1 ;;;; -*- coding: utf-8; -*-
2 * new feature: SIMPLE-ARRAY-VECTOR provides access to the underlying
3 data vector of an multidimensional SIMPLE-ARRAY.
4 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
5 * bug fix: lexical type declarations are now correctly reported by
6 SB-CLTL2. (reported by Larry D'Anna)
7 * bug fix: STRING-TO-OCTETS did not handle :START properly when
8 using UTF-8 as external format. (thanks to Luis Oliveira)
9 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
10 are signalled correctly. (thanks to Stas Boukarev)
12 changes in sbcl-1.0.23 relative to 1.0.22:
13 * enhancement: when disassembling method functions, disassembly
14 for the associated fast function is also produced.
15 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
17 * optimization: printing with *PRINT-PRETTY* true is now more
18 efficient as long as the object being printed doesn't require
19 special handling by the pretty printer.
20 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
21 now interact correctly with type declarations.
22 * partial bug fix: PCL detects infinite recursion during wrapper
23 validation. (thanks to Attila Lendvai)
24 * bug fix: #426; nested function calls are inlined properly.
25 Previously if FOO was an inline function, in calls of the form
26 (FOO (FOO ...)) the outer call was not inlined.
27 * bug fix: long long arguments passed in registers to alien
28 functions on PowerPC Linux and NetBSD are now handled correctly
29 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
31 changes in sbcl-1.0.22 relative to 1.0.21:
32 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
33 for the shared object in the current directory, but passes the native
34 namestring of the designated pathname to the operation system's shared
35 object loading function as-it.
36 * minor incompatible change: --disable-debugger toplevel option now takes
37 effect before processing of initialization files and --eval or --load
39 * new feature: new commandline argument: --script, which supports
40 shebang lines. See documentation for details. (based on work by
42 * new feature: SAVE-LISP-AND-DIE can save current values of
43 --dynamic-space-size and --control-stack-size in the executable core,
44 causing it to skip normal runtime option processing. See documentation
45 for details. (thanks to Zach Beane)
46 * enhancement: inoccous calls to EVAL or generic functions dispatching
47 on subclasses of eg. STREAM no longer cause compiler notes to appear.
48 * enhancement: the system no longer resignals errors from --load and
49 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
50 associated with the original error to be lost. (thanks to Ariel
52 * enhancement: :DONT-SAVE keyword argument has been added to
53 LOAD-SHARED-OBJECT for controlling interaction with
54 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
55 the effects of an earlier LOAD-SHARED-OBJECT call.
56 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
57 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
58 provided. (reported by Cedric St-Jean)
59 * bug fix: circularity handling in the reader did not treat raw
60 structure slots correctly. (reported by Cedric St-Jean)
61 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
62 file descriptors when there were none.
63 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
64 was broken. (reporteed by Willem Broekema)
65 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
66 pathnames without a directory.
67 * bug fix: COMPILE-FILE did not warn about undefined variable
68 references at toplevel, and LOAD of the resulting fasl did
70 * bug fix: functions with non-required arguments used to end up with
71 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
72 * bug fix: redefining a function with non-required arguments didn't
73 update the system's knowledge about its call signature properly.
74 * bug fix: fixed #431; incompatible alien record type redefinitions
75 are detected and handled. (reported by Neil Haven)
76 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
77 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
79 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
80 after alien stack frames.
81 * bug fix: strings passed to mk{s,d}temp lacked null termination
83 changes in sbcl-1.0.21 relative to 1.0.20:
84 * new feature: the compiler is able to track the effective type of a
85 generic function across method addition and removal.
86 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
87 appropriately typed structure slots without locking.
88 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
89 information like those gathered by TIME using a programming-friendly
91 * new feature: TIME reports time taken even if the form performs a
92 non-local transfer of control.
93 * enhancement: reduced conservativism on GENCGC platforms: on
94 average 45% less pages pinned (measured from SBCL self build).
95 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
96 mutate constant symbols or violate declaimed type of the symbol.
97 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
98 sleeping on the semaphore in heavily contested semaphores.
99 * bug fix: semaphores and condition variables were not interrupt
101 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
102 owned by other threads anymore.
103 * bug fix: FIND on lists called KEY outside the specified
104 subsequence. (reported by budden)
105 * bug fix: LOG doesn't use single-float intermediate results when
106 given mixed integer and double-float arguments, leading to better
107 precision. (reported by Bob Felts)
108 * bug fix: LOG with base zero returned values of inconsistent type.
110 changes in sbcl-1.0.20 relative to 1.0.19:
111 * minor incompatible change: OPTIMIZE qualities
112 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
113 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
114 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
115 * documentation: some slot access efficiency guidelines have been
116 added to the user manual.
117 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
118 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
119 as ASSOC and MEMEBER.
120 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
121 to more efficient EQ-comparison versions more often.
122 * optimization: enhanced derivation of DOLIST iteration variable type
124 * optimization: constant folding of simple (LIST ...) forms as DOLIST
126 * optimization: runtime lookup of function definitions can be
127 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
128 * optimization: compiler is able to derive the return type of
129 (AREF (THE STRING X) Y) as being CHARACTER.
130 * optimization: CLRHASH on empty hash-tables no longer does pointless
131 work. (thanks to Alec Berryman)
132 * optimization: REPLACE deftransforms don't punt when :START1 or
134 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
135 no longer emits a code deletion note. (reported by Andrew
137 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
138 functions with an odd number of &OPTIONAL arguments, a &REST
139 argument, and one or more &KEY arguments at the call site.
140 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
141 arguments appeared at call sites as well.
142 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
143 READ-LINE &co can no longer read from them afterwards. (reported
145 * bug fix: fixed #427: unused local aliens no longer cause compiler
146 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
148 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
149 alien-stack leakage. (reported by Andy Hefner)
150 * bug fix: PROGV signals an error when an attempt to violate declared
151 type of a variable or bind a constant is made.
152 * bug fix: SET signals an error when an attempt to violate declared
153 type of a variable is made.
154 * bug fix: restart computation during the execution of a restart
155 test function no longer causes infinite recursion. (reported by
157 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
158 signals an error. (thanks to Pierre Mai)
159 * bug fix: EXPORT left symbol unexported in conflict situations.
160 (thanks to Michael Weber)
161 * bug fix: correctly handle name conflicts involving CL:NIL.
162 (thanks to Michael Weber)
163 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
164 conflicts arising from USEing package with conflicting symbols
165 correctly. (thanks to Michael Weber)
167 changes in sbcl-1.0.19 relative to 1.0.18:
168 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
169 warnings that go otherwise unhandled will be muffled if they are
170 of the type that's the value of this variable.
171 * optimization: stack allocation is slightly more efficient on x86
173 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
174 where a raw slot always is initialized using the initform whose
175 type is not know sufficiently well a compile-time are now compiled
176 correctly. (reported by John Morrison)
177 * bug fix: compiler no longer makes erronous assumptions in the
178 presense of non-foldable SATISFIES types.
179 * bug fix: stack analysis missed cleanups of dynamic-extent
180 arguments in non-let-converted calls to local functions.
181 * improvements to the Windows port:
182 ** adjusted address spaces for building on both Win32 and
183 Win64. (thanks for John Connors)
184 * fixed some bugs revealed by Paul Dietz' test suite:
185 ** interval arithmetic during type derivation used inexact integer
186 to single-float coercions.
187 ** arithmetic operations involving large integers and single
188 floats give the same results in compiled and interpreted code.
189 ** deriving the result type of COERCE no longer signals an error
190 if the derived type of the second argument is a MEMBER type
191 containing invalid type specifiers.
192 ** ADJOIN with constant NIL as second argument works correctly.
193 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
195 changes in sbcl-1.0.18 relative to 1.0.17:
196 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
197 profiles only the current thread.
198 * minor incompatible change: changes to SYMBOL-VALUE of constants
199 defined with DEFCONSTANT now signal an error.
200 * enhancement: SB-SPROF now has support for wallclock profiling,
201 and is also able to profile specific threads. REPORT output
202 has also additional sorting options.
203 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
205 * optimization: structure allocation has been improved
206 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
207 ** out of line constructors are ~10% faster.
208 ** inline constructors are ~15% faster.
209 ** inline constructors are capable of dynamic extent allocation
210 (generally on x86 and x86-64, in some cases on other platforms
212 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
214 * optimization: file compiler is now able to coalesce non-circular
215 lists, non-base strings, and bit-vectors. Additionally, constants
216 are never referenced through SYMBOL-VALUE at runtime.
217 * optimization: code defining methods on PRINT-OBJECT (and other
218 generic functions in the COMMON-LISP package) now loads faster.
219 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
221 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
222 fasl is now deleted. (reported by Attila Lendvai)
223 * bug fix: READ-LINE always returned NIL for the last line in files.
224 (reported by Yoshinori Tahara)
225 * bug fix: more accurate disassembly annotations of foreign function
226 calls. (thanks to Andy Hefner)
227 * bug fix: trimming non-simple strings and non-string string
228 designators when the there is nothing to trim works properly.
229 (thanks to James Knight)
230 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
231 (thanks to Travis Cross)
232 * fixed some bugs revealed by Paul Dietz' test suite:
233 ** NIL is a valid function name (regression at 1.0.13.38)
234 ** FILL on lists was missing its return value (regression at 1.0.12.27)
235 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
236 fill pointers properly (regression at 1.0.12.23)
237 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
238 displacement indices properly (regression at 1.0.12.23)
240 changes in sbcl-1.0.17 relative to 1.0.16:
241 * temporary regression: user code can no longer allocate closure
242 variable storage on stack, due to bug 419 without explicitly
243 requesting it. Please consult sbcl-devel for advice if you need to
244 use this feature in the meanwhile.
245 * new feature: runtime argument --control-stack-size can be used to
246 adjust thread default control stack size.
247 * enhancement: improved TIME output
248 ** all times are reported using the measured accuracy (milliseconds
249 for real and GC times, microseconds for everything else.)
250 ** processor cycle counts on x86 and x86-64.
251 ** interpreted forms are counted for both evaluator modes.
252 ** number of lambdas converted by the compiler is reported.
253 ** CPU percentage report (computed from real and total run time.)
254 ** more comprehensive run time reporting, using a condenced format
255 ** interperted form, lambda, and page fault counts are omitted
257 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
259 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
260 * optimization: two argument forms of LAST are upto ~10% faster
261 in normal SPEED policies.
262 * optimization: NCONC no longer needs to heap cons its &REST list
263 in normal SPEED policies.
264 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
266 * bug fix: UNION and NUNION work with :TEST-NOT once more,
267 regression since 1.0.9.1. (thanks to Eric Marsden)
268 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
269 since 1.0.15.36. (thanks to Paul Khuong)
270 * bug fix: LAST when always returned the whole list when given a bignum
271 as the second argument.
272 * bug fix: dynamic extent allocation of nested lists and vectors
273 could leak to otherwise accessible parts.
274 * bug fix: invalid optimization of heap-allocated alien variable
276 * bug fix: fasl header checking is less vulnerable to different
277 platform word lengths.
278 * bug fix: more correct assembler syntax for GNU binutils
279 2.18.50.0.4 support. (thanks to Marijn Schouten)
280 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
282 * bug fix: the fopcompiler can handle LOCALLY forms (with no
283 declarations) successfully. (reported by Attila Lendvai)
285 changes in sbcl-1.0.16 relative to 1.0.15:
286 * minor incompatible change: revert the changes to sb-posix's error
287 signaling added in 1.0.14.
288 * minor incompatible change: change PROBE-FILE back to returning
289 NIL whenever we can't get a truename, as was the case before 1.0.14.
290 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
291 inherits from ERROR instead of just CONDITION.
292 * new feature: SB-INTROSPECT can provide source locations for instances
293 as well. (thanks to Tobian Ritterweiler)
294 * optimization: binding special variables now generates smaller code
295 on threaded platforms.
296 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
297 and cases where no :TEST is given but the compiler can infer that
298 the element to search is of type (OR FIXNUM (NOT NUMBER)).
299 * optimization: better LOGNOT on fixnums.
300 * optimization: modular arithmetic for a particular requested width
301 is implemented using a tagged representation unless a better
302 representation is available.
303 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
304 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
305 for wellformedness and returns a specialized vector. (reported by
306 Francois-Rene Rideau)
307 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
308 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
309 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
310 unscheduling timer at the same time as another timer fires.
311 * bug fix: don't reschedule timers for dead threads.
312 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
313 * bug fix: copying output from RUN-PROGRAM to a stream signalled
314 bogus errors if select() was interrupted.
315 * enhancement: add support for fcntl's struct flock to SB-POSIX.
317 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
318 * enhancement: cleaner backtraces for interactive interrupts, as
319 well as other cases where the interesting frames used to be
320 obscured by interrupt handling frames.
321 * enhancement: untracing a whole package using (UNTRACE "FOO") is
322 now supported, and tracing a whole package using (TRACE "FOO") now
323 traces SETF-functions as well.
324 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
325 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
326 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
327 * unreadably printed representation of hash-tables now includes
329 * bug fix: partially fixed #188: type propagation from assignments
330 is now more efficient.
331 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
332 'DOUBLE-FLOAT) are not flushable.
333 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
334 full address of the object, and none of the tag bits.
335 * bug fix: readably printing hash-tables now respects other printer
336 control variables. (reported by Cedric St-Jean)
337 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
338 keyword with MAKE-HASH-TABLE.
339 * bug fix: export SB-POSIX:MKSTEMP.
340 * bug fix: SORT was not interrupt safe.
341 * bug fix: XREF accounts for the last node of each basic-block as
343 * bug fix: MAKE-INSTANCE optimizations interacted badly with
344 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
345 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
346 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
347 create obsolete instances in the case of redefinition or
348 obsoletion of a superclass. (thanks to Andy Hefner)
349 * bug fix: Support for the Alpha architecture has been revived; it had
350 suffered somewhat from lack of maintenance since sbcl-1.0.
351 * improvements to the Windows port:
352 ** The system detects the codepage to use at startup, instead of
353 using the value from the saved core. (thanks to Kei Suzuki)
355 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
356 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
357 (see documentation for details.)
358 * revived support for OpenBSD (contributed by Josh Elsasser)
359 * partially fixed bug #108: ROOM no longer suffers from occasional
360 (AVER (SAP= CURRENT END)) failures.
361 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
362 bodies no longer cause a WARNING to be signalled. (reported by
364 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
365 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
366 the bug in TRUNCATE.) (reported by Michael Weber)
367 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
368 no samples. (reported by Andy Hefner)
369 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
370 no longer appear as (NIL ...) frames in backtraces.
371 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
372 is now actually exported from SB-EXT as documented. (reported by
374 * bug fix: sb-aclrepl now correctly understands how to inspect
375 single-floats on 64-bit platforms where single-floats are not boxed.
376 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
377 yet been finalized. (reported by Levente Meszaros)
378 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
379 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
380 SBCL-specific optimize qualities.
382 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
383 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
384 an executable in the search path, and does so in the child
385 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
386 been removed; it can be found in the CVS history, for somebody who
387 needs that search behavior (see the manual).
388 * minor incompatible change: compiler policy re. weakening type
389 checks has changed: now type checks are weakened only if SAFETY < 2
391 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
392 unparsing of directory pathnames as files. Analogously,
393 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
394 filename to parse into a directory pathname.
395 * enhancement: implicit generic function creation now signals a
396 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
397 users can bind handlers for and muffle around calls to LOAD.
398 * enhancement: RUN-PROGRAM allows unicode arguments and environments
399 to be used (using the default stream external format), and allows
400 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
401 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
402 strings and vectors whose element-type or simplicity is not fully
403 known at compile-time.
404 * optimization: STRING-TRIM and related functions no longer allocate
405 a new string when no trimming needs to be performed. These functions
406 are also faster than before when the input string has been declared
408 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
409 * optimization: READ-LINE is significantly faster for files containing
411 * optimization: non-open coded uses of character comparison operators
412 (e.g. char=) no longer cons when called with more than one parameter
413 on platforms supporting dynamic-extent allocation.
414 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
415 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
416 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
418 * bug fix: some sequence functions elided bounds checking when
420 * bug fix: too liberal weakening of union-type checks when SPEED >
422 * bug fix: more bogus fixnum declarations in ROOM implementation
425 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
426 * new feature: MAKE-HASH-TABLE now experimentally accepts a
427 :SYNCHRONIZED argument, which makes the hash-table safe for
428 concurrent accesses (but not iteration.) See also:
429 SB-EXT:WITH-LOCKED-HASH-TABLE, and
430 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
431 * optimization: CONCATENATE on strings is an order of magnitude faster
432 in code compiled with (> SPEED SPACE).
433 * optimization: SUBSEQ is ~50% faster on lists.
434 * bug fix: bug 417 fixed -- source location reporting is now more robust.
435 * bug fix: SUBSEQ on a list will now correctly signal an error if if
436 END is smaller then START.
437 * bug fix: SB-PROFILE will no longer report extra consing for nested
438 calls to profiled functions.
439 * bug fix: ROOM implementation had bogus fixnum declarations which
440 could cause type-errors when calling ROOM on large images.
441 * bug fix: if file compilation is aborted, the partial fasl is now
442 deleted, and COMPILE-FILE returns NIL as the primary value.
443 * bug fix: number of thread safety issues relating to SBCL's internal
444 hash-table usage have been fixed.
445 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
446 be returned from its body when the values were being returned
447 using unknown-values return convection and the W-P-O was wrapped
448 inside an UNWIND-PROTECT.
449 * bug fix: sb-posix should now compile again under Windows, enabling
452 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
453 * incompatible change: hash-table accessor functions are no longer
454 automatically protected by locks. Concurrent accesses on the same hash-table
455 from multiple threads can give inconsistent results or even corrupt the
456 hash-table completely. Multi-threaded applications should do their own
457 locking at the correct granularity. In the current implementation it is
458 still safe to have multiple readers access the same table, but it's not
459 guaranteed that this property will be maintained in future releases.
460 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
461 and will signal an error at runtime.
462 * enhancement: SB-THREAD package now exports a semaphore interface.
463 * enhancement: CONS can now stack-allocate on x86 and
464 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
466 * enhancement: nested lists can now be stack allocated on
467 platforms providing stack allocation support.
468 * enhancement: dynamic-extent support has been extended to support
469 cases where there are multiple possible sources for the stack
471 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
472 if the mutex is uncontested on Linux.
473 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
474 as specified by the CLHS. (thanks to Tobias Rittweiler)
475 * bug fix: NaN comparison now works on x86-64.
476 * bug fix: CLOSE :ABORT T on a stream with pending output now
478 * bug fix: instances of non-standard metaclasses using standard
479 instance structure protocol sometimes missed the slot type checks
481 * bug fix: known functions can be inlined in high-debug code.
483 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
484 * minor incompatible change: the MSI installer on Windows no longer
485 associates .lisp and .fasl files with the installed SBCL.
486 * minor incompatible change: :UNIX is no longer present in *FEATURES*
487 on Windows. (thanks to Luis Oliviera)
488 * new feature: SB-CLTL2 contrib module now implements
489 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
490 * optimization: scavenging weak pointers is now more efficient,
491 requiring O(1) instead of O(N) per weak pointer to identify
492 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
493 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
494 method bodies using constant slot-names when the first argument is
495 a specializer parameter for the method.
496 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
497 names, when the class of the instance is a direct instance
498 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
499 as slow as the constant slot-name case.)
500 * optimization: member type construction is now O(N) instead
502 * optimization: UNION and NUNION are now O(N+M) for large
503 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
505 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
506 now more readable in environments like Slime which display it.
507 (thanks to Tobias C. Rittweiler)
508 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
509 was able to infer the correct argument type for the object on which
510 the CAS operation was being performed.
511 * bug fix: copy propagation interfered with parallel assignment
512 semantics in local calls. (reported by Paul Khuong)
513 * bug fix: the signed modular fixnum shift compiled to wrong code on
514 x86 and x86-64. (spotted by a slight modification to some of
517 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
518 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
519 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
520 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
522 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
523 and generic function now print the object identity for anonymous
524 instances. (thanks to Attila Lendvai)
525 * enhancement: as an extension to MOP, SBCL now supports SETF of
526 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
527 (thanks to Attila Lendvai)
528 * enhancement: sb-cover has an alternate annotation mode which puts
529 the form annotations at the car of the form, instead of annotating
530 it completely. The benefit of this mode is that it shows better
531 which forms were instrumented by the compiler.
532 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
533 by printer control variables. (reported by Dan Corkill)
534 * bug fix: system leaked memory when delayed output was performed by
535 the OS in smaller chunks then expected. (thanks to David Smith)
536 * bug fix: system leaked memory when file streams were not closed
538 * bug fix: large objects written to slow streams that were modified
539 after the write could end up with the modified state written to
540 the underlying file descriptor.
541 * bug fix: multiple threads operating in parallel on the same stream
542 could cause buffer-overflows.
543 * bug fix: source location information is stored correctly
544 (broken since 1.0.6). This bug would generally show up as the
545 Slime debugger highlighting the wrong form when the sldb-show-source
547 * bug fix: Branch forms are again annotated as branches in the sb-cover
549 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
550 on x86-64. (reported by Gregory Vanuxem)
551 * bug fix: EQUALP could return wrong results for structures with raw
552 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
553 word). (reported by Vjacheslav Fyodorov)
554 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
557 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
558 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
559 atomic compare-and-swap operations on threaded platforms.
560 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
561 allows assigning a global minimum value to optimization qualities
562 (overriding proclamations and declarations).
563 * enhancement: closed over variables can be stack-allocated on x86
565 * performance bug fix: GETHASH and (SETF GETHASH) are once again
567 * optimization: slot definition lookup is now O(1). This speeds up
568 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
569 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
570 * optimization: ASSOC and MEMBER can now be open-coded for all
571 combinations of keyword arguments when second argument is constant
572 and SPEED >= SPACE. In other cases a specialized version is
574 * bug fix: using obsoleted structure instances with TYPEP and
575 generic functions now signals a sensible error.
576 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
577 (reported by Kristoffer Kvello)
578 * bug fix: backtrace construction is now more careful when making
579 lisp-objects from pointers on the stack, to avoid creating bogus
580 objects that can be seen by the GC.
581 * bug fix: defaulting of values in contexts expecting more than 7
582 variables now works on x86-64. (reported by Christopher Laux)
583 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
585 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
586 as the property-list of a symbol.
587 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
588 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
589 situations requested, are once again file-compileable. (reported
592 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
593 * MOP improvement: support for user-defined subclasses of
594 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
595 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
596 the expansion of DEFMETHOD, is responsible for generating a form
597 which creates a list of specializers when evaluated. Additional
598 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
599 debugging and introspective support.
600 * minor incompatible change: the (unsupported) spinlock interface
601 has changed: free spinlock now has the value NIL, and a held spinlock
602 has the owning thread as its value.
603 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
604 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
606 * enhancement: name of a socket-stream is now "a socket" instead of
608 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
609 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
610 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
611 for generic functions. (Thanks to Tobias C. Rittweiler)
612 * optimization: bignum printing speed has been improved by 20-40%
613 (depending on the bignum size.)
614 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
616 * bug fix: the cache used by the CLOS to store precomputed effective
617 methods, slot offsets, and constant return values is now thread and
619 * bug fix: generic function dispatch function updating is now thread
620 and interrupt safe (in the sense that the known issues have been
622 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
623 * bug fix: interning EQL-specializers is now thread and interrupt safe.
624 * bug fix: asdf systems with dependencies to the SB-POSIX or
625 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
626 * bug fix: interrupt safety of applicable method computation has been
629 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
630 * new contrib: sb-cover, an experimental code coverage tool, is included
632 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
633 significantly faster.
634 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
635 produce more efficient code on x86-64 if the compiler can determine
636 that (+ INDEX <constant>) does not require a bounds check and FOO
637 has an element type at least 8 bits wide.
638 * enhancement: a new, experimental synchronous timeout facility is
639 provided. Refer to SB-SYS:WITH-DEADLINE for details.
640 * enhancement: when a symbol name conflict error arises, the
641 conflicting symbols are always printed with a package prefix.
642 (thanks to Kevin Reid)
643 * enhancement: stepping is now once again supported on the SPARC and
644 MIPS platforms. (It is also now more likely to work on CheneyGC
646 * enhancement: sb-sprof can now also track and report accurate call
648 * bug fixes: the treatment of non-standard subclasses of
649 SB-MOP:SPECIALIZER is more correct.
650 * incompatible change: PURIFY no longer copies the data from the
651 dynamic space into the static and read-only spaces on platforms
652 that use the generational garbage collector
653 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
655 * bug fix: GC race condition occasionally resulting in crashes with
656 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
658 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
660 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
661 a STYLE-WARNING to be signalled (regression from 1.0.4.)
662 * bug fix: an asynchronous interrupt could previously leave the
663 system running with GC inhibited.
664 * bug fix: a DECLARE form evaluated at top-level now causes an error
665 rather than silently (or verbosely) returning NIL.
666 * bug fix: trying to dissassemble functions compiled at high DEBUG could
667 result in a error being signalled due to source form lookup errors.
668 (reported by Peter Graves)
670 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
671 * incompatible change: removed writer methods for host-ent-name,
672 host-ent-addresses -- changing the values did not update the DNS
674 * minor incompatible change: changed experimental JOIN-THREAD interface
675 * documentation: the manual now lists reader and writer methods
676 in class slot documentation sections. (thanks to Richard M Kreuter)
677 * documentation: unwinding from asyncronous events has been
678 documented as unsafe.
679 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
680 in multithreaded application code.
681 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
682 platforms. (thanks to James Anderson for the optimization hint)
683 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
685 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
686 variants no longer cons.
687 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
688 their NOT- variants no longer cons.
689 * optimization: EQUAL hash tables no longer use SXHASH for objects
690 of all data types, but instead use an EQL hash for types for which
691 EQUAL is the same as EQL.
692 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
693 are significantly faster.
694 * optimization: new STRING-OUTPUT-STREAM implementation is much
695 faster for large outputs and conses 30% less on average.
696 * enhancement: XREF information is now collected for references made
697 to global variables using SYMBOL-VALUE and a constant argument.
698 * enhancement: SIGINT now causes a specific condition
699 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
700 * bug fix: FILE-POSITION used to signal an error in some cases where
701 ANSI requires it to return NIL.
702 * bug fix: ADJUST-ARRAY is now interrupt-safe.
703 * bug fix: adding and removing fd-handlers is now interrupt-safe.
704 * bug fix: inlined calls to C now ensure 16byte stack alignment on
706 * bug fix: bad type declaration in the CLOS implementation has
707 been fixed. (reported by James Anderson)
708 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
709 (thanks to Lutz Euler)
710 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
711 required by the CLHS. (reported by Samium Gromoff)
712 * bug fix: dead unbound variable references now signal an error.
713 (reported by Marco Monteiro)
714 * bug fix: / with an unused value was being deleted in safe code.
715 (thanks to Marco Monteiro and Kevin Reid)
716 * bug fix: number of characters that can be written onto a single
717 line in a file is unlimited.
718 * bug fix: some GC deadlocks caused by asynchronous interrupts have
719 been fixed by inhibiting interrupts for when GC is disabled.
720 * bug fix: some interrupt safety issues with GETHASH, (SETF
721 GETHASH), CLRHASH and REMHASH have been fixed.
722 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
723 specifier no longer causes infinite recursion.
724 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
725 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
726 * bug fix: an &environment argument with the correct variable information
727 is passed to macros that are expanded during byte compilation.
728 (reported by Samium Gromoff)
729 * bug fix: a logic error was causing the hash function for bignums to
730 have excessive amounts of collisions. (reported by Faré Rideau)
731 * bug fix: modifying the contents of an array could change the return
732 value of SXHASH on that array, which is only allowed for strings
733 and bit vectors. (bug introduced in 0.9.16)
734 * improvement: the x86-64/darwin port now passes all tests (except
735 for the debugger tests) but should still be considered
736 experimental until this is fixed.
737 * improvement: a style-warning is signaled for CASE (etc) clauses with
738 duplicate keys. (thanks to Kevin Reid)
739 * improvement: macos/x86 and macos/x86-64 now use mach exception
740 handlers for dealing with illegal instructions (for trapping and
741 error handling) and memory protection violations (for GC).
743 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
744 * new platform: experimental support for x86-64/darwin (MacOS).
745 * incompatible change: the thread-safe (on most platforms) getaddrinfo
746 and getnameinfo sockets functions are used instead of gethostbyaddr
747 and gethostbyname, on platforms where the newer functions are available.
748 As a result, the ALIASES field of HOST-ENT will always be NIL on these
750 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
751 don't print the startup banner, but behave as if --noinform was passed
752 as a command line argument. (thanks to Kevin Reid)
753 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
754 * optimization: code using alien values with undeclared types is much faster.
755 * optimization: the compiler is now able to open code SEARCH in more cases.
756 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
757 * bug fix: using standardized COMMON-LISP special variables as loop
758 variables no longer signals bogus package lock violations. (reported
760 * bug fix: declaring local loop variables to be of a range-limited type
761 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
762 (reported by Andras Simon)
763 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
764 bugs remain on x86-64.)
765 * bug fix: the #= and ## reader macros now interact reasonably with
766 funcallable instances.
767 * bug fix: type-checks for function arguments were compiled using the
768 compiler policy settings of the wrong lexical environment (in 1.0.2
770 * bug fix: SHADOW accepts characters as string designators, as required
771 by the spec (thanks to Eric Marsden)
772 * bug fix: fixed GC safety issues when foreign functions are called with
773 non-base strings as arguments
774 * bug fix: more consistent error output for fatal-compiler-warnings, like
776 * bug fix: functions with proper names were showing up as NIL in
777 backtraces (reported by Edi Weitz, regression in 1.0.2)
779 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
780 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
781 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
782 produce more efficient code on the x86 if the compiler can determine
783 that (+ INDEX <constant>) does not require a bounds check and FOO
784 has an element type at least 8 bits wide.
785 * bug fix: references to undefined variables are handled the same way
786 in toplevel forms as in the normal compiler
787 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
789 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
790 (thanks to Jon Buller)
791 * improvement: faster compilation times for complex functions
792 * improvement: added readlink support to SB-POSIX (thanks to Richard
795 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
796 * improvement: experimental support for mach exception handling on
797 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
799 * improvement: support for GBK external format.
800 (thanks to Chun Tian (binghe))
801 * improvement: the debugger now displays variables that have been closed
802 over, in code compiled with (DEBUG 2) or higher
803 * improvement: support for executable cores on NetBSD (thanks to
805 * new feature: added a RESTART-FRAME debugger command
806 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
807 be used to provide an implementation for FILE-POSITION on Gray streams
808 (thanks to Eric Marsden)
809 * new feature: add syslog, openlog and closelog support to SB-POSIX
810 (thanks to Richard Kreuter)
811 * optimization: the function call overhead in code compiled with
812 a high DEBUG optimization setting is significantly
813 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
814 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
815 (thanks to Tony Martinez)
816 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
817 evaluated once (reported by Kevin Reid)
818 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
819 works again (thanks to Kilian Sprotte)
820 * bug fix: an error is signaled for tagbodies with duplicate tags
821 (thanks to Stephen Wilson)
822 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
823 * bug fix: Win32 port can now handle foreign code unwinding Lisp
824 stack frames from alien callbacks.
825 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
826 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
827 65k pages (thanks to David Woodhouse)
828 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
830 changes in sbcl-1.0.1 relative to sbcl-1.0:
831 * new platform: FreeBSD/x86-64, including support for threading.
832 * new feature: the compiler stores cross-referencing information
833 about function calls (who-calls), macroexpansion (who-macroexpands)
834 and special variables (who-binds, who-sets, who-references) for code
835 compiled with (< SPACE 3). This information is available through the
836 sb-introspect contrib.
837 * new feature: users may subclass SEQUENCE, and have instances of
838 these classes interoperate with standard Common Lisp functions if
839 a number of methods are defined. (This feature is experimental
840 and the interface subject to change based on feedback from SBCL
841 users and the general community)
842 * improvement: sb-sprof traces call stacks to an arbitrary depth on
843 x86/x86-64, rather than the previous fixed depth of 8
844 * improvement: another pthread back-end of mutex "pthread-futex".
845 * bug fix: non-ascii command-line arguments are processed correctly
846 (thanks to Yaroslav Kavenchuk)
847 * bug fix: non-required arguments were not passed correctly when a method
848 defined using DEFMETHOD was called from a mop-generated method using
849 CALL-NEXT-METHOD (reported by Pascal Costanza)
850 * bug fix: recursion is now permitted in accessors through
851 SLOT-UNBOUND. (reported by Pascal Costanza)
852 * bug fix: an error was signaled at startup if the HOME environment
853 variable was defined, but had an empty value (reported by Peter Van Eynde)
854 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
855 signal a PROGRAM-ERROR, not a BUG.
856 * bug fix: SB-POSIX:READDIR works when built with large file support.
857 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
859 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
860 for better automatic indentation support. (thanks to Matt Pillsbury)
861 * optimization: loading generic functions no longer takes O(n^2) time,
862 proportional to the amount of methods in the generic function
863 (reported by Todd Sabin and Jeremy Brown)
864 * optimization: the FIND and POSITION family of sequence functions
865 are significantly faster on arrays whose element types have been
867 * improvements to the Windows port:
868 ** Intermittent heap corruption problems have been fixed. (thanks
869 to Alastair Bridgewater)
870 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
872 ** Lisp is able to unwind foreign exception frames from alien
873 callbacks. (thanks to Alastair Bridgewater)
875 changes in sbcl-1.0 relative to sbcl-0.9.18:
876 * improvement: experimental support for threading on FreeBSD/x86.
877 (thanks to NIIMI Satoshi)
878 * improvement: runtime option --dynamic-space-size can be used
879 to set the size of the dynamic space reserved on startup.
880 * improvement: floating point modes in effect are now saved in
881 core, and restored on startup.
882 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
883 startup, not time since first call to GET-INTERNAL-REAL-TIME.
884 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
885 threads are not running after *SAVE-HOOKS* have run.
886 * improvement: writes to CLOS instance slots are type-checked in code
887 compiled with (SAFETY 3)
888 * improvement: floating-point exception handling on FreeBSD (thanks to
890 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
891 (thanks to Zach Beane)
892 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
894 * improvement: added support for the Shift-JIS external format.
895 (contributed by NIIMI Satoshi)
896 * improvement: callbacks are supported on Linux/PPC. (thanks to
898 * bug fix: compiler bug triggered by a (non-standard) VALUES
899 declaration in a LET* was fixed. (reported by Kaersten Poeck)
900 * bug fix: file compiler no longer confuses validated and already
901 dumped structurres. (reported by Kaersten Poeck)
902 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
903 fill-pointer signals a type-error as required. (thanks to
905 * bug fix: disassemly of funcallable instances works.
906 * bug fix: single stepping on PPC.
907 * bug fix: fix thread-safety problems in the type system (generally
908 manifesting as nonsensical errors like "STRING is a bad type specifier
909 for sequences" or "The value 1 is not of type FIXNUM")
910 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
911 (reported by Josip Gracin)
912 * bug fix: an error is signaled for attempts to displace arrays with
913 incompatible element types (thanks to Mario Mommer)
914 * bug fix: more correct handling of wide characters in debug info
915 (bug reported by Attila Lendvai and fixed by Juho Snellman)
916 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
917 and don't cause extra consing
918 * optimization: MAP and MAP-INTO are significantly faster on vectors
919 whose elements types have been declared.
920 * Improvements to SB-SPROF:
921 ** Support for allocation profiling
922 ** Reduced profiling overhead, especially for long profiling runs
923 * Improvements to the Windows port:
924 ** floating point exceptions are now reported correctly.
925 ** stack exhaustion detection works partially.
926 ** more accurate GET-INTERNAL-REAL-TIME.
927 ** better RUN-PROGRAM behaviour in case of errors in the child
929 ** PROBE-FILE now simplifies pathnames correctly.
930 ** DIRECTORY now works correctly with :WILD-INFERIORS.
931 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
932 1st 1970 (although time timezone data will be unreliable).
933 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
935 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
936 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
937 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
939 * bug fix: two potential GC deadlocks affecting threaded builds.
940 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
941 returns T (reported by Anton Kazennikov)
942 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
943 with non-variable places
944 * bug fix: remove a race condition in the setting of
945 funcallable-instance functions, this should make threaded CLOS
946 code more stable against memory faults.
947 * bug fix: corruption of specials when unbinding is interrupted by an
948 asynchronous unwind (reported by Hannu Koivisto)
949 * improvement: the debugger will now also display local variables that
950 are only used once, for code compiled with a DEBUG optimization quality
953 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
954 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
955 * incompatible change: External-format support for FFI calls. The
956 SB-ALIEN:C-STRING no longer implies an ASCII
957 external-format. Instead the string is subject to external-format
958 conversion. Additionally return values of type SB-ALIEN:C-STRING
959 are by default of type (SIMPLE-ARRAY CHARACTER), not
960 SIMPLE-BASE-STRING. For an alien type that behaves like the old
961 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
962 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
963 * incompatible change: SB-EXT package no longer contains the
964 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
965 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
967 * incompatible change: the single-stepper is no longer available
968 on Alpha, Mips and Sparc platforms.
969 * minor incompatible change: the direct superclasses of
970 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
971 not (STANDARD-OBJECT FUNCTION). This makes the
972 class-precedence-lists of GENERIC-FUNCTION and
973 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
975 * minor incompatible change: the default stream external format on
976 non-unicode SBCL is no longer detected from the locale, but is always
978 * new feature: Add a version of evaluator that uses an interpreter instead
979 of the compiler. EVAL still uses the compiler by default, to switch it
980 to use the interpreter, set the value of the variable
981 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
982 * minor incompatible change: the single-stepper REPL has been merged
983 with the normal debugger (see the "Stepping" heading of the debugger help
984 for more details). The debugger command STEP will no longer switch
985 to the single-stepper REPL.
986 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
987 for a type now works.
988 * bug fix: loading of default sysinit file works. (thanks to Leonid
990 * bug fix: better detection of circularities in the file-compiler.
991 (reported by Marco Monteiro)
992 * bug fix: the CL pathname functions now work with files that have
993 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
994 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
995 systems with Unix98 pty semantics.
996 * bug fix: ASDF-INSTALL will now work with bsd tar.
997 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
999 * bug fix: timers expiring in dead threads no longer cause a
1000 type-error (reported by Paul "Nonny Mouse").
1001 * bug fix: thanks to more lightweight single-stepper instrumentation,
1002 code compiled with (DEBUG 3) will compile and execute significantly faster,
1003 and will have more accurate type-inferencing than before
1004 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1005 whose bindings are modified
1006 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1007 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1008 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1009 CL:SOFTWARE-VERSION with proper version
1011 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1012 * feature: implemented the READER-METHOD-CLASS and
1013 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1014 as specified by AMOP.
1015 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1017 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1018 (thanks to Lutz Euler)
1019 * optimization: hashing of general arrays and vectors has been
1020 improved. (reported by Any Fingerhut)
1021 * enhancement: SB-INTROSPECT is now able to find definitions of
1022 profiled functions. (thanks to Troels Henriksen)
1023 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1024 as well, allowing compiler-macros for SETF-functions to expand.
1025 * enhancement: step-instrumentation no longer wraps known
1026 single-value functions with multiple-value context, allowing
1027 better type inference.
1028 * fixed bug #337: use of MAKE-METHOD in method combination now works
1029 even in the presence of user-defined method classes. (reported by
1030 Bruno Haible and Pascal Costanza)
1031 * fixed bug #339(c): if there are applicable methods not part of any
1032 long-form method-combination group, call INVALID-METHOD-ERROR.
1033 (reported by Bruno Haible)
1034 * fixed bug #361: the :FUNCTION initarg in the protocol for
1035 initialization of methods can now be used to override
1036 internally-produced optimized functions. (reported by Bruno
1038 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1039 system-provided lambda expression no longer cause warnings about
1040 unbound #:|pv-table| symbols.
1041 * bug fix: improved the handling of type declarations and the
1042 detection of violations for keyword arguments with non-constant
1044 * bug fix: potentially erronous calls to PATHNAME and
1045 MERGE-PATHNAMES were being flushed in some cases.
1046 (reported by Richard Kreuter)
1047 * bug fix: compiled calls to TYPEP were mishandling obsolete
1048 instances. (reported by James Bielman and Attila Lendvai)
1049 * bug fix: format strings with ~> without matching ~< no longer
1050 trigger an AVER, but signal an understandable error instead.
1051 (reported by Antonio Martinez)
1052 * bug fix: specifying an output-file without a pathname-type for
1053 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1054 of input-file instead of "fasl". (reported by Robert Dodier)
1055 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1056 (reported by James Y Knight).
1057 * bug fix: compiler-macros-function did not consider the environment
1058 argument for shadowing by local functions.
1059 * bug fix: compiler-macros expansion was inhibited by local INLINE
1061 * bug fix: inline expansions of known functions were subject to
1062 step-instrumentation in high debug policies, leading to problems
1063 with type-inference.
1064 * bug fix: compiler failed to differentiate between different CONS
1065 types in some cases.
1066 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1067 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1068 * bug fix: #368: incorrect use of expressed vs. upgraded array
1070 * bug fix: #306a: more precise unions of array types.
1071 * thread-safety improvements:
1072 ** CONDITION-WAIT could return early on Linux, if the thread was
1073 interrupted and subsequently continued with SIGCONT.
1074 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1076 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1077 * added support for the ucs-2 external format. (contributed by Ivan
1079 * minor incompatible change: pretty printing of objects of type
1080 (cons symbol) is, in the default pprint-dispatch-table, now
1081 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1083 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1084 called later in a class's lifetime, possibly as late as when the
1085 first instance of the class is created. Previously,
1086 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1087 class became finalizeable.
1088 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1089 for multibyte external-format streams. (thanks to "vbzoli")
1090 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1091 with :ALLOCATION :CLASS inherited from superclasses of the
1093 * fixed bug: anonymous classes can now be created using the :NAME
1094 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1095 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1096 * fixed bug: core-files saved with :EXECUTABLE T can again be
1097 executed when SBCL_HOME isn't set. (reported by James Knight)
1098 * fixed bug: toplevel LOCALLY forms with declarations could
1099 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1100 * fixed bug: printing from several different threads using different
1101 values of *print-case* could cause invalid output, due to
1102 some internal special variables of the printer not being bound
1103 thread-locally (reported by Max Mikhanosha)
1104 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1105 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1106 (reported by Pascal Costanza)
1107 * fixed bug: FINALIZE-INHERITANCE is called from
1108 REINITIALIZE-INSTANCE on classes when the class has previously
1109 been finalized, as required by AMOP.
1110 * minor code generation optimizations:
1111 ** better register allocation in CLOS dispatching functions
1112 ** overflow detection when coercing signed bytes to fixnums on x86-64
1113 ** is now implemented with one IMUL instruction instead of three shifts
1114 ** more efficient bit-vector access on x86 and x86-64
1115 ** more efficient access to raw structure slots on x86 and x86-64
1116 * fixed some bugs revealed by Paul Dietz' test suite:
1117 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1118 return its argument.
1120 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1121 * feature: thread support on Solaris/x86, and experimental thread support
1123 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1125 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1126 default method for SLOT-UNBOUND.
1127 * minor incompatible change: prevent the user from specializing the
1128 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1129 somewhat counter to the intent of the protocol, I (CSR) think, and
1130 additionally it just doesn't work in SBCL as currently
1131 implemented, thanks to optimizations (that are always valid for
1132 the other three SLOT-VALUEish functions, but not for the setter).
1133 * minor incompatible change: the :SB-LDB feature is now enabled by
1134 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1135 the low-level debugger.
1136 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1137 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1138 down to an SB-SYS:FD-STREAM, allowing direct communication in
1139 more cases. Temporary files and pipes are still used as a fallback
1141 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1142 on the non-reentrant C functions localtime(3) and gmtime(3).
1143 * bug fix: global optimization policy was not visible in LOCALLY and
1145 * bug fix: class objects can be used as specializers in methods.
1146 (reported by Pascal Costanza)
1147 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1148 with a type error. (reported by blitz_ on #lisp)
1149 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1150 failed with a type error. (reported by Pascal Bourguignon)
1151 * bug fix: merging pathnames against defaults with :DIRECTORY
1152 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1153 (reported by James Y Knight)
1154 * bug fix: saving large (>2GB) cores on x86-64 now works
1155 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1156 constant index and a (simple-array (signed-byte 32)) array
1157 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1158 error (patch by Robert J. Macomber)
1159 * feature: TIME now displays the amount of run-time spent in GC
1160 * bug fix: The debugger now does a better job of respecting
1161 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1162 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1163 workaround for bug 403.)
1164 * bug fix: floating point exception handling now works on FreeBSD
1165 (thanks to NIIMI Satoshi)
1166 * fixed some bugs revealed by Paul Dietz' test suite:
1167 ** MISC.641: LET-conversion were not supposed to work in late
1169 * improvements to the Win32/x86 port:
1170 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1171 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1173 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1174 * new feature: source path information is generated for macro-expansion
1175 errors for use in IDE's like Slime (thanks to Helmut Eller)
1176 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1177 * bug fix: compiler does not loop forever on an invalid type in
1179 * improvement: compilation of most CLOS applications is significantly
1181 * optimization: added a limited bytecode compiler for simple toplevel
1182 forms, speeding up compilation and FASL loading
1183 * bug fix: the statistical profiler now properly distinguishes anonymous
1186 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1187 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1188 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1189 into subclasses of the finalized class. Now user calls to
1190 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1191 system return before any subclasses are finalized.
1192 * minor incompatible change: The reader no longer ignores errors
1193 regarding non-existent packages in #+ and #- feature tests.
1194 * new feature: command line options --no-sysinit, --no-userinit to
1195 inhibit loading the corresponding init files
1196 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1197 for sending data through UDP sockets (thanks to François-René Rideau)
1198 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1199 error is signalled instead (thanks to François-René Rideau)
1200 * minor incompatible change: Error signalling behaviour of lexical
1201 operations violating package locks has changed slightly. Refer to
1202 documentation on package locks for details.
1203 * enhancement: EVAL can process IF-expressions without resorting to the
1205 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1206 constant-folded properly due to missing out-of-line functions.
1207 (reported by Mika Pihlajamäki)
1208 * bug fix: LISTEN sometimes returned T even in cases where no data was
1209 immediately available from the stream
1210 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1211 were reversed. (reported by Levente Mészáros)
1212 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1213 when given a vector displaced to an adjustable array. (reported
1215 * bug fix: derivation of float boundaries from numbers outside the
1216 appropriate float range (reported by John Wiseman)
1217 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1219 * bug fix: precomputing cache entries for generic functions with
1220 some subclasses of specializers as yet invalid does not attempt to
1221 fill a cache line with a negative offset. (reported by Levente
1223 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1224 allow condition class objects as arguments to DOCUMENTATION and
1225 (SETF DOCUMENTATION); only find and set documentation for
1226 structure names for the STRUCTURE doc-type. (suggested by Gary
1228 * improvements to the Win32/x86 port:
1229 ** (user-homedir-pathname) and default initialization file
1230 locations now know about the user's "Documents and Settings"
1231 directory (thanks to Yaroslav Kavenchuk)
1232 ** run-program is implemented (thanks to Mike Thomas)
1233 ** sockets support (thanks to Timothy Ritchey)
1234 ** better backtrace support (thanks to Alastair Bridgewater)
1235 ** sb-grovel supported
1236 ** asdf-install and sb-posix work somewhat
1237 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1238 * improvements to the Solaris/x86 port:
1239 ** works on Solaris 11/Solaris Express
1240 ** floating-point exception handling support
1241 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1242 * fixed some bugs revealed by Paul Dietz' test suite:
1243 ** REMOVE-METHOD returns its generic function argument even when
1244 the method is not one of the generic functions' methods.
1245 ** objects declared as MEMBER types can be admissible arguments to
1246 structure accessors.
1247 ** printing characters should simply be printed by the FORMAT ~:C
1249 ** compiler failure when compiling functions with hairy constant
1250 defaults for optional parameters.
1251 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1252 function, which is already optimized.
1254 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1255 * new platform: experimental support for SBCL x86/Darwin, including
1256 MacOS X 10.4.5 on Intel.
1257 * new feature: Unicode character names are now known to the system
1258 (through CHAR-NAME and NAME-CHAR).
1259 * new feature: the filesystem location of SBCL's core file is
1260 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1261 * minor incompatible change: the contrib modules SB-POSIX and
1262 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1263 this change is to make it easier to distribute
1264 location-independent binaries.
1265 * bug fix: as implied by AMOP, standardized classes no longer have
1266 slots named by external symbols of public packages. (reported by
1268 * optimization: calling functions via constant symbols -- as in
1269 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1270 via the function object as in (FUNCALL #'FOO).
1271 * optimization: on x86 and x86-64, the calling convention for the
1272 case of dealing with an unknown number of values has been altered
1273 to be friendlier to the prediction heuristics implemented,
1274 particularly with reference to CALL and RET pairing. (thanks to
1275 Alastair Bridgewater)
1276 * optimization: on x86, the code for access to thread-local symbol
1277 values and binding/unbinding of thread-local symbols is smaller.
1278 (thanks to Alastair Bridgewater)
1279 * enhancement: CONSTANTP is now able to determine constantness of
1280 more complex forms, including calls to constant-foldable standardized
1281 functions and some special forms beyond QUOTE.
1282 * fixed bug: occasional GC crashes on Solaris/x86
1283 * optimization: x86-64 supports stack allocation of results of simple
1284 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1285 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1286 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1287 (thanks to James Knight)
1288 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1289 by the standard (thanks to Stephen Compall)
1291 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1292 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1293 be used for bundling the runtime and the core file into one
1294 executable binary. This feature is not currently supported on all SBCL
1295 platforms. (thanks to James Bielman and NIIMI Satoshi)
1296 * new feature: a generational or ephemeral garbage collector is now
1297 the default on the PowerPC platform (both Linux and Darwin). The
1298 old Cheney (stop and copy) collector is a build-time option.
1299 (thanks to Cyrus Harmon, after Raymond Toy)
1300 * minor incompatible change: the method by which SBCL finds its
1301 contributed modules has changed; it no longer relies on symbolic
1302 links from an $SBCL_HOME/systems directory, but searches directly
1303 in the subdirectories of $SBCL_HOME.
1304 * enhancement: the dynamic heap size on the Linux/PPC platform is
1305 markedly larger, even using the older Cheney garbage collector.
1306 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1307 certain complicated string types. (reported by Gary King)
1308 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1309 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1310 * fixed bug: DOCUMENTATION on structures no longer signals an
1311 error if no structure type of the right name exists. (reported by
1313 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1314 applicable non-standard primary or :AROUND method on
1315 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1316 no longer cause unbound variable errors. (reported by Kevin Reid)
1317 * fixed bug: metacircle resolution in cases where methods have slots
1318 added before the slots from STANDARD-METHOD. (reported by Jean
1320 * fixed bug: the Power PC assembly code for calling into lisp
1321 sometimes computed its return address wrongly (depending on the
1322 15th bit of the address, fixed at link time). This used to
1323 manifest itself in a segmentation violation while building PCL.
1324 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1325 many others over the years)
1326 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1327 the stack pointer (treated as a signed integer) is negative.
1328 (thanks to Peter van Eynde)
1330 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1331 * new platform: experimental support for the Windows operating
1332 system has been added. (thanks to Alastair Bridgewater)
1333 * fixed several bugs in and robustified the PPC FFI (including
1334 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1335 * bug fix: allow non-simple string symbol names (reported by Paul
1337 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1339 * bug fix: saving a core corrupted callbacks on x86/x86-64
1340 * bug fix: closed a loophole in metacircularity detection and
1341 grounding in the PCL implementation of CLOS.
1342 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1343 (reported by tomppa on #lisp)
1344 * bug fix: clear the direction flag on Lisp -> C transitions, as
1345 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1346 (reported by Andrej Grozin and Hendrik Maryns)
1347 * optimization: major improvements to GC efficiency on GENCGC platforms
1348 * optimization: faster implementation of EQUAL
1349 * optimization: emit more efficient opcodes for some common
1350 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1351 * optimization: several other minor code-generation improvements on x86-64
1352 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1354 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1355 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1356 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1357 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1358 by ANSI as a generic function, and for consistency (SETF
1359 GENERIC-FUNCTION-NAME) is treated likewise.
1360 * fixed bug #233.b: make constraint propagation notice when a variable
1361 value is changed after it is referenced but before it is used
1362 * fixed bug #296: no more arbitrary behaviour from filenames with
1363 odd characters as --load arguments. (reported by Adam Warner)
1364 * fixed bug #390: :CHARACTER-SET pathname components now work as
1365 expected. (reported by Tim Daly Jr)
1366 * fixed bug #391: complicated :TYPE intersections in slot
1367 definitions no longer cause an error in PCL internals.
1368 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1369 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1370 returning the number of octets which would be written to the
1371 file-stream. (thanks to Robert J. Macomber)
1372 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1373 arguments to the internal stream functions. (thanks to David
1375 * bug fix: stack exhaustion is now handled on the x86/SunOS
1376 platform. (thanks to Daisuke Homma)
1377 * bug fix: more accurate ROOM results on GENCGC platforms
1378 * optimization: improved type inference for arithmetic-for
1379 index variables in LOOP
1380 * optimization: faster floating-point SQRT on x86-64
1381 * fixed some bugs revealed by Paul Dietz' test suite:
1382 ** DOCUMENTATION returns NIL instead of "" for method combinations
1383 that don't have a docstring
1385 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1386 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1387 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1388 therefore are not customizeable by user code (as seems to be at
1389 least permitted and maybe required by AMOP). As a consolation,
1390 however, the SBCL implementation of these functions now calls
1391 REINITIALIZE-INSTANCE as specified by AMOP.
1392 * bug fix: slot-definition documentation is propagated as per ANSI
1393 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1394 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1395 argument, as do eslotd initialization methods. (from Pascal
1396 Costanza's "Closer" project)
1397 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1398 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1400 * bug fix: it is now possible to have more than one subclass of
1401 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1402 (reported by Bruno Haible, Pascal Costanza and others)
1403 * bug fix: the dependent update protocol now works for generic
1404 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1405 and Pascal Costanza)
1406 * bug fix: condition-class instances corresponding to
1407 DEFINE-CONDITION forms are now created eagerly. (reported by
1408 Kalle Olavi Niemitalo on comp.lang.lisp)
1409 * bug fix: floating point printing is more accurate in some
1410 circumstances. (thanks to Simon Alexander)
1411 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1412 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1413 * bug fix: callbacks on OS X now preserve stack-alignment.
1414 (thanks to Cyrus Harmon)
1415 * enhancement: the x86-64 disassembler is much better at
1416 disassembling SSE instructions. (thanks to Lutz Euler)
1417 * enhancement: CHECK-TYPE on a variable now allows type-inference
1418 (thanks to Vincent Arkesteijn)
1419 * optimization: improved performance of EUC-JP external format.
1420 (thanks to NIIMI Satoshi)
1421 * optimization: performance improvements to IO on file streams of
1422 :ELEMENT-TYPE CHARACTER
1423 * optimization: much faster memory allocation on x86-64
1424 * optimization: faster garbage collections (latency and throughput)
1426 * optimization: faster bignum subtraction and division on x86 and x86-64
1427 (thanks to Lutz Euler)
1428 * optimization: more accurate type-inference for loops that use a
1429 floating point index variable or a negative step.
1431 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1432 * bug fix: add a workaround to SBCL looping infinitely at startup on
1433 Linux kernels with apparently buggy implementations of personality().
1434 (thanks to Svein Ove Aas)
1435 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1436 (thanks to David Lichteblau)
1437 * bug fix: local bindings shadow global symbol macros for
1438 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1439 * optimization: non-open coded uses of numeric comparison operators
1440 (e.g. >) no longer cons when called with more than one parameter
1441 on platforms supporting dynamic-extent allocation.
1442 * enhancement: saving cores with foreign code loaded is now supported
1443 on MIPS/Linux in addition to the previously supported platforms.
1444 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1445 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1446 * bug fix: division by zero in sb-sprof when no samples were collected
1447 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1448 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1449 (thanks to Alexander Kjeldaas)
1450 * new feature: ignoring the return values of destructive functions
1451 like NREVERSE signals a compile-time style-warning.
1452 (thanks to Kevin Reid)
1453 * new feature: the alignment of alien structure fields can be
1454 explicitly specified. (thanks to Cyrus Harmon)
1455 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1456 and dump core on SIGQUIT
1458 ** incompatible change: threads do not inherit values of specials
1459 from their parents (see manual)
1460 ** bug fix: threads stacks belonging to dead threads are freed by the
1461 next exiting thread, no need to gc to collect thread stacks anymore
1462 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1463 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1464 ** bug fix: dynamic variable and thread start related gc lossage
1465 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1467 * fixed some bugs revealed by Paul Dietz' test suite:
1468 ** SUBTYPEP is slightly more accurate on heinously complicated
1469 CONS types where some of the members have uncertain (in the
1470 NIL, NIL sense) type relationships to each other.
1472 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1473 no longer purified unless :PURIFY T is explicitly specified.
1474 ** Non-purified cores are significantly smaller than before
1476 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1477 * new feature: timers based on Zach Beane's excellent timer package
1478 * added support for the following external formats: koi8-u,
1479 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1480 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1481 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1482 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1483 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1484 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1485 * incompatible change: a threaded SBCL will no longer revert to
1486 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1487 * bug fix: interrupts are disabled until startup is complete; no
1488 more sigsegvs when receiving a signal to soon
1489 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1491 * bug fix: add a workaround for the memory randomization features in
1492 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1493 workaround will only be in effect on systems with the proc filesystem
1495 * bug fix: printing objects of type HASH-TABLE signals a
1496 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1498 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1499 have been running for over 50 days. (reported by Gilbert Baumann)
1500 * bug fix: the logic for getting names of functions gets less
1501 confused when confronded with alternate-metaclass
1502 funcallable-instances. (reported by Cyrus Harmon)
1503 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1504 consistent, even on internal alternate-metaclass objects.
1505 * bug fix: flush closure information collected by physical
1506 environment analysis prepass before the main pass. (bug reported
1508 * bug fix: compiler pack phase does not modify a hashtable, which is
1509 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1510 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1511 may be expanded. (reported by Alexander <malishev> on c.l.l)
1512 * bug fix: no more style-warnings when compiling code using LAST
1513 under high-SPEED low-SPACE settings. (reported by David Wallin
1515 * fixed some incompatibilities between SBCL's MOP and the MOP
1517 ** the METAOBJECT class is now implemented;
1518 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1519 STANDARD-OBJECT, as required;
1520 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1521 now compatible (as required by VALIDATE-SUPERCLASS); there
1522 remains a consistency requirement over the presence or absence
1523 of the FUNCTION class in the superclass list of finalized
1524 classes; see the manual for more details;
1525 ** the :AROUND method for COMPUTE-SLOTS on
1526 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1527 requested slot ordering.
1529 ** bug fix: parent thread now can be gc'ed even with a live
1531 ** bug fix: binding a special with PROGV to no value is not
1532 the same as that symbol not having been bound (thanks to
1534 ** bug fix: binding specials is thread safe (thanks to
1536 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1537 and SB-SPROF do not die with 'no handler for signal XX in
1538 interrupt_handle_now(..)' anymore
1539 ** bug fix: WITH-TIMEOUT works with multiple running threads
1540 * fixed some bugs revealed by Paul Dietz' test suite:
1541 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1542 the :method-class keyword argument.
1544 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1545 * new port: the Solaris operating system on x86 processors is now
1546 mostly supported, though some rough edges in the environment
1547 remain. (thanks to Daisuke Homma)
1548 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1549 space for its heap. (thanks to Thiemo Seufer)
1550 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1551 closures. (thanks to Thiemo Seufer)
1552 * minor incompatible change: eof selects abort in the debugger.
1553 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1554 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1556 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1557 deprecated, and will go away in a future revision of SBCL.
1558 * minor incompatible change: GC-ON and GC-OFF are no longer
1559 implemented with a counter, it does not matter how many times gc
1560 is switched on or off
1561 * bug fix: discriminating functions for generic function classes
1562 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1563 make invalid assumptions about method precedence order. (reported
1565 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1566 platforms. (thanks to Thiemo Seufer)
1567 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1568 * bug fix: correct pathname printing with printer escaping is on.
1569 (thanks to Kevin Reid)
1570 * bug fix: complex VOP definitions in "user-space" no longer trigger
1571 package locks. (reported by Zach Beane)
1572 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1573 for funcallable-instance functions, and is no different from
1575 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1576 characters in the current readtable. (reported by Nicholas Neuss)
1577 * bug fix: SUBTYPEP on various CONS types returns more a more
1578 accurate acknowledgment of its certainty.
1579 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1580 lists in some cases. This partially fixes bug 384.
1581 * flush all standard streams before prompting in the REPL and the
1583 * bug fix: signal handling and triggering gc do not conflict
1584 directly anymore, in particular a high frequency sb-sprof does
1585 not prevent gc from running
1586 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1587 approximation for timezone and DST information between the
1588 universal time epoch and the smallest negative 32-bit time_t.
1589 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1590 year 1899 so long as the encoded time is non-negative.
1591 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1592 the correct answer for non-constant rotate amounts when performing
1593 an inline 32-bit rotation.
1595 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1596 there is only one thread in the session
1597 ** bug fix: memory leak for streams created in one thread and
1598 written to in another
1599 ** bug fix: lockup when compiled with gcc4
1600 ** bug fix: race that allows the gc to be triggered when gc is
1602 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1603 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1605 ** bug fix: finished threads are gc'ed properly
1606 * fixed some bugs revealed by Paul Dietz' test suite:
1607 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1608 has arguments with a different set of applicable methods from
1609 the orignal arguments.
1610 ** The type error thrown by MAP now has a correct expected-type
1612 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1613 name a compiled function.
1614 ** Three MISC tests where a large bignum was improperly coerced to
1615 a float that couldn't represent that bignum during type
1616 derivation were fixed.
1617 ** SUBTYPEP can now handle types involving the negation of a
1618 list-form FUNCTION type.
1619 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1620 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1621 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1623 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1624 * New feature: Experimental support for bivalent streams: streams
1625 opened with :element-type :default now allow character and binary
1626 (unsigned-byte 8) I/O
1627 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1628 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1630 * bug fix: no more highly sporadic "couldn't check whether ~S is
1631 readable" when reading a stream and an interrupt hits in the middle
1632 of a select system call
1633 * compiler better recognizes complex arrays (reported by Eduardo
1635 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1636 for variables on SBCL built with linkage-tables. (reported by Luis
1638 * various error reporting improvements.
1639 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1640 (Thanks to Thiemo Seufer)
1641 * bug fix: nested reader invokations work correctly
1642 * bug fix: it is possible to have simultaneous references to foreign
1643 code and foreign data with the same name.
1645 ** added x86-64 support
1646 ** incompatible change: the threading api now works with thread
1647 objects instead of thread ids
1648 ** bug fix: threads are protected from signals and interruption when
1649 starting up or going down
1650 ** bug fix: a race where an exiting thread could lose its stack to gc
1651 ** bug fix: don't halt on infinite error in threads if possible
1652 ** fixed numerous gc deadlocks introduced in the pthread merge
1653 ** bug fix: fixed thread safety issues in read and print
1654 ** bug fix: debugger doesn't hang on session lock if interrupted at
1655 an inappropriate moment
1656 ** bug fix: run-program is now thread safe(r)
1657 ** bug fix: inner with-recursive-lock no longer releases the mutex
1658 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1659 * fixed some bugs revealed by Paul Dietz' test suite:
1660 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1661 EXPECTED-TYPE slots filled.
1662 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1664 ** MAKE-CONDITION accepts classes as type-designators.
1665 ** COMPILE may never return NIL.
1666 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1667 range before calling Unix time functions
1669 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1670 * numerous signal handling fixes to increase stability
1671 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1672 * minor incompatible change: we now correctly canonize default
1673 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1674 as per the MOP, rather than the historical (INITARG INITFUNCTION
1675 INITFORM). (reported by Bruno Haible)
1676 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1677 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1678 for more information.
1679 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1680 pathname is a directory pathname.
1681 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1682 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1684 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1685 :METACLASS argument in addition to a class name. (reported by
1686 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1687 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1688 (reported by Bruno Haible; patch from Gerd Moellmann)
1689 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1691 * bug fix: more cleanups to the floating point exception handling on
1692 x86-64 (thanks to James Knight)
1693 * bug fix: the compiler does not try to constant fold calls of
1694 COERCE and BIT-* functions when they return freshly constructed
1695 objects. (reported by Nikodemus Siivola and Paul Dietz)
1696 * optimization: improved the MIPS versions of generic (in the
1697 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1698 * optimization: direct conversion of (unsigned-byte 32) to floats on
1699 the PowerPC platform.
1700 * optimization: structure instances with raw slots now use less
1701 memory, and probably show better memory locality. (thanks to
1703 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1704 are treated as requests for stack allocation on the x86-64,
1705 Alpha-32, PPC and SPARC platforms.
1706 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1707 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1708 (thanks to Tony Martinez)
1710 ** gcing a dead thread can no longer lead to lockups
1711 ** threads block signals until they are set up properly
1712 ** errno is no longer shared by threads
1713 ** interrupt-thread restores the eflags register on x86
1714 ** fixed some lockups due to gc/thread interaction
1715 * fixed some bugs revealed by Paul Dietz' test suite:
1716 ** invalid dotted lists no longer raise a read error when
1717 *READ-SUPPRESS* is T
1718 ** signal an error if a symbol that names a declaration is used
1719 as the name of a type, or vice versa
1720 ** allow using the (declare (typespec var*)) abbreviation for
1721 (declare (type typespec var*)) with all type specifiers
1722 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1723 ** always return NIL from PPRINT-POP when OBJECT is NIL
1724 ** don't signal errors when pretty-printing malformed LABELS,
1725 FLET or MACROLET forms
1726 ** declarations in a DOLIST body are also in scope for the
1728 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1730 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1733 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1734 * fixed cross-compiler leakages that prevented building a 32-bit
1735 target with a 64-bit host compiler.
1736 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1737 opened with :IF-EXISTS :OVERWRITE.
1738 * fixed bug 281: error for an invalid qualifier in a short-form method
1739 combination method is not signalled until the faulty method is called.
1740 * bug fix: iteration variable type inferrer failed to deal with open
1741 intervals. (reported by Alan Shields)
1742 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1743 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1744 or more runtime options were provided to the sbcl binary.
1745 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1746 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1747 * optimization: there's now a fast-path for fixnum arguments in the
1748 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1749 * optimization: the code generated on x86-64 is more compact thanks
1750 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1751 * bug fix: floating-point exception handling now partly works on
1752 x86-64. (Thanks to James Knight)
1753 * improvement to the MIPS backend from Thiemo Seufer: C-style
1754 64-bit long long arguments and return values to alien functions
1756 * fixed some bugs revealed by Paul Dietz' test suite:
1757 ** the type-error signalled from WARN has a filled-in DATUM slot.
1758 ** the type-error required when a stream is not associated with
1759 a file has the stream as its datum.
1760 ** type-errors on single-floats on x86-64 no longer have
1761 :INVALID-OBJECT as the datum
1762 ** the type-errors signalled for invalid function names now have
1763 a correct expected type
1764 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1765 for structures defined with a :TYPE. Documentation strings for
1766 typed structures are no longer immediately discarded
1767 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1768 broadcast streams and synonym streams. FILE-LENGTH now also works
1769 on broadcast streams.
1771 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1772 * incompatible change: the --noprogrammer option, deprecated since
1773 version 0.7.5, has been removed. Please use the equivalent
1774 --disable-debugger option instead.
1775 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1777 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1778 inoperational for a while now) has been completely removed.
1779 * null lexical environments are now printed as #<NULL-LEXENV>,
1780 significantly reducing the amount of clutter in typical backtraces.
1781 * documentation on weak pointers, finalization, and after GC hooks
1782 has been added to the manual.
1783 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1784 as other specialized array types, is much faster. SUBSEQ and
1785 COPY-SEQ on such arrays have also been sped up.
1786 * optimization: EQL is now more efficient when at least other argument
1787 is known to be of type (OR FIXNUM (NOT NUMBER)).
1788 * fixed bug: the runtime is now less vulnerable to changes in the
1789 size of the SBCL object on OS X, and virtual memory is reserved for
1790 all spaces that need to be at a fixed address.
1791 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1792 * fixed bug: finalizers and after GC hooks that cause consing are now
1794 * fixed bug: compiler error messages and summaries are now printed to
1795 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1796 * fixed inference of the upper bound of an iteration variable.
1797 (reported by Rajat Datta).
1798 * fixed bug 211e: calling local functions with duplicated constant
1799 keyword argument no longer causes a bogus style warning about an
1801 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1802 ftype declaration to be disregarded. (reported by Dave Roberts)
1803 * fixed bug 373: caused by erronous compilation of references to alien
1804 variables in the runtime on ppc/darwin.
1805 * fixed bug 376: CONJUGATE type deriver.
1806 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1807 (reported by Baughn on #lisp)
1808 * a cross-compiler bug on non-x86 platforms has been identified and
1809 fixed. (thanks to Bruno Haible)
1810 * improvements to the MIPS runtime code for increased stability.
1811 (thanks to Thiemo Seufer)
1812 * increased the maximimum compact environment size to allow
1813 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1814 * partial workaround for bug 135: don't name the function we're
1815 calling for hairy cases of EVAL, so as not to accumulate one environment
1816 entry per call to EVAL. (reported by Kevin Reid)
1817 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1818 * optimization: type testing for non-vector arrays should be faster.
1819 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1821 * bug fix: setting 31st element of a bit vector to zero did not work
1823 * bug fix: redefining a class definition which failed due to a
1824 previous accessor / function clash now works (but see BUGS entry
1825 #380 for more problems in this area). (thanks to Zach Beane)
1826 * the long form of DEFINE-METHOD-COMBINATION disables method group
1827 checking when given a single method group with pattern *. (thanks
1829 * on x86 compiler supports stack allocation of results of simple
1830 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1831 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1832 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1833 * fixed some bugs related to Unicode integration:
1834 ** the restarts for recovering from input and output encoding
1835 errors only appear when there is in fact such an error to
1837 * fixed some bugs revealed by Paul Dietz' test suite:
1838 ** MISC.549 and similar: late transformation of unsafe type
1839 assertions into derived types caused unexpected code
1841 ** SCALE-FLOAT type deriver is less wrong.
1842 ** type derivers for EXP, LOG and similar functions compute result
1843 types for complex arguments better.
1844 ** (MISC.563) CONJUGATE type deriver works for very restricted
1846 ** out-of-line type testers for character strings are available.
1847 ** EQUAL compiler transform understands specialness of objects
1849 ** accessing double-floats stored on the stack now works on x86-64.
1850 ** debugger internals could sometimes create invalid lispobjs,
1851 resulting in GC crashes.
1852 ** MISC.548: type check weakening can convert required type into
1854 ** initialization forms for bindings are not in scope of free special
1857 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1858 * incompatible change: thread support for non-NPTL systems has
1859 been removed - locking is buggy and unreliable. A threaded
1860 SBCL build will now warn at startup and refuse to create
1861 new threads, unless futex support is detected in the kernel
1862 * incompatible change: the top level REPL now has only an ABORT
1863 restart associated with it, not TOPLEVEL and ABORT as it used to.
1864 TOP and TOPLEVEL are now available as debugger commands for
1865 returning to the top level.
1866 * incompatible change: forms evaluated in the REPL now use the
1867 global optimization policy.
1868 * incompatible change: user- and system-initialization files are
1869 no longer processed with LOAD, but by READ and EVAL; hence the
1870 global optimization policy, startup package, readtable, etc,
1872 * The .fasl file format number has been incremented because of
1873 various incompatible changes.
1874 * internal entry point details and argument counts no longer appear
1875 in backtraces unless explicitly requested by setting
1876 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1877 * built-in and standard functions no longer have names like "top
1878 level local call to FOO".
1879 * fixed bug 32: functions defined in non-null lexical environments
1880 now have more legible printed representation
1881 * fixed bug 33: functions defined in non-null lexical environemnts
1882 are now more amenable to inspection by INSPECT.
1883 * workaround for bug 354: XEPs no longer appear in backtraces unless
1884 explicitly requested.
1885 * fixed bug: receiving the signal which results from attempting to
1886 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1887 SIGBUS on other *BSDs) on architectures where the C stack is also
1888 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1889 to debugger confusion or wild instability. (reported by Cheuksan
1891 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1892 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1893 (reported by Lutz Euler)
1894 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1895 compiled code. (reported by Paul Dietz)
1896 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1897 specialized parameter to a method to something that is not TYPEP
1898 the specializer is now possible.
1899 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1900 face of package deletion.
1901 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1902 pathname. (reported by David Tolpin on c.l.l)
1903 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1904 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1905 STRING-STREAM and FILE-STREAM, respectively.
1906 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1907 than previously; in addition, its test suite is now run on build.
1908 (thanks to Luke Gorrie)
1909 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1910 * optimization: added a immediate representation for single-floats
1912 * fixed some bugs revealed by Paul Dietz' test suite:
1913 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1914 ** attempting to create a package with a colliding nickname causes
1915 correctable errors to be signalled.
1916 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1917 ** several x86-64 backend bugs related to sign-extension of immediate
1920 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1921 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1922 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1924 * fixed loading of multiply forward-referenced layouts.
1925 (thanks to Cheuksan Wang)
1926 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1927 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1928 (see COMPILE-FILE documentation for details of :PRINT option)
1929 * fixed bugs 19 and 317: fixed-format floating point printing is
1930 more accurate. This also fixes a bug reported by Adam Warner
1931 related to the ~@F format directive.
1932 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1934 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1935 dispatch table if the from-char is a dispatch macro character.
1936 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1937 by the wordsize no longer ignore the last word. (reported by Lutz
1939 * fixed bug in type checking of dynamic-extent variables. (reported
1941 * optimization: sequence traversal functions use their freedom to
1942 coerce function designators to functions.
1943 * optimization: code with many calls to CLOS methods specialized on
1944 CLOS classes has had redundant type checks removed.
1945 * optimization: added declarations to speed up operations that access
1946 the internal character database (for example STRING-UPCASE)
1947 * optimization: comparison operations between floats and sufficiently small
1948 fixnums no longer create extra rationals
1949 * fixed some bugs related to Unicode integration:
1950 ** portions of multibyte characters at the end of buffers for
1951 character-based file input are correctly transferred to the
1952 start of the buffer at the next read.
1953 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1954 passing it through to OPEN.
1955 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1956 argument given to internal calls to OPEN.
1957 ** the built-in comment readers (introduced by character sequences
1958 ";" and "#|") are more forgiving to encoding errors; they will
1959 STYLE-WARN and then attempt to resync the stream at a character
1960 boundary. (thanks to Teemu Kalvas)
1961 * fixed some bugs in the x86-64 port:
1962 ** Negative short int return values from c-calls are sign-extended
1964 ** The stack is aligned to 16-bytes for c-calls, as required by
1965 the ABI. (thanks to Cheuksan Wang)
1966 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1967 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1968 for enums. (thanks to Vincent Arkesteijn)
1969 ** Multiple small optimizations and bugfixes for floating point
1971 * fixed some bugs revealed by Paul Dietz' test suite:
1972 ** Space, Tab, Linefeed, Return and Page have the invalid
1973 secondary constituent character trait.
1974 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1976 ** WITH-INPUT-FROM-STRING should only update the index place on
1978 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1979 works more reliably.
1980 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1981 with one-dimensional arrays with fill pointers.
1982 ** TYPE-OF failed on a complex with an integer realpart and a
1984 ** compiler failure during type inference for the code of form
1985 (IF (EQL X (THE ...)) ...) (MISC.535).
1987 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1988 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1989 platform. The port supports 61-bit fixnums, large memory spaces
1990 and reloading shared object files.
1991 * enhancement: saving cores with foreign code loaded is now
1992 supported on ppc/Darwin in addition to the previously supported
1994 * enhancement: the statistical profiler now walks deeper into the
1995 call stack for more meaningful call-graphs and accrued time
1996 reports (x86/x86-64 only). It also now reports time spent in
1998 * enhancement: it is now possible to trace most individual methods
1999 of a generic function in addition to tracing the generic function
2001 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2003 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2004 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2005 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2006 produces an error. (thanks to Vincent Arkesteijn)
2007 * bug fix: NAMESTRING on pathnames with :WILD components in their
2008 directories works correctly. (thanks to Artem V. Andreev)
2009 * fixed bug 125: compiler preserves identity of closures. (reported
2011 * bug fixed: functions with &REST arguments sometimes failed with
2012 "Undefined function" when compiled with (DEBUG 3). (reported by
2014 * bug fix: overflow during compiling of setting element of a bit
2015 vector with constant index and value. (reported by Timmy Douglas)
2016 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2017 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2018 * on x86 compiler partially supports stack allocation of dynamic-extent
2020 * GO and RETURN-FROM do not check the extent of their exit points
2021 when compiled with SAFETY 0.
2022 * fixed some bugs related to Unicode integration:
2023 ** encoding and decoding errors are now much more robustly
2024 handled; it should now be possible to recover even from invalid
2025 input or output to the terminal. (thanks to Teemu Kalvas)
2026 ** provided a first cut at implementing STRING-TO-OCTETS and
2027 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2028 ** altered the SB-MD5 contributed module slightly, changing the
2029 interface just enough for it to be supportable for builds where
2030 lisp characters are not eight bits.
2031 * fixed some bugs revealed by Paul Dietz' test suite:
2032 ** the FORMATTER-generated functions for ~V[ conditionals require
2033 the correct number of arguments.
2034 ** READ-FROM-STRING returns the mandated second value when applied
2035 to displaced strings.
2036 ** the #\Rubout and #\Backspace characters are treated as invalid
2037 constituent characters by the tokenizer.
2039 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2040 * new feature: reloading changed shared object files with
2041 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2042 * new feature: references to foreign variables and functions
2043 can now be compiled and loaded before the corresponding shared
2044 object file is loaded, as long as the foreign definitions are
2045 available at runtime.
2046 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2047 supported. (thanks to Dan Debertin)
2048 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2049 just keywords. (thanks to Vincent Arkesteijn)
2050 * fixed bug #331: structure-class instances corresponding to
2051 DEFSTRUCT forms are now created eagerly.
2052 * fixed bug #345: backtraces from calls to undefined functions work
2053 on x86 as well. Related bug #61 is now also partially fixed on x86
2054 (backtraces from throws to unknown catch tags.)
2055 * bug fix: lambda-list parsing is now stricter vrt. order and number
2056 of lambda-list keywords.
2057 * bug fix: as specified by AMOP, an error is signalled if a
2058 class-option appears multiple times in a DEFCLASS form. (reported
2060 * bug fix: RANDOM can be compiled when the compiler derives the type
2061 of its numeric argument as a disjoint set of small integers.
2062 (reported by Paul Dietz)
2063 * bug fix: starting a core saved with shared objects loaded when
2064 those objects are not available no longer causes threaded SBCL to
2065 hang. (reported by Sean Ross)
2066 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2068 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2069 argument, which must always be NIL. (reported by Kalle Niemitalo)
2070 * bug fix: printing 1.0d+23 no longer results in an error.
2071 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2072 * bug fix: structure slot setters preserve evaluation order. (thanks
2074 * fixed some bugs related to Unicode integration:
2075 ** RUN-PROGRAM can allow its child to take input from a Lisp
2076 stream. (reported by Stefan Scholl)
2077 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2078 locales. (reported by Ken Causey)
2079 * fixed some bugs revealed by Paul Dietz' test suite:
2080 ** INCF, DECF and REMF evaluate their place form as specified in
2082 ** FORMATTER expands ~{ ~} iteration directives with V or #
2083 parameters correctly.
2084 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2085 consequent uses no arguments correctly.
2086 ** the system has a partial understanding of the (COMPLEX RATIO)
2088 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2091 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2092 * new feature: a build-time option (controlled by the :SB-UNICODE
2093 keyword feature, enabled by default) for building the system with
2094 support for the entire 21-bit character space defined by the
2096 * new feature: the system now has rudimentary external-format
2097 support; the primary user-visible change at this time is that
2098 characters with the high bit set (such as lower-case-e-acute) will
2099 print correctly to a terminal in a UTF-8 environment.
2100 * minor incompatible change: BASE-CHAR no longer names a class;
2101 however, CHARACTER continues to do so, as required by ANSI.
2102 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2103 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2104 has been moved to the SB-EXT package (temporarily re-exported from
2106 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2107 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2109 * fixed bug #351: better error-handlind and reporting for malformed
2111 * fixed bug #350: bignum-printing is now more memory-efficient,
2112 allowing printing of very large bignums, eg. (expt 2 10000000).
2113 (reported by Bruno Haible)
2114 * fixed bug #302: better primitive-type selection for intersection
2116 * fixed bug #308: non-graphic characters now all have names, as
2117 required. (reported by Bruno Haible)
2118 * bug fix: redefining a class with different superclasses now correctly
2119 removes it from the direct-subclasses of its previous superclasses.
2120 (reported by David Morse)
2121 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2122 new value now works. (reported by Bruno Haible)
2123 * bug fix: correct canonicalization of multiple non-standard slot
2124 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2125 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2126 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2127 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2128 now exists, an signals an error.
2129 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2130 during macroexpansion for non-iteration variables as well. (reported
2131 by Bruno Haible for CMUCL)
2132 * bug fix: Cyclic structures and unprintable objects in compiler
2133 messages no longer cause errors. (reported by Bruno Haible)
2134 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2135 and READ-FROM-STRING all now return a primary value of NIL if
2136 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2137 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2138 (reported by Bruno Haible for CMUCL)
2139 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2140 do not designate a condition. (reported by Bruno Haible for
2142 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2143 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2144 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2145 permitting resolution in favour of any of the conflicting symbols.
2146 (reported by Bruno Haible for CMUCL)
2147 * bug fix: EQUAL compiler optimizations is less aggressive on
2148 strings which can potentially compare true despite having distinct
2149 specialized array element types.
2150 * bug fix: unit enumerations can be defined without dividing by
2151 zero. (thanks to Vincent Arkesteijn)
2152 * FORMAT compile-time argument count checking has been enhanced.
2153 (report from Bruno Haible for CMUCL)
2154 * a partial workaround for the bug 262: the compiler does not try to
2155 inline-expand a local function doing RETURN-FROM from a deleted
2156 BLOCK. (thanks to Peter Denno for the bug report and to David
2157 Wragg for the simple test case)
2158 * fixed some bugs revealed by Paul Dietz' test suite:
2159 ** RENAME-PACKAGE allows all package designators as new package
2161 ** constraint propagation and lambda variable substitution are
2162 more cautious in dealing with partially deleted code.
2163 ** compiler sometimes lost reoptimization passes.
2164 ** CERROR, given a condition as condition designator, passes its
2165 remaining arguments to the continue format control without
2167 ** Case-altering FORMAT directives work correctly on non-ASCII
2169 ** The REAL type specifier handles bounds outside the floating
2170 point ranges without signalling FLOATING-POINT-OVERFLOW.
2171 ** Functions with IR1-transformations can create intercomponent
2172 references to global functions.
2173 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2175 ** FORMAT ~R treats a nil value for its first parameter correctly.
2177 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2178 * enhancement: saving cores with foreign code loaded is now
2179 supported on x86/NetBSD and sparc/Linux in addition to the previously
2180 supported platforms.
2181 * bug fix: on some platforms repeated installations caused multiple
2182 copies of HTML documentation to be installed -- should not happen
2183 any more. (reported by Stefan Scholl)
2184 * bug fix: parsing self-recursive alien record types multiple times
2185 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2186 original patch by Helmut Eller for CMUCL)
2187 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2188 (thanks to Richard Kreuter)
2189 * bug fix: defining classes whose accessors are methods on existing
2190 generic functions in other (locked) packages no longer signals
2191 bogus package lock violations. (reported by François-René Rideau)
2192 * bug fix: special variables as DEFMETHOD parameters no longer have
2193 associated bogus type declarations. (reported by David Wragg and
2195 * bug fix: read-write consistency on streams of element-type
2196 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2197 * bug fix: redefiniton of the only method of a generic function with
2198 no DEFGENERIC no longer emits a full WARNING. In addition,
2199 redefinition of generic functions with no DEFGENERIC to an
2200 incompatible lambda list now signals an error. (thanks to Zach
2202 * bug fix: DEFGENERIC now works even when there's a function of the
2203 same name in an enclosing lexical environment. (thanks to Zach
2205 * fixed compiler failure, caused by instrumenting code during
2206 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2207 * optimization: added loop analysis and improved register allocation
2208 * fixed some bugs revealed by Paul Dietz' test suite:
2209 ** POSITION on displaced vectors with non-zero displacement
2210 returns the right answer.
2211 ** (SIMPLE-STRING) is a valid type specifier for sequence
2213 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2215 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2216 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2218 ** The justification version of the FORMAT ~< directive treats
2219 non-zero minpad parameter correctly.
2221 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2222 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2223 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2224 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2225 the supported interface.
2226 * new feature: Single-stepping of code compiled with DEBUG 2 or
2227 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2228 * new feature: saving cores with foreign code loaded is now
2229 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2230 Timothy Moore's work for CMUCL)
2231 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2232 parameters to * instead of NIL if no initform is supplied.
2233 (reported by Johan Bockgaard)
2234 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2235 correctly when FUNCALL appears as the car of the form. Note:
2236 despite this FUNCALL forms are not currently subject to
2237 compiler-macro expansion. (port of Raymond Toy's fix for the
2238 same from CMUCL, reported by Johan Bockgaard)
2239 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2240 (thanks for Teemu Kalvas)
2241 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2242 case instead of terminating the iteration (thanks for Julian
2243 Squires, Sean Champ and Raymond Toy)
2244 * bug fix: incorrect expansion of defgeneric that caused a style
2245 warning. (thanks for Zach Beane)
2246 * bug fix: slot accessor effective method computation works properly
2247 for classes with multiple non-standard applicable methods on
2248 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2249 * on x86 compiler supports stack allocation of results of LIST and
2250 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2251 CMUCL implementation by Gerd Moellmann)
2252 * fixed some bugs revealed by Paul Dietz' test suite:
2253 ** FORMAT strings with both the ~<~:;~> form of the justification
2254 directive and pretty-printing directives cause an error.
2256 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2257 * incompatible change: the internal functions
2258 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2259 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2260 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2261 instead of the old functions.
2262 * new feature: on platforms where "dladdr" is available foreign
2263 function names now appear in backtraces. (based on Helmut Eller's
2265 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2266 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2268 * documentation: ASDF manual is now distributed with SBCL.
2269 * bug fix: SBCL can now load its contributed modules with REQUIRE
2270 even if the system-provided entries have been removed from
2271 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2273 * bug fix: backtraces involving undefined functions or assembly
2274 routines are more informative. (thanks to Brian Downing)
2275 * bug fix: mutually referent alien structures now work correctly.
2276 (reported by Rick Taube)
2277 * bug fix: structures defined by WITH-ALIEN can be referred to
2278 within other definitions in the same WITH-ALIEN.
2279 * bug fix: division operators (MOD, TRUNCATE and the like) with
2280 constant zero divisors and integer dividends no longer generate
2282 * bug fix: provide default methods for INPUT-STREAM-P and
2283 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2284 * bug fix: improve the detection and resolution of MOP metacycles
2285 (where effective method computation depends on the generic function
2286 having its effective method computed). (reported by Bruno Haible)
2287 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2288 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2289 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2290 * bug fix: pathnames with relative directory components can be
2291 represented relative to default pathnames.
2292 * optimization: in taking the GCD of bignums, reduce the two bignums
2293 to approximately the same size (using Euclid's algorithm) before
2294 applying the more sophisticated binary GCD. (thanks to Juho
2296 * optimization: COUNT on bitvectors now operates word-at-a-time.
2297 * optimization: ASH with a positive, but not necessarily constant,
2298 (leftwards) shift, when in a modular context, is compiled to a
2300 * fixed some bugs revealed by Paul Dietz' test suite:
2301 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2302 if the corresponding argument is NIL.
2303 ** FORMAT directives accepting a minpad parameter treat negative
2305 ** literal commas as character parameters to FORMAT directives are
2307 ** literal spaces directly after ~<Newline> directives within a
2308 format-logical-block (~:< ~@:>) do not induce :FILL-style
2309 conditional newlines.
2310 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2311 as specified: it no longer includes conditional newlines).
2312 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2314 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2315 :PER-LINE-PREFIX argument does not evaluate to a string.
2317 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2318 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2319 the manual for details; for now, package locks can be disabled by
2320 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2321 if no major problems are found then it is likely that they will be
2322 compiled in unconditionally.
2323 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2324 unsupported operators on all platforms. To load a shared library
2325 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2326 object file, link it into a shared library outside of SBCL and
2327 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2329 * bug workaround: ROOM T and the GC were not working together
2330 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2331 reporting facility aren't preserved. That reporting has been
2332 disabled, so now until and unless someone figures out how to make it
2333 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2334 * minor incompatible change: as threatened around sbcl-0.8.0, the
2335 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2336 The symbols are also exported from SB-PCL for backwards
2337 compatibility, but more so than before SB-PCL should be treated as
2338 an implementation-internal package.
2339 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2341 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2342 It is known as of this release not to work on the Alpha, however.
2343 * fixed bug #167: errors signalled due to illegal syntax in method
2344 bodies are now more legible.
2345 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2346 designators and can hence be used with TYPEP.
2347 * fixed bug #333: CHECK-TYPE now ensures that the type error
2348 signalled, if any, has the right object to be accessed by
2349 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2350 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2351 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2353 * fixed bug #334: programmatic addition of slots using specialized
2354 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2355 and :ALLOCATION :CLASS slots.
2356 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2357 float exponents. (rereported by Peter Seibel)
2358 * fixed a bug: #\Space (and other whitespace characters) are no
2359 longer considered to be macro characters in standard syntax by
2360 GET-MACRO-CHARACTER.
2361 * fixed bug: initialization of condition class metaobjects no longer
2362 causes an instance of the condition to be created. (reported by
2364 * fixed bug: it is now possible to have slots such that
2365 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2366 description is neither :INSTANCE nor :CLASS.
2367 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2368 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2369 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2370 non-standard SLOT-DEFINITION-ALLOCATION values.
2371 * fixed bugs: various race conditions handling exiting threads.
2372 CL-PPCRE's thread torture test now passes.
2373 * fixed arguably-a-bug: GC time proportional to number of threads in
2374 system even when most of them are idle
2375 * optimization: improved performance of BIT and SBIT on bit-vectors.
2376 * .fasl file incompatibility: The fasl file version number has
2377 been incremented because of changes associated with package locks.
2379 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2380 * minor incompatible change: the system no longer provides
2381 optimization hints (conditions which are TYPEP
2382 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2383 mode; these hints are still emitted when higher SPEED optimization
2385 * new contrib module: a sampling profiler (profiling by statistical
2386 sampling, rather than by instrumenting functions) is available as
2387 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2388 * the behaviour of the standard function ED is now customizeable by
2389 third parties through a hook variable: see ED's documentation
2390 string for information on the protocol.
2391 * the compiler no longer emits efficiency notes for (FUNCALL X)
2392 when the type of X is uncertain under default optimization
2394 * fixed bug 276: mutating a binding of a specialized parameter to a
2395 method to something that is not TYPEP the specializer is now
2397 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2398 exhibit the required behaviour on the x86 platform. (thanks to
2399 Peter van Eynde, Eric Marsden and Bruno Haible)
2400 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2401 even for difficult arguments. (reported by Peter Graves)
2402 * fixed bug 141a: the backquote printer now descends quoted
2404 * fixed another bug in backquote printing: no more destructive
2405 modification of the form's list structure. (reported by Brian
2407 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2408 move between its address being taken and the call to
2409 interrupt_thread, fixing a crashing race condition.
2410 * the SB-POSIX contrib implementation has been adjusted so that it
2411 no longer exhibits ridiculously poor performance when constructing
2412 instances corresponding to C structs.
2414 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2415 * minor incompatible change: the sb-grovel contrib now treats C
2416 structures as alien (in the sense of SB-ALIEN) objects rather than
2417 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2418 has implications for memory management of client code
2419 (sb-grovel-returned objects must now be manually managed) and for
2420 type safety (alien objects now have full types).
2421 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2422 used to control emission of compiler diagnostics, rather than the
2423 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2424 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2425 quality should be considered deprecated.
2426 * install.sh now installs the user manual as well
2427 * (not quite a new documentable feature, but worth considering in
2428 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2429 beginnings of a semantically meaningful condition hierarchy is
2430 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2432 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2433 designator as the defaults argument.
2434 * fixed bug: Displaced arrays whose displaced-to array has become
2435 too small now cause ARRAY-DIMENSION to signal an error, providing
2436 for safer bounds-checking. (reported by Bruno Haible)
2437 * fixed bug: DEFCLASS slot definitions with identical :READER and
2438 :WRITER names now signal a reasonable error. (reported by Thomas
2440 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2442 * fixed bug: Invalid slot specification errors now print correctly.
2443 (thanks to Zach Beane)
2444 * fixed bug 320: Shared to local slot value transfers in class
2445 redefinitions now happen corrently from superclasses as
2446 well. (reported by Bruno Haible)
2447 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2449 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2450 improper lists correctly. (reported by Bruno Haible)
2451 * fixed bug 313: source-transform for <fun-name> was erroneously
2452 applied to a call of a value of a variable with name <fun-name>.
2453 (reported by Antonio Menezes Leitao)
2454 * fixed bug 307: The obsolete instance protocol ensures that
2455 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2456 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2457 remove dead unknown-values globs from the middle of the stack.
2458 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2459 easier to resume long computations after using *BREAK-ON-SIGNALS*
2460 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2461 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2462 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2463 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2464 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2465 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2467 * fixed a (fairly theoretical) bug in string printing: if
2468 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2469 does not have array-element-type equal to the most general string
2471 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2472 function name. (thanks to Nikodemus Siivola)
2473 * fixed bug: compile-time format string checker failed on
2474 non-closed ~{. (reported by Thomas F Burdick)
2475 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2476 don't warn on legal loop code involving more than one aggregate
2477 boolean. (thanks to Nikodemus Siivola)
2478 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2479 in keyword position to MAKE-PACKAGE was not being reported as
2481 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2482 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2484 * fixed bugs 280 and 312: the checking for multiple definitions in a
2485 file is less likely to become confused by uses of inline
2487 * fixed bug: the #S reader macro performs the keyword coercion
2488 specified for slot names. (reported by Kalle Niemitalo)
2489 * fixed bug: lambda lists may contain symbols whose names start with
2490 & but are not lambda-list-keywords; their occurrence triggers a
2492 * fixed bug 321: define-method-combination argument lambda lists do
2493 not cause invalid code to be generated when &optional or &aux
2494 variables have default values. (reported by Bruno Haible)
2495 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2496 have CLOS classes; this appears also to have fixed reported
2497 instability in stack exhaustion detection.
2498 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2499 MUFFLE-WARNING when no associated restart is present is now
2501 * optimization: rearranged the expansion of various defining macros
2502 so that each expands into only one top-level form in a
2503 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2505 * optimization: used a previously little-used slot in symbols to
2506 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2508 * fixed some bugs revealed by Paul Dietz' test suite:
2509 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2511 ** Fixed an optimization bug related to inheritance of initforms
2512 from local to shared slots.
2513 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2514 ** CAST optimizer forgot to flush argument derived type.
2515 ** print/read consistency on floats is now orders of magnitude
2516 more likely. (thanks also to Bruno Haible for a similar report
2518 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2519 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2520 for objects of type REAL. Make it so.
2521 ** SXHASH is specified (infelicitously) to respect similarity,
2522 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2523 it so. (thanks to Markus Ziegler)
2524 ** on the Alpha, the compiler succeeds in compiling functions
2525 returning a known number of arguments greater than 63.
2526 ** fixed handling of invalid NIL arguments in keyword position
2528 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2530 ** provided a readably-printable representation for RANDOM-STATE
2532 ** ensured that pathnames, where they have a namestring, always
2533 print using #P"..." syntax.
2535 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2536 * Support for the forthcoming 2.0 version of the NetBSD kernel
2537 running on x86 hardware has been added. (thanks to Perry
2538 E. Metzger most immediately, and others for their past work)
2539 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2540 current release version; SBCL's OpenBSD support had been broken
2541 since about the time of OpenBSD's switch to ELF binary format.)
2542 * [placeholder for DX summary]
2543 ** user code with &REST lists declared dynamic-extent, under high
2544 speed or space and low safety and debug optimization policy.
2545 * The manual has been converted to Texinfo format and the debugger
2546 chapter from the cmucl manual has been added.
2547 * A facility has been added to extract documentation strings from
2548 sbcl and store them as Texinfo-formatted snippets for inclusion in
2549 the manual (via Texinfo's @include directive)
2550 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2551 the test case to Dave Roberts)
2552 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2553 pointers. (reported by Sean Ross)
2554 * bug fix: PROFILE output is printed nicely even for large numerical
2555 values. (thanks to Zach Beane)
2556 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2557 greater than 32 handle EOF correctly.
2558 * bug fix: on X86 an immediate argument of the IMUL instruction is
2559 correctly printed in disassembly. (thanks to Lutz Euler)
2560 * bug fix: class slots in redefined classes preserve their old
2561 values. (thanks to Bruno Haible and Nikodemus Siivola)
2562 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2563 succeeds. (reported by Marco Baringer)
2564 * fixed some bugs revealed by Paul Dietz' test suite:
2565 ** READ-SEQUENCE now works on ECHO-STREAMs.
2566 ** RATIONALIZE works more according to its specification. (thanks
2569 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2570 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2571 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2572 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2573 should matter to you only if you rebind the printer control
2574 variables and then find you want different bindings in the
2575 debugger than in the ordinary execution of your program.)
2576 * The runtime build system has been tweaked to support building
2577 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2578 assembler and linker. (thanks to Nikodemus Siivola)
2579 * Unbound, undefined, undeclared variables now trigger full
2580 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2581 more often programmer error than deliberate exploitation of undefined
2583 * optimization: the hash algorithm for strings has changed to one
2584 that is less vulnerable to spurious collisions. (thanks to Juho
2586 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2587 needless bounds checking. (thanks to Juho Snellman)
2588 * optimization: implemented multiplication as a modular
2589 (UNSIGNED-BYTE 32) operation on the PPC backend.
2590 * fixed some bugs revealed by Paul Dietz' test suite:
2591 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2593 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2594 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2595 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2597 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2599 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2601 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2602 their output stream on EOF from read.
2603 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2604 have been read to end-of-file.
2605 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2607 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2608 description of determination of which consecutive characters
2610 ** Printing the "Space" character with escaping on now yields "#\\ ",
2611 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2612 ** Reading floating-point numbers with *READ-BASE* set to a number
2613 less than 10 works correctly.
2614 ** Reading floating-point numbers with *READ-BASE* set to a number
2615 more than 10 works correctly.
2616 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2617 the readtable currently in effect.
2619 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2620 * minor incompatible change: parsing of namestrings on a physical
2621 (Unix) host has changed; numbers after the final #\. in a
2622 namestring are no longer interpreted as a version field. This is
2623 intented to be largely invisible to the user, except that the
2624 meaning of the namestring "*.*.*" has changed: it now refers to a
2625 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2626 should usually be replaced by
2627 (make-pathname :name :wild :type :wild :version :wild)
2628 with the added benefit that this is more likely to be portable.
2629 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2630 option to OPEN now signals an error if the file being opened
2631 exists; this may have an impact on existing code.
2632 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2633 signals received, on the PowerPC platforms (both Linux and
2634 Darwin). (thanks to Pierre Mai for pointing out the location of
2636 * several fixes on OS X: The system now builds and runs cleanly on
2637 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2638 (thanks to Brian Mastenbrook)
2639 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2640 second-resolution: integer multiples of 1/3600 between -24 and 24.
2641 (thanks to Vincent Arkesteijn)
2642 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2643 type when called with 1 argument; PEEK-CHAR checked type of
2644 PEEK-TYPE only after having read first character from a
2645 stream. (reported by Peter Graves)
2646 * bug fix: the garbage collector now has much better locality
2647 behaviour, and in particular no longer treats all memory as being
2648 exhausted when many small objects point to each other in a deeply
2650 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2651 recognized as being TYPEP their class.
2652 * bug fix: the PUSHNEW documentation string has been corrected.
2653 (thanks to Vincent Arkesteijn)
2654 * bug fix: defaulting of the value for the last of an atypically
2655 large number of multiple values being bound was not being
2656 performed correctly on the Alpha or PPC platforms
2657 * optimization: implemented multiplication as a modular
2658 (UNSIGNED-BYTE 32) operation on the x86 backend.
2659 * optimization: SEARCH on simple-base-strings can now be open-coded.
2660 (see also contrib/compiler-extras.lisp for inspiration for
2661 teaching the compiler about the Boyer-Moore algorithm).
2662 * value, returned by MAX (and MIN) called with several EQUALP, but
2663 not EQL, arguments now does not depend on compiler settings.
2664 * fixed some bugs revealed by Paul Dietz' test suite:
2665 ** in stack analysis liveness information is propagated from
2666 non-local entry points.
2667 ** pathwise CAST removing failed when the CAST node did not start
2669 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2670 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2672 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2673 host is already defined.
2674 ** RENAME-FILE works on streams instead of signalling an internal
2676 ** PEEK-CHAR uses the current readtable when determining whether
2677 or not a character is whitespace.
2678 ** MERGE-PATHNAMES handles the case when the pathname does not
2679 specify a name while the default-pathname specifies a version.
2680 ** Pathnames now stand a better chance of respecting print/read
2682 ** Attempting to use standardized file system operators with a
2683 pathname with invalid :DIRECTORY components signals a
2685 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2687 ** DIRECTORY on logical pathnames is more correct.
2688 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2689 signal a TYPE-ERROR if their argument is not a stream.
2690 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2691 designator argument does not designate a stream.
2692 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2693 examining the synonym.
2694 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2696 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2697 element-type larger than ([UN]SIGNED-BYTE 32).
2699 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2700 * When built with the :SB-FUTEX feature, threaded builds now take
2701 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2702 for faster/more reliable mutex and condition variable support.
2703 * Incompatible change (but one you probably shouldn't have been using
2704 anyway): the interface and code for arbitrating between multiple
2705 threads in the same user session has been redesigned.
2706 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2707 error when called without an explicit environment argument.
2708 (thanks to Nikodemus Siivola)
2709 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2710 confused about their position. (thanks to Adam Warner and Gerd
2712 * bug fix: Pretty printing backquoted forms with unquotations in the
2713 argument list position of various code constructs such as LAMBDA
2714 now works correctly. (reported by Paul Dietz)
2715 * bug fix: Pretty printing unquotations no longer loses all
2716 stream position information.
2717 * optimization: performance of string output streams is now less
2718 poor for multiple small sequence writes.
2719 * optimization: performance of CSUBTYPEP in the presence of complex
2720 expressions involving CONS and NOT many times has been improved.
2721 (reported by Paul Dietz)
2722 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2724 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2725 streams for the REPL, for the debugger, and so on. (thanks to
2727 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2728 * fixed some bugs revealed by Paul Dietz' test suite:
2729 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2730 is ignored if it is too small, rather than propagating through
2731 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2732 ** extremely complex negations of CONS types were not being
2733 sufficiently canonicalized, leading to inconsistencies in
2735 ** VALUES tranformer lost derived type.
2737 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2738 * fixed a bootstrapping bug: the build process no longer assumes
2739 that the various BOOLE-related constants have the same value in
2740 host and target lisps. (noted by Paul Dietz' test suite on an
2741 SBCL binary built from CLISP)
2742 * The system can now be dynamically linked on the MIPS platform,
2743 which enables dynamic loading of foreign code from Lisp. (thanks
2744 to Ralf Baechle for discussions on the MIPS ABI)
2745 * The system now records debugging information for its own source
2746 files in a filesystem-position-independent manner, relative to
2747 the "SYS" logical host.
2748 * fixed a compiler bug: MV-LET convertion did not check references
2749 to the "max args" entry point. (reported by Brian Downing)
2750 * tweaked disassembly notes to be less confident about proclaiming
2751 some instruction as an LRA. (thanks to Brian Downing)
2752 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2753 now each have their own history, command character, and other
2754 characteristics. (thanks to David Lichteblau)
2755 * fixed some bugs revealed by Paul Dietz' test suite:
2756 ** compiler failure in compiling LOGAND expressions including a
2758 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2759 shift greater than 32.
2760 ** FLUSH-DEST did not mark blocks for type check regeneration.
2761 ** HANDLER-CASE failed to accept declarations in handler clauses
2762 in some circumstances.
2764 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2765 * New code in contrib/sb-introspect (still probably not entirely
2766 stable yet) provides some support for smart Lisp development
2767 environments like SLIME.
2768 * The conditions signalled for errors occurring when loading .fasl
2769 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2770 in a way which should help ASDF recover gracefully.
2771 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2772 stuff has been cleaned up. If you code contrib/ stuff, this might
2773 affect you, and you can look at contrib/README, contrib/STANDARDS,
2774 and/or the 0.8.4.27 diff to check.
2775 * In full calls the compiler now does not generate checks for declared
2776 argument types for all arguments.
2777 * various threading fixes
2778 ** and some experimental patches which didn't make it into
2779 the main tree for this release, but which are shipped in
2780 contrib/experimental-thread.patch as a possible fix for some
2781 failures (deadlock, spinning...) in GC-intensive multithreaded
2783 * fixed PPC build problem (source code incompatibility of different
2784 library versions): added offsetof() hackery which attempts to divine
2785 where glibc maintainers put uc_mcontext today
2786 * fixed bug 282: compiler does not trust type assertions while passing
2787 arguments to a full call.
2788 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2789 &OPTIONAL VALUES type parameter.
2790 * fix bug 214: algorithm for noting rejected templates is now more
2791 similar to that of template seletion. (also reported by rydis on
2793 * fixed bug 141b: printing backquoted information readably and prettily
2794 inserts a space where necessary.
2795 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2796 (CHARACTER 10) are now reported as errors, rather than propagated
2797 as unknown types. (reported by piso on #lisp)
2798 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2799 with values NIL and :ERROR. (thanks to Milan Zamazal)
2800 * fixed bug 191c: CLOS now does proper keyword argument checking as
2801 described in CLHS 7.6.5 and 7.6.5.1.
2802 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2803 longer raise an error; further, using a list as a for-as-arithmetic
2804 counter now raises a meaningful error.
2805 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2806 now understood by sequence creation functions such as MAKE-SEQUENCE
2808 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2809 a STRING-INPUT-STREAM.
2810 * compiler enhancement: SIGNUM is now better able to derive the type
2812 * type declarations inside WITH-SLOTS are checked. (reported by
2814 * fixed some bugs revealed by Paul Dietz' test suite:
2815 ** incorrect optimization of TRUNCATE for a positive first
2816 argument and negative second.
2817 ** compiler failure in let-convertion during flushing dead code.
2818 ** compiler failure while deriving type of TRUNCATE on an
2819 interval, containing 0.
2820 ** ASH of a negative bignum by a negative bignum count now returns
2822 ** intersection of CONS types now canonicalizes properly, fixing
2823 inconsistencies in SUBTYPEP.
2825 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2826 * incompatible change: The --disable-debugger command line
2827 option now clobbers the debugger at a more fundamental
2828 level, by redefining #'INVOKE-DEBUGGER instead of by
2829 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2830 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2831 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2832 enter the debugger REPL and then suffer recursive errors
2833 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2834 while under the new system, BREAK in --disable-debugger mode
2835 terminates the system just as an unhandled error would.
2836 * fixed compiler performance when processing loops with a step >1;
2837 * bug fix: DOCUMENTATION now retrieves generic function
2838 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2839 support has been systematized, and now supports the methods
2840 specified by ANSI, along with a default method and a method for
2841 slot documentation. (reported by Nathan Froyd)
2842 * bug fix: effective methods associated with a generic function are
2843 no longer cached over a change of that generic function's method
2844 combination. (reported by Andreas Fuchs)
2845 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2846 names a non-existent directory. (thanks to Andreas Fuchs)
2847 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2848 the PPC platform, lead to stack corruption; this has been fixed.
2849 (reported by Rainer Joswig)
2850 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2851 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2852 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2853 arithmetic in general, is now much more reliable on the PPC
2855 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2856 platform now returns the right answer.
2857 * optimization: restored some effective method precomputation in
2858 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2859 precomputation is now tunable.
2860 * optimization: compiler-internal data structure use has been
2861 reviewed, and changes have been made that should improve the
2862 performance of the compiler by about 20%.
2863 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2864 simple-base-strings and simple-bit-vectors is improved.
2865 * optimization: the optimization of 32-bit logical and arithmetic
2866 functions introduced in version 0.8.3 on the x86 has been
2867 implemented on the mips, ppc and sparc platforms; an
2868 implementation of the same facility, but for 64-bit arithmetic,
2869 has been added for the alpha.
2870 * microoptimization: the compiler is better able to make use of the
2871 x86 LEA instruction for multiplication by constants.
2872 * bug fix: in some situations compiler did not report usage of
2873 generic arithmetic in (SPEED 3) policy.
2874 * bug 145b fix: compiler used wrong type specifier while converting
2875 MEMBER-types to numeric.
2876 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2878 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2879 index argument now works properly on the Alpha platform.
2880 * bug fix: floating point exception treatment on the Alpha platform
2882 * bug fix: FILE-POSITION works much better on string input and
2883 output streams. (thanks to Nikodemus Siivola)
2884 * bug fix: many threading/garbage collection symptoms sorted.
2885 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2886 might be pseudo-atomic.
2887 * internal change: Stopping for GC is now done with signals not
2888 ptrace. GC is now done in whichever thread wanted it, instead of
2890 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2892 * bug fix: result form in DO is not contained in the implicit
2894 * incompatible change: ICR structure is changed; the value part of
2895 CONTINUATION is now called LVAR; corresponding functions are
2896 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2897 * added type deriver for ISQRT (thanks to Robert E. Brown).
2898 * bug fix: better support for loading from the command line when an
2899 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2901 * fixed some bugs revealed by Paul Dietz' test suite:
2902 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2903 ** ROUND and FROUND now give the right answer when given very
2904 small float arguments.
2905 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2907 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2908 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2909 ** LCM should return a non-negative integer.
2910 ** PARSE-INTEGER returned the index of a terminator instead of the
2911 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2912 ** PARSE-INTEGER returned an incorrect index being applied to a
2914 ** LCM with two arguments of 0 returns 0 rather than signalling
2916 ** unsigned addition of a 32-bit constant with the high bit set no
2917 longer causes an internal compiler error on the x86.
2918 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2919 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2920 with negative last argument.
2921 ** byte specifiers with zero size and position no longer cause
2922 an error during type derivation.
2923 ** bignum multiplication on the Alpha platform now returns the
2925 * porting: The system now builds on SuSE AMD64, although it still
2926 generates a 32-bit binary.
2927 * .fasl file incompatibility: The fasl file version number has
2928 been incremented (because of the changes to internal compiler
2929 data structures referred to above).
2931 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2932 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2933 more accurately, on the Darwin kernel running on PowerPC hardware.
2934 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2935 * Compiler code deletion notes now signal a condition of type
2936 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2937 an associated MUFFLE-WARNING restart.
2938 * The compiler now performs limited argument count validation of
2939 constant format strings in FORMAT, and where appropriate in ERROR,
2940 CERROR and WARN. (thanks to Gerd Moellmann)
2941 * New ASDF-INSTALL contrib can be used for automatic download and
2942 installation of third-party Lisp code from CCLAN or other sites
2944 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2945 which forces another thread to execute a function supplied by the
2947 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2948 now accept and act upon their :ELEMENT-TYPE keyword argument.
2949 (reported by Martin Atzmueller, Edi Weitz)
2950 * bug fix: FILE-POSITION now accepts position designators up to
2951 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2952 is the greater. (thanks to Patrik Nordebo)
2953 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2954 Kalle Olavi Niemitalo)
2955 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2956 printable. (reported by Eric Marsden)
2957 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2958 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2959 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2960 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2961 (thanks to Stig Sandoe)
2962 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2963 UNREAD-CHAR now correctly includes the unread character in the
2964 target sequence. (thanks to Gerd Moellmann)
2965 * bug fix in threaded builds: the system can now be suspended and
2966 resumed by shell job control with minimal disruption.
2967 * bug fixes in times and timezones >2038AD
2968 * better handling of "where is GNU make?" problem in build scripts
2969 (thanks to Nikodemus Siivola)
2970 * new optimization: inside a named function any reference to a
2971 function with the same name is considered to be a self-reference;
2972 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2973 optimization quality.
2974 * new optimization on x86: logical functions and + now have
2975 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2976 used when the result is truncated to 32 bits.
2977 * VALUES declaration is partially enabled.
2978 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2979 * fixed some bugs revealed by Paul Dietz' test suite:
2980 ** The system now obeys the constraint imposed by
2981 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2982 types form a lattice under type intersection.
2983 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2984 ** ASSOC now ignores NIL elements in an alist.
2985 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2986 and (1+ MOST-POSITIVE-FIXNUM) answers.
2987 ** The addition of a method with invalid qualifiers to a generic
2988 function does not cause an error to be signalled immediately;
2989 a warning is signalled, and the error is generated only on
2990 calling the generic function.
2991 * changed .fasl file version number, in part to add type codes for
2992 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2993 obscure ANSI requirements
2995 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2996 * fixed bug 148: failure to inline-expand a local function left
2997 garbage, confusing the compiler.
2998 * fixed bugs 3cd: structure slot readers perform type check if the
2999 slot can have an invalid value (i.e. it is either not initialized
3000 or can be written with a less specific slot writer).
3001 * bug fix: the compiler now traps array references to elements off
3002 the end of an array; previously, the bounds checking in some
3003 circumstances could go off-by-one.
3004 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3006 * type declarations for array element types now obey the description
3007 on the CLHS page "Declaration TYPE", as per discussions on
3008 sbcl-help around 2003-05-08. This means that a declaration
3009 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3010 declaration, all references to BAR will be asserted or assumed
3011 (with THE, so dependent on compiler policy) to involve objects of
3012 type FOO. Note that no such declaration is implied in
3013 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3014 * declared types of functions from the "Conditions"
3015 chapter. (reported by Paul Dietz)
3016 * bug fix: CERROR accepts a function as its first argument.
3017 * bug fix: NTH an NTHCDR accept a bignum as index
3018 arguments. (reported by Adam Warner)
3019 * optimization: character compare routines now optimize comparing
3020 against a constant character. (reported by Gilbert Baumann)
3021 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3022 argument now works properly on the MIPS platform.
3023 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3024 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3025 declared to have a functional type, does not cause an error
3027 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3028 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3029 * fixed bug: if last continuation of a deleted block has a
3030 destination, this destination should be deleted too. (reported by
3032 * fixed a bug in the bootstrap process: the host compiler's values
3033 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3034 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3035 test case from Patrik Nordebo)
3036 * improved the ability of the disassembler on the PPC platform to
3037 provide helpful disassembly notes.
3038 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3039 the class in more cases than previously.
3040 * bug fix: FILE-POSITION now understands :START and :END for
3041 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3042 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3043 8) data. (thanks to David Lichteblau)
3044 * bug fix: it is possible to add a method to a generic function
3045 without lambda list.
3046 * bug fix: reader failed to signal END-OF-FILE inside an
3047 object representation. (reported by Nikodemus Siivola)
3048 * fixed some bugs revealed by Paul Dietz' test suite:
3049 ** LAST and [N]BUTLAST should accept a bignum.
3050 ** condition slot accessors are methods.
3051 ** (VECTOR NIL) is a subtype of STRING.
3053 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3054 * minor incompatible change: some nonsensical specialized lambda
3055 lists (used in DEFMETHOD) which were previously ignored now signal
3057 * minor incompatible change: the system is now aware of the types of
3058 variables in the COMMON-LISP package, and will signal errors for
3059 most violations of these type constraints (where previously they
3060 were silently accepted).
3061 * minor incompatible change: COMPILE-FILE now uses the freedom
3062 afforded (ANSI 3.2.2.3) to use derived function types for
3063 functions defined in the same file. This also permits the system
3064 to warn on static type mismatches and function
3065 redefinition. (Currently it does not work with high DEBUG level.)
3066 * minor incompatible change: VALUES declaration is disabled.
3067 * When issuing notes, the compiler now signals a condition of type
3068 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3069 restart for use in user handlers. It is expected that the
3070 COMPILER-NOTE condition will eventually become a condition
3071 supertype to a hierarchy of note types, which will then be
3072 handleable in a similar fashion. However, at the moment, no such
3073 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3074 but it's an implementation detail, not a classification for the
3076 * Changes in type checking closed the following bugs:
3077 ** type checking of unused values (192b, 194d, 203);
3078 ** template selection based on unsafe type assertions (192c, 236);
3079 ** type checking in branches (194bc).
3080 * A short form of VALUES type specifier has ANSI meaning (it has
3081 increased the number of situations when SBCL cannot perform type
3083 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3084 keywords or constants is permissible.
3085 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3086 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3087 classes. (thanks to Antonio Martinez)
3088 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3089 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3090 argument to a pretty-print stream. Instead, it leaves any such
3091 operations to DESCRIBE-OBJECT methods.
3092 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3094 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3095 lambda lists. (thanks to David Lichteblau)
3096 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3097 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3098 (thanks to Gerd Moellmann)
3099 * fixed bug 246: increased compilation speed of long
3100 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3102 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3104 * DEFCONSTANT now throws a condition of type
3105 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3106 constant to a non-EQL value; CONTINUE and ABORT restarts
3107 respectively change and preserve the value.
3108 * fixed bug 63: The code walker, part of the implementation of CLOS,
3109 is now better at handling symbol macros.
3110 * bug fix: There is no longer an internal implementation type named
3111 CL:LENGTH. (reported by Raymond Toy)
3112 * bug fix: In macro-like defining macros/special operators the
3113 implicit block does not enclose the lambda list.
3114 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3115 enclosed in parentheses) are not suitable as type specifiers, and
3116 their use properly signals an error now.
3117 * bug fix: An argument count mismatch for a type specifier in code
3118 being compiled no longer causes an unhandled error at compile
3119 time, but signals a compile-time warning.
3120 * fixed simple vector readable printing
3121 * bug fix: DESCRIBE takes more care over whether the class
3122 precedence list slot of a class is bound before accessing it.
3123 (reported by Markus Krummenacker)
3124 * bug fix: FORMATTER can successfully compile pretty-printer format
3125 strings which use variants of the ~* directive inside.
3126 * bug fix: SEARCH now applies its TEST predicate to the elements of
3127 the arguments in the correct order. (thanks to Wolfhard Buss)
3128 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3129 in type inference. (thanks to Robert E. Brown)
3130 * bug fix: Reading in symbols with an explicit package name of ""
3131 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3132 * fixed some bugs revealed by Paul Dietz' test suite:
3133 ** NIL is now allowed as a structure slot name.
3134 ** Arbitrary numbers, not just REALs, are allowed in certain
3135 circumstances in LOOP for-as-arithmetic clauses.
3136 ** Multiple class redefinitions before slot access no longer
3137 causes a type error.
3138 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3139 association between the name and a class.
3140 ** Generic functions with non-standard method-combination and over
3141 five methods all of which return constants no longer return NIL
3142 after the first few invocations. (thanks to Gerd Moellmann)
3143 ** CALL-NEXT-METHOD with no arguments now passes the original
3144 values of the arguments, even in the presence of assignment.
3145 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3147 ** DISASSEMBLE works with closures and funcallable instances.
3148 ** ADD-METHOD now returns the generic function, not the new method.
3149 ** FIND-METHOD signals an error if the lengths of the specializers
3150 is incompatible with the generic function, even if the ERRORP
3152 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3153 which its argument is a member.
3154 ** DEFCLASS only redefines the class named by its class-name
3155 argument if that name is the proper name of the class;
3156 otherwise, it creates a new class.
3157 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3158 of the UNBOUND-SLOT condition to the name of the slot.
3159 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3160 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3161 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3162 SLOT-MAKUNBOUND in the specified fashion.
3164 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3165 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3166 cross-compilation host. As a consequence, we can now bootstrap our
3167 way up to SBCL starting with a bare gcc toolchain and human-readable
3168 source code (first the source to CLISP, then the source to SBCL).
3169 * A contributed module containing a partial implementation of the
3170 simple-streams interface has been included. (thanks to Rudi
3172 * A contributed module implementing the RFC1321 Message Digest
3173 Algorithm, known as MD5, has been included.
3174 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3175 no longer has any effect, as the code controlled by this feature
3176 has been deleted. (As far as we know, no-one has ever built using
3177 this feature, and its semantics were confused in any case).
3178 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3179 work on conditions (as required by the ANSI specification),
3180 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3181 expected behaviour on conditions. Users should note, however,
3182 that such behaviour is not required by the ANSI specification,
3183 and so use of this behaviour may render their code unportable.
3184 * fixed some bugs revealed by Paul Dietz' test suite:
3185 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3187 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3188 ** SXHASH on condition objects no longer returns NIL.
3189 ** :ALLOCATION :CLASS slots are better treated; their values are
3190 updated on class redefinition, and initforms inherited from
3191 superclasses are applied.
3192 ** REMOVE-METHOD returns its generic function argument even when
3193 no method was removed.
3194 ** SHARED-INITIALIZE now initializes the values of the requested
3195 slots, including those with :ALLOCATION :CLASS.
3196 ** ALLOCATE-INSTANCE now works on structure classes defined via
3197 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3199 ** SLOT-EXISTS-P now works on conditions, as well as structures
3201 ** MAKE-LOAD-FORM now has the required methods on
3202 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3203 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3204 keyword, and now implements the SLOT-NAMES argument.
3205 ** methods with &OPTIONAL arguments no longer allow too many
3206 arguments to be passed in the call without error.
3207 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3208 option is consistent with the required arguments of the generic
3209 function lambda list.
3210 * bug fix: REQUIRE accepts a string designator. (Thanks to
3212 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3213 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3214 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3215 * bug fix: The compiler now checks for duplicated variables in macro
3217 * bug fix: SETQ on globals returns the correct value.
3218 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3219 not a COMPILER-ERROR (followed by some other strange error on
3220 choosing the CONTINUE restart).
3221 * bug fix: make.sh and friends are now more consistent in the way that
3222 they look for GNU "make".
3224 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3225 * experimental native threads support (on x86 Linux >=2.4 only).
3226 This is not compiled in by default: you need to add :SB-THREAD to
3227 the target features. See the "Beyond ANSI" chapter of the manual
3229 * fix for longstanding nonANSIism: The old distinction between
3230 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3231 The return value from CL:FIND-CLASS is now a CLOS class, and
3232 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3233 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3234 * An interface to the MetaObject Protocol, as described in Kiczales,
3235 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3236 MIT Press, 1991, is available from the SB-MOP package.
3237 * incompatible change: the SB-PCL package should now be considered
3238 a private implementation detail, and no longer a semi-private MOP
3240 * minor incompatible change: due to rearrangement for threads, the
3241 control stack and binding stack are now allocated at arbitrary
3242 addresses instead of being hardcoded per-port. Users affected by
3243 this probably have to be doing advanced things with shared
3244 libraries, and will know who they are.
3245 * minor incompatible change: Previously, all --eval forms used were
3246 processed with READ before any of them were processed with EVAL.
3247 Now each --eval form is processed with both READ and EVAL before
3248 the next --eval form is processed. (Thus package operations like
3249 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3250 work as the user might reasonably expect.)
3251 * minor incompatible change: *STANDARD-INPUT* is now only an
3252 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3254 * minor incompatible change: Y-OR-N-P is now character-oriented,
3255 not line oriented. Also, YES-OR-NO-P now works without errors.
3256 (thanks to Antonio Martinez)
3257 * sb-aclrepl module improvements: an integrated inspector, added
3258 repl features, and a bug fix to :trace command.
3259 * Known functions, which cannot be open coded by the backend, are
3260 considered to be able to check types of their arguments. (fixing
3261 a bug report by Nathan J. Froyd)
3262 * fixed a bug in computing method discriminating functions: It is
3263 now possible to define methods specialized on classes which have
3264 forward-referenced superclasses. (thanks to Gerd Moellmann)
3265 * fixed evaluation order in optional entries (reported by Gilbert
3267 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3268 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3269 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3270 required-by-AMOP NAME argument, as well as CLASS and
3271 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3272 * fixed bug 20: DEFMETHOD can define methods using names that are
3273 not the proper names of classes to designate class specializers.
3274 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3275 Unix file descriptors, instead of blowing up. (thanks to Antonio
3277 * Garbage collection refactoring: user-visible change is that a
3278 call to the GC function during WITHOUT-GCING will not do garbage
3279 collection until the end of the WITHOUT-GCING. If you were doing
3280 this you were probably losing anyway.
3281 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3282 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3283 TYPEP the latter but not the former.
3284 * The compiler issues a full WARNING for calls to undefined functions
3285 with names from the CL package.
3286 * MAP-INTO for a vector destination is open coded. (reported by
3287 Brian Downing on c.l.l)
3288 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3289 documentation string.
3290 * fixed some bugs revealed by Paul Dietz' test suite:
3291 ** COPY-ALIST now signals an error if its argument is a dotted
3293 ** Condition slots are now accessed more correctly in the presence
3294 of multiple initargs for a given slot.
3295 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3296 exclude restarts of the same name associated with a different
3298 ** DEFCLASS of forward-referenced classes with another
3299 forward-referenced class in the superclasses list no longer
3301 ** Condition slots are now initialized once each, not multiple
3302 times. (thanks to Gerd Moellmann)
3303 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3305 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3306 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3307 arguments contain duplicated elements.
3308 ** RESTART-CASE understands local macros.
3309 ** RESTART-CASE associates exactly its own restarts with a condition.
3310 ** ENDP in safe mode checks its argument to be of type LIST.
3311 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3312 in question is unbound.
3313 ** Optimized MAKE-INSTANCE functions no longer cause internal
3314 assertion failures in the presence of duplicate initargs.
3315 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3316 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3317 propagated even in the ctor optimized implementation of
3319 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3321 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3322 * a better implementation of SXHASH on (simple) bit vectors,
3323 measured both in execution speed and in distribution of results
3324 over the positive fixnums, has been installed. Likewise, a better
3325 implementation of EQUAL for simple bit vectors is now available.
3326 * fixed CEILING optimization for a divisor of form 2^k.
3327 * fixed bug 240 (emitting extra style warnings "using the lexical
3328 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3329 by Antonio Martinez)
3330 * fixed SXHASH, giving different results for NIL depending on type
3331 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3332 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3333 variable. (found by Rolf Wester)
3334 * SBCL does not ignore type declarations for special
3335 variables. (reported by rif on c.l.l 2003-03-05)
3336 * some bug fixes in contrib/sb-aclrepl/
3337 * fixed some bugs revealed by Paul Dietz' test suite:
3338 ** a bug in the CONS type specifier, whereby the CAR and CDR
3339 types got intertwined, has been fixed;
3340 ** the type system is now able to reason about the interaction
3341 between INTEGER and RATIO types more completely;
3342 ** APPEND, [N]REVERSE and NRECONC check that those their
3343 arguments, which must be proper lists, are really so;
3344 ** An array specialized to be unable to hold elements has been
3345 implemented, as required -- yes, really -- by ANSI;
3346 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3347 on malformed property lists;
3349 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3350 * incompatible packaging change: in line with Unix convention,
3351 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3352 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3353 * REQUIRE and PROVIDE are now optionally capable of doing something
3354 useful. See the documentation string for REQUIRE.
3355 * infrastructure for a managed SBCL contrib system: contributed
3356 modules in this release include:
3357 ** the ASDF system definition facility;
3358 ** an interface to the BSD Sockets API;
3359 ** an ACL-like convenience interface to the repl;
3360 (thanks to Kevin Rosenberg)
3361 ** an implementation of ROTATE-BYTE, with efficient implementation
3363 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3364 gives a closer approximation to the right answer than previously.
3365 (thanks to Raymond Toy)
3366 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3367 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3368 optional environment arguments, as required by ANSI.
3369 * fixed bugs in other functions taking environment objects, allowing
3370 calls with an explicit NIL environment argument to be compiled
3372 * fixed bug 228: primary return values from
3373 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3374 COMPILE or FUNCTION.
3375 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3376 :TYPE will no longer signal errors on innocuous objects.
3377 * fixed bug 231b: SETQ is better at respecting type declarations in
3378 the lexical environment.
3379 * fixed a bug in DEFCLASS: classes named by symbols with no or
3380 unprintable packages can now be defined.
3381 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3382 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3383 * fixed compiler failure related to checking types of functions.
3384 (reported by Robert E. Brown)
3385 * the compiler is now much more consistent in its error-checking
3386 treatment of bounding index arguments to sequence functions: in
3387 (SAFETY 3) code, errors will be signalled in almost all cases if
3388 invalid sequence bounding indices are passed to functions defined
3389 by ANSI to operate on sequences.
3390 * fixed a bug in the build procedure: documentation of SBCL-specific
3391 packages is now preserved and available in the final Lisp image.
3392 * lifted FDEFINITION lookup out of loops in the implementation of
3393 many list operations. (thanks to Robert E. Brown)
3394 * fixed a bug in the reader: the #n# reader macro now works for
3395 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3396 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3397 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3398 E. Brown and Tony Martinez respectively)
3399 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3400 SLOT-BOUNDP: in particular, it is now able to optimize them much
3401 better, and is now not vulnerable to having packages renamed.
3402 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3403 always called when a slot is not present in an instance. (thanks
3405 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3406 longer signal an error on unreading a character following EOF on
3407 the previous constituent stream. (thanks to Tony Martinez)
3408 * fixed some bugs revealed by Paul Dietz' test suite:
3409 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3410 not just nonnegative fixnums;
3411 ** the logical bit-array operators such as BIT-AND now accept an
3412 explicit NIL for their "opt-arg" argument (to indicate a
3413 freshly-consed result bit-array);
3414 ** ELT now signals an error on an invalid sequence index in safe
3416 ** the type system is now cleverer about negations of numeric
3417 types, and consequently understands the BIGNUM and RATIO types
3419 ** the type system is now cleverer about the interaction between
3420 INTEGER and RATIO types: while bugs still remain, many more
3421 cases are accurately computed;
3422 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3423 if it is in the last clause;
3424 ** CONSTANTLY now correctly returns a side-effect-free function in
3426 ** DECLARE is no longer treated as a special-operator; in
3427 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3428 * incremented fasl file version number due to the change in the
3429 DEFSTRUCT-SLOT-DESCRIPTION structure.
3431 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3432 * minor incompatible change: code processed by the "interpreter" or
3433 EVAL now has a compilation optimization policy of (DEBUG 2)
3434 (changed from (DEBUG 1)) to improve debuggability of interactive
3435 development, and to allow the use of the debug RETURN command in
3437 * an experimental implementation of the RETURN command for the
3438 debugger has been included. (thanks to Frederik Kuivinen)
3439 * fixed bug 62: constraints were not propagated into a loop.
3440 * fixed bug in embedded calls of SORT (reported and investigated by
3442 * fixed some bugs revealed by Paul Dietz' test suite:
3443 ** printing and reading of arrays with some dimensions having
3444 length 0 (thanks to Gerd Moellmann);
3445 ** BOA constructor with &AUX argument without a default value does
3446 not cause a type error;
3447 ** CONSTANTP now returns true for all self-evaluating objects.
3449 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3450 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3451 accessors that are related by inheritance, as specified in the
3452 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3453 Valtteri Vuorikoski)
3454 * The compiler is now able to inline functions that were defined in
3455 a complex lexical environment (e.g. inside a MACROLET).
3456 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3458 * Support for the upcoming FreeBSD-5.0 release has been included.
3459 (thanks to Dag-Erling Smorgrav)
3460 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3461 effect when it is not in a toplevel context.
3462 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3463 stand a better chance of being correct. (thanks to Gerd
3465 * fixed bug in COERCE, which now signals an error on coercing a
3466 rational to a bounded real type which excludes the expected
3468 * The compiler is now able to derive types more accurately from the
3469 COERCE and COMPILE functions.
3470 * fixed bug 223: functional binding is considered to be constant
3471 only for symbols in the CL package.
3472 * fixed bug 231: SETQ did not check the type of a variable being set
3473 (reported by Robert E. Brown)
3474 * A new optimization for MAKE-INSTANCE has been included, fixing
3475 various bugs (including relating to :ALLOCATION :CLASS slots and
3476 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3478 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3479 ** As required by ANSI, LOOP now disallows anonymous collection
3480 clauses such as COLLECT I in conjunction with aggregate boolean
3481 clauses such as THEREIS (= I 1);
3482 ** LOOP now signals an error when any variable is reused in the
3483 same loop (including the potentially useful construct analogous
3484 to WITH A = 1 WITH A = (1+ A);
3485 ** IT is only a special loop symbol within the first clause of a
3486 conditional loop clause;
3487 ** LOOP with a typed iteration variable over a hashtable now
3488 signals a type error iff it should.
3489 * fixed some other bugs revealed by Paul Dietz' test suite:
3490 ** FILE-STREAM now names the class previously known as FD-STREAM;
3491 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3492 argument) no longer signals an error;
3493 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3494 of the slot symbol, rather than using the current package
3495 ((:CONC-NAME "") continues to intern the slot's name in the
3497 * incremented fasl file version number, because of the incompatible
3498 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3499 of the new implementation of DEFINE-COMPILER-MACRO.
3501 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3502 * Support for building SBCL for MIPS platforms running in
3503 little-endian mode has now been checked in, and basic
3504 functionality on said platforms verified.
3505 * minor incompatible change: PCL now records the pathname of a file
3506 in which methods and the like are defined, rather than its
3508 * minor incompatible change: TRUENAME now considers the truename of
3509 a file naming a directory to be the pathname with :DIRECTORY
3510 component indicating that directory.
3511 * minor incompatible change: a NAMED clause in the extended form of
3512 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3513 reason for the previous behaviour is unclear.
3514 * more systematization and improvement of CLOS and MOP conformance
3515 in PCL (thanks to Gerd Moellman and Pierre Mai):
3516 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3518 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3519 primary methods with no specializers;
3520 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3522 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3523 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3524 FORWARD-REFERENCED-CLASSes; error reporting on
3525 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3527 ** SXHASH on CLOS instances now uses a slot internal to the
3528 instance to return different numbers on distinct instances,
3529 while preserving the same return value through invocations of
3531 ** DEFMETHOD signals errors when methods with longer incongruent
3532 lambda lists are added to generic functions;
3533 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3534 CLASS, as specified in AMOP;
3535 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3536 on the order returned by the primary method for classes of
3537 class STANDARD-CLASS;
3538 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3539 * fixed some bugs shown by Paul Dietz' test suite:
3540 ** DOLIST puts its body in TAGBODY;
3541 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3543 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3544 value producing form;
3545 ** if more variables are given to PROGV than values, extra
3546 variables are bound and made to have no value;
3547 ** NSUBSTITUTE on list arguments gets the right answer with
3549 ** ELT signals an error of type TYPE-ERROR when the index argument
3550 is not a valid sequence index;
3551 ** LOOP signals (at macroexpansion time) an error of type
3552 PROGRAM-ERROR when duplicate variable names are found;
3553 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3554 ** FUNCALL of special-operators now cause an error of type
3556 ** PSETQ now works as required in the presence of side-effecting
3557 symbol-macro places;
3558 ** NCONC accepts any object as its last argument;
3559 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3561 ** loop-for-as-package does not require a package to be explicitely
3563 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3565 * fixed bug 166: compiler preserves "there is a way to go"
3566 invariant when deleting code.
3567 * fixed bug 172: macro lambda lists with required arguments after
3568 &REST arguments now cause an error to be signalled. (thanks to
3570 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3571 bug. (thanks to Matthew Danish)
3572 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3574 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3575 arguments are lexically rebound. (thanks to Gerd Moellmann and
3577 * fixed bug 194: error messages are now more informative when there
3578 is no primary method applicable in a call to a generic
3579 function. (thanks to Gerd Moellmann)
3580 * fixed bug in command line argument checking (thanks to Julian
3582 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3583 (thanks to Matthew Danish)
3584 * incremented fasl file version number, because of the
3585 SXHASH-related changes in the layout of CLOS data structures
3587 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3588 * minor incompatible change: The runtime (the Unix executable named
3589 "sbcl") is now much pickier about the .core files it will load.
3590 Essentially it now requires .core files to descend from the same
3591 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3592 as the runtime does. (The intent is to prevent the crashes which
3593 can occur, and which can even be reported as mysterious failures,
3594 when people patch the sources or change the build parameters
3595 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3596 sbcl and .core files.)
3597 * fixed bug: VALUES-LIST is no longer optimized away.
3598 * fixed bug 142: The FFI conversion of C string values to Lisp
3599 string values no longer conses excessively. (thanks to Nathan
3600 Froyd porting Raymond Toy's fix to CMU CL)
3601 * began to systematize and improve MOP conformance in PCL (thanks to
3602 Nathan Froyd, Gerd Moellman and Pierre Mai):
3603 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3605 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3606 ** FINALIZE-INHERITANCE is now called on class finalization;
3607 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3608 argument precedence order.
3609 * fixed bug 202: The compiler no longer fails on functions whose
3610 derived types contradict their declared type.
3611 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3612 so it can be non-toplevel.
3613 * The fasl file version number has changed (because of the new
3614 implementation of DEFMACRO).
3615 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3616 safe code, that any length requirement by their type-specifier
3617 argument is valid. The exceptions to this are described in bug
3619 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3620 functions now check the types of their inputs as required by ANSI.
3621 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3622 introduced symbol is DECLAREd to be SPECIAL.
3623 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3624 * fixed bug: :COUNT argument to sequence functions may be negative
3625 * fixed bug: body of DO-SYMBOLS may contain declarations
3626 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3627 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3628 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3629 symbol macro only once
3630 * fixed printing of call frame when argument list is unavailable
3631 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3632 * compiler no longer signals WARNING on unknown keyword
3635 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3636 * A beta-quality port to the mips architecture running Linux,
3637 based on the old CMUCL backend, has been made. It has been tested
3638 on a big-endian kernel, and works sufficiently well to be able to
3639 rebuild itself; it has not been tested in little-endian mode.
3640 * fixed an inconsistency between gencgc.c and purify.c which made
3641 dumping/loading .core files unreliable
3642 * fixed bug 120a: The compiler now deals correctly with IFs where
3643 the consequent is the same as the alternative, instead of
3644 misderiving the return type. (thanks to Alexey Dejneka)
3645 * fixed bug 113: Logical pathnames are now dumpable (the logical
3646 host is resolved at load-time, throwing an error if it is not
3648 * fixed bug 174: FORMAT's error message is slightly clearer when a
3649 non-printing character is used in a format directive.
3650 * fixed several bugs in compiler checking of type declarations, i.e.
3651 violations of the Python "declarations are assertions" principle
3652 (thanks to Alexey Dejneka)
3653 * fixed several bugs in PCL's error checking (thanks to Gerd
3655 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3657 * fixed bug in compilation of functions as first class values
3658 (thanks to Antonio Martinez-Shotton)
3659 * The compiler's handling TYPE-ERRORs which it can prove will
3660 inevitably happen at runtime has been cleaned up and corrected
3661 in several ways. (thanks to Alexey Dejneka)
3662 * improved argument type checking for various basic arithmetic
3663 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3664 lavished on them in the past that they can be compiled in many
3665 ways in different special cases
3666 * fixed bug 181: compiler checks validity of user supplied type
3668 * cleaned up code flushing in optimization: Function calls which
3669 should signal errors for safety purposes (e.g. which ANSI says
3670 should signal errors when their arguments are of incorrect type)
3671 are no longer optimized away.
3672 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3673 * incremented fasl file version number, because changes in the
3674 implementation of sequence functions like COERCE caused
3675 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3676 in old inline expansions) to become undefined. (Actually these
3677 changes were later undone, so we might very well be binary
3678 compatible with 0.7.7 after all, but leaving the version number
3679 incremented seemed like the simplest and most conservative
3682 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3683 * An alpha-quality port to the parisc architecture running Linux,
3684 based on the old CMUCL backend, has been made. This, even more so
3685 than the other backends, should be considered still a work in
3686 progress; known problems include that the Linux kernel in 64-bit
3687 mode does not propagate the correct sigcontext structure to
3688 userspace, and consequently SBCL on a parisc64 kernel will not
3690 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3691 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3692 suppressed.) Also now that the compiler is looking at declarations
3693 in the environment, it checks optimization declarations as well,
3694 and suppresses inlining when (> DEBUG SPEED).
3695 * More fixes have been made to treatment of floating point exception
3696 treatment and other Unix signals. In particular, floating point
3697 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3698 * The detection and handling of control stack exhaustion (infinite
3699 or very deeply nested recursion) has changed. Stack exhaustion
3700 detection is now done by write-protecting pages at the OS level
3701 and applies at all optimization settings; when found, a
3702 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3703 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3704 be caught using IGNORE-ERRORS.
3705 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3706 correctly when transforming recursive LABELS functions to LETs.
3707 (thanks to Alexey Dejneka)
3708 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3709 that are names of constants or global variables.
3710 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3711 alien routines with docstrings.
3712 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3713 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3715 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3716 (thanks to Eric Marsden)
3717 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3718 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3719 * Bug fix: The SPARC backend can now compile functions involving
3720 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3721 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3722 to a functions expecting a PATHNAME-DESIGNATOR.
3723 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3724 lambda lists. (thanks to Alexey Dejneka)
3725 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3726 functions so that EXPORTing the name of the class doesn't cause
3727 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3728 to Antonio Martinez for reporting this)
3729 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3730 constant-folded without failing an assertion. (thanks to Einar
3731 Floystad Dorum for reporting this)
3732 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3733 types are now dealt with more correctly by the compiler.
3734 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3735 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3736 behaviour analogous to RENAME-FILE. This puts its behaviour more
3737 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3739 * The fasl file version number has changed again. (because of the
3740 bug fix involving the names of PCL MAKE-INSTANCE functions)
3742 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3743 * bug fix: Floating point exceptions are treated much more
3744 consistently on the x86/Linux and PPC/Linux platforms.
3745 * Array initialization with :INITIAL-ELEMENT is now much faster for
3746 cases when the compiler cannot open code the array creation, but
3747 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3748 array accesses have also seen a speed increase.
3749 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3750 specified. (This was at the root of some bad interactions between
3751 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3753 * bug fix: Internal error arguments for undefined functions are now
3754 computed correctly on the PPC/Linux platform.
3755 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3756 Raymond Toy's patch for CMU CL)
3757 * Support for the Solaris 9 operating environment has been included
3758 (thanks to Daniel Merritt)
3759 * A very ugly but hopefully complete draft of the missing FFI chapter
3760 of the manual has been created by reformatting the corresponding
3761 CMU CL manual chapter into (currently very ugly and incoherent)
3762 DocBook and bringing it up to date for SBCL behavior. Thus, the
3763 manual is now essentially complete, at least by my extreme
3764 once-and-only-once standards, whereby it's acceptable to refer to
3765 the doc strings of SB-EXT functions as the primary documentation.
3766 * The fasl file version number has changed again, due to cleanup of
3767 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3768 is no longer a static symbol.)
3770 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3771 * SBCL now builds with OpenMCL (version 0.12) as the
3772 cross-compilation host; also, more progress has been made toward
3773 bootstrapping under CLISP.
3774 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3776 * bug 158 fixed: The compiler can now deal with integer loop
3777 increments different from 1; fixing this turned out also to fix
3779 * bug 169 fixed: no more bogus warnings about using lexical bindings
3780 despite the presence of perfectly good SPECIAL declarations (thanks
3781 to David Lichteblau)
3782 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3783 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3784 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3786 * bug 180 fixed: Method combination specifications no longer ignore
3787 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3788 * bug fix: Structure type predicate functions now check their argument
3789 count as they should.
3790 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3791 correctly. (thanks to Pierre Mai)
3792 * minor incompatible change: The --noprogrammer option is deprecated
3793 in favor of the new --disable-debugger option, which is very similar.
3794 (The major difference is that it takes effect at a slightly different
3795 time at startup, causing handling of errors in --sysinit and
3796 --userinit files will be affected differently.) The
3797 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3798 been added to allow this functionality to be controlled from ordinary
3799 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3800 maintainers, who might want to run non-interactive scripts to
3801 build SBCL cores which will later be used interactively.)
3802 * minor incompatible change: The LOAD function no longer, when given
3803 a wild pathname to load, loads all files matching that pathname.
3804 Instead, an error of type FILE-ERROR is signalled.
3806 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3807 * bug 147 fixed: The compiler preserves its block link/count
3808 invariants more correctly now so that it doesn't crash. (thanks
3810 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3812 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3813 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3814 no-value-for-this-character as NIL (as specified by ANSI).
3815 * HOST-NAMESTRING on physical pathnames now returns a string that is
3816 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3817 (thanks to Christophe Rhodes)
3818 * The Alpha port handles icache flushing more correctly. (thanks to
3820 * More progress has been made toward bootstrapping under CLISP. (thanks
3821 to Christophe Rhodes)
3822 * The fasl file format has changed again, because dynamic loading
3823 on OpenBSD (which has non-ELF object files) motivated some cleanups
3824 in the way that foreign symbols are transformed and passed around.
3825 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3826 is no longer treated as whitespace by the reader, but instead as
3827 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3828 |A
\7fB|, instead of A as it used to.
3830 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3831 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3832 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3833 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3834 this as of 0.7.1.45, but was left out of the previous news section
3835 (thanks to Dan Barlow)
3836 * SBCL now runs on the Solaris operating system on SPARC architectures
3837 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3838 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3839 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3840 * SPARC backend cleanups, allowing builds of cores optimized for V8
3841 and V9 SPARCS, and also emission of code targeted to a particular
3842 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3844 * SBCL is closer to bootstrapping under CLISP, thanks to various
3845 fixes by Christophe Rhodes.
3846 * The fasl file format has changed again, to allow the compiler's
3847 INFO database to support symbol macros.
3848 * The user manual (in doc/) is formatted into HTML more nicely.
3849 (thanks to coreythomas)
3850 * The system is smarter about SUBTYPEP relationships, especially
3851 those involving NOT types (including types such as ATOM which are
3852 represented internally using NOT types). Thus SUBTYPEP is less
3853 likely to return (VALUES NIL NIL) in general, and in particular
3854 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3856 * The fasl file format has changed again, because the internal
3857 representation of types now includes a new slot to support the new
3858 SUBTYPEP-of-NOT-types logic.
3859 * (not a change in the main branch of SBCL, but a related prototype
3860 which can hopefully be merged into the main branch of SBCL in the
3861 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3862 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3863 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3864 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3865 dispatch macro character combinations. (thanks to Alexey Dejneka)
3866 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3867 that superficially look like logical namestrings correctly.
3868 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3870 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3871 default supplied-p) for &optional and &key arguments. (thanks to
3874 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3875 * incompatible change: The compiler is now less aggressive about
3876 tail call optimization, doing it only when (> SPACE DEBUG) or
3877 (> SPEED DEBUG). (This is an incompatible change because there are
3878 programs which relied on the old CMU-CL-style behavior to optimize
3879 away their unbounded recursion which will now die of stack overflow.)
3880 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3881 for non-GENCGC systems has been increased to 20M (since that
3882 seems much closer to the likely performance optimum for modern
3883 systems than the old 4M value was)
3884 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3885 and *DEBUG-PRINT-LEVEL*
3886 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3887 of CMU CL's support for SPARC, and various endianness and other
3888 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3889 * new syntactic sugar for the Unix command line: --load foo.bar is now
3890 an alternate notation for --eval '(load "foo.bar")'.
3892 ** The system now detects stack overflow and handles it gracefully,
3893 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3894 optimization settings. (This is a good thing in general, and
3895 its introduction in this version should be particularly timely
3896 for anyone whose code fails because of suppression of tail
3898 ** The system now hunts for the C variable "environ" in a more
3899 devious way, to avoid segfaults when the C library version differs
3900 between compile time and run time. (thanks to Christophe Rhodes)
3901 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3902 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3903 ** The compiler no longer issues bogus style warnings for undefined
3904 classes in the same source file as the DEFCLASSes which defined
3905 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3907 ** fixes in CONDITION class precedence list for undefined function
3908 errors (thanks to Alexei Dejneka)
3909 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3910 correctly. (thanks to Dan Barlow)
3911 ** portability fixes aiming at bootstrapping under CLISP (thanks
3912 to Dave McDonald and Christophe Rhodes)
3913 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3914 ** fixes in type translation and and type inference (thanks to
3916 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3917 ** various fixes in the new ports (thanks to Dan Barlow)
3918 * several changes related to debugging:
3919 ** suppression of tail recursion, as noted above
3920 ** stack overflow detection, as noted above
3921 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3922 is now the default. (For some time encapsulation has been more
3923 reliable than the breakpoint-based :ENCAPSULATE NIL
3924 implementation, at least on X86 systems; and I just noticed that
3925 encapsulation also seems closer to the spirit of the ANSI
3928 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3930 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3931 up properly again. (There was a packaging bug in 0.7.0 which
3932 left their definitions in SB-SYS::LOAD-FOREIGN and
3933 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3934 vital for most things which interface to C-level interfaces,
3935 like extensions working with sockets or databases or
3936 Perl-compatible regexes or whatever, and the need to fix
3937 this bug is the main reason that 0.7.1 was released so
3939 ** DEFGENERIC is now choosier about the methods it redefines, so that
3940 reLOADing a previously-LOADed file containing DEFGENERICs does
3941 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3942 cycle now works as it should. (thanks to Alexey Dejneka)
3943 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3944 fixed by Christophe Rhodes. (He actually submitted this patch
3945 months ago, and I delayed until after 0.7.0.)
3946 ** Bug 111 (internal compiler confusion about runtime checks on
3947 FUNCTION types) was fixed by Alexey Dejneka.
3948 * Some internal cleanups (getting rid of variables which aren't
3949 needed now that the byte interpreter is gone) caused the fasl
3950 file format number to change again.
3952 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3953 * major incompatible change: The default fasl file extension, i.e. the
3954 default extension for files produced by COMPILE-FILE, has changed
3955 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3957 ** There are many changes in the implementation of the compiler.
3958 SBCL is now essentially a compiler-only implementation of ANSI
3959 Common Lisp. EVAL still "interprets" a few special cases, but
3960 almost all the interesting cases are handled by creating
3961 a LAMBDA expression, calling COMPILE on it, then calling
3962 FUNCALL on the result.
3963 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3964 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3965 Since the code is newer, there might still be some new bugs
3966 (though not as many as before Martin Atzmueller's fixes:-). But
3967 the new code is substantially simpler and clearer, and hopefully
3968 any remaining bugs will be simpler, less fundamental, and more
3969 fixable then the bugs in the old code.
3970 ** The revised compiler is still a little unsteady on its feet.
3972 *** The debugging information it produces (particularly the names
3973 of FUNCTION objects) is sometimes much less useful than what
3974 the old compiler produced.
3975 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3976 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3977 within a MACROLET) has been temporarily weakened.
3978 ** There are new compiler optimizations for various functions:
3979 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3980 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3981 *** the math functions TRUNCATE, FLOOR, and CEILING
3982 *** the function-of-all-trades COERCE
3983 Mostly these should be transparent, but there's one
3984 potentially-annoying problem (bug 117): when the compiler
3985 inline-expands a function and does type analysis on the result,
3986 it can create control paths which have type mismatches, and
3987 when it can't prove that those control paths aren't taken,
3988 it will issue WARNINGs about the type mismatches. This is
3989 a particular problem in practice for the new sequence functions.
3990 It's not clear how this should be fixed, and for now, a
3991 workaround is given in the entry for 117 in the BUGS file.
3992 ** (Because of the interaction between the two previous items --
3993 occasional inlining problems and new inline expansions -- some
3994 of the new sequence function optimizations won't really kick in
3995 completely until debugging information, and then inlining, are
3996 straightened out in some future version.)
3997 * minor incompatible changes:
3998 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3999 DIRECTORY no longer implicitly promotes NIL slots of its
4000 pathname argument to :WILD. In particular, when you ask for the
4001 contents of a directory (which you used to be able to do without
4002 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4003 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4004 ** changes in behavior that ANSI explicitly defines to be
4005 implementation dependent:
4006 *** The new compiler-only implementation still conforms with ANSI,
4007 but acts a little different than before. Besides the obvious
4008 changes in performance tradeoffs (that the cost per form passed
4009 to EVAL has gone up, and the cost per form executed by EVAL
4010 has gone down), the behavior of the system changes a little
4011 because there are no longer any interpreted function objects.
4012 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4013 e.g. doing COMPILE on the output of interactive DEFUN is
4015 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4017 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4018 has changed from (:CL) to NIL.
4019 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4020 e.g. #\Newline and #\Tab, don't have names specified in the
4021 ANSI Common Lisp standard, is now based on their ASCII symbolic
4022 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4023 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4024 are no longer used for output.
4025 ** changes in internal implementation constants:
4026 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4027 4 million. (If your application spends a lot of time GCing and
4028 you have a lot of RAM, you might want to experiment with
4029 increasing it even more.)
4030 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4031 However, almost all old code should still continue to work without
4032 immediate update, as SB-C-CALL is now a (deprecated) nickname
4034 ** Old operator names in the style DEF-FOO are now deprecated in
4035 favor of new corresponding names DEFINE-FOO, for consistency with
4036 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4037 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4038 internal symbols, but a few supported extensions like
4039 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4040 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4041 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4042 etc. as you get deeper into recursive calls to the debugger
4043 command loop, instead of the old "5]", "5]]", "5]]]"
4044 sequence. (I was motivated to do this when squabbles between
4045 ILISP and SBCL left me very deeply nested in the debugger. In the
4046 short term, this change will probably provoke more ILISP/SBCL
4047 squabbles, but hopefully it will be an improvement in the long run.)
4048 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4049 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4050 function, since declaiming return types involving aliens is
4051 (1) annoyingly messy to do by hand and (2) vital to efficient
4052 compilation of code which calls such functions.
4053 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4054 longer reexported by the SB-EXT package. They're solely useful
4055 for alien code, so it seems more logical that you should get
4056 them from the SB-ALIEN package, not in SB-EXT.
4057 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4058 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4059 features. Instead, the code that they used to control is always
4060 built into the system.
4061 * many other bug fixes
4062 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4063 advantage of the new EVAL-WHEN stuff and to clean them up in
4064 general, and they are now more ANSI-compliant in a number of
4065 ways. Martin Atzmueller is responsible for a lot of this.
4066 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4068 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4069 *** correct ERROR type for various file operations
4070 *** some fixes for Lisp streams
4071 *** DEFMETHOD syntax checking
4072 *** changing old weird representation of debug information as
4073 strings (which, among their other deficiencies, don't transform
4074 correctly when you rename packages, and don't change their
4075 print representation when you change things like *PACKAGE*
4076 and *PRINT-LENGTH*) to symbols and lists of symbols
4077 He also made several improvements and fixed several bugs in DESCRIBE.
4078 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4080 *** misbehavior of WRITE-STRING/WRITE-LINE
4081 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4082 and several other LOOP problems as well
4083 *** DIRECTORY when similar filenames are present
4084 *** DEFGENERIC with :METHOD options
4085 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4086 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4087 *** argument ordering in FIND with :TEST option
4088 *** mishandled package designator argument in APROPOS-LIST
4089 *** various problems in the backquote readmacro
4090 *** a bug in APROPOS
4091 *** probably some others that I'm not describing very well here,
4092 since the CVS log documents them by reference to sbcl-devel
4093 messages, and the SourceForge archives aren't working well.:-(
4094 ** Dan Barlow improved the Alpha port (and is making progress on the
4095 PPC port, for those of you who think different).
4096 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4097 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4098 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4099 continues to work on the SPARC port (for those of you in a position
4100 to look down upon our little PC-compatible boxes from a great height).
4101 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4102 and uses that copy, rather than the current dynamic value, when
4103 it's trying to decide whether to truncate output. Thus e.g.
4104 (let ((*print-lines* 50))
4105 (pprint-logical-block (stream nil)
4107 (let ((*print-lines* 8))
4108 (print (aref possiblybigthings i) stream)))))
4109 should now truncate the logical block only at 50 lines, instead of
4110 often truncating it at 8 lines, as it did before.
4111 * The doc/cmucl/ directory, containing old CMU CL documentation
4112 from the time of the fork, is no longer part of the base system.
4113 SourceForge has shut down its anonymous FTP service, and with it
4114 my original plan for distributing the old CMU CL documentation
4115 there. For now, if you need these files you can download an old
4116 SBCL source release and extract them from it.
4117 * The fasl file version number changed again, for dozens of reasons,
4118 some of which are apparent above.
4120 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4121 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4122 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4123 patch, so that DISASSEMBLE output is much nicer.
4124 * The code in the SB-PROFILE package now seems reasonably stable.
4125 I still haven't decided what the final interface should look like
4126 (I'd like PROFILE to interact cleanly with TRACE, since both
4127 facilities use function encapsulation) but if you have a need
4128 for profiling now, you can probably use it successfully with
4129 the current CMU-CL-style interface.
4130 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4131 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4132 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4133 link now returns the dangling link itself, and for similar
4134 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4135 link itself. (In these cases the old code signalled an error and
4136 looped endlessly, respectively.) Thus, DIRECTORY now works even
4137 in the presence of dangling and cyclic symbolic links.
4138 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4139 is now a supported extension again, since the consensus on
4140 sbcl-devel was that it can be useful for ordinary development
4141 work, not just for debugging SBCL itself.
4142 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4143 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4144 that currently-defined functions might be redefined later with
4145 different return types.
4146 * Hash tables can be printed readably, as inspired by CMU CL code
4147 of Eric Marsden and SBCL code of Martin Atzmueller.
4148 * better error handling in CLOS method combination, thanks to
4149 Martin Atzmueller porting Pierre Mai's CMU CL patches
4150 * more overflow fixes for >16Mbyte I/O buffers
4151 * A bug in READ has been fixed, so that now a single Ctrl-D
4152 character suffices to cause end-of-file on character streams.
4153 In particular, now you only need one Ctrl-D at the command
4154 line (not two) to exit SBCL.
4155 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4157 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4158 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4159 does the right thing.
4160 * The implementation of some type tests, especially for CONDITION
4161 types, is now tidier and maybe faster, due to CMU CL code
4162 originally by Douglas Crosher, ported by Martin Atzmueller.
4163 * Some math functions have been fixed, and there are new
4164 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4165 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4166 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4167 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4168 likely to be built into sbcl-0.7.0.)
4169 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4170 CL, ported to SBCL by Martin Atzmueller)
4171 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4172 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4173 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4174 * There's a new slam.sh hack to shorten the edit/compile/debug
4175 cycle for low-level changes to SBCL itself, and a new
4176 :SB-AFTER-XC-CORE target feature to control the generation of
4177 the after-xc.core file needed by slam.sh.
4178 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4179 COMPILE-FILE is no longer supported, so that now every function
4180 gets an entry point, so that block compilation looks a little
4181 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4182 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4183 returns the number of bytes consed since the system started,
4184 rather than the number consed since the first time the function
4185 was called. (The new definition parallels ANSI functions like
4186 CL:GET-INTERNAL-RUN-TIME.)
4187 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4188 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4189 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4190 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4191 * Fasl file version numbers are now independent of the target CPU,
4192 since historically most system changes which required version
4193 number changes have affected all CPUs equally. Similarly,
4194 the byte fasl file version is now equal to the ordinary
4197 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4198 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4199 is no longer recognized. I apologize for this, because it was
4200 listed in SB-EXT as a supported extension, but I found that
4201 its existing behavior was poorly specified, as well as incorrectly
4202 specified, and it looked like too much of a mess to straighten it
4203 out. I have enough on my hands trying to get ANSI stuff to work..
4204 * many patches ported from CMU CL by Martin Atzmueller, with
4205 half a dozen bug fixes in pretty-printing and the debugger, and
4206 half a dozen others elsewhere
4207 * fixed bug 13: Floating point infinities are now supported again.
4208 They might still be a little bit flaky, but thanks to bug reports
4209 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4210 as flaky as they were.
4211 * The --noprogrammer command line option is now supported. (Its
4212 behavior is slightly different in detail from what the old man
4213 page claimed it would do, but it's still appropriate under the
4214 same circumstances that the man page talks about.)
4215 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4216 are now supported, and enabled by default. Thus, the compiler can
4217 handle many floating point and complex operations much less
4218 inefficiently. (Thus e.g. you can implement a complex FFT
4220 * The compiler now detects type mismatches between DECLAIM FTYPE
4221 and DEFUN better, and implements CHECK-TYPE more correctly, and
4222 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4223 thanks to patches from Martin Atzmueller.
4224 * various fixes to make the cross-compiler more portable to
4225 ANSI-conforming-but-different cross-compilation hosts (notably
4226 Lispworks for Windows, following bug reports from Arthur Lemmens)
4227 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4228 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4229 fixed thanks to Pierre Mai's CMU CL patches.
4230 * fixes to keep the system from overflowing internal counters when
4231 it tries to use i/o buffers larger than 16M bytes
4232 * fixed bug 45a: Various internal functions required to support
4233 complex special functions have been merged from CMU CL sources.
4234 (When I was first setting up SBCL, I misunderstood a compile-time
4235 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4236 * improved support for type intersection and union, fixing bug 12
4237 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4238 more obscure bugs as well
4239 * some steps toward byte-compiling non-performance-critical
4240 parts of the system, courtesy of patches from Martin Atzmueller
4241 * Christophe Rhodes has made some debian packages of sbcl at
4242 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4243 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4244 stable, but are nonetheless usable. When he's ready, I'd be happy
4245 to add them to the SourceForge "File Releases" section. (And if
4246 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4247 * new fasl file format version number (because of changes in
4248 internal representation of (OR ..) types to accommodate the new
4249 support for (AND ..) types, among other things)
4251 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4252 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4254 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4256 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4257 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4258 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4259 * DESCRIBE now works on structure objects again.
4260 * Most function call argument type mismatches are now handled as
4261 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4262 know whether the function will be redefined before the call is
4263 executed. (The compiler could flag local calls with full WARNINGs,
4264 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4265 it doesn't keep track of enough information to know whether calls
4266 are local in this sense.)
4267 * Compiler output is now more verbose, with messages truncated
4268 later than before. (There should be some supported way for users
4269 to override the default verbosity, but I haven't decided how to
4270 provide it yet, so this behavior is still controlled by the internal
4271 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4272 src/compiler/ir1util.lisp.)
4273 * Fasl file format version numbers have increased again, because
4274 support for the Gray streams extension changes the layout of the
4275 system's STREAM objects.
4276 * The Gray subclassable streams extension now works, thanks to a
4277 patch from Martin Atzmueller.
4278 * The full LOAD-FOREIGN extension (not just the primitive
4279 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4280 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4281 but like most other programs, it defaults to copying the Unix
4282 environment from the original process instead of starting the
4283 new process in an empty environment.
4284 * Extensions which manipulate the Unix environment now support
4285 an :ENVIRONMENT keyword option which doesn't smash case or
4286 do other bad things. The CMU-CL-style :ENV option is retained
4287 for porting convenience.
4288 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4289 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4291 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4293 * A patch from Martin Atzmueller seems to have solved the SIGINT
4294 problem, and as far as we know, signal-handling now works cleanly.
4295 (If you find any new bugs, please report them!)
4296 * The system no longer defaults Lisp source file names to types
4297 ".l", ".cl", or ".lsp", but only to ".lisp".
4298 * The compiler no longer uses special default file extensions for
4299 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4300 seems to expect a single default extension for all compiled code,
4301 and there's no compelling reason to try to stretch the standard
4302 to allow two different extensions.) Instead, byte-compiled files
4303 default to the same extension as native-compiled files.
4304 * Fasl file format version numbers have increased again, because
4305 a rearrangement of internal implementation packages made some
4306 dumped symbols in old fasl files unreadable in new cores.
4307 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4308 many fewer weird special cases.
4309 * Bug #17 (differing COMPILE-FILE behavior between logical and
4310 physical pathnames) has been fixed, and some related misbehavior too,
4311 thanks to a patch from Martin Atzmueller.
4312 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4313 by Tim Moore, ported to SBCL by Martin Atzmueller.
4314 * Martin Atzmueller fixed several filesystem-related problems,
4315 including bug #36, in part by porting CMU CL patches, which were
4316 written in part by Paul Werkowski.
4317 * More compiler warnings in src/runtime/ are gone, thanks to
4318 more patches from Martin Atzmueller.
4319 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4322 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4324 * DESCRIBE now works on CONDITION objects.
4325 * The debugger now handles errors which arise when trying to print
4326 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4328 * The build system now uses an additional file, customize-target-features.lisp,
4329 to allow local modifications to the target *FEATURES* list. (The point of
4330 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4331 debugging features enabled, without having to worry about propagating it
4332 into everyone's system when I do a "cvs update".) When no
4333 customize-target-features.lisp file exists, the target *FEATURES* list
4334 should be constructed the same way as before.
4335 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4336 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4337 warn about behavior which is undefined under the ANSI spec. Note: This
4338 is specified by ANSI, but it's not very popular with programmers.
4339 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4340 macro in the SBCL sources for an example of a workaround which you
4341 might use to make portable ANSI-standard code which does what you want.
4342 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4343 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4344 when it's used as a non-toplevel form. (This is required in order
4345 to implement the DEFCONSTANT-EQX macro.)
4346 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4347 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4348 funny things, due to bugs in the implementation of EVAL-WHEN
4349 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4350 non-toplevel DEFCONSTANTs are uncommon.)
4351 * The core file version number and fasl file version number have been
4352 incremented, because the old noncompliant DEFCONSTANT behavior involved
4353 calling functions which no longer exist, and because I also took the
4354 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4355 * fixed bug 1 (error handling before read-eval-print loop starts), and
4356 redid debugger restarts and related debugger commands somewhat while
4358 ** The QUIT debugger command is gone, since it did something
4359 rather different than the SB-EXT:QUIT command, and since it never
4360 worked properly outside the main toplevel read/eval/print loop.
4361 Invoking the new TOPLEVEL restart provides the same functionality.
4362 ** The GO debugger command is also gone, since you can just invoke
4363 the CONTINUE restart directly instead.
4364 ** The TOP debugger command is also gone, since it's redundant with the
4365 FRAME 0 command, and since it interfered with abbreviations for the
4367 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4369 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4370 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4371 some porting work by Martin Atzmueller)
4372 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4373 with a boatload of patches which he ported to SBCL. Now that those
4375 ** The system tries to make sure that its low-priority messages
4376 are prefixed by semicolons, to help people who like to use
4377 syntax highlighting in their ILISP buffer. (This patch
4378 was originally due to Raymond Toy.)
4379 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4380 patches originally written by Raymond Toy.
4381 ** The compiler understands coercion between single-value and
4382 multiple-VALUES type expressions better, getting rid of some very
4383 weird behavior, thanks to patches originally by Robert MacLachlan
4384 and Douglas Crosher.
4385 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4386 lambda lists, thanks to a patch originally by Pierre Mai.
4387 ** The system no longer bogusly warns about "abbreviated type
4389 ** The compiler gets less confused by inlining and RETURN-FROM,
4390 thanks to some patches originally by Tim Moore.
4391 ** The system no longer hangs when dumping circular lists to fasl
4392 files, thanks to a patch originally from Douglas Crosher.
4393 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4394 undefined function error.
4395 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4396 in structure slot initforms) for now, documented workaround instead:-|
4397 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4398 * fixed bug 5: added stubs for various Gray stream functions called
4399 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4400 installed, at least appropriate type errors are generated
4401 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4402 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4403 consistently in DEFMETHOD forms.
4404 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4405 it doesn't seem to affect SBCL after all
4406 * The C runtime system now builds with better optimization and many
4407 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4409 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4411 * The system is now under CVS at SourceForge (instead of the
4412 CVS repository on my home machine).
4413 * The new signal handling code has been tweaked to treat register
4414 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4415 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4416 downstream, e.g. in debug-int.lisp, has implicit dependencies
4417 on the unsignedness of integer representation of machine words,
4418 and that caused the system to bomb out with infinite regress
4419 when trying to recover from type errors involving signed values,
4420 e.g. (BUTLAST '(1 2 3) -1).)
4421 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4422 (This was one of the bugs Peter Van Eynde reported back in July.)
4423 * The system now uses code inspired by Colin Walters' O(N)
4424 implementation of MAP (from the cmucl-imp@cons.org mailing
4425 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4426 inline the MAP operation, and there is more than one
4427 sequence argument to the MAP call (so that it can't just
4428 do ETYPECASE once and for all based on the type of the
4429 single sequence argument). (The old non-inline implementation
4430 of the general M-argument sequence-of-length-N case required
4431 O(M*N*N) time when any of the sequence arguments were LISTs.)
4432 * The QUIT :UNIX-CODE keyword argument has been renamed to
4433 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4434 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4436 * Raymond Wiker's patches to port dynamic loading from Linux to
4437 FreeBSD have been added.
4438 * The BUGS file is now more nearly up to date, thanks in large part
4439 to Martin Atzmueller's review of it.
4440 * The debugger now flushes standard output streams before it begins
4441 its output ("debugger invoked" and so forth).
4442 * The core version number and fasl file version number have both
4443 been incremented, because of incompatible changes in the layout
4445 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4446 used to not be called for a saved Lisp image.)
4447 * Martin Atzmueller's version of a patch to fix a compiler crash,
4448 as posted on sbcl-devel 13 September 2000, has been installed.
4449 * Instead of installing Martin Atzmueller's patch for the
4450 compiler transform for SUBSEQ, I deleted the compiler transform,
4451 and transforms for some similar consing operations.
4452 * A bug in signal handling which kept TRACE from working on OpenBSD
4454 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4455 away by constant folding
4456 * The system now defines its address space constants in one place
4457 (in the Lisp sources), and propagates them automatically elsewhere
4458 (through GENESIS and the sbcl.h file). Therefore, patching the
4459 address map is less unnecessarily tedious and error-prone. The
4460 Lisp names of address space constants have also been systematized.
4461 * CVS tags like dollar-Header-dollar have been removed from
4462 the sources, because they have never saved me trouble and
4463 they've been source of trouble working with patches and other
4464 diff-related operations.
4465 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4466 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4468 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4470 * The system has been ported to OpenBSD.
4471 * The system now compiles with a simple "sh make.sh" on the systems
4472 that it's supported on. I.e., now you no longer need to tweak
4473 text in the target-features.lisp-expr and symlinks in src/runtime/
4474 by hand, the make.sh takes care of it for you.
4475 * The system is no longer so grossly inefficient when compiling code
4476 involving vectors implemented as general (not simple) vectors (VECTOR T),
4477 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4478 which dares to use the various sequence functions on non-simple
4479 vectors, takes less of a performance hit.
4480 * There is now a primitive type predicate VECTOR-T-P
4481 to test for the (VECTOR T) type, so that e.g.
4482 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4483 can now be compiled with some semblance of efficiency. (The old code
4484 turned the type declaration into a full call to %TYPEP at runtime!)
4485 * AREF on (VECTOR T) is still not fast, since it's still compiled
4486 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4487 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4488 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4489 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4490 without making full calls to SUBTYPEP at runtime.
4491 (Some analogous efficiency issues for non-simple vectors specialized to
4492 element types other than T, or for non-simple multidimensional arrays,
4493 have not been addressed. They could almost certainly be handled the
4494 same way if anyone is motivated to do so.)
4495 * The changes in array handling break binary compatibility, so
4496 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4497 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4498 instead of (VALUES T T).
4499 * By following the instructions that Dan Barlow posted to sbcl-devel
4500 on 2 July 2000, I was able to enable primitive dynamic object
4501 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4502 functionality is not implemented (since it calls ld to resolve
4503 library references automatically, requiring RUN-PROGRAM for its
4504 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4505 try to resolve library references) is now supported.
4506 * The system now flushes the standard output streams when it terminates,
4507 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4508 them at several other probably-convenient times, e.g. in each pass of
4509 the toplevel read-eval-print loop, and after evaluating a form given
4510 as an "--eval" command-line option. (These changes were motivated by a
4511 discussion of stream flushing issues on cmucl-imp in August 2000.)
4512 * The source transform for TYPEP of array types no longer assumes
4513 that an array whose element type is a not-yet-defined type
4514 is implemented as an array of T, but instead punts, so that the
4515 type will be interpreted at runtime.
4516 * There is now some support for cross-compiling in make.sh: each of
4517 the phases of make.sh has its own script. (This should be transparent
4518 to people doing ordinary, non-cross-compile builds.)
4519 * Since my laptop doesn't have hundreds of megabytes of memory like
4520 my desktop machine, I became more motivated to do some items on
4521 my to-do list in order to reduce the size of the system a little:
4522 ** Arrange for various needed-only-at-cold-init things to be
4523 uninterned after cold init. To support this, those things have
4524 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4525 symbols with such names are now uninterned after cold init).
4526 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4527 fasl files for cold load.
4528 ** Remove the old compiler structure pooling code (which used to
4529 be conditional on the target feature :SB-ALLOC) completely.
4530 ** Redo the representation of some data in cold init to be more compact.
4531 (I also looked into supporting byte compiled code at bootstrap time,
4532 which would probably reduce the size of the system a lot, but that
4533 looked too complicated, so I punted for now.)
4534 * The maximum signal nesting depth in the src/runtime/ support code has
4535 been reduced from 4096 to 256. (I don't know any reason for the very
4536 large old value. If the new smaller value turns out to break something,
4537 I'll probably just bump it back up.)
4538 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4540 * Many, many bugs reported by Peter Van Eynde have been added to
4541 the BUGS list; some have even been fixed.
4542 * While enabling dynamic object file loading, I tried to make the
4543 code easier to understand, renaming various functions and variables
4544 with less ambiguous names, and changing some function calling
4545 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4546 * While trying to figure out how to do the OpenBSD port, I tried to
4547 clean up some of the code in src/runtime/. In particular, I dropped
4548 support for non-POSIX signal handling, added various comments,
4549 tweaked the code to reduce the number of compilation warnings, and
4550 renamed some files to increase consistency.
4551 * To support the new automatic configuration functionality in make.sh,
4552 the source file target-features.lisp-expr has been replaced with the
4553 source file base-target-features.lisp-expr and the machine-generated
4554 file local-target-features.lisp-expr.
4555 * fixed a stupid quoting error in make.sh so that using CMU CL
4556 "lisp -batch" as cross-compilation host works again
4558 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4560 * DESCRIBE no longer tries to call itself recursively to describe
4561 bound/fbound values, so that it no longer fails on symbols which are
4562 bound to themselves (like keywords, T, and NIL).
4563 * DESCRIBE now works on generic functions.
4564 * The printer now prints less-screwed-up representations of closures
4565 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4566 * A private symbol is used instead of the :EMPTY keyword previously
4567 used to mark empty slots in hash tables. Thus
4568 (DEFVAR *HT* (MAKE-HASH-TABLE))
4569 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4570 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4571 now does what ANSI says that it should. (You can still get
4572 similar noncompliant behavior if bang on the hash table
4573 implementation with all the symbols you get back from
4574 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4575 This breaks binary compatibility, since tests for equality to
4576 :EMPTY are wired into things like the macroexpansion of
4577 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4579 * There's now a minimal placeholder implementation for CL:STEP,
4580 as required by ANSI.
4581 * An obscure bug in the interaction of the normal compiler, the byte
4582 compiler, inlining, and structure predicates has been patched
4583 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4584 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4586 * Missing ordinary arguments in a macro call are now detected even
4587 when the macro lambda list contains &KEY or &REST.
4588 * The debugger no longer complains about encountering the top of the
4589 stack when you type "FRAME 0" to explicitly instruct it to go to
4590 the top of the stack. And it now prints the frame you request even
4591 if it's the current frame (instead of saying "You are here.").
4592 * As specified by ANSI, the system now always prints keywords
4593 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4595 * The default initial SIZE of HASH-TABLEs is now smaller.
4596 * Type information from CLOS class dispatch is now propagated
4597 into DEFMETHOD bodies, so that e.g.
4598 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4600 is now basically equivalent to
4601 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4602 (DECLARE (TYPE SINGLE-FLOAT X))
4604 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4605 operation, without having to do run-time type dispatch.
4606 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4607 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4608 * Since I don't seem to be making big file reorganizations very often
4609 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4610 is overflowing my ability to conveniently back them up), I've finally
4611 checked the system into CVS. (The CVS repository is on my home system,
4612 not at SourceForge -- putting it on SourceForge might come later.)
4613 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4614 high-level GC-NOTIFY-FOO functions send their output. (There's
4615 still very little control of where low-level verbose GC functions
4616 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4617 less than it used to -- the GC-NOTIFY-FOO functions are now under
4618 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4619 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4620 in only one place in the source code, and propagates it automatically
4621 everywhere that it's needed. Thus e.g. when I bump the version from
4622 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4623 * The C source files now include boilerplate legalese and documentation
4624 at the head of each file (just as the Lisp source files already did).
4625 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4626 to his page will be replaced with a link to his new CLiki service.
4628 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4630 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4631 * The build process now looks for GNU make under the default name "gmake",
4632 instead of "make" as it used to. If GNU make is not available as "gmake"
4633 on your system, you can change this default behavior by setting the
4634 GNUMAKE environment variable.
4635 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4636 can build without error under CMU CL.
4638 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4640 * There is now a partial SBCL user manual (with some new text and some
4641 text cribbed from the CMU CL manual).
4642 * The beginnings of a profiler have been added (starting with the
4643 CMU CL profiler and simplifying and cleaning up). Eventually the
4644 main interface should be through the TRACE macro, but for now,
4645 it's still accessed through vaguely CMU-CL-style functions and macros
4646 exported from the package SB-PROFILE.
4647 * Some problems left over from porting CMU CL to the new
4648 cross-compilation bootstrap process have been cleaned up:
4649 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4650 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4651 part when getting rid of the compiler *BACKEND* stuff.)
4652 ** The value of *NULL-TYPE* was screwed up, because it was
4653 being initialized before the type system knew the final
4654 definition of the 'NULL type. This screwed up several key
4655 optimizations in the compiler, causing inefficiency in all sorts
4656 of places. (I found it because I wanted to understand why
4657 GET-INTERNAL-RUN-TIME was consing.)
4658 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4659 PROCLAIM FTYPE information. Unfortunately this broke binary
4660 compatibility again, since now the forms output by DEFGENERIC
4661 to refer to functions which didn't exist in 0.6.3.
4662 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4663 can use the new (as of 0.6.3) transform for SOME into MAP into
4665 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4666 values to (INTEGER 0 1000000), so that the time code will no longer
4667 occasionally get blown up by Linux returning 1000000 microseconds
4668 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4669 its output conform to the ANSI spec. (Alas, this makes its output
4670 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4671 * A full call to MAP NIL with a single sequence argument no longer conses.
4672 * fixes to problems pointed out by Martin Atzmueller:
4673 * The manual page no longer talks about multiprocessing as though
4674 it were currently supported.
4675 * The ILISP support patches have been removed from the distribution,
4676 because as of version 5.10.1, ILISP now supports SBCL without us
4677 having to maintain patches.
4678 * added a modified version of Raymond Toy's recent CMU CL patch for
4679 EQUALP comparison of HASH-TABLE
4681 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4683 * The system still can't cross-compile itself with
4684 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4685 entails), but at least it can compile more of itself that way
4686 than it used to be able to, and various buglets which were uncovered
4687 by trying to cross-compile itself that way have now been fixed.
4688 * This release breaks binary compatibility again. This time
4689 at least I've incremented the FASL file format version to 2, so that the
4690 problem can be detected reliably instead of just causing weird errors.
4691 * various new style warnings:
4692 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4693 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4694 ** using the lexical binding of a variable named in the *FOO* style
4695 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4696 as specified by ANSI.
4697 * *RANDOM-STATE* is no longer automatically initialized from
4698 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4699 default behavior of the system is to repeat its behavior every time
4700 it's run. If you'd like to change this behavior, you can always
4701 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4702 old convention there was no comparably easy way to get the system to
4703 repeat its behavior every time it was run.
4704 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4705 FTYPE declarations has been removed, in favor of their ANSI
4706 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4707 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4708 the types of their sequence arguments can be determined at compile time.
4709 This is done through a new open code expansion for MAP which eliminates
4710 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4711 argument types can be determined at compile time.
4712 * The optimizer now transforms COERCE into an identity operation when it
4713 can prove that the coerced object is already of the correct type. (This
4714 can be a win for machine generated code, including the output of other
4715 optimization transforms, such as the MAP transform above.)
4716 * Credit information has been moved from source file headers into CREDITS.
4717 * Source file headers have been made more standard.
4718 * The CASE macro now compiles without complaining even when it has
4721 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4723 * (Note that the way that the PCL macroexpansions were rewritten
4724 to accommodate the change in DEFGENERIC below breaks binary
4725 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4726 not run under sbcl-0.6.2. Once we get out of alpha releases,
4727 i.e. hit release 1.0.0, we'll probably try to maintain binary
4728 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4729 and sbcl-1.4.4. Until then, however, it might be fairly common
4730 for maintenance releases to break binary compatibility.)
4731 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4733 * The reporting of unhandled signals has been changed to print some
4734 explanatory text as well as the report form. (Previously only
4735 the report form was printed.)
4736 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4737 it defines, so that the compiler no longer issues undefined function
4738 warnings for compiled-but-not-yet-loaded generic functions.
4739 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4740 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4741 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4742 section of the ANSI spec are supported.
4743 * The "" nickname for the "KEYWORD" package has been removed.
4744 The reader still handles symbol tokens which begin with a package marker
4745 as keywords, but it doesn't expose its mechanism for doing so in the
4746 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4747 * The system now issues STYLE-WARNINGs for contradictory TYPE
4748 proclamations. (Warnings for contradictory FTYPE proclamations would
4749 be nice too, but those can't be done usefully unless the type system
4750 is made smarter about FUNCTION types.)
4751 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4752 systematized, so that "*target-*.lisp is supposed to exist only on the
4753 target and imply that there's a related file which exists on the
4754 host, and *host-*.lisp is supposed to exist only on the host and imply
4755 that there's a related file which exists on the target. This involves a
4756 lot of renaming. Hopefully the acute confusion caused by the renaming
4757 will be justified by the reduction in chronic confusion..
4758 ** runtime-type.lisp -> early-target-type.lisp
4759 ** target-type.lisp -> late-target-type.lisp
4760 ** early-host-format.lisp -> early-format.lisp
4761 ** late-host-format.lisp -> late-format.lisp
4762 ** host-error.lisp -> misc-error.lisp
4763 ** early-error.lisp -> early-target-error.lisp
4764 ** late-error.lisp -> late-target-error.lisp
4765 ** host-defboot.lisp -> early-defboot.lisp
4766 ** code/misc.lisp -> code/target-misc.lisp
4767 ** code/host-misc.lisp -> code/misc.lisp
4768 ** code/numbers.lisp -> code/target-numbers.lisp
4769 ** code/early-numbers.lisp -> numbers.lisp
4770 ** early-host-type.lisp -> early-type.lisp
4771 ** late-host-type.lisp -> late-type.lisp
4772 ** host-typep.lisp -> typep.lisp
4773 ** load.lisp -> target-load.lisp
4774 ** host-load.lisp -> load.lisp
4775 ** host-disassem.lisp -> disassem.lisp
4776 ** host-insts.lisp -> insts.lisp
4777 ** byte-comp.lisp -> target-byte-comp.lisp
4778 ** host-byte-comp.lisp -> byte-comp.lisp
4779 ** host-signal.lisp -> signal.lisp
4780 ** host-defstruct.lisp -> defstruct.lisp
4781 ** late-target-type.lisp -> deftypes-for-target.lisp
4782 Furthermore, several other previously target-only files foo.lisp (e.g.
4783 hash-table.lisp and random.lisp) have been split into a target-and-host
4784 foo.lisp file and a target-only target-foo.lisp file, with their key type
4785 definitions in the target-and-host part, so that the cross-compiler will
4786 know more about target types.
4787 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4788 gone away. In their place are various *BACKEND-FOO* variables
4789 corresponding to the slots of the old structure.
4790 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4791 into the target SBCL has been fixed.
4792 * The chill.lisp system for loading cold code into a running SBCL
4794 * Support for the CMU CL "scavenger hook" extension has been removed.
4795 (It was undocumented and unused in the CMU CL sources that SBCL was
4796 derived from, and stale in sbcl-0.6.1.)
4797 * Various errors in the cross-compiler type system were detected
4798 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4799 (enabling various consistency checks). Many of them were fixed,
4800 but some hard problems remain, so the compiler is back to
4801 running without *TYPE-SYSTEM-INITIALIZED* for now.
4802 * As part of the cross-compiler type system cleanup, I implemented
4803 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4804 * I have started adding UNCROSS calls throughout the type system
4805 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4806 on cross-compiler input files will be able to go away, and various
4808 * CONSTANTP now returns true for quoted forms (as explicitly required
4811 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4813 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4814 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4815 (MAKE-LIST -1). In the longer term, it ought to become true
4816 that declarations are assertions even at SAFETY 1. For now, it's not
4817 quite true even at SAFETY 3, but it's at least more nearly true..
4818 (Note that this change seems to increases the size of the system by
4819 O(5%) and to decrease the speed of the compiler by 20% or more.)
4820 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4821 for the problem of printing dozens of lines of distracting information
4822 about low-level system machinery as part of the top stack frame
4823 on entry to the debugger when an undefined function was called.
4824 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4825 is not reset to COMMON-LISP-USER.
4826 * Compilation of stuff related to dyncount.lisp has been made conditional
4827 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4828 smaller. The various dyncount-related symbols have been moved into
4829 a new "SB-DYNCOUNT" package.
4830 * tty-inspect.lisp has been renamed to inspect.lisp.
4831 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4832 feature has gone away. (When we eventually port to other flavors of
4833 libc and/or Unix, we'll try to make the differences between flavors
4834 invisible at the user level.)
4835 * Various other *FEATURES* tags, and/or their associated conditionals,
4836 have been removed if obsolescent, or given better documentation, or
4837 sometimes given more-mnemonic names.
4839 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4841 * tidied up "make.sh" script
4842 * tidied up system directory structure
4843 * better "clean.sh" behavior
4844 * added doc/FOR-CMUCL-DEVELOPERS
4845 * many many small tweaks to output format, e.g. removing possibly-confusing
4846 trailing #\. character in DESCRIBE-INSTANCE
4847 * (EQUALP #\A 'A) no longer signals an error.
4848 * new hashing code, including EQUALP hashing
4849 * tidied up Lisp initialization and toplevel
4850 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4851 * command line argument processing
4852 * added POSIX-GETENV function to deal with Unix-ish environment variables
4853 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4854 terminating SBCL on EOF
4855 * non-verbose GC by default
4856 * There is no more "sbcl" shell script; the sbcl file is now the C
4857 runtime executable (just like CMU CL).
4858 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4860 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4861 symbol and package information as Lisp native symbol and package objects
4862 instead of strings naming symbols and strings naming packages. This way,
4863 whenever packages are renamed (as in warm init), debug information is
4864 transformed along with everything else.
4865 * tweaked the optimization policy declarations which control the building
4866 of SBCL itself. Now, among other things, the system no longer saves
4867 source location debugging information. (This helps two problems at once
4868 by reducing SBCL size and by keeping SBCL from trying to look for its
4869 sources -- which may not exist -- when reporting errors.)
4870 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4871 debugging and testing purposes
4872 * cleaned up printing, making the printer call PRINT-OBJECT for
4873 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4874 methods, giving nearly-ANSI behavior
4875 * converted almost all special variables to use *FOO* naming convention
4876 * deleted PARSE-TIME functionality, since it can be done portably
4877 * moved some files out of cold init into warm init
4878 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4880 * regularized formatting of source files
4881 * added an install.sh script
4882 * fixed ridiculous memory usage of cross-compiler by making
4883 compiler/alloc.lisp not try to do pooling unless it can hook
4884 itself into the GC of the cross-compilation host. Now the system
4885 builds nicely on my old laptop.
4886 * added :SB-ALLOC in target-features.lisp-expr
4887 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4888 was not implemented)
4889 * re-did condition handling and note reporting in the compiler. Notes
4890 are no longer handled by signalling conditions. Style warnings
4891 and warnings are handled more correctly and reported in such a way
4892 that it's easy to find one or the other in your output (so that you
4893 can e.g. figure out which of many problems caused COMPILE-FILE to
4895 * changed the severity of several compiler warnings from full WARNING
4896 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4897 compiler note reporting so that it doesn't use the condition system
4898 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4899 COMPILE-FILE command)
4900 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4901 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4902 no longer supported, since their implementation was deeply intertwingled
4903 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4904 * removed bogus "support" for compiler macros named (SETF FOO), and
4905 removed the compiler macro for SETF INFO (but only after making a fool
4906 of myself on the cmucl-imp mailing list by posting a bogus patch for
4907 DEFINE-COMPILER-MACRO..)
4908 * Compiled files containing forms which have side effects on the Lisp
4909 reader (such as DEFPACKAGE forms) are now handled more correctly.
4910 (Compiler queuing of top level lambdas has been suppressed by setting
4911 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4912 * deleted various currently-unused source files, e.g. gengc.lisp. They
4913 may be added back at some point e.g. when porting to other architectures,
4914 but until they are it's distracting to distribute them and to try to
4916 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4917 there were so many of them they're just distractions, and UNCROSS is
4918 known to be able to handle the current sources
4919 * moved PROFILE functionality into TRACE, so that it will be clear
4920 how the wrapping and unwrapping of functions when you profile them
4921 interacts with the wrapping and unwrapping of functions when you
4922 trace them. (Actually, the functionality isn't there yet, but at least
4923 the interface specification is there. Hopefully, the functionality will
4924 arrive with some future maintenance release.)
4925 * removed host-oops.lisp
4926 * changed signature of QUIT function to allow UNIX-CODE argument
4927 * fixed READ-SEQUENCE bug
4928 * tweaked verbose GC output so that it looks more like the progress
4929 output that ANSI specifies for functions like LOAD
4930 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4931 * added <http://sbcl.sourceforge.com> to the banner information printed by