from Martin Atzmueller.
4270 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4271 by Tim Moore, ported to SBCL by Martin Atzmueller.
4272 * Martin Atzmueller fixed several filesystem-related problems,
4273 including bug #36, in part by porting CMU CL patches, which were
4274 written in part by Paul Werkowski.
4275 * More compiler warnings in src/runtime/ are gone, thanks to
4276 more patches from Martin Atzmueller.
4277 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4280 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4282 * DESCRIBE now works on CONDITION objects.
4283 * The debugger now handles errors which arise when trying to print
4284 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4286 * The build system now uses an additional file, customize-target-features.lisp,
4287 to allow local modifications to the target *FEATURES* list. (The point of
4288 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4289 debugging features enabled, without having to worry about propagating it
4290 into everyone's system when I do a "cvs update".) When no
4291 customize-target-features.lisp file exists, the target *FEATURES* list
4292 should be constructed the same way as before.
4293 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4294 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4295 warn about behavior which is undefined under the ANSI spec. Note: This
4296 is specified by ANSI, but it's not very popular with programmers.
4297 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4298 macro in the SBCL sources for an example of a workaround which you
4299 might use to make portable ANSI-standard code which does what you want.
4300 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4301 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4302 when it's used as a non-toplevel form. (This is required in order
4303 to implement the DEFCONSTANT-EQX macro.)
4304 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4305 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4306 funny things, due to bugs in the implementation of EVAL-WHEN
4307 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4308 non-toplevel DEFCONSTANTs are uncommon.)
4309 * The core file version number and fasl file version number have been
4310 incremented, because the old noncompliant DEFCONSTANT behavior involved
4311 calling functions which no longer exist, and because I also took the
4312 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4313 * fixed bug 1 (error handling before read-eval-print loop starts), and
4314 redid debugger restarts and related debugger commands somewhat while
4316 ** The QUIT debugger command is gone, since it did something
4317 rather different than the SB-EXT:QUIT command, and since it never
4318 worked properly outside the main toplevel read/eval/print loop.
4319 Invoking the new TOPLEVEL restart provides the same functionality.
4320 ** The GO debugger command is also gone, since you can just invoke
4321 the CONTINUE restart directly instead.
4322 ** The TOP debugger command is also gone, since it's redundant with the
4323 FRAME 0 command, and since it interfered with abbreviations for the
4325 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4327 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4328 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4329 some porting work by Martin Atzmueller)
4330 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4331 with a boatload of patches which he ported to SBCL. Now that those
4333 ** The system tries to make sure that its low-priority messages
4334 are prefixed by semicolons, to help people who like to use
4335 syntax highlighting in their ILISP buffer. (This patch
4336 was originally due to Raymond Toy.)
4337 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4338 patches originally written by Raymond Toy.
4339 ** The compiler understands coercion between single-value and
4340 multiple-VALUES type expressions better, getting rid of some very
4341 weird behavior, thanks to patches originally by Robert MacLachlan
4342 and Douglas Crosher.
4343 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4344 lambda lists, thanks to a patch originally by Pierre Mai.
4345 ** The system no longer bogusly warns about "abbreviated type
4347 ** The compiler gets less confused by inlining and RETURN-FROM,
4348 thanks to some patches originally by Tim Moore.
4349 ** The system no longer hangs when dumping circular lists to fasl
4350 files, thanks to a patch originally from Douglas Crosher.
4351 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4352 undefined function error.
4353 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4354 in structure slot initforms) for now, documented workaround instead:-|
4355 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4356 * fixed bug 5: added stubs for various Gray stream functions called
4357 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4358 installed, at least appropriate type errors are generated
4359 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4360 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4361 consistently in DEFMETHOD forms.
4362 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4363 it doesn't seem to affect SBCL after all
4364 * The C runtime system now builds with better optimization and many
4365 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4367 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4369 * The system is now under CVS at SourceForge (instead of the
4370 CVS repository on my home machine).
4371 * The new signal handling code has been tweaked to treat register
4372 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4373 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4374 downstream, e.g. in debug-int.lisp, has implicit dependencies
4375 on the unsignedness of integer representation of machine words,
4376 and that caused the system to bomb out with infinite regress
4377 when trying to recover from type errors involving signed values,
4378 e.g. (BUTLAST '(1 2 3) -1).)
4379 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4380 (This was one of the bugs Peter Van Eynde reported back in July.)
4381 * The system now uses code inspired by Colin Walters' O(N)
4382 implementation of MAP (from the cmucl-imp@cons.org mailing
4383 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4384 inline the MAP operation, and there is more than one
4385 sequence argument to the MAP call (so that it can't just
4386 do ETYPECASE once and for all based on the type of the
4387 single sequence argument). (The old non-inline implementation
4388 of the general M-argument sequence-of-length-N case required
4389 O(M*N*N) time when any of the sequence arguments were LISTs.)
4390 * The QUIT :UNIX-CODE keyword argument has been renamed to
4391 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4392 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4394 * Raymond Wiker's patches to port dynamic loading from Linux to
4395 FreeBSD have been added.
4396 * The BUGS file is now more nearly up to date, thanks in large part
4397 to Martin Atzmueller's review of it.
4398 * The debugger now flushes standard output streams before it begins
4399 its output ("debugger invoked" and so forth).
4400 * The core version number and fasl file version number have both
4401 been incremented, because of incompatible changes in the layout
4403 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4404 used to not be called for a saved Lisp image.)
4405 * Martin Atzmueller's version of a patch to fix a compiler crash,
4406 as posted on sbcl-devel 13 September 2000, has been installed.
4407 * Instead of installing Martin Atzmueller's patch for the
4408 compiler transform for SUBSEQ, I deleted the compiler transform,
4409 and transforms for some similar consing operations.
4410 * A bug in signal handling which kept TRACE from working on OpenBSD
4412 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4413 away by constant folding
4414 * The system now defines its address space constants in one place
4415 (in the Lisp sources), and propagates them automatically elsewhere
4416 (through GENESIS and the sbcl.h file). Therefore, patching the
4417 address map is less unnecessarily tedious and error-prone. The
4418 Lisp names of address space constants have also been systematized.
4419 * CVS tags like dollar-Header-dollar have been removed from
4420 the sources, because they have never saved me trouble and
4421 they've been source of trouble working with patches and other
4422 diff-related operations.
4423 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4424 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4426 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4428 * The system has been ported to OpenBSD.
4429 * The system now compiles with a simple "sh make.sh" on the systems
4430 that it's supported on. I.e., now you no longer need to tweak
4431 text in the target-features.lisp-expr and symlinks in src/runtime/
4432 by hand, the make.sh takes care of it for you.
4433 * The system is no longer so grossly inefficient when compiling code
4434 involving vectors implemented as general (not simple) vectors (VECTOR T),
4435 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4436 which dares to use the various sequence functions on non-simple
4437 vectors, takes less of a performance hit.
4438 * There is now a primitive type predicate VECTOR-T-P
4439 to test for the (VECTOR T) type, so that e.g.
4440 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4441 can now be compiled with some semblance of efficiency. (The old code
4442 turned the type declaration into a full call to %TYPEP at runtime!)
4443 * AREF on (VECTOR T) is still not fast, since it's still compiled
4444 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4445 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4446 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4447 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4448 without making full calls to SUBTYPEP at runtime.
4449 (Some analogous efficiency issues for non-simple vectors specialized to
4450 element types other than T, or for non-simple multidimensional arrays,
4451 have not been addressed. They could almost certainly be handled the
4452 same way if anyone is motivated to do so.)
4453 * The changes in array handling break binary compatibility, so
4454 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4455 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4456 instead of (VALUES T T).
4457 * By following the instructions that Dan Barlow posted to sbcl-devel
4458 on 2 July 2000, I was able to enable primitive dynamic object
4459 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4460 functionality is not implemented (since it calls ld to resolve
4461 library references automatically, requiring RUN-PROGRAM for its
4462 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4463 try to resolve library references) is now supported.
4464 * The system now flushes the standard output streams when it terminates,
4465 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4466 them at several other probably-convenient times, e.g. in each pass of
4467 the toplevel read-eval-print loop, and after evaluating a form given
4468 as an "--eval" command-line option. (These changes were motivated by a
4469 discussion of stream flushing issues on cmucl-imp in August 2000.)
4470 * The source transform for TYPEP of array types no longer assumes
4471 that an array whose element type is a not-yet-defined type
4472 is implemented as an array of T, but instead punts, so that the
4473 type will be interpreted at runtime.
4474 * There is now some support for cross-compiling in make.sh: each of
4475 the phases of make.sh has its own script. (This should be transparent
4476 to people doing ordinary, non-cross-compile builds.)
4477 * Since my laptop doesn't have hundreds of megabytes of memory like
4478 my desktop machine, I became more motivated to do some items on
4479 my to-do list in order to reduce the size of the system a little:
4480 ** Arrange for various needed-only-at-cold-init things to be
4481 uninterned after cold init. To support this, those things have
4482 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4483 symbols with such names are now uninterned after cold init).
4484 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4485 fasl files for cold load.
4486 ** Remove the old compiler structure pooling code (which used to
4487 be conditional on the target feature :SB-ALLOC) completely.
4488 ** Redo the representation of some data in cold init to be more compact.
4489 (I also looked into supporting byte compiled code at bootstrap time,
4490 which would probably reduce the size of the system a lot, but that
4491 looked too complicated, so I punted for now.)
4492 * The maximum signal nesting depth in the src/runtime/ support code has
4493 been reduced from 4096 to 256. (I don't know any reason for the very
4494 large old value. If the new smaller value turns out to break something,
4495 I'll probably just bump it back up.)
4496 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4498 * Many, many bugs reported by Peter Van Eynde have been added to
4499 the BUGS list; some have even been fixed.
4500 * While enabling dynamic object file loading, I tried to make the
4501 code easier to understand, renaming various functions and variables
4502 with less ambiguous names, and changing some function calling
4503 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4504 * While trying to figure out how to do the OpenBSD port, I tried to
4505 clean up some of the code in src/runtime/. In particular, I dropped
4506 support for non-POSIX signal handling, added various comments,
4507 tweaked the code to reduce the number of compilation warnings, and
4508 renamed some files to increase consistency.
4509 * To support the new automatic configuration functionality in make.sh,
4510 the source file target-features.lisp-expr has been replaced with the
4511 source file base-target-features.lisp-expr and the machine-generated
4512 file local-target-features.lisp-expr.
4513 * fixed a stupid quoting error in make.sh so that using CMU CL
4514 "lisp -batch" as cross-compilation host works again
4516 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4518 * DESCRIBE no longer tries to call itself recursively to describe
4519 bound/fbound values, so that it no longer fails on symbols which are
4520 bound to themselves (like keywords, T, and NIL).
4521 * DESCRIBE now works on generic functions.
4522 * The printer now prints less-screwed-up representations of closures
4523 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4524 * A private symbol is used instead of the :EMPTY keyword previously
4525 used to mark empty slots in hash tables. Thus
4526 (DEFVAR *HT* (MAKE-HASH-TABLE))
4527 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4528 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4529 now does what ANSI says that it should. (You can still get
4530 similar noncompliant behavior if bang on the hash table
4531 implementation with all the symbols you get back from
4532 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4533 This breaks binary compatibility, since tests for equality to
4534 :EMPTY are wired into things like the macroexpansion of
4535 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4537 * There's now a minimal placeholder implementation for CL:STEP,
4538 as required by ANSI.
4539 * An obscure bug in the interaction of the normal compiler, the byte
4540 compiler, inlining, and structure predicates has been patched
4541 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4542 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4544 * Missing ordinary arguments in a macro call are now detected even
4545 when the macro lambda list contains &KEY or &REST.
4546 * The debugger no longer complains about encountering the top of the
4547 stack when you type "FRAME 0" to explicitly instruct it to go to
4548 the top of the stack. And it now prints the frame you request even
4549 if it's the current frame (instead of saying "You are here.").
4550 * As specified by ANSI, the system now always prints keywords
4551 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4553 * The default initial SIZE of HASH-TABLEs is now smaller.
4554 * Type information from CLOS class dispatch is now propagated
4555 into DEFMETHOD bodies, so that e.g.
4556 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4558 is now basically equivalent to
4559 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4560 (DECLARE (TYPE SINGLE-FLOAT X))
4562 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4563 operation, without having to do run-time type dispatch.
4564 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4565 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4566 * Since I don't seem to be making big file reorganizations very often
4567 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4568 is overflowing my ability to conveniently back them up), I've finally
4569 checked the system into CVS. (The CVS repository is on my home system,
4570 not at SourceForge -- putting it on SourceForge might come later.)
4571 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4572 high-level GC-NOTIFY-FOO functions send their output. (There's
4573 still very little control of where low-level verbose GC functions
4574 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4575 less than it used to -- the GC-NOTIFY-FOO functions are now under
4576 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4577 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4578 in only one place in the source code, and propagates it automatically
4579 everywhere that it's needed. Thus e.g. when I bump the version from
4580 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4581 * The C source files now include boilerplate legalese and documentation
4582 at the head of each file (just as the Lisp source files already did).
4583 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4584 to his page will be replaced with a link to his new CLiki service.
4586 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4588 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4589 * The build process now looks for GNU make under the default name "gmake",
4590 instead of "make" as it used to. If GNU make is not available as "gmake"
4591 on your system, you can change this default behavior by setting the
4592 GNUMAKE environment variable.
4593 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4594 can build without error under CMU CL.
4596 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4598 * There is now a partial SBCL user manual (with some new text and some
4599 text cribbed from the CMU CL manual).
4600 * The beginnings of a profiler have been added (starting with the
4601 CMU CL profiler and simplifying and cleaning up). Eventually the
4602 main interface should be through the TRACE macro, but for now,
4603 it's still accessed through vaguely CMU-CL-style functions and macros
4604 exported from the package SB-PROFILE.
4605 * Some problems left over from porting CMU CL to the new
4606 cross-compilation bootstrap process have been cleaned up:
4607 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4608 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4609 part when getting rid of the compiler *BACKEND* stuff.)
4610 ** The value of *NULL-TYPE* was screwed up, because it was
4611 being initialized before the type system knew the final
4612 definition of the 'NULL type. This screwed up several key
4613 optimizations in the compiler, causing inefficiency in all sorts
4614 of places. (I found it because I wanted to understand why
4615 GET-INTERNAL-RUN-TIME was consing.)
4616 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4617 PROCLAIM FTYPE information. Unfortunately this broke binary
4618 compatibility again, since now the forms output by DEFGENERIC
4619 to refer to functions which didn't exist in 0.6.3.
4620 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4621 can use the new (as of 0.6.3) transform for SOME into MAP into
4623 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4624 values to (INTEGER 0 1000000), so that the time code will no longer
4625 occasionally get blown up by Linux returning 1000000 microseconds
4626 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4627 its output conform to the ANSI spec. (Alas, this makes its output
4628 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4629 * A full call to MAP NIL with a single sequence argument no longer conses.
4630 * fixes to problems pointed out by Martin Atzmueller:
4631 * The manual page no longer talks about multiprocessing as though
4632 it were currently supported.
4633 * The ILISP support patches have been removed from the distribution,
4634 because as of version 5.10.1, ILISP now supports SBCL without us
4635 having to maintain patches.
4636 * added a modified version of Raymond Toy's recent CMU CL patch for
4637 EQUALP comparison of HASH-TABLE
4639 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4641 * The system still can't cross-compile itself with
4642 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4643 entails), but at least it can compile more of itself that way
4644 than it used to be able to, and various buglets which were uncovered
4645 by trying to cross-compile itself that way have now been fixed.
4646 * This release breaks binary compatibility again. This time
4647 at least I've incremented the FASL file format version to 2, so that the
4648 problem can be detected reliably instead of just causing weird errors.
4649 * various new style warnings:
4650 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4651 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4652 ** using the lexical binding of a variable named in the *FOO* style
4653 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4654 as specified by ANSI.
4655 * *RANDOM-STATE* is no longer automatically initialized from
4656 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4657 default behavior of the system is to repeat its behavior every time
4658 it's run. If you'd like to change this behavior, you can always
4659 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4660 old convention there was no comparably easy way to get the system to
4661 repeat its behavior every time it was run.
4662 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4663 FTYPE declarations has been removed, in favor of their ANSI
4664 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4665 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4666 the types of their sequence arguments can be determined at compile time.
4667 This is done through a new open code expansion for MAP which eliminates
4668 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4669 argument types can be determined at compile time.
4670 * The optimizer now transforms COERCE into an identity operation when it
4671 can prove that the coerced object is already of the correct type. (This
4672 can be a win for machine generated code, including the output of other
4673 optimization transforms, such as the MAP transform above.)
4674 * Credit information has been moved from source file headers into CREDITS.
4675 * Source file headers have been made more standard.
4676 * The CASE macro now compiles without complaining even when it has
4679 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4681 * (Note that the way that the PCL macroexpansions were rewritten
4682 to accommodate the change in DEFGENERIC below breaks binary
4683 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4684 not run under sbcl-0.6.2. Once we get out of alpha releases,
4685 i.e. hit release 1.0.0, we'll probably try to maintain binary
4686 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4687 and sbcl-1.4.4. Until then, however, it might be fairly common
4688 for maintenance releases to break binary compatibility.)
4689 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4691 * The reporting of unhandled signals has been changed to print some
4692 explanatory text as well as the report form. (Previously only
4693 the report form was printed.)
4694 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4695 it defines, so that the compiler no longer issues undefined function
4696 warnings for compiled-but-not-yet-loaded generic functions.
4697 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4698 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4699 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4700 section of the ANSI spec are supported.
4701 * The "" nickname for the "KEYWORD" package has been removed.
4702 The reader still handles symbol tokens which begin with a package marker
4703 as keywords, but it doesn't expose its mechanism for doing so in the
4704 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4705 * The system now issues STYLE-WARNINGs for contradictory TYPE
4706 proclamations. (Warnings for contradictory FTYPE proclamations would
4707 be nice too, but those can't be done usefully unless the type system
4708 is made smarter about FUNCTION types.)
4709 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4710 systematized, so that "*target-*.lisp is supposed to exist only on the
4711 target and imply that there's a related file which exists on the
4712 host, and *host-*.lisp is supposed to exist only on the host and imply
4713 that there's a related file which exists on the target. This involves a
4714 lot of renaming. Hopefully the acute confusion caused by the renaming
4715 will be justified by the reduction in chronic confusion..
4716 ** runtime-type.lisp -> early-target-type.lisp
4717 ** target-type.lisp -> late-target-type.lisp
4718 ** early-host-format.lisp -> early-format.lisp
4719 ** late-host-format.lisp -> late-format.lisp
4720 ** host-error.lisp -> misc-error.lisp
4721 ** early-error.lisp -> early-target-error.lisp
4722 ** late-error.lisp -> late-target-error.lisp
4723 ** host-defboot.lisp -> early-defboot.lisp
4724 ** code/misc.lisp -> code/target-misc.lisp
4725 ** code/host-misc.lisp -> code/misc.lisp
4726 ** code/numbers.lisp -> code/target-numbers.lisp
4727 ** code/early-numbers.lisp -> numbers.lisp
4728 ** early-host-type.lisp -> early-type.lisp
4729 ** late-host-type.lisp -> late-type.lisp
4730 ** host-typep.lisp -> typep.lisp
4731 ** load.lisp -> target-load.lisp
4732 ** host-load.lisp -> load.lisp
4733 ** host-disassem.lisp -> disassem.lisp
4734 ** host-insts.lisp -> insts.lisp
4735 ** byte-comp.lisp -> target-byte-comp.lisp
4736 ** host-byte-comp.lisp -> byte-comp.lisp
4737 ** host-signal.lisp -> signal.lisp
4738 ** host-defstruct.lisp -> defstruct.lisp
4739 ** late-target-type.lisp -> deftypes-for-target.lisp
4740 Furthermore, several other previously target-only files foo.lisp (e.g.
4741 hash-table.lisp and random.lisp) have been split into a target-and-host
4742 foo.lisp file and a target-only target-foo.lisp file, with their key type
4743 definitions in the target-and-host part, so that the cross-compiler will
4744 know more about target types.
4745 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4746 gone away. In their place are various *BACKEND-FOO* variables
4747 corresponding to the slots of the old structure.
4748 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4749 into the target SBCL has been fixed.
4750 * The chill.lisp system for loading cold code into a running SBCL
4752 * Support for the CMU CL "scavenger hook" extension has been removed.
4753 (It was undocumented and unused in the CMU CL sources that SBCL was
4754 derived from, and stale in sbcl-0.6.1.)
4755 * Various errors in the cross-compiler type system were detected
4756 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4757 (enabling various consistency checks). Many of them were fixed,
4758 but some hard problems remain, so the compiler is back to
4759 running without *TYPE-SYSTEM-INITIALIZED* for now.
4760 * As part of the cross-compiler type system cleanup, I implemented
4761 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4762 * I have started adding UNCROSS calls throughout the type system
4763 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4764 on cross-compiler input files will be able to go away, and various
4766 * CONSTANTP now returns true for quoted forms (as explicitly required
4769 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4771 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4772 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4773 (MAKE-LIST -1). In the longer term, it ought to become true
4774 that declarations are assertions even at SAFETY 1. For now, it's not
4775 quite true even at SAFETY 3, but it's at least more nearly true..
4776 (Note that this change seems to increases the size of the system by
4777 O(5%) and to decrease the speed of the compiler by 20% or more.)
4778 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4779 for the problem of printing dozens of lines of distracting information
4780 about low-level system machinery as part of the top stack frame
4781 on entry to the debugger when an undefined function was called.
4782 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4783 is not reset to COMMON-LISP-USER.
4784 * Compilation of stuff related to dyncount.lisp has been made conditional
4785 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4786 smaller. The various dyncount-related symbols have been moved into
4787 a new "SB-DYNCOUNT" package.
4788 * tty-inspect.lisp has been renamed to inspect.lisp.
4789 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4790 feature has gone away. (When we eventually port to other flavors of
4791 libc and/or Unix, we'll try to make the differences between flavors
4792 invisible at the user level.)
4793 * Various other *FEATURES* tags, and/or their associated conditionals,
4794 have been removed if obsolescent, or given better documentation, or
4795 sometimes given more-mnemonic names.
4797 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4799 * tidied up "make.sh" script
4800 * tidied up system directory structure
4801 * better "clean.sh" behavior
4802 * added doc/FOR-CMUCL-DEVELOPERS
4803 * many many small tweaks to output format, e.g. removing possibly-confusing
4804 trailing #\. character in DESCRIBE-INSTANCE
4805 * (EQUALP #\A 'A) no longer signals an error.
4806 * new hashing code, including EQUALP hashing
4807 * tidied up Lisp initialization and toplevel
4808 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4809 * command line argument processing
4810 * added POSIX-GETENV function to deal with Unix-ish environment variables
4811 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4812 terminating SBCL on EOF
4813 * non-verbose GC by default
4814 * There is no more "sbcl" shell script; the sbcl file is now the C
4815 runtime executable (just like CMU CL).
4816 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4818 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4819 symbol and package information as Lisp native symbol and package objects
4820 instead of strings naming symbols and strings naming packages. This way,
4821 whenever packages are renamed (as in warm init), debug information is
4822 transformed along with everything else.
4823 * tweaked the optimization policy declarations which control the building
4824 of SBCL itself. Now, among other things, the system no longer saves
4825 source location debugging information. (This helps two problems at once
4826 by reducing SBCL size and by keeping SBCL from trying to look for its
4827 sources -- which may not exist -- when reporting errors.)
4828 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4829 debugging and testing purposes
4830 * cleaned up printing, making the printer call PRINT-OBJECT for
4831 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4832 methods, giving nearly-ANSI behavior
4833 * converted almost all special variables to use *FOO* naming convention
4834 * deleted PARSE-TIME functionality, since it can be done portably
4835 * moved some files out of cold init into warm init
4836 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4838 * regularized formatting of source files
4839 * added an install.sh script
4840 * fixed ridiculous memory usage of cross-compiler by making
4841 compiler/alloc.lisp not try to do pooling unless it can hook
4842 itself into the GC of the cross-compilation host. Now the system
4843 builds nicely on my old laptop.
4844 * added :SB-ALLOC in target-features.lisp-expr
4845 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4846 was not implemented)
4847 * re-did condition handling and note reporting in the compiler. Notes
4848 are no longer handled by signalling conditions. Style warnings
4849 and warnings are handled more correctly and reported in such a way
4850 that it's easy to find one or the other in your output (so that you
4851 can e.g. figure out which of many problems caused COMPILE-FILE to
4853 * changed the severity of several compiler warnings from full WARNING
4854 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4855 compiler note reporting so that it doesn't use the condition system
4856 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4857 COMPILE-FILE command)
4858 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4859 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4860 no longer supported, since their implementation was deeply intertwingled
4861 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4862 * removed bogus "support" for compiler macros named (SETF FOO), and
4863 removed the compiler macro for SETF INFO (but only after making a fool
4864 of myself on the cmucl-imp mailing list by posting a bogus patch for
4865 DEFINE-COMPILER-MACRO..)
4866 * Compiled files containing forms which have side effects on the Lisp
4867 reader (such as DEFPACKAGE forms) are now handled more correctly.
4868 (Compiler queuing of top level lambdas has been suppressed by setting
4869 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4870 * deleted various currently-unused source files, e.g. gengc.lisp. They
4871 may be added back at some point e.g. when porting to other architectures,
4872 but until they are it's distracting to distribute them and to try to
4874 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4875 there were so many of them they're just distractions, and UNCROSS is
4876 known to be able to handle the current sources
4877 * moved PROFILE functionality into TRACE, so that it will be clear
4878 how the wrapping and unwrapping of functions when you profile them
4879 interacts with the wrapping and unwrapping of functions when you
4880 trace them. (Actually, the functionality isn't there yet, but at least
4881 the interface specification is there. Hopefully, the functionality will
4882 arrive with some future maintenance release.)
4883 * removed host-oops.lisp
4884 * changed signature of QUIT function to allow UNIX-CODE argument
4885 * fixed READ-SEQUENCE bug
4886 * tweaked verbose GC output so that it looks more like the progress
4887 output that ANSI specifies for functions like LOAD
4888 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4889 * added <http://sbcl.sourceforge.com> to the banner information printed by