1 ;;;; -*- coding: utf-8; -*-
2 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
3 appropriately typed structure slots without locking.
4 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
5 information like those gathered by TIME using a programming-friendly
7 * new feature: TIME reports time taken even if the form performs a
8 non-local transfer of control.
9 * enhancement: reduced conservativism on GENCGC platforms: on
10 average 45% less pages pinned (measured from SBCL self build).
11 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
12 mutate constant symbols or violate declaimed type of the symbol.
13 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
14 sleeping on the semaphore in heavily contested semaphores.
15 * bug fix: semaphores and condition variables were not interrupt
17 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
18 owned by other threads anymore.
19 * bug fix: FIND on lists called KEY outside the specified
20 subsequence. (reported by budden)
21 * bug fix: LOG doesn't use single-float intermediate results when
22 given mixed integer and double-float arguments, leading to better
23 precision. (reported by Bob Felts)
24 * bug fix: LOG with base zero returned values of inconsistent type.
25 * new feature: have the compiler track the effective type of a generic
26 function across method addition and removal.
28 changes in sbcl-1.0.20 relative to 1.0.19:
29 * minor incompatible change: OPTIMIZE qualities
30 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
31 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
32 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
33 * documentation: some slot access efficiency guidelines have been
34 added to the user manual.
35 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
36 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
38 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
39 to more efficient EQ-comparison versions more often.
40 * optimization: enhanced derivation of DOLIST iteration variable type
42 * optimization: constant folding of simple (LIST ...) forms as DOLIST
44 * optimization: runtime lookup of function definitions can be
45 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
46 * optimization: compiler is able to derive the return type of
47 (AREF (THE STRING X) Y) as being CHARACTER.
48 * optimization: CLRHASH on empty hash-tables no longer does pointless
49 work. (thanks to Alec Berryman)
50 * optimization: REPLACE deftransforms don't punt when :START1 or
52 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
53 no longer emits a code deletion note. (reported by Andrew
55 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
56 functions with an odd number of &OPTIONAL arguments, a &REST
57 argument, and one or more &KEY arguments at the call site.
58 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
59 arguments appeared at call sites as well.
60 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
61 READ-LINE &co can no longer read from them afterwards. (reported
63 * bug fix: fixed #427: unused local aliens no longer cause compiler
64 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
66 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
67 alien-stack leakage. (reported by Andy Hefner)
68 * bug fix: PROGV signals an error when an attempt to violate declared
69 type of a variable or bind a constant is made.
70 * bug fix: SET signals an error when an attempt to violate declared
71 type of a variable is made.
72 * bug fix: restart computation during the execution of a restart
73 test function no longer causes infinite recursion. (reported by
75 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
76 signals an error. (thanks to Pierre Mai)
77 * bug fix: EXPORT left symbol unexported in conflict situations.
78 (thanks to Michael Weber)
79 * bug fix: correctly handle name conflicts involving CL:NIL.
80 (thanks to Michael Weber)
81 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
82 conflicts arising from USEing package with conflicting symbols
83 correctly. (thanks to Michael Weber)
85 changes in sbcl-1.0.19 relative to 1.0.18:
86 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
87 warnings that go otherwise unhandled will be muffled if they are
88 of the type that's the value of this variable.
89 * optimization: stack allocation is slightly more efficient on x86
91 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
92 where a raw slot always is initialized using the initform whose
93 type is not know sufficiently well a compile-time are now compiled
94 correctly. (reported by John Morrison)
95 * bug fix: compiler no longer makes erronous assumptions in the
96 presense of non-foldable SATISFIES types.
97 * bug fix: stack analysis missed cleanups of dynamic-extent
98 arguments in non-let-converted calls to local functions.
99 * improvements to the Windows port:
100 ** adjusted address spaces for building on both Win32 and
101 Win64. (thanks for John Connors)
102 * fixed some bugs revealed by Paul Dietz' test suite:
103 ** interval arithmetic during type derivation used inexact integer
104 to single-float coercions.
105 ** arithmetic operations involving large integers and single
106 floats give the same results in compiled and interpreted code.
107 ** deriving the result type of COERCE no longer signals an error
108 if the derived type of the second argument is a MEMBER type
109 containing invalid type specifiers.
110 ** ADJOIN with constant NIL as second argument works correctly.
111 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
113 changes in sbcl-1.0.18 relative to 1.0.17:
114 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
115 profiles only the current thread.
116 * minor incompatible change: changes to SYMBOL-VALUE of constants
117 defined with DEFCONSTANT now signal an error.
118 * enhancement: SB-SPROF now has support for wallclock profiling,
119 and is also able to profile specific threads. REPORT output
120 has also additional sorting options.
121 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
123 * optimization: structure allocation has been improved
124 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
125 ** out of line constructors are ~10% faster.
126 ** inline constructors are ~15% faster.
127 ** inline constructors are capable of dynamic extent allocation
128 (generally on x86 and x86-64, in some cases on other platforms
130 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
132 * optimization: file compiler is now able to coalesce non-circular
133 lists, non-base strings, and bit-vectors. Additionally, constants
134 are never referenced through SYMBOL-VALUE at runtime.
135 * optimization: code defining methods on PRINT-OBJECT (and other
136 generic functions in the COMMON-LISP package) now loads faster.
137 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
139 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
140 fasl is now deleted. (reported by Attila Lendvai)
141 * bug fix: READ-LINE always returned NIL for the last line in files.
142 (reported by Yoshinori Tahara)
143 * bug fix: more accurate disassembly annotations of foreign function
144 calls. (thanks to Andy Hefner)
145 * bug fix: trimming non-simple strings and non-string string
146 designators when the there is nothing to trim works properly.
147 (thanks to James Knight)
148 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
149 (thanks to Travis Cross)
150 * fixed some bugs revealed by Paul Dietz' test suite:
151 ** NIL is a valid function name (regression at 1.0.13.38)
152 ** FILL on lists was missing its return value (regression at 1.0.12.27)
153 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
154 fill pointers properly (regression at 1.0.12.23)
155 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
156 displacement indices properly (regression at 1.0.12.23)
158 changes in sbcl-1.0.17 relative to 1.0.16:
159 * temporary regression: user code can no longer allocate closure
160 variable storage on stack, due to bug 419 without explicitly
161 requesting it. Please consult sbcl-devel for advice if you need to
162 use this feature in the meanwhile.
163 * new feature: runtime argument --control-stack-size can be used to
164 adjust thread default control stack size.
165 * enhancement: improved TIME output
166 ** all times are reported using the measured accuracy (milliseconds
167 for real and GC times, microseconds for everything else.)
168 ** processor cycle counts on x86 and x86-64.
169 ** interpreted forms are counted for both evaluator modes.
170 ** number of lambdas converted by the compiler is reported.
171 ** CPU percentage report (computed from real and total run time.)
172 ** more comprehensive run time reporting, using a condenced format
173 ** interperted form, lambda, and page fault counts are omitted
175 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
177 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
178 * optimization: two argument forms of LAST are upto ~10% faster
179 in normal SPEED policies.
180 * optimization: NCONC no longer needs to heap cons its &REST list
181 in normal SPEED policies.
182 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
184 * bug fix: UNION and NUNION work with :TEST-NOT once more,
185 regression since 1.0.9.1. (thanks to Eric Marsden)
186 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
187 since 1.0.15.36. (thanks to Paul Khuong)
188 * bug fix: LAST when always returned the whole list when given a bignum
189 as the second argument.
190 * bug fix: dynamic extent allocation of nested lists and vectors
191 could leak to otherwise accessible parts.
192 * bug fix: invalid optimization of heap-allocated alien variable
194 * bug fix: fasl header checking is less vulnerable to different
195 platform word lengths.
196 * bug fix: more correct assembler syntax for GNU binutils
197 2.18.50.0.4 support. (thanks to Marijn Schouten)
198 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
200 * bug fix: the fopcompiler can handle LOCALLY forms (with no
201 declarations) successfully. (reported by Attila Lendvai)
203 changes in sbcl-1.0.16 relative to 1.0.15:
204 * minor incompatible change: revert the changes to sb-posix's error
205 signaling added in 1.0.14.
206 * minor incompatible change: change PROBE-FILE back to returning
207 NIL whenever we can't get a truename, as was the case before 1.0.14.
208 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
209 inherits from ERROR instead of just CONDITION.
210 * new feature: SB-INTROSPECT can provide source locations for instances
211 as well. (thanks to Tobian Ritterweiler)
212 * optimization: binding special variables now generates smaller code
213 on threaded platforms.
214 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
215 and cases where no :TEST is given but the compiler can infer that
216 the element to search is of type (OR FIXNUM (NOT NUMBER)).
217 * optimization: better LOGNOT on fixnums.
218 * optimization: modular arithmetic for a particular requested width
219 is implemented using a tagged representation unless a better
220 representation is available.
221 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
222 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
223 for wellformedness and returns a specialized vector. (reported by
224 Francois-Rene Rideau)
225 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
226 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
227 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
228 unscheduling timer at the same time as another timer fires.
229 * bug fix: don't reschedule timers for dead threads.
230 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
231 * bug fix: copying output from RUN-PROGRAM to a stream signalled
232 bogus errors if select() was interrupted.
233 * enhancement: add support for fcntl's struct flock to SB-POSIX.
235 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
236 * enhancement: cleaner backtraces for interactive interrupts, as
237 well as other cases where the interesting frames used to be
238 obscured by interrupt handling frames.
239 * enhancement: untracing a whole package using (UNTRACE "FOO") is
240 now supported, and tracing a whole package using (TRACE "FOO") now
241 traces SETF-functions as well.
242 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
243 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
244 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
245 * unreadably printed representation of hash-tables now includes
247 * bug fix: partially fixed #188: type propagation from assignments
248 is now more efficient.
249 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
250 'DOUBLE-FLOAT) are not flushable.
251 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
252 full address of the object, and none of the tag bits.
253 * bug fix: readably printing hash-tables now respects other printer
254 control variables. (reported by Cedric St-Jean)
255 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
256 keyword with MAKE-HASH-TABLE.
257 * bug fix: export SB-POSIX:MKSTEMP.
258 * bug fix: SORT was not interrupt safe.
259 * bug fix: XREF accounts for the last node of each basic-block as
261 * bug fix: MAKE-INSTANCE optimizations interacted badly with
262 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
263 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
264 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
265 create obsolete instances in the case of redefinition or
266 obsoletion of a superclass. (thanks to Andy Hefner)
267 * bug fix: Support for the Alpha architecture has been revived; it had
268 suffered somewhat from lack of maintenance since sbcl-1.0.
269 * improvements to the Windows port:
270 ** The system detects the codepage to use at startup, instead of
271 using the value from the saved core. (thanks to Kei Suzuki)
273 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
274 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
275 (see documentation for details.)
276 * revived support for OpenBSD (contributed by Josh Elsasser)
277 * partially fixed bug #108: ROOM no longer suffers from occasional
278 (AVER (SAP= CURRENT END)) failures.
279 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
280 bodies no longer cause a WARNING to be signalled. (reported by
282 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
283 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
284 the bug in TRUNCATE.) (reported by Michael Weber)
285 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
286 no samples. (reported by Andy Hefner)
287 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
288 no longer appear as (NIL ...) frames in backtraces.
289 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
290 is now actually exported from SB-EXT as documented. (reported by
292 * bug fix: sb-aclrepl now correctly understands how to inspect
293 single-floats on 64-bit platforms where single-floats are not boxed.
294 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
295 yet been finalized. (reported by Levente Meszaros)
296 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
297 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
298 SBCL-specific optimize qualities.
300 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
301 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
302 an executable in the search path, and does so in the child
303 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
304 been removed; it can be found in the CVS history, for somebody who
305 needs that search behavior (see the manual).
306 * minor incompatible change: compiler policy re. weakening type
307 checks has changed: now type checks are weakened only if SAFETY < 2
309 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
310 unparsing of directory pathnames as files. Analogously,
311 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
312 filename to parse into a directory pathname.
313 * enhancement: implicit generic function creation now signals a
314 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
315 users can bind handlers for and muffle around calls to LOAD.
316 * enhancement: RUN-PROGRAM allows unicode arguments and environments
317 to be used (using the default stream external format), and allows
318 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
319 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
320 strings and vectors whose element-type or simplicity is not fully
321 known at compile-time.
322 * optimization: STRING-TRIM and related functions no longer allocate
323 a new string when no trimming needs to be performed. These functions
324 are also faster than before when the input string has been declared
326 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
327 * optimization: READ-LINE is significantly faster for files containing
329 * optimization: non-open coded uses of character comparison operators
330 (e.g. char=) no longer cons when called with more than one parameter
331 on platforms supporting dynamic-extent allocation.
332 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
333 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
334 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
336 * bug fix: some sequence functions elided bounds checking when
338 * bug fix: too liberal weakening of union-type checks when SPEED >
340 * bug fix: more bogus fixnum declarations in ROOM implementation
343 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
344 * new feature: MAKE-HASH-TABLE now experimentally accepts a
345 :SYNCHRONIZED argument, which makes the hash-table safe for
346 concurrent accesses (but not iteration.) See also:
347 SB-EXT:WITH-LOCKED-HASH-TABLE, and
348 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
349 * optimization: CONCATENATE on strings is an order of magnitude faster
350 in code compiled with (> SPEED SPACE).
351 * optimization: SUBSEQ is ~50% faster on lists.
352 * bug fix: bug 417 fixed -- source location reporting is now more robust.
353 * bug fix: SUBSEQ on a list will now correctly signal an error if if
354 END is smaller then START.
355 * bug fix: SB-PROFILE will no longer report extra consing for nested
356 calls to profiled functions.
357 * bug fix: ROOM implementation had bogus fixnum declarations which
358 could cause type-errors when calling ROOM on large images.
359 * bug fix: if file compilation is aborted, the partial fasl is now
360 deleted, and COMPILE-FILE returns NIL as the primary value.
361 * bug fix: number of thread safety issues relating to SBCL's internal
362 hash-table usage have been fixed.
363 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
364 be returned from its body when the values were being returned
365 using unknown-values return convection and the W-P-O was wrapped
366 inside an UNWIND-PROTECT.
367 * bug fix: sb-posix should now compile again under Windows, enabling
370 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
371 * incompatible change: hash-table accessor functions are no longer
372 automatically protected by locks. Concurrent accesses on the same hash-table
373 from multiple threads can give inconsistent results or even corrupt the
374 hash-table completely. Multi-threaded applications should do their own
375 locking at the correct granularity. In the current implementation it is
376 still safe to have multiple readers access the same table, but it's not
377 guaranteed that this property will be maintained in future releases.
378 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
379 and will signal an error at runtime.
380 * enhancement: SB-THREAD package now exports a semaphore interface.
381 * enhancement: CONS can now stack-allocate on x86 and
382 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
384 * enhancement: nested lists can now be stack allocated on
385 platforms providing stack allocation support.
386 * enhancement: dynamic-extent support has been extended to support
387 cases where there are multiple possible sources for the stack
389 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
390 if the mutex is uncontested on Linux.
391 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
392 as specified by the CLHS. (thanks to Tobias Rittweiler)
393 * bug fix: NaN comparison now works on x86-64.
394 * bug fix: CLOSE :ABORT T on a stream with pending output now
396 * bug fix: instances of non-standard metaclasses using standard
397 instance structure protocol sometimes missed the slot type checks
399 * bug fix: known functions can be inlined in high-debug code.
401 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
402 * minor incompatible change: the MSI installer on Windows no longer
403 associates .lisp and .fasl files with the installed SBCL.
404 * minor incompatible change: :UNIX is no longer present in *FEATURES*
405 on Windows. (thanks to Luis Oliviera)
406 * new feature: SB-CLTL2 contrib module now implements
407 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
408 * optimization: scavenging weak pointers is now more efficient,
409 requiring O(1) instead of O(N) per weak pointer to identify
410 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
411 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
412 method bodies using constant slot-names when the first argument is
413 a specializer parameter for the method.
414 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
415 names, when the class of the instance is a direct instance
416 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
417 as slow as the constant slot-name case.)
418 * optimization: member type construction is now O(N) instead
420 * optimization: UNION and NUNION are now O(N+M) for large
421 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
423 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
424 now more readable in environments like Slime which display it.
425 (thanks to Tobias C. Rittweiler)
426 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
427 was able to infer the correct argument type for the object on which
428 the CAS operation was being performed.
429 * bug fix: copy propagation interfered with parallel assignment
430 semantics in local calls. (reported by Paul Khuong)
431 * bug fix: the signed modular fixnum shift compiled to wrong code on
432 x86 and x86-64. (spotted by a slight modification to some of
435 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
436 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
437 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
438 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
440 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
441 and generic function now print the object identity for anonymous
442 instances. (thanks to Attila Lendvai)
443 * enhancement: as an extension to MOP, SBCL now supports SETF of
444 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
445 (thanks to Attila Lendvai)
446 * enhancement: sb-cover has an alternate annotation mode which puts
447 the form annotations at the car of the form, instead of annotating
448 it completely. The benefit of this mode is that it shows better
449 which forms were instrumented by the compiler.
450 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
451 by printer control variables. (reported by Dan Corkill)
452 * bug fix: system leaked memory when delayed output was performed by
453 the OS in smaller chunks then expected. (thanks to David Smith)
454 * bug fix: system leaked memory when file streams were not closed
456 * bug fix: large objects written to slow streams that were modified
457 after the write could end up with the modified state written to
458 the underlying file descriptor.
459 * bug fix: multiple threads operating in parallel on the same stream
460 could cause buffer-overflows.
461 * bug fix: source location information is stored correctly
462 (broken since 1.0.6). This bug would generally show up as the
463 Slime debugger highlighting the wrong form when the sldb-show-source
465 * bug fix: Branch forms are again annotated as branches in the sb-cover
467 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
468 on x86-64. (reported by Gregory Vanuxem)
469 * bug fix: EQUALP could return wrong results for structures with raw
470 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
471 word). (reported by Vjacheslav Fyodorov)
472 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
475 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
476 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
477 atomic compare-and-swap operations on threaded platforms.
478 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
479 allows assigning a global minimum value to optimization qualities
480 (overriding proclamations and declarations).
481 * enhancement: closed over variables can be stack-allocated on x86
483 * performance bug fix: GETHASH and (SETF GETHASH) are once again
485 * optimization: slot definition lookup is now O(1). This speeds up
486 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
487 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
488 * optimization: ASSOC and MEMBER can now be open-coded for all
489 combinations of keyword arguments when second argument is constant
490 and SPEED >= SPACE. In other cases a specialized version is
492 * bug fix: using obsoleted structure instances with TYPEP and
493 generic functions now signals a sensible error.
494 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
495 (reported by Kristoffer Kvello)
496 * bug fix: backtrace construction is now more careful when making
497 lisp-objects from pointers on the stack, to avoid creating bogus
498 objects that can be seen by the GC.
499 * bug fix: defaulting of values in contexts expecting more than 7
500 variables now works on x86-64. (reported by Christopher Laux)
501 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
503 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
504 as the property-list of a symbol.
505 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
506 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
507 situations requested, are once again file-compileable. (reported
510 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
511 * MOP improvement: support for user-defined subclasses of
512 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
513 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
514 the expansion of DEFMETHOD, is responsible for generating a form
515 which creates a list of specializers when evaluated. Additional
516 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
517 debugging and introspective support.
518 * minor incompatible change: the (unsupported) spinlock interface
519 has changed: free spinlock now has the value NIL, and a held spinlock
520 has the owning thread as its value.
521 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
522 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
524 * enhancement: name of a socket-stream is now "a socket" instead of
526 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
527 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
528 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
529 for generic functions. (Thanks to Tobias C. Rittweiler)
530 * optimization: bignum printing speed has been improved by 20-40%
531 (depending on the bignum size.)
532 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
534 * bug fix: the cache used by the CLOS to store precomputed effective
535 methods, slot offsets, and constant return values is now thread and
537 * bug fix: generic function dispatch function updating is now thread
538 and interrupt safe (in the sense that the known issues have been
540 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
541 * bug fix: interning EQL-specializers is now thread and interrupt safe.
542 * bug fix: asdf systems with dependencies to the SB-POSIX or
543 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
544 * bug fix: interrupt safety of applicable method computation has been
547 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
548 * new contrib: sb-cover, an experimental code coverage tool, is included
550 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
551 significantly faster.
552 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
553 produce more efficient code on x86-64 if the compiler can determine
554 that (+ INDEX <constant>) does not require a bounds check and FOO
555 has an element type at least 8 bits wide.
556 * enhancement: a new, experimental synchronous timeout facility is
557 provided. Refer to SB-SYS:WITH-DEADLINE for details.
558 * enhancement: when a symbol name conflict error arises, the
559 conflicting symbols are always printed with a package prefix.
560 (thanks to Kevin Reid)
561 * enhancement: stepping is now once again supported on the SPARC and
562 MIPS platforms. (It is also now more likely to work on CheneyGC
564 * enhancement: sb-sprof can now also track and report accurate call
566 * bug fixes: the treatment of non-standard subclasses of
567 SB-MOP:SPECIALIZER is more correct.
568 * incompatible change: PURIFY no longer copies the data from the
569 dynamic space into the static and read-only spaces on platforms
570 that use the generational garbage collector
571 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
573 * bug fix: GC race condition occasionally resulting in crashes with
574 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
576 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
578 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
579 a STYLE-WARNING to be signalled (regression from 1.0.4.)
580 * bug fix: an asynchronous interrupt could previously leave the
581 system running with GC inhibited.
582 * bug fix: a DECLARE form evaluated at top-level now causes an error
583 rather than silently (or verbosely) returning NIL.
584 * bug fix: trying to dissassemble functions compiled at high DEBUG could
585 result in a error being signalled due to source form lookup errors.
586 (reported by Peter Graves)
588 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
589 * incompatible change: removed writer methods for host-ent-name,
590 host-ent-addresses -- changing the values did not update the DNS
592 * minor incompatible change: changed experimental JOIN-THREAD interface
593 * documentation: the manual now lists reader and writer methods
594 in class slot documentation sections. (thanks to Richard M Kreuter)
595 * documentation: unwinding from asyncronous events has been
596 documented as unsafe.
597 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
598 in multithreaded application code.
599 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
600 platforms. (thanks to James Anderson for the optimization hint)
601 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
603 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
604 variants no longer cons.
605 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
606 their NOT- variants no longer cons.
607 * optimization: EQUAL hash tables no longer use SXHASH for objects
608 of all data types, but instead use an EQL hash for types for which
609 EQUAL is the same as EQL.
610 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
611 are significantly faster.
612 * optimization: new STRING-OUTPUT-STREAM implementation is much
613 faster for large outputs and conses 30% less on average.
614 * enhancement: XREF information is now collected for references made
615 to global variables using SYMBOL-VALUE and a constant argument.
616 * enhancement: SIGINT now causes a specific condition
617 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
618 * bug fix: FILE-POSITION used to signal an error in some cases where
619 ANSI requires it to return NIL.
620 * bug fix: ADJUST-ARRAY is now interrupt-safe.
621 * bug fix: adding and removing fd-handlers is now interrupt-safe.
622 * bug fix: inlined calls to C now ensure 16byte stack alignment on
624 * bug fix: bad type declaration in the CLOS implementation has
625 been fixed. (reported by James Anderson)
626 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
627 (thanks to Lutz Euler)
628 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
629 required by the CLHS. (reported by Samium Gromoff)
630 * bug fix: dead unbound variable references now signal an error.
631 (reported by Marco Monteiro)
632 * bug fix: / with an unused value was being deleted in safe code.
633 (thanks to Marco Monteiro and Kevin Reid)
634 * bug fix: number of characters that can be written onto a single
635 line in a file is unlimited.
636 * bug fix: some GC deadlocks caused by asynchronous interrupts have
637 been fixed by inhibiting interrupts for when GC is disabled.
638 * bug fix: some interrupt safety issues with GETHASH, (SETF
639 GETHASH), CLRHASH and REMHASH have been fixed.
640 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
641 specifier no longer causes infinite recursion.
642 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
643 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
644 * bug fix: an &environment argument with the correct variable information
645 is passed to macros that are expanded during byte compilation.
646 (reported by Samium Gromoff)
647 * bug fix: a logic error was causing the hash function for bignums to
648 have excessive amounts of collisions. (reported by Faré Rideau)
649 * bug fix: modifying the contents of an array could change the return
650 value of SXHASH on that array, which is only allowed for strings
651 and bit vectors. (bug introduced in 0.9.16)
652 * improvement: the x86-64/darwin port now passes all tests (except
653 for the debugger tests) but should still be considered
654 experimental until this is fixed.
655 * improvement: a style-warning is signaled for CASE (etc) clauses with
656 duplicate keys. (thanks to Kevin Reid)
657 * improvement: macos/x86 and macos/x86-64 now use mach exception
658 handlers for dealing with illegal instructions (for trapping and
659 error handling) and memory protection violations (for GC).
661 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
662 * new platform: experimental support for x86-64/darwin (MacOS).
663 * incompatible change: the thread-safe (on most platforms) getaddrinfo
664 and getnameinfo sockets functions are used instead of gethostbyaddr
665 and gethostbyname, on platforms where the newer functions are available.
666 As a result, the ALIASES field of HOST-ENT will always be NIL on these
668 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
669 don't print the startup banner, but behave as if --noinform was passed
670 as a command line argument. (thanks to Kevin Reid)
671 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
672 * optimization: code using alien values with undeclared types is much faster.
673 * optimization: the compiler is now able to open code SEARCH in more cases.
674 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
675 * bug fix: using standardized COMMON-LISP special variables as loop
676 variables no longer signals bogus package lock violations. (reported
678 * bug fix: declaring local loop variables to be of a range-limited type
679 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
680 (reported by Andras Simon)
681 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
682 bugs remain on x86-64.)
683 * bug fix: the #= and ## reader macros now interact reasonably with
684 funcallable instances.
685 * bug fix: type-checks for function arguments were compiled using the
686 compiler policy settings of the wrong lexical environment (in 1.0.2
688 * bug fix: SHADOW accepts characters as string designators, as required
689 by the spec (thanks to Eric Marsden)
690 * bug fix: fixed GC safety issues when foreign functions are called with
691 non-base strings as arguments
692 * bug fix: more consistent error output for fatal-compiler-warnings, like
694 * bug fix: functions with proper names were showing up as NIL in
695 backtraces (reported by Edi Weitz, regression in 1.0.2)
697 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
698 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
699 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
700 produce more efficient code on the x86 if the compiler can determine
701 that (+ INDEX <constant>) does not require a bounds check and FOO
702 has an element type at least 8 bits wide.
703 * bug fix: references to undefined variables are handled the same way
704 in toplevel forms as in the normal compiler
705 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
707 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
708 (thanks to Jon Buller)
709 * improvement: faster compilation times for complex functions
710 * improvement: added readlink support to SB-POSIX (thanks to Richard
713 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
714 * improvement: experimental support for mach exception handling on
715 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
717 * improvement: support for GBK external format.
718 (thanks to Chun Tian (binghe))
719 * improvement: the debugger now displays variables that have been closed
720 over, in code compiled with (DEBUG 2) or higher
721 * improvement: support for executable cores on NetBSD (thanks to
723 * new feature: added a RESTART-FRAME debugger command
724 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
725 be used to provide an implementation for FILE-POSITION on Gray streams
726 (thanks to Eric Marsden)
727 * new feature: add syslog, openlog and closelog support to SB-POSIX
728 (thanks to Richard Kreuter)
729 * optimization: the function call overhead in code compiled with
730 a high DEBUG optimization setting is significantly
731 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
732 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
733 (thanks to Tony Martinez)
734 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
735 evaluated once (reported by Kevin Reid)
736 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
737 works again (thanks to Kilian Sprotte)
738 * bug fix: an error is signaled for tagbodies with duplicate tags
739 (thanks to Stephen Wilson)
740 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
741 * bug fix: Win32 port can now handle foreign code unwinding Lisp
742 stack frames from alien callbacks.
743 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
744 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
745 65k pages (thanks to David Woodhouse)
746 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
748 changes in sbcl-1.0.1 relative to sbcl-1.0:
749 * new platform: FreeBSD/x86-64, including support for threading.
750 * new feature: the compiler stores cross-referencing information
751 about function calls (who-calls), macroexpansion (who-macroexpands)
752 and special variables (who-binds, who-sets, who-references) for code
753 compiled with (< SPACE 3). This information is available through the
754 sb-introspect contrib.
755 * new feature: users may subclass SEQUENCE, and have instances of
756 these classes interoperate with standard Common Lisp functions if
757 a number of methods are defined. (This feature is experimental
758 and the interface subject to change based on feedback from SBCL
759 users and the general community)
760 * improvement: sb-sprof traces call stacks to an arbitrary depth on
761 x86/x86-64, rather than the previous fixed depth of 8
762 * improvement: another pthread back-end of mutex "pthread-futex".
763 * bug fix: non-ascii command-line arguments are processed correctly
764 (thanks to Yaroslav Kavenchuk)
765 * bug fix: non-required arguments were not passed correctly when a method
766 defined using DEFMETHOD was called from a mop-generated method using
767 CALL-NEXT-METHOD (reported by Pascal Costanza)
768 * bug fix: recursion is now permitted in accessors through
769 SLOT-UNBOUND. (reported by Pascal Costanza)
770 * bug fix: an error was signaled at startup if the HOME environment
771 variable was defined, but had an empty value (reported by Peter Van Eynde)
772 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
773 signal a PROGRAM-ERROR, not a BUG.
774 * bug fix: SB-POSIX:READDIR works when built with large file support.
775 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
777 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
778 for better automatic indentation support. (thanks to Matt Pillsbury)
779 * optimization: loading generic functions no longer takes O(n^2) time,
780 proportional to the amount of methods in the generic function
781 (reported by Todd Sabin and Jeremy Brown)
782 * optimization: the FIND and POSITION family of sequence functions
783 are significantly faster on arrays whose element types have been
785 * improvements to the Windows port:
786 ** Intermittent heap corruption problems have been fixed. (thanks
787 to Alastair Bridgewater)
788 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
790 ** Lisp is able to unwind foreign exception frames from alien
791 callbacks. (thanks to Alastair Bridgewater)
793 changes in sbcl-1.0 relative to sbcl-0.9.18:
794 * improvement: experimental support for threading on FreeBSD/x86.
795 (thanks to NIIMI Satoshi)
796 * improvement: runtime option --dynamic-space-size can be used
797 to set the size of the dynamic space reserved on startup.
798 * improvement: floating point modes in effect are now saved in
799 core, and restored on startup.
800 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
801 startup, not time since first call to GET-INTERNAL-REAL-TIME.
802 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
803 threads are not running after *SAVE-HOOKS* have run.
804 * improvement: writes to CLOS instance slots are type-checked in code
805 compiled with (SAFETY 3)
806 * improvement: floating-point exception handling on FreeBSD (thanks to
808 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
809 (thanks to Zach Beane)
810 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
812 * improvement: added support for the Shift-JIS external format.
813 (contributed by NIIMI Satoshi)
814 * improvement: callbacks are supported on Linux/PPC. (thanks to
816 * bug fix: compiler bug triggered by a (non-standard) VALUES
817 declaration in a LET* was fixed. (reported by Kaersten Poeck)
818 * bug fix: file compiler no longer confuses validated and already
819 dumped structurres. (reported by Kaersten Poeck)
820 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
821 fill-pointer signals a type-error as required. (thanks to
823 * bug fix: disassemly of funcallable instances works.
824 * bug fix: single stepping on PPC.
825 * bug fix: fix thread-safety problems in the type system (generally
826 manifesting as nonsensical errors like "STRING is a bad type specifier
827 for sequences" or "The value 1 is not of type FIXNUM")
828 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
829 (reported by Josip Gracin)
830 * bug fix: an error is signaled for attempts to displace arrays with
831 incompatible element types (thanks to Mario Mommer)
832 * bug fix: more correct handling of wide characters in debug info
833 (bug reported by Attila Lendvai and fixed by Juho Snellman)
834 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
835 and don't cause extra consing
836 * optimization: MAP and MAP-INTO are significantly faster on vectors
837 whose elements types have been declared.
838 * Improvements to SB-SPROF:
839 ** Support for allocation profiling
840 ** Reduced profiling overhead, especially for long profiling runs
841 * Improvements to the Windows port:
842 ** floating point exceptions are now reported correctly.
843 ** stack exhaustion detection works partially.
844 ** more accurate GET-INTERNAL-REAL-TIME.
845 ** better RUN-PROGRAM behaviour in case of errors in the child
847 ** PROBE-FILE now simplifies pathnames correctly.
848 ** DIRECTORY now works correctly with :WILD-INFERIORS.
849 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
850 1st 1970 (although time timezone data will be unreliable).
851 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
853 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
854 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
855 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
857 * bug fix: two potential GC deadlocks affecting threaded builds.
858 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
859 returns T (reported by Anton Kazennikov)
860 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
861 with non-variable places
862 * bug fix: remove a race condition in the setting of
863 funcallable-instance functions, this should make threaded CLOS
864 code more stable against memory faults.
865 * bug fix: corruption of specials when unbinding is interrupted by an
866 asynchronous unwind (reported by Hannu Koivisto)
867 * improvement: the debugger will now also display local variables that
868 are only used once, for code compiled with a DEBUG optimization quality
871 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
872 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
873 * incompatible change: External-format support for FFI calls. The
874 SB-ALIEN:C-STRING no longer implies an ASCII
875 external-format. Instead the string is subject to external-format
876 conversion. Additionally return values of type SB-ALIEN:C-STRING
877 are by default of type (SIMPLE-ARRAY CHARACTER), not
878 SIMPLE-BASE-STRING. For an alien type that behaves like the old
879 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
880 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
881 * incompatible change: SB-EXT package no longer contains the
882 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
883 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
885 * incompatible change: the single-stepper is no longer available
886 on Alpha, Mips and Sparc platforms.
887 * minor incompatible change: the direct superclasses of
888 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
889 not (STANDARD-OBJECT FUNCTION). This makes the
890 class-precedence-lists of GENERIC-FUNCTION and
891 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
893 * minor incompatible change: the default stream external format on
894 non-unicode SBCL is no longer detected from the locale, but is always
896 * new feature: Add a version of evaluator that uses an interpreter instead
897 of the compiler. EVAL still uses the compiler by default, to switch it
898 to use the interpreter, set the value of the variable
899 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
900 * minor incompatible change: the single-stepper REPL has been merged
901 with the normal debugger (see the "Stepping" heading of the debugger help
902 for more details). The debugger command STEP will no longer switch
903 to the single-stepper REPL.
904 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
905 for a type now works.
906 * bug fix: loading of default sysinit file works. (thanks to Leonid
908 * bug fix: better detection of circularities in the file-compiler.
909 (reported by Marco Monteiro)
910 * bug fix: the CL pathname functions now work with files that have
911 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
912 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
913 systems with Unix98 pty semantics.
914 * bug fix: ASDF-INSTALL will now work with bsd tar.
915 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
917 * bug fix: timers expiring in dead threads no longer cause a
918 type-error (reported by Paul "Nonny Mouse").
919 * bug fix: thanks to more lightweight single-stepper instrumentation,
920 code compiled with (DEBUG 3) will compile and execute significantly faster,
921 and will have more accurate type-inferencing than before
922 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
923 whose bindings are modified
924 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
925 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
926 ** replace dummy implementations of CL:MACHINE-INSTANCE and
927 CL:SOFTWARE-VERSION with proper version
929 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
930 * feature: implemented the READER-METHOD-CLASS and
931 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
932 as specified by AMOP.
933 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
935 * optimization: faster LOGCOUNT implementation on x86 and x86-64
936 (thanks to Lutz Euler)
937 * optimization: hashing of general arrays and vectors has been
938 improved. (reported by Any Fingerhut)
939 * enhancement: SB-INTROSPECT is now able to find definitions of
940 profiled functions. (thanks to Troels Henriksen)
941 * enhancement: compiler-macro expansion applies now to FUNCALL forms
942 as well, allowing compiler-macros for SETF-functions to expand.
943 * enhancement: step-instrumentation no longer wraps known
944 single-value functions with multiple-value context, allowing
945 better type inference.
946 * fixed bug #337: use of MAKE-METHOD in method combination now works
947 even in the presence of user-defined method classes. (reported by
948 Bruno Haible and Pascal Costanza)
949 * fixed bug #339(c): if there are applicable methods not part of any
950 long-form method-combination group, call INVALID-METHOD-ERROR.
951 (reported by Bruno Haible)
952 * fixed bug #361: the :FUNCTION initarg in the protocol for
953 initialization of methods can now be used to override
954 internally-produced optimized functions. (reported by Bruno
956 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
957 system-provided lambda expression no longer cause warnings about
958 unbound #:|pv-table| symbols.
959 * bug fix: improved the handling of type declarations and the
960 detection of violations for keyword arguments with non-constant
962 * bug fix: potentially erronous calls to PATHNAME and
963 MERGE-PATHNAMES were being flushed in some cases.
964 (reported by Richard Kreuter)
965 * bug fix: compiled calls to TYPEP were mishandling obsolete
966 instances. (reported by James Bielman and Attila Lendvai)
967 * bug fix: format strings with ~> without matching ~< no longer
968 trigger an AVER, but signal an understandable error instead.
969 (reported by Antonio Martinez)
970 * bug fix: specifying an output-file without a pathname-type for
971 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
972 of input-file instead of "fasl". (reported by Robert Dodier)
973 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
974 (reported by James Y Knight).
975 * bug fix: compiler-macros-function did not consider the environment
976 argument for shadowing by local functions.
977 * bug fix: compiler-macros expansion was inhibited by local INLINE
979 * bug fix: inline expansions of known functions were subject to
980 step-instrumentation in high debug policies, leading to problems
982 * bug fix: compiler failed to differentiate between different CONS
984 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
985 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
986 * bug fix: #368: incorrect use of expressed vs. upgraded array
988 * bug fix: #306a: more precise unions of array types.
989 * thread-safety improvements:
990 ** CONDITION-WAIT could return early on Linux, if the thread was
991 interrupted and subsequently continued with SIGCONT.
992 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
994 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
995 * added support for the ucs-2 external format. (contributed by Ivan
997 * minor incompatible change: pretty printing of objects of type
998 (cons symbol) is, in the default pprint-dispatch-table, now
999 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1001 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1002 called later in a class's lifetime, possibly as late as when the
1003 first instance of the class is created. Previously,
1004 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1005 class became finalizeable.
1006 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1007 for multibyte external-format streams. (thanks to "vbzoli")
1008 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1009 with :ALLOCATION :CLASS inherited from superclasses of the
1011 * fixed bug: anonymous classes can now be created using the :NAME
1012 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1013 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1014 * fixed bug: core-files saved with :EXECUTABLE T can again be
1015 executed when SBCL_HOME isn't set. (reported by James Knight)
1016 * fixed bug: toplevel LOCALLY forms with declarations could
1017 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1018 * fixed bug: printing from several different threads using different
1019 values of *print-case* could cause invalid output, due to
1020 some internal special variables of the printer not being bound
1021 thread-locally (reported by Max Mikhanosha)
1022 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1023 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1024 (reported by Pascal Costanza)
1025 * fixed bug: FINALIZE-INHERITANCE is called from
1026 REINITIALIZE-INSTANCE on classes when the class has previously
1027 been finalized, as required by AMOP.
1028 * minor code generation optimizations:
1029 ** better register allocation in CLOS dispatching functions
1030 ** overflow detection when coercing signed bytes to fixnums on x86-64
1031 ** is now implemented with one IMUL instruction instead of three shifts
1032 ** more efficient bit-vector access on x86 and x86-64
1033 ** more efficient access to raw structure slots on x86 and x86-64
1034 * fixed some bugs revealed by Paul Dietz' test suite:
1035 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1036 return its argument.
1038 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1039 * feature: thread support on Solaris/x86, and experimental thread support
1041 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1043 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1044 default method for SLOT-UNBOUND.
1045 * minor incompatible change: prevent the user from specializing the
1046 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1047 somewhat counter to the intent of the protocol, I (CSR) think, and
1048 additionally it just doesn't work in SBCL as currently
1049 implemented, thanks to optimizations (that are always valid for
1050 the other three SLOT-VALUEish functions, but not for the setter).
1051 * minor incompatible change: the :SB-LDB feature is now enabled by
1052 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1053 the low-level debugger.
1054 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1055 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1056 down to an SB-SYS:FD-STREAM, allowing direct communication in
1057 more cases. Temporary files and pipes are still used as a fallback
1059 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1060 on the non-reentrant C functions localtime(3) and gmtime(3).
1061 * bug fix: global optimization policy was not visible in LOCALLY and
1063 * bug fix: class objects can be used as specializers in methods.
1064 (reported by Pascal Costanza)
1065 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1066 with a type error. (reported by blitz_ on #lisp)
1067 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1068 failed with a type error. (reported by Pascal Bourguignon)
1069 * bug fix: merging pathnames against defaults with :DIRECTORY
1070 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1071 (reported by James Y Knight)
1072 * bug fix: saving large (>2GB) cores on x86-64 now works
1073 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1074 constant index and a (simple-array (signed-byte 32)) array
1075 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1076 error (patch by Robert J. Macomber)
1077 * feature: TIME now displays the amount of run-time spent in GC
1078 * bug fix: The debugger now does a better job of respecting
1079 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1080 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1081 workaround for bug 403.)
1082 * bug fix: floating point exception handling now works on FreeBSD
1083 (thanks to NIIMI Satoshi)
1084 * fixed some bugs revealed by Paul Dietz' test suite:
1085 ** MISC.641: LET-conversion were not supposed to work in late
1087 * improvements to the Win32/x86 port:
1088 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1089 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1091 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1092 * new feature: source path information is generated for macro-expansion
1093 errors for use in IDE's like Slime (thanks to Helmut Eller)
1094 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1095 * bug fix: compiler does not loop forever on an invalid type in
1097 * improvement: compilation of most CLOS applications is significantly
1099 * optimization: added a limited bytecode compiler for simple toplevel
1100 forms, speeding up compilation and FASL loading
1101 * bug fix: the statistical profiler now properly distinguishes anonymous
1104 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1105 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1106 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1107 into subclasses of the finalized class. Now user calls to
1108 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1109 system return before any subclasses are finalized.
1110 * minor incompatible change: The reader no longer ignores errors
1111 regarding non-existent packages in #+ and #- feature tests.
1112 * new feature: command line options --no-sysinit, --no-userinit to
1113 inhibit loading the corresponding init files
1114 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1115 for sending data through UDP sockets (thanks to François-René Rideau)
1116 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1117 error is signalled instead (thanks to François-René Rideau)
1118 * minor incompatible change: Error signalling behaviour of lexical
1119 operations violating package locks has changed slightly. Refer to
1120 documentation on package locks for details.
1121 * enhancement: EVAL can process IF-expressions without resorting to the
1123 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1124 constant-folded properly due to missing out-of-line functions.
1125 (reported by Mika Pihlajamäki)
1126 * bug fix: LISTEN sometimes returned T even in cases where no data was
1127 immediately available from the stream
1128 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1129 were reversed. (reported by Levente Mészáros)
1130 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1131 when given a vector displaced to an adjustable array. (reported
1133 * bug fix: derivation of float boundaries from numbers outside the
1134 appropriate float range (reported by John Wiseman)
1135 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1137 * bug fix: precomputing cache entries for generic functions with
1138 some subclasses of specializers as yet invalid does not attempt to
1139 fill a cache line with a negative offset. (reported by Levente
1141 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1142 allow condition class objects as arguments to DOCUMENTATION and
1143 (SETF DOCUMENTATION); only find and set documentation for
1144 structure names for the STRUCTURE doc-type. (suggested by Gary
1146 * improvements to the Win32/x86 port:
1147 ** (user-homedir-pathname) and default initialization file
1148 locations now know about the user's "Documents and Settings"
1149 directory (thanks to Yaroslav Kavenchuk)
1150 ** run-program is implemented (thanks to Mike Thomas)
1151 ** sockets support (thanks to Timothy Ritchey)
1152 ** better backtrace support (thanks to Alastair Bridgewater)
1153 ** sb-grovel supported
1154 ** asdf-install and sb-posix work somewhat
1155 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1156 * improvements to the Solaris/x86 port:
1157 ** works on Solaris 11/Solaris Express
1158 ** floating-point exception handling support
1159 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1160 * fixed some bugs revealed by Paul Dietz' test suite:
1161 ** REMOVE-METHOD returns its generic function argument even when
1162 the method is not one of the generic functions' methods.
1163 ** objects declared as MEMBER types can be admissible arguments to
1164 structure accessors.
1165 ** printing characters should simply be printed by the FORMAT ~:C
1167 ** compiler failure when compiling functions with hairy constant
1168 defaults for optional parameters.
1169 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1170 function, which is already optimized.
1172 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1173 * new platform: experimental support for SBCL x86/Darwin, including
1174 MacOS X 10.4.5 on Intel.
1175 * new feature: Unicode character names are now known to the system
1176 (through CHAR-NAME and NAME-CHAR).
1177 * new feature: the filesystem location of SBCL's core file is
1178 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1179 * minor incompatible change: the contrib modules SB-POSIX and
1180 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1181 this change is to make it easier to distribute
1182 location-independent binaries.
1183 * bug fix: as implied by AMOP, standardized classes no longer have
1184 slots named by external symbols of public packages. (reported by
1186 * optimization: calling functions via constant symbols -- as in
1187 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1188 via the function object as in (FUNCALL #'FOO).
1189 * optimization: on x86 and x86-64, the calling convention for the
1190 case of dealing with an unknown number of values has been altered
1191 to be friendlier to the prediction heuristics implemented,
1192 particularly with reference to CALL and RET pairing. (thanks to
1193 Alastair Bridgewater)
1194 * optimization: on x86, the code for access to thread-local symbol
1195 values and binding/unbinding of thread-local symbols is smaller.
1196 (thanks to Alastair Bridgewater)
1197 * enhancement: CONSTANTP is now able to determine constantness of
1198 more complex forms, including calls to constant-foldable standardized
1199 functions and some special forms beyond QUOTE.
1200 * fixed bug: occasional GC crashes on Solaris/x86
1201 * optimization: x86-64 supports stack allocation of results of simple
1202 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1203 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1204 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1205 (thanks to James Knight)
1206 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1207 by the standard (thanks to Stephen Compall)
1209 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1210 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1211 be used for bundling the runtime and the core file into one
1212 executable binary. This feature is not currently supported on all SBCL
1213 platforms. (thanks to James Bielman and NIIMI Satoshi)
1214 * new feature: a generational or ephemeral garbage collector is now
1215 the default on the PowerPC platform (both Linux and Darwin). The
1216 old Cheney (stop and copy) collector is a build-time option.
1217 (thanks to Cyrus Harmon, after Raymond Toy)
1218 * minor incompatible change: the method by which SBCL finds its
1219 contributed modules has changed; it no longer relies on symbolic
1220 links from an $SBCL_HOME/systems directory, but searches directly
1221 in the subdirectories of $SBCL_HOME.
1222 * enhancement: the dynamic heap size on the Linux/PPC platform is
1223 markedly larger, even using the older Cheney garbage collector.
1224 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1225 certain complicated string types. (reported by Gary King)
1226 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1227 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1228 * fixed bug: DOCUMENTATION on structures no longer signals an
1229 error if no structure type of the right name exists. (reported by
1231 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1232 applicable non-standard primary or :AROUND method on
1233 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1234 no longer cause unbound variable errors. (reported by Kevin Reid)
1235 * fixed bug: metacircle resolution in cases where methods have slots
1236 added before the slots from STANDARD-METHOD. (reported by Jean
1238 * fixed bug: the Power PC assembly code for calling into lisp
1239 sometimes computed its return address wrongly (depending on the
1240 15th bit of the address, fixed at link time). This used to
1241 manifest itself in a segmentation violation while building PCL.
1242 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1243 many others over the years)
1244 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1245 the stack pointer (treated as a signed integer) is negative.
1246 (thanks to Peter van Eynde)
1248 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1249 * new platform: experimental support for the Windows operating
1250 system has been added. (thanks to Alastair Bridgewater)
1251 * fixed several bugs in and robustified the PPC FFI (including
1252 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1253 * bug fix: allow non-simple string symbol names (reported by Paul
1255 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1257 * bug fix: saving a core corrupted callbacks on x86/x86-64
1258 * bug fix: closed a loophole in metacircularity detection and
1259 grounding in the PCL implementation of CLOS.
1260 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1261 (reported by tomppa on #lisp)
1262 * bug fix: clear the direction flag on Lisp -> C transitions, as
1263 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1264 (reported by Andrej Grozin and Hendrik Maryns)
1265 * optimization: major improvements to GC efficiency on GENCGC platforms
1266 * optimization: faster implementation of EQUAL
1267 * optimization: emit more efficient opcodes for some common
1268 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1269 * optimization: several other minor code-generation improvements on x86-64
1270 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1272 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1273 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1274 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1275 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1276 by ANSI as a generic function, and for consistency (SETF
1277 GENERIC-FUNCTION-NAME) is treated likewise.
1278 * fixed bug #233.b: make constraint propagation notice when a variable
1279 value is changed after it is referenced but before it is used
1280 * fixed bug #296: no more arbitrary behaviour from filenames with
1281 odd characters as --load arguments. (reported by Adam Warner)
1282 * fixed bug #390: :CHARACTER-SET pathname components now work as
1283 expected. (reported by Tim Daly Jr)
1284 * fixed bug #391: complicated :TYPE intersections in slot
1285 definitions no longer cause an error in PCL internals.
1286 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1287 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1288 returning the number of octets which would be written to the
1289 file-stream. (thanks to Robert J. Macomber)
1290 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1291 arguments to the internal stream functions. (thanks to David
1293 * bug fix: stack exhaustion is now handled on the x86/SunOS
1294 platform. (thanks to Daisuke Homma)
1295 * bug fix: more accurate ROOM results on GENCGC platforms
1296 * optimization: improved type inference for arithmetic-for
1297 index variables in LOOP
1298 * optimization: faster floating-point SQRT on x86-64
1299 * fixed some bugs revealed by Paul Dietz' test suite:
1300 ** DOCUMENTATION returns NIL instead of "" for method combinations
1301 that don't have a docstring
1303 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1304 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1305 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1306 therefore are not customizeable by user code (as seems to be at
1307 least permitted and maybe required by AMOP). As a consolation,
1308 however, the SBCL implementation of these functions now calls
1309 REINITIALIZE-INSTANCE as specified by AMOP.
1310 * bug fix: slot-definition documentation is propagated as per ANSI
1311 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1312 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1313 argument, as do eslotd initialization methods. (from Pascal
1314 Costanza's "Closer" project)
1315 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1316 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1318 * bug fix: it is now possible to have more than one subclass of
1319 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1320 (reported by Bruno Haible, Pascal Costanza and others)
1321 * bug fix: the dependent update protocol now works for generic
1322 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1323 and Pascal Costanza)
1324 * bug fix: condition-class instances corresponding to
1325 DEFINE-CONDITION forms are now created eagerly. (reported by
1326 Kalle Olavi Niemitalo on comp.lang.lisp)
1327 * bug fix: floating point printing is more accurate in some
1328 circumstances. (thanks to Simon Alexander)
1329 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1330 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1331 * bug fix: callbacks on OS X now preserve stack-alignment.
1332 (thanks to Cyrus Harmon)
1333 * enhancement: the x86-64 disassembler is much better at
1334 disassembling SSE instructions. (thanks to Lutz Euler)
1335 * enhancement: CHECK-TYPE on a variable now allows type-inference
1336 (thanks to Vincent Arkesteijn)
1337 * optimization: improved performance of EUC-JP external format.
1338 (thanks to NIIMI Satoshi)
1339 * optimization: performance improvements to IO on file streams of
1340 :ELEMENT-TYPE CHARACTER
1341 * optimization: much faster memory allocation on x86-64
1342 * optimization: faster garbage collections (latency and throughput)
1344 * optimization: faster bignum subtraction and division on x86 and x86-64
1345 (thanks to Lutz Euler)
1346 * optimization: more accurate type-inference for loops that use a
1347 floating point index variable or a negative step.
1349 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1350 * bug fix: add a workaround to SBCL looping infinitely at startup on
1351 Linux kernels with apparently buggy implementations of personality().
1352 (thanks to Svein Ove Aas)
1353 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1354 (thanks to David Lichteblau)
1355 * bug fix: local bindings shadow global symbol macros for
1356 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1357 * optimization: non-open coded uses of numeric comparison operators
1358 (e.g. >) no longer cons when called with more than one parameter
1359 on platforms supporting dynamic-extent allocation.
1360 * enhancement: saving cores with foreign code loaded is now supported
1361 on MIPS/Linux in addition to the previously supported platforms.
1362 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1363 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1364 * bug fix: division by zero in sb-sprof when no samples were collected
1365 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1366 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1367 (thanks to Alexander Kjeldaas)
1368 * new feature: ignoring the return values of destructive functions
1369 like NREVERSE signals a compile-time style-warning.
1370 (thanks to Kevin Reid)
1371 * new feature: the alignment of alien structure fields can be
1372 explicitly specified. (thanks to Cyrus Harmon)
1373 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1374 and dump core on SIGQUIT
1376 ** incompatible change: threads do not inherit values of specials
1377 from their parents (see manual)
1378 ** bug fix: threads stacks belonging to dead threads are freed by the
1379 next exiting thread, no need to gc to collect thread stacks anymore
1380 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1381 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1382 ** bug fix: dynamic variable and thread start related gc lossage
1383 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1385 * fixed some bugs revealed by Paul Dietz' test suite:
1386 ** SUBTYPEP is slightly more accurate on heinously complicated
1387 CONS types where some of the members have uncertain (in the
1388 NIL, NIL sense) type relationships to each other.
1390 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1391 no longer purified unless :PURIFY T is explicitly specified.
1392 ** Non-purified cores are significantly smaller than before
1394 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1395 * new feature: timers based on Zach Beane's excellent timer package
1396 * added support for the following external formats: koi8-u,
1397 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1398 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1399 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1400 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1401 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1402 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1403 * incompatible change: a threaded SBCL will no longer revert to
1404 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1405 * bug fix: interrupts are disabled until startup is complete; no
1406 more sigsegvs when receiving a signal to soon
1407 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1409 * bug fix: add a workaround for the memory randomization features in
1410 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1411 workaround will only be in effect on systems with the proc filesystem
1413 * bug fix: printing objects of type HASH-TABLE signals a
1414 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1416 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1417 have been running for over 50 days. (reported by Gilbert Baumann)
1418 * bug fix: the logic for getting names of functions gets less
1419 confused when confronded with alternate-metaclass
1420 funcallable-instances. (reported by Cyrus Harmon)
1421 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1422 consistent, even on internal alternate-metaclass objects.
1423 * bug fix: flush closure information collected by physical
1424 environment analysis prepass before the main pass. (bug reported
1426 * bug fix: compiler pack phase does not modify a hashtable, which is
1427 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1428 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1429 may be expanded. (reported by Alexander <malishev> on c.l.l)
1430 * bug fix: no more style-warnings when compiling code using LAST
1431 under high-SPEED low-SPACE settings. (reported by David Wallin
1433 * fixed some incompatibilities between SBCL's MOP and the MOP
1435 ** the METAOBJECT class is now implemented;
1436 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1437 STANDARD-OBJECT, as required;
1438 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1439 now compatible (as required by VALIDATE-SUPERCLASS); there
1440 remains a consistency requirement over the presence or absence
1441 of the FUNCTION class in the superclass list of finalized
1442 classes; see the manual for more details;
1443 ** the :AROUND method for COMPUTE-SLOTS on
1444 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1445 requested slot ordering.
1447 ** bug fix: parent thread now can be gc'ed even with a live
1449 ** bug fix: binding a special with PROGV to no value is not
1450 the same as that symbol not having been bound (thanks to
1452 ** bug fix: binding specials is thread safe (thanks to
1454 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1455 and SB-SPROF do not die with 'no handler for signal XX in
1456 interrupt_handle_now(..)' anymore
1457 ** bug fix: WITH-TIMEOUT works with multiple running threads
1458 * fixed some bugs revealed by Paul Dietz' test suite:
1459 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1460 the :method-class keyword argument.
1462 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1463 * new port: the Solaris operating system on x86 processors is now
1464 mostly supported, though some rough edges in the environment
1465 remain. (thanks to Daisuke Homma)
1466 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1467 space for its heap. (thanks to Thiemo Seufer)
1468 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1469 closures. (thanks to Thiemo Seufer)
1470 * minor incompatible change: eof selects abort in the debugger.
1471 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1472 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1474 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1475 deprecated, and will go away in a future revision of SBCL.
1476 * minor incompatible change: GC-ON and GC-OFF are no longer
1477 implemented with a counter, it does not matter how many times gc
1478 is switched on or off
1479 * bug fix: discriminating functions for generic function classes
1480 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1481 make invalid assumptions about method precedence order. (reported
1483 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1484 platforms. (thanks to Thiemo Seufer)
1485 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1486 * bug fix: correct pathname printing with printer escaping is on.
1487 (thanks to Kevin Reid)
1488 * bug fix: complex VOP definitions in "user-space" no longer trigger
1489 package locks. (reported by Zach Beane)
1490 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1491 for funcallable-instance functions, and is no different from
1493 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1494 characters in the current readtable. (reported by Nicholas Neuss)
1495 * bug fix: SUBTYPEP on various CONS types returns more a more
1496 accurate acknowledgment of its certainty.
1497 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1498 lists in some cases. This partially fixes bug 384.
1499 * flush all standard streams before prompting in the REPL and the
1501 * bug fix: signal handling and triggering gc do not conflict
1502 directly anymore, in particular a high frequency sb-sprof does
1503 not prevent gc from running
1504 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1505 approximation for timezone and DST information between the
1506 universal time epoch and the smallest negative 32-bit time_t.
1507 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1508 year 1899 so long as the encoded time is non-negative.
1509 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1510 the correct answer for non-constant rotate amounts when performing
1511 an inline 32-bit rotation.
1513 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1514 there is only one thread in the session
1515 ** bug fix: memory leak for streams created in one thread and
1516 written to in another
1517 ** bug fix: lockup when compiled with gcc4
1518 ** bug fix: race that allows the gc to be triggered when gc is
1520 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1521 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1523 ** bug fix: finished threads are gc'ed properly
1524 * fixed some bugs revealed by Paul Dietz' test suite:
1525 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1526 has arguments with a different set of applicable methods from
1527 the orignal arguments.
1528 ** The type error thrown by MAP now has a correct expected-type
1530 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1531 name a compiled function.
1532 ** Three MISC tests where a large bignum was improperly coerced to
1533 a float that couldn't represent that bignum during type
1534 derivation were fixed.
1535 ** SUBTYPEP can now handle types involving the negation of a
1536 list-form FUNCTION type.
1537 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1538 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1539 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1541 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1542 * New feature: Experimental support for bivalent streams: streams
1543 opened with :element-type :default now allow character and binary
1544 (unsigned-byte 8) I/O
1545 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1546 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1548 * bug fix: no more highly sporadic "couldn't check whether ~S is
1549 readable" when reading a stream and an interrupt hits in the middle
1550 of a select system call
1551 * compiler better recognizes complex arrays (reported by Eduardo
1553 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1554 for variables on SBCL built with linkage-tables. (reported by Luis
1556 * various error reporting improvements.
1557 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1558 (Thanks to Thiemo Seufer)
1559 * bug fix: nested reader invokations work correctly
1560 * bug fix: it is possible to have simultaneous references to foreign
1561 code and foreign data with the same name.
1563 ** added x86-64 support
1564 ** incompatible change: the threading api now works with thread
1565 objects instead of thread ids
1566 ** bug fix: threads are protected from signals and interruption when
1567 starting up or going down
1568 ** bug fix: a race where an exiting thread could lose its stack to gc
1569 ** bug fix: don't halt on infinite error in threads if possible
1570 ** fixed numerous gc deadlocks introduced in the pthread merge
1571 ** bug fix: fixed thread safety issues in read and print
1572 ** bug fix: debugger doesn't hang on session lock if interrupted at
1573 an inappropriate moment
1574 ** bug fix: run-program is now thread safe(r)
1575 ** bug fix: inner with-recursive-lock no longer releases the mutex
1576 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1577 * fixed some bugs revealed by Paul Dietz' test suite:
1578 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1579 EXPECTED-TYPE slots filled.
1580 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1582 ** MAKE-CONDITION accepts classes as type-designators.
1583 ** COMPILE may never return NIL.
1584 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1585 range before calling Unix time functions
1587 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1588 * numerous signal handling fixes to increase stability
1589 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1590 * minor incompatible change: we now correctly canonize default
1591 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1592 as per the MOP, rather than the historical (INITARG INITFUNCTION
1593 INITFORM). (reported by Bruno Haible)
1594 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1595 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1596 for more information.
1597 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1598 pathname is a directory pathname.
1599 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1600 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1602 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1603 :METACLASS argument in addition to a class name. (reported by
1604 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1605 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1606 (reported by Bruno Haible; patch from Gerd Moellmann)
1607 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1609 * bug fix: more cleanups to the floating point exception handling on
1610 x86-64 (thanks to James Knight)
1611 * bug fix: the compiler does not try to constant fold calls of
1612 COERCE and BIT-* functions when they return freshly constructed
1613 objects. (reported by Nikodemus Siivola and Paul Dietz)
1614 * optimization: improved the MIPS versions of generic (in the
1615 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1616 * optimization: direct conversion of (unsigned-byte 32) to floats on
1617 the PowerPC platform.
1618 * optimization: structure instances with raw slots now use less
1619 memory, and probably show better memory locality. (thanks to
1621 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1622 are treated as requests for stack allocation on the x86-64,
1623 Alpha-32, PPC and SPARC platforms.
1624 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1625 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1626 (thanks to Tony Martinez)
1628 ** gcing a dead thread can no longer lead to lockups
1629 ** threads block signals until they are set up properly
1630 ** errno is no longer shared by threads
1631 ** interrupt-thread restores the eflags register on x86
1632 ** fixed some lockups due to gc/thread interaction
1633 * fixed some bugs revealed by Paul Dietz' test suite:
1634 ** invalid dotted lists no longer raise a read error when
1635 *READ-SUPPRESS* is T
1636 ** signal an error if a symbol that names a declaration is used
1637 as the name of a type, or vice versa
1638 ** allow using the (declare (typespec var*)) abbreviation for
1639 (declare (type typespec var*)) with all type specifiers
1640 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1641 ** always return NIL from PPRINT-POP when OBJECT is NIL
1642 ** don't signal errors when pretty-printing malformed LABELS,
1643 FLET or MACROLET forms
1644 ** declarations in a DOLIST body are also in scope for the
1646 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1648 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1651 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1652 * fixed cross-compiler leakages that prevented building a 32-bit
1653 target with a 64-bit host compiler.
1654 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1655 opened with :IF-EXISTS :OVERWRITE.
1656 * fixed bug 281: error for an invalid qualifier in a short-form method
1657 combination method is not signalled until the faulty method is called.
1658 * bug fix: iteration variable type inferrer failed to deal with open
1659 intervals. (reported by Alan Shields)
1660 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1661 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1662 or more runtime options were provided to the sbcl binary.
1663 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1664 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1665 * optimization: there's now a fast-path for fixnum arguments in the
1666 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1667 * optimization: the code generated on x86-64 is more compact thanks
1668 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1669 * bug fix: floating-point exception handling now partly works on
1670 x86-64. (Thanks to James Knight)
1671 * improvement to the MIPS backend from Thiemo Seufer: C-style
1672 64-bit long long arguments and return values to alien functions
1674 * fixed some bugs revealed by Paul Dietz' test suite:
1675 ** the type-error signalled from WARN has a filled-in DATUM slot.
1676 ** the type-error required when a stream is not associated with
1677 a file has the stream as its datum.
1678 ** type-errors on single-floats on x86-64 no longer have
1679 :INVALID-OBJECT as the datum
1680 ** the type-errors signalled for invalid function names now have
1681 a correct expected type
1682 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1683 for structures defined with a :TYPE. Documentation strings for
1684 typed structures are no longer immediately discarded
1685 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1686 broadcast streams and synonym streams. FILE-LENGTH now also works
1687 on broadcast streams.
1689 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1690 * incompatible change: the --noprogrammer option, deprecated since
1691 version 0.7.5, has been removed. Please use the equivalent
1692 --disable-debugger option instead.
1693 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1695 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1696 inoperational for a while now) has been completely removed.
1697 * null lexical environments are now printed as #<NULL-LEXENV>,
1698 significantly reducing the amount of clutter in typical backtraces.
1699 * documentation on weak pointers, finalization, and after GC hooks
1700 has been added to the manual.
1701 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1702 as other specialized array types, is much faster. SUBSEQ and
1703 COPY-SEQ on such arrays have also been sped up.
1704 * optimization: EQL is now more efficient when at least other argument
1705 is known to be of type (OR FIXNUM (NOT NUMBER)).
1706 * fixed bug: the runtime is now less vulnerable to changes in the
1707 size of the SBCL object on OS X, and virtual memory is reserved for
1708 all spaces that need to be at a fixed address.
1709 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1710 * fixed bug: finalizers and after GC hooks that cause consing are now
1712 * fixed bug: compiler error messages and summaries are now printed to
1713 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1714 * fixed inference of the upper bound of an iteration variable.
1715 (reported by Rajat Datta).
1716 * fixed bug 211e: calling local functions with duplicated constant
1717 keyword argument no longer causes a bogus style warning about an
1719 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1720 ftype declaration to be disregarded. (reported by Dave Roberts)
1721 * fixed bug 373: caused by erronous compilation of references to alien
1722 variables in the runtime on ppc/darwin.
1723 * fixed bug 376: CONJUGATE type deriver.
1724 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1725 (reported by Baughn on #lisp)
1726 * a cross-compiler bug on non-x86 platforms has been identified and
1727 fixed. (thanks to Bruno Haible)
1728 * improvements to the MIPS runtime code for increased stability.
1729 (thanks to Thiemo Seufer)
1730 * increased the maximimum compact environment size to allow
1731 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1732 * partial workaround for bug 135: don't name the function we're
1733 calling for hairy cases of EVAL, so as not to accumulate one environment
1734 entry per call to EVAL. (reported by Kevin Reid)
1735 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1736 * optimization: type testing for non-vector arrays should be faster.
1737 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1739 * bug fix: setting 31st element of a bit vector to zero did not work
1741 * bug fix: redefining a class definition which failed due to a
1742 previous accessor / function clash now works (but see BUGS entry
1743 #380 for more problems in this area). (thanks to Zach Beane)
1744 * the long form of DEFINE-METHOD-COMBINATION disables method group
1745 checking when given a single method group with pattern *. (thanks
1747 * on x86 compiler supports stack allocation of results of simple
1748 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1749 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1750 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1751 * fixed some bugs related to Unicode integration:
1752 ** the restarts for recovering from input and output encoding
1753 errors only appear when there is in fact such an error to
1755 * fixed some bugs revealed by Paul Dietz' test suite:
1756 ** MISC.549 and similar: late transformation of unsafe type
1757 assertions into derived types caused unexpected code
1759 ** SCALE-FLOAT type deriver is less wrong.
1760 ** type derivers for EXP, LOG and similar functions compute result
1761 types for complex arguments better.
1762 ** (MISC.563) CONJUGATE type deriver works for very restricted
1764 ** out-of-line type testers for character strings are available.
1765 ** EQUAL compiler transform understands specialness of objects
1767 ** accessing double-floats stored on the stack now works on x86-64.
1768 ** debugger internals could sometimes create invalid lispobjs,
1769 resulting in GC crashes.
1770 ** MISC.548: type check weakening can convert required type into
1772 ** initialization forms for bindings are not in scope of free special
1775 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1776 * incompatible change: thread support for non-NPTL systems has
1777 been removed - locking is buggy and unreliable. A threaded
1778 SBCL build will now warn at startup and refuse to create
1779 new threads, unless futex support is detected in the kernel
1780 * incompatible change: the top level REPL now has only an ABORT
1781 restart associated with it, not TOPLEVEL and ABORT as it used to.
1782 TOP and TOPLEVEL are now available as debugger commands for
1783 returning to the top level.
1784 * incompatible change: forms evaluated in the REPL now use the
1785 global optimization policy.
1786 * incompatible change: user- and system-initialization files are
1787 no longer processed with LOAD, but by READ and EVAL; hence the
1788 global optimization policy, startup package, readtable, etc,
1790 * The .fasl file format number has been incremented because of
1791 various incompatible changes.
1792 * internal entry point details and argument counts no longer appear
1793 in backtraces unless explicitly requested by setting
1794 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1795 * built-in and standard functions no longer have names like "top
1796 level local call to FOO".
1797 * fixed bug 32: functions defined in non-null lexical environments
1798 now have more legible printed representation
1799 * fixed bug 33: functions defined in non-null lexical environemnts
1800 are now more amenable to inspection by INSPECT.
1801 * workaround for bug 354: XEPs no longer appear in backtraces unless
1802 explicitly requested.
1803 * fixed bug: receiving the signal which results from attempting to
1804 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1805 SIGBUS on other *BSDs) on architectures where the C stack is also
1806 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1807 to debugger confusion or wild instability. (reported by Cheuksan
1809 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1810 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1811 (reported by Lutz Euler)
1812 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1813 compiled code. (reported by Paul Dietz)
1814 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1815 specialized parameter to a method to something that is not TYPEP
1816 the specializer is now possible.
1817 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1818 face of package deletion.
1819 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1820 pathname. (reported by David Tolpin on c.l.l)
1821 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1822 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1823 STRING-STREAM and FILE-STREAM, respectively.
1824 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1825 than previously; in addition, its test suite is now run on build.
1826 (thanks to Luke Gorrie)
1827 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1828 * optimization: added a immediate representation for single-floats
1830 * fixed some bugs revealed by Paul Dietz' test suite:
1831 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1832 ** attempting to create a package with a colliding nickname causes
1833 correctable errors to be signalled.
1834 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1835 ** several x86-64 backend bugs related to sign-extension of immediate
1838 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1839 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1840 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1842 * fixed loading of multiply forward-referenced layouts.
1843 (thanks to Cheuksan Wang)
1844 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1845 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1846 (see COMPILE-FILE documentation for details of :PRINT option)
1847 * fixed bugs 19 and 317: fixed-format floating point printing is
1848 more accurate. This also fixes a bug reported by Adam Warner
1849 related to the ~@F format directive.
1850 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1852 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1853 dispatch table if the from-char is a dispatch macro character.
1854 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1855 by the wordsize no longer ignore the last word. (reported by Lutz
1857 * fixed bug in type checking of dynamic-extent variables. (reported
1859 * optimization: sequence traversal functions use their freedom to
1860 coerce function designators to functions.
1861 * optimization: code with many calls to CLOS methods specialized on
1862 CLOS classes has had redundant type checks removed.
1863 * optimization: added declarations to speed up operations that access
1864 the internal character database (for example STRING-UPCASE)
1865 * optimization: comparison operations between floats and sufficiently small
1866 fixnums no longer create extra rationals
1867 * fixed some bugs related to Unicode integration:
1868 ** portions of multibyte characters at the end of buffers for
1869 character-based file input are correctly transferred to the
1870 start of the buffer at the next read.
1871 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1872 passing it through to OPEN.
1873 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1874 argument given to internal calls to OPEN.
1875 ** the built-in comment readers (introduced by character sequences
1876 ";" and "#|") are more forgiving to encoding errors; they will
1877 STYLE-WARN and then attempt to resync the stream at a character
1878 boundary. (thanks to Teemu Kalvas)
1879 * fixed some bugs in the x86-64 port:
1880 ** Negative short int return values from c-calls are sign-extended
1882 ** The stack is aligned to 16-bytes for c-calls, as required by
1883 the ABI. (thanks to Cheuksan Wang)
1884 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1885 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1886 for enums. (thanks to Vincent Arkesteijn)
1887 ** Multiple small optimizations and bugfixes for floating point
1889 * fixed some bugs revealed by Paul Dietz' test suite:
1890 ** Space, Tab, Linefeed, Return and Page have the invalid
1891 secondary constituent character trait.
1892 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1894 ** WITH-INPUT-FROM-STRING should only update the index place on
1896 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1897 works more reliably.
1898 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1899 with one-dimensional arrays with fill pointers.
1900 ** TYPE-OF failed on a complex with an integer realpart and a
1902 ** compiler failure during type inference for the code of form
1903 (IF (EQL X (THE ...)) ...) (MISC.535).
1905 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1906 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1907 platform. The port supports 61-bit fixnums, large memory spaces
1908 and reloading shared object files.
1909 * enhancement: saving cores with foreign code loaded is now
1910 supported on ppc/Darwin in addition to the previously supported
1912 * enhancement: the statistical profiler now walks deeper into the
1913 call stack for more meaningful call-graphs and accrued time
1914 reports (x86/x86-64 only). It also now reports time spent in
1916 * enhancement: it is now possible to trace most individual methods
1917 of a generic function in addition to tracing the generic function
1919 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1921 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1922 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1923 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1924 produces an error. (thanks to Vincent Arkesteijn)
1925 * bug fix: NAMESTRING on pathnames with :WILD components in their
1926 directories works correctly. (thanks to Artem V. Andreev)
1927 * fixed bug 125: compiler preserves identity of closures. (reported
1929 * bug fixed: functions with &REST arguments sometimes failed with
1930 "Undefined function" when compiled with (DEBUG 3). (reported by
1932 * bug fix: overflow during compiling of setting element of a bit
1933 vector with constant index and value. (reported by Timmy Douglas)
1934 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1935 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1936 * on x86 compiler partially supports stack allocation of dynamic-extent
1938 * GO and RETURN-FROM do not check the extent of their exit points
1939 when compiled with SAFETY 0.
1940 * fixed some bugs related to Unicode integration:
1941 ** encoding and decoding errors are now much more robustly
1942 handled; it should now be possible to recover even from invalid
1943 input or output to the terminal. (thanks to Teemu Kalvas)
1944 ** provided a first cut at implementing STRING-TO-OCTETS and
1945 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1946 ** altered the SB-MD5 contributed module slightly, changing the
1947 interface just enough for it to be supportable for builds where
1948 lisp characters are not eight bits.
1949 * fixed some bugs revealed by Paul Dietz' test suite:
1950 ** the FORMATTER-generated functions for ~V[ conditionals require
1951 the correct number of arguments.
1952 ** READ-FROM-STRING returns the mandated second value when applied
1953 to displaced strings.
1954 ** the #\Rubout and #\Backspace characters are treated as invalid
1955 constituent characters by the tokenizer.
1957 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1958 * new feature: reloading changed shared object files with
1959 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1960 * new feature: references to foreign variables and functions
1961 can now be compiled and loaded before the corresponding shared
1962 object file is loaded, as long as the foreign definitions are
1963 available at runtime.
1964 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1965 supported. (thanks to Dan Debertin)
1966 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1967 just keywords. (thanks to Vincent Arkesteijn)
1968 * fixed bug #331: structure-class instances corresponding to
1969 DEFSTRUCT forms are now created eagerly.
1970 * fixed bug #345: backtraces from calls to undefined functions work
1971 on x86 as well. Related bug #61 is now also partially fixed on x86
1972 (backtraces from throws to unknown catch tags.)
1973 * bug fix: lambda-list parsing is now stricter vrt. order and number
1974 of lambda-list keywords.
1975 * bug fix: as specified by AMOP, an error is signalled if a
1976 class-option appears multiple times in a DEFCLASS form. (reported
1978 * bug fix: RANDOM can be compiled when the compiler derives the type
1979 of its numeric argument as a disjoint set of small integers.
1980 (reported by Paul Dietz)
1981 * bug fix: starting a core saved with shared objects loaded when
1982 those objects are not available no longer causes threaded SBCL to
1983 hang. (reported by Sean Ross)
1984 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1986 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1987 argument, which must always be NIL. (reported by Kalle Niemitalo)
1988 * bug fix: printing 1.0d+23 no longer results in an error.
1989 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1990 * bug fix: structure slot setters preserve evaluation order. (thanks
1992 * fixed some bugs related to Unicode integration:
1993 ** RUN-PROGRAM can allow its child to take input from a Lisp
1994 stream. (reported by Stefan Scholl)
1995 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1996 locales. (reported by Ken Causey)
1997 * fixed some bugs revealed by Paul Dietz' test suite:
1998 ** INCF, DECF and REMF evaluate their place form as specified in
2000 ** FORMATTER expands ~{ ~} iteration directives with V or #
2001 parameters correctly.
2002 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2003 consequent uses no arguments correctly.
2004 ** the system has a partial understanding of the (COMPLEX RATIO)
2006 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2009 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2010 * new feature: a build-time option (controlled by the :SB-UNICODE
2011 keyword feature, enabled by default) for building the system with
2012 support for the entire 21-bit character space defined by the
2014 * new feature: the system now has rudimentary external-format
2015 support; the primary user-visible change at this time is that
2016 characters with the high bit set (such as lower-case-e-acute) will
2017 print correctly to a terminal in a UTF-8 environment.
2018 * minor incompatible change: BASE-CHAR no longer names a class;
2019 however, CHARACTER continues to do so, as required by ANSI.
2020 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2021 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2022 has been moved to the SB-EXT package (temporarily re-exported from
2024 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2025 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2027 * fixed bug #351: better error-handlind and reporting for malformed
2029 * fixed bug #350: bignum-printing is now more memory-efficient,
2030 allowing printing of very large bignums, eg. (expt 2 10000000).
2031 (reported by Bruno Haible)
2032 * fixed bug #302: better primitive-type selection for intersection
2034 * fixed bug #308: non-graphic characters now all have names, as
2035 required. (reported by Bruno Haible)
2036 * bug fix: redefining a class with different superclasses now correctly
2037 removes it from the direct-subclasses of its previous superclasses.
2038 (reported by David Morse)
2039 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2040 new value now works. (reported by Bruno Haible)
2041 * bug fix: correct canonicalization of multiple non-standard slot
2042 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2043 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2044 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2045 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2046 now exists, an signals an error.
2047 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2048 during macroexpansion for non-iteration variables as well. (reported
2049 by Bruno Haible for CMUCL)
2050 * bug fix: Cyclic structures and unprintable objects in compiler
2051 messages no longer cause errors. (reported by Bruno Haible)
2052 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2053 and READ-FROM-STRING all now return a primary value of NIL if
2054 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2055 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2056 (reported by Bruno Haible for CMUCL)
2057 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2058 do not designate a condition. (reported by Bruno Haible for
2060 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2061 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2062 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2063 permitting resolution in favour of any of the conflicting symbols.
2064 (reported by Bruno Haible for CMUCL)
2065 * bug fix: EQUAL compiler optimizations is less aggressive on
2066 strings which can potentially compare true despite having distinct
2067 specialized array element types.
2068 * bug fix: unit enumerations can be defined without dividing by
2069 zero. (thanks to Vincent Arkesteijn)
2070 * FORMAT compile-time argument count checking has been enhanced.
2071 (report from Bruno Haible for CMUCL)
2072 * a partial workaround for the bug 262: the compiler does not try to
2073 inline-expand a local function doing RETURN-FROM from a deleted
2074 BLOCK. (thanks to Peter Denno for the bug report and to David
2075 Wragg for the simple test case)
2076 * fixed some bugs revealed by Paul Dietz' test suite:
2077 ** RENAME-PACKAGE allows all package designators as new package
2079 ** constraint propagation and lambda variable substitution are
2080 more cautious in dealing with partially deleted code.
2081 ** compiler sometimes lost reoptimization passes.
2082 ** CERROR, given a condition as condition designator, passes its
2083 remaining arguments to the continue format control without
2085 ** Case-altering FORMAT directives work correctly on non-ASCII
2087 ** The REAL type specifier handles bounds outside the floating
2088 point ranges without signalling FLOATING-POINT-OVERFLOW.
2089 ** Functions with IR1-transformations can create intercomponent
2090 references to global functions.
2091 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2093 ** FORMAT ~R treats a nil value for its first parameter correctly.
2095 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2096 * enhancement: saving cores with foreign code loaded is now
2097 supported on x86/NetBSD and sparc/Linux in addition to the previously
2098 supported platforms.
2099 * bug fix: on some platforms repeated installations caused multiple
2100 copies of HTML documentation to be installed -- should not happen
2101 any more. (reported by Stefan Scholl)
2102 * bug fix: parsing self-recursive alien record types multiple times
2103 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2104 original patch by Helmut Eller for CMUCL)
2105 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2106 (thanks to Richard Kreuter)
2107 * bug fix: defining classes whose accessors are methods on existing
2108 generic functions in other (locked) packages no longer signals
2109 bogus package lock violations. (reported by François-René Rideau)
2110 * bug fix: special variables as DEFMETHOD parameters no longer have
2111 associated bogus type declarations. (reported by David Wragg and
2113 * bug fix: read-write consistency on streams of element-type
2114 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2115 * bug fix: redefiniton of the only method of a generic function with
2116 no DEFGENERIC no longer emits a full WARNING. In addition,
2117 redefinition of generic functions with no DEFGENERIC to an
2118 incompatible lambda list now signals an error. (thanks to Zach
2120 * bug fix: DEFGENERIC now works even when there's a function of the
2121 same name in an enclosing lexical environment. (thanks to Zach
2123 * fixed compiler failure, caused by instrumenting code during
2124 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2125 * optimization: added loop analysis and improved register allocation
2126 * fixed some bugs revealed by Paul Dietz' test suite:
2127 ** POSITION on displaced vectors with non-zero displacement
2128 returns the right answer.
2129 ** (SIMPLE-STRING) is a valid type specifier for sequence
2131 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2133 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2134 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2136 ** The justification version of the FORMAT ~< directive treats
2137 non-zero minpad parameter correctly.
2139 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2140 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2141 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2142 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2143 the supported interface.
2144 * new feature: Single-stepping of code compiled with DEBUG 2 or
2145 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2146 * new feature: saving cores with foreign code loaded is now
2147 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2148 Timothy Moore's work for CMUCL)
2149 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2150 parameters to * instead of NIL if no initform is supplied.
2151 (reported by Johan Bockgaard)
2152 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2153 correctly when FUNCALL appears as the car of the form. Note:
2154 despite this FUNCALL forms are not currently subject to
2155 compiler-macro expansion. (port of Raymond Toy's fix for the
2156 same from CMUCL, reported by Johan Bockgaard)
2157 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2158 (thanks for Teemu Kalvas)
2159 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2160 case instead of terminating the iteration (thanks for Julian
2161 Squires, Sean Champ and Raymond Toy)
2162 * bug fix: incorrect expansion of defgeneric that caused a style
2163 warning. (thanks for Zach Beane)
2164 * bug fix: slot accessor effective method computation works properly
2165 for classes with multiple non-standard applicable methods on
2166 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2167 * on x86 compiler supports stack allocation of results of LIST and
2168 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2169 CMUCL implementation by Gerd Moellmann)
2170 * fixed some bugs revealed by Paul Dietz' test suite:
2171 ** FORMAT strings with both the ~<~:;~> form of the justification
2172 directive and pretty-printing directives cause an error.
2174 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2175 * incompatible change: the internal functions
2176 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2177 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2178 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2179 instead of the old functions.
2180 * new feature: on platforms where "dladdr" is available foreign
2181 function names now appear in backtraces. (based on Helmut Eller's
2183 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2184 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2186 * documentation: ASDF manual is now distributed with SBCL.
2187 * bug fix: SBCL can now load its contributed modules with REQUIRE
2188 even if the system-provided entries have been removed from
2189 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2191 * bug fix: backtraces involving undefined functions or assembly
2192 routines are more informative. (thanks to Brian Downing)
2193 * bug fix: mutually referent alien structures now work correctly.
2194 (reported by Rick Taube)
2195 * bug fix: structures defined by WITH-ALIEN can be referred to
2196 within other definitions in the same WITH-ALIEN.
2197 * bug fix: division operators (MOD, TRUNCATE and the like) with
2198 constant zero divisors and integer dividends no longer generate
2200 * bug fix: provide default methods for INPUT-STREAM-P and
2201 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2202 * bug fix: improve the detection and resolution of MOP metacycles
2203 (where effective method computation depends on the generic function
2204 having its effective method computed). (reported by Bruno Haible)
2205 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2206 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2207 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2208 * bug fix: pathnames with relative directory components can be
2209 represented relative to default pathnames.
2210 * optimization: in taking the GCD of bignums, reduce the two bignums
2211 to approximately the same size (using Euclid's algorithm) before
2212 applying the more sophisticated binary GCD. (thanks to Juho
2214 * optimization: COUNT on bitvectors now operates word-at-a-time.
2215 * optimization: ASH with a positive, but not necessarily constant,
2216 (leftwards) shift, when in a modular context, is compiled to a
2218 * fixed some bugs revealed by Paul Dietz' test suite:
2219 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2220 if the corresponding argument is NIL.
2221 ** FORMAT directives accepting a minpad parameter treat negative
2223 ** literal commas as character parameters to FORMAT directives are
2225 ** literal spaces directly after ~<Newline> directives within a
2226 format-logical-block (~:< ~@:>) do not induce :FILL-style
2227 conditional newlines.
2228 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2229 as specified: it no longer includes conditional newlines).
2230 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2232 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2233 :PER-LINE-PREFIX argument does not evaluate to a string.
2235 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2236 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2237 the manual for details; for now, package locks can be disabled by
2238 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2239 if no major problems are found then it is likely that they will be
2240 compiled in unconditionally.
2241 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2242 unsupported operators on all platforms. To load a shared library
2243 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2244 object file, link it into a shared library outside of SBCL and
2245 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2247 * bug workaround: ROOM T and the GC were not working together
2248 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2249 reporting facility aren't preserved. That reporting has been
2250 disabled, so now until and unless someone figures out how to make it
2251 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2252 * minor incompatible change: as threatened around sbcl-0.8.0, the
2253 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2254 The symbols are also exported from SB-PCL for backwards
2255 compatibility, but more so than before SB-PCL should be treated as
2256 an implementation-internal package.
2257 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2259 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2260 It is known as of this release not to work on the Alpha, however.
2261 * fixed bug #167: errors signalled due to illegal syntax in method
2262 bodies are now more legible.
2263 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2264 designators and can hence be used with TYPEP.
2265 * fixed bug #333: CHECK-TYPE now ensures that the type error
2266 signalled, if any, has the right object to be accessed by
2267 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2268 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2269 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2271 * fixed bug #334: programmatic addition of slots using specialized
2272 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2273 and :ALLOCATION :CLASS slots.
2274 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2275 float exponents. (rereported by Peter Seibel)
2276 * fixed a bug: #\Space (and other whitespace characters) are no
2277 longer considered to be macro characters in standard syntax by
2278 GET-MACRO-CHARACTER.
2279 * fixed bug: initialization of condition class metaobjects no longer
2280 causes an instance of the condition to be created. (reported by
2282 * fixed bug: it is now possible to have slots such that
2283 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2284 description is neither :INSTANCE nor :CLASS.
2285 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2286 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2287 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2288 non-standard SLOT-DEFINITION-ALLOCATION values.
2289 * fixed bugs: various race conditions handling exiting threads.
2290 CL-PPCRE's thread torture test now passes.
2291 * fixed arguably-a-bug: GC time proportional to number of threads in
2292 system even when most of them are idle
2293 * optimization: improved performance of BIT and SBIT on bit-vectors.
2294 * .fasl file incompatibility: The fasl file version number has
2295 been incremented because of changes associated with package locks.
2297 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2298 * minor incompatible change: the system no longer provides
2299 optimization hints (conditions which are TYPEP
2300 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2301 mode; these hints are still emitted when higher SPEED optimization
2303 * new contrib module: a sampling profiler (profiling by statistical
2304 sampling, rather than by instrumenting functions) is available as
2305 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2306 * the behaviour of the standard function ED is now customizeable by
2307 third parties through a hook variable: see ED's documentation
2308 string for information on the protocol.
2309 * the compiler no longer emits efficiency notes for (FUNCALL X)
2310 when the type of X is uncertain under default optimization
2312 * fixed bug 276: mutating a binding of a specialized parameter to a
2313 method to something that is not TYPEP the specializer is now
2315 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2316 exhibit the required behaviour on the x86 platform. (thanks to
2317 Peter van Eynde, Eric Marsden and Bruno Haible)
2318 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2319 even for difficult arguments. (reported by Peter Graves)
2320 * fixed bug 141a: the backquote printer now descends quoted
2322 * fixed another bug in backquote printing: no more destructive
2323 modification of the form's list structure. (reported by Brian
2325 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2326 move between its address being taken and the call to
2327 interrupt_thread, fixing a crashing race condition.
2328 * the SB-POSIX contrib implementation has been adjusted so that it
2329 no longer exhibits ridiculously poor performance when constructing
2330 instances corresponding to C structs.
2332 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2333 * minor incompatible change: the sb-grovel contrib now treats C
2334 structures as alien (in the sense of SB-ALIEN) objects rather than
2335 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2336 has implications for memory management of client code
2337 (sb-grovel-returned objects must now be manually managed) and for
2338 type safety (alien objects now have full types).
2339 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2340 used to control emission of compiler diagnostics, rather than the
2341 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2342 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2343 quality should be considered deprecated.
2344 * install.sh now installs the user manual as well
2345 * (not quite a new documentable feature, but worth considering in
2346 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2347 beginnings of a semantically meaningful condition hierarchy is
2348 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2350 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2351 designator as the defaults argument.
2352 * fixed bug: Displaced arrays whose displaced-to array has become
2353 too small now cause ARRAY-DIMENSION to signal an error, providing
2354 for safer bounds-checking. (reported by Bruno Haible)
2355 * fixed bug: DEFCLASS slot definitions with identical :READER and
2356 :WRITER names now signal a reasonable error. (reported by Thomas
2358 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2360 * fixed bug: Invalid slot specification errors now print correctly.
2361 (thanks to Zach Beane)
2362 * fixed bug 320: Shared to local slot value transfers in class
2363 redefinitions now happen corrently from superclasses as
2364 well. (reported by Bruno Haible)
2365 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2367 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2368 improper lists correctly. (reported by Bruno Haible)
2369 * fixed bug 313: source-transform for <fun-name> was erroneously
2370 applied to a call of a value of a variable with name <fun-name>.
2371 (reported by Antonio Menezes Leitao)
2372 * fixed bug 307: The obsolete instance protocol ensures that
2373 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2374 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2375 remove dead unknown-values globs from the middle of the stack.
2376 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2377 easier to resume long computations after using *BREAK-ON-SIGNALS*
2378 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2379 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2380 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2381 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2382 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2383 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2385 * fixed a (fairly theoretical) bug in string printing: if
2386 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2387 does not have array-element-type equal to the most general string
2389 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2390 function name. (thanks to Nikodemus Siivola)
2391 * fixed bug: compile-time format string checker failed on
2392 non-closed ~{. (reported by Thomas F Burdick)
2393 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2394 don't warn on legal loop code involving more than one aggregate
2395 boolean. (thanks to Nikodemus Siivola)
2396 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2397 in keyword position to MAKE-PACKAGE was not being reported as
2399 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2400 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2402 * fixed bugs 280 and 312: the checking for multiple definitions in a
2403 file is less likely to become confused by uses of inline
2405 * fixed bug: the #S reader macro performs the keyword coercion
2406 specified for slot names. (reported by Kalle Niemitalo)
2407 * fixed bug: lambda lists may contain symbols whose names start with
2408 & but are not lambda-list-keywords; their occurrence triggers a
2410 * fixed bug 321: define-method-combination argument lambda lists do
2411 not cause invalid code to be generated when &optional or &aux
2412 variables have default values. (reported by Bruno Haible)
2413 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2414 have CLOS classes; this appears also to have fixed reported
2415 instability in stack exhaustion detection.
2416 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2417 MUFFLE-WARNING when no associated restart is present is now
2419 * optimization: rearranged the expansion of various defining macros
2420 so that each expands into only one top-level form in a
2421 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2423 * optimization: used a previously little-used slot in symbols to
2424 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2426 * fixed some bugs revealed by Paul Dietz' test suite:
2427 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2429 ** Fixed an optimization bug related to inheritance of initforms
2430 from local to shared slots.
2431 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2432 ** CAST optimizer forgot to flush argument derived type.
2433 ** print/read consistency on floats is now orders of magnitude
2434 more likely. (thanks also to Bruno Haible for a similar report
2436 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2437 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2438 for objects of type REAL. Make it so.
2439 ** SXHASH is specified (infelicitously) to respect similarity,
2440 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2441 it so. (thanks to Markus Ziegler)
2442 ** on the Alpha, the compiler succeeds in compiling functions
2443 returning a known number of arguments greater than 63.
2444 ** fixed handling of invalid NIL arguments in keyword position
2446 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2448 ** provided a readably-printable representation for RANDOM-STATE
2450 ** ensured that pathnames, where they have a namestring, always
2451 print using #P"..." syntax.
2453 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2454 * Support for the forthcoming 2.0 version of the NetBSD kernel
2455 running on x86 hardware has been added. (thanks to Perry
2456 E. Metzger most immediately, and others for their past work)
2457 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2458 current release version; SBCL's OpenBSD support had been broken
2459 since about the time of OpenBSD's switch to ELF binary format.)
2460 * [placeholder for DX summary]
2461 ** user code with &REST lists declared dynamic-extent, under high
2462 speed or space and low safety and debug optimization policy.
2463 * The manual has been converted to Texinfo format and the debugger
2464 chapter from the cmucl manual has been added.
2465 * A facility has been added to extract documentation strings from
2466 sbcl and store them as Texinfo-formatted snippets for inclusion in
2467 the manual (via Texinfo's @include directive)
2468 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2469 the test case to Dave Roberts)
2470 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2471 pointers. (reported by Sean Ross)
2472 * bug fix: PROFILE output is printed nicely even for large numerical
2473 values. (thanks to Zach Beane)
2474 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2475 greater than 32 handle EOF correctly.
2476 * bug fix: on X86 an immediate argument of the IMUL instruction is
2477 correctly printed in disassembly. (thanks to Lutz Euler)
2478 * bug fix: class slots in redefined classes preserve their old
2479 values. (thanks to Bruno Haible and Nikodemus Siivola)
2480 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2481 succeeds. (reported by Marco Baringer)
2482 * fixed some bugs revealed by Paul Dietz' test suite:
2483 ** READ-SEQUENCE now works on ECHO-STREAMs.
2484 ** RATIONALIZE works more according to its specification. (thanks
2487 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2488 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2489 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2490 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2491 should matter to you only if you rebind the printer control
2492 variables and then find you want different bindings in the
2493 debugger than in the ordinary execution of your program.)
2494 * The runtime build system has been tweaked to support building
2495 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2496 assembler and linker. (thanks to Nikodemus Siivola)
2497 * Unbound, undefined, undeclared variables now trigger full
2498 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2499 more often programmer error than deliberate exploitation of undefined
2501 * optimization: the hash algorithm for strings has changed to one
2502 that is less vulnerable to spurious collisions. (thanks to Juho
2504 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2505 needless bounds checking. (thanks to Juho Snellman)
2506 * optimization: implemented multiplication as a modular
2507 (UNSIGNED-BYTE 32) operation on the PPC backend.
2508 * fixed some bugs revealed by Paul Dietz' test suite:
2509 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2511 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2512 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2513 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2515 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2517 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2519 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2520 their output stream on EOF from read.
2521 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2522 have been read to end-of-file.
2523 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2525 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2526 description of determination of which consecutive characters
2528 ** Printing the "Space" character with escaping on now yields "#\\ ",
2529 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2530 ** Reading floating-point numbers with *READ-BASE* set to a number
2531 less than 10 works correctly.
2532 ** Reading floating-point numbers with *READ-BASE* set to a number
2533 more than 10 works correctly.
2534 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2535 the readtable currently in effect.
2537 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2538 * minor incompatible change: parsing of namestrings on a physical
2539 (Unix) host has changed; numbers after the final #\. in a
2540 namestring are no longer interpreted as a version field. This is
2541 intented to be largely invisible to the user, except that the
2542 meaning of the namestring "*.*.*" has changed: it now refers to a
2543 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2544 should usually be replaced by
2545 (make-pathname :name :wild :type :wild :version :wild)
2546 with the added benefit that this is more likely to be portable.
2547 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2548 option to OPEN now signals an error if the file being opened
2549 exists; this may have an impact on existing code.
2550 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2551 signals received, on the PowerPC platforms (both Linux and
2552 Darwin). (thanks to Pierre Mai for pointing out the location of
2554 * several fixes on OS X: The system now builds and runs cleanly on
2555 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2556 (thanks to Brian Mastenbrook)
2557 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2558 second-resolution: integer multiples of 1/3600 between -24 and 24.
2559 (thanks to Vincent Arkesteijn)
2560 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2561 type when called with 1 argument; PEEK-CHAR checked type of
2562 PEEK-TYPE only after having read first character from a
2563 stream. (reported by Peter Graves)
2564 * bug fix: the garbage collector now has much better locality
2565 behaviour, and in particular no longer treats all memory as being
2566 exhausted when many small objects point to each other in a deeply
2568 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2569 recognized as being TYPEP their class.
2570 * bug fix: the PUSHNEW documentation string has been corrected.
2571 (thanks to Vincent Arkesteijn)
2572 * bug fix: defaulting of the value for the last of an atypically
2573 large number of multiple values being bound was not being
2574 performed correctly on the Alpha or PPC platforms
2575 * optimization: implemented multiplication as a modular
2576 (UNSIGNED-BYTE 32) operation on the x86 backend.
2577 * optimization: SEARCH on simple-base-strings can now be open-coded.
2578 (see also contrib/compiler-extras.lisp for inspiration for
2579 teaching the compiler about the Boyer-Moore algorithm).
2580 * value, returned by MAX (and MIN) called with several EQUALP, but
2581 not EQL, arguments now does not depend on compiler settings.
2582 * fixed some bugs revealed by Paul Dietz' test suite:
2583 ** in stack analysis liveness information is propagated from
2584 non-local entry points.
2585 ** pathwise CAST removing failed when the CAST node did not start
2587 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2588 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2590 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2591 host is already defined.
2592 ** RENAME-FILE works on streams instead of signalling an internal
2594 ** PEEK-CHAR uses the current readtable when determining whether
2595 or not a character is whitespace.
2596 ** MERGE-PATHNAMES handles the case when the pathname does not
2597 specify a name while the default-pathname specifies a version.
2598 ** Pathnames now stand a better chance of respecting print/read
2600 ** Attempting to use standardized file system operators with a
2601 pathname with invalid :DIRECTORY components signals a
2603 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2605 ** DIRECTORY on logical pathnames is more correct.
2606 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2607 signal a TYPE-ERROR if their argument is not a stream.
2608 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2609 designator argument does not designate a stream.
2610 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2611 examining the synonym.
2612 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2614 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2615 element-type larger than ([UN]SIGNED-BYTE 32).
2617 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2618 * When built with the :SB-FUTEX feature, threaded builds now take
2619 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2620 for faster/more reliable mutex and condition variable support.
2621 * Incompatible change (but one you probably shouldn't have been using
2622 anyway): the interface and code for arbitrating between multiple
2623 threads in the same user session has been redesigned.
2624 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2625 error when called without an explicit environment argument.
2626 (thanks to Nikodemus Siivola)
2627 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2628 confused about their position. (thanks to Adam Warner and Gerd
2630 * bug fix: Pretty printing backquoted forms with unquotations in the
2631 argument list position of various code constructs such as LAMBDA
2632 now works correctly. (reported by Paul Dietz)
2633 * bug fix: Pretty printing unquotations no longer loses all
2634 stream position information.
2635 * optimization: performance of string output streams is now less
2636 poor for multiple small sequence writes.
2637 * optimization: performance of CSUBTYPEP in the presence of complex
2638 expressions involving CONS and NOT many times has been improved.
2639 (reported by Paul Dietz)
2640 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2642 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2643 streams for the REPL, for the debugger, and so on. (thanks to
2645 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2646 * fixed some bugs revealed by Paul Dietz' test suite:
2647 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2648 is ignored if it is too small, rather than propagating through
2649 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2650 ** extremely complex negations of CONS types were not being
2651 sufficiently canonicalized, leading to inconsistencies in
2653 ** VALUES tranformer lost derived type.
2655 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2656 * fixed a bootstrapping bug: the build process no longer assumes
2657 that the various BOOLE-related constants have the same value in
2658 host and target lisps. (noted by Paul Dietz' test suite on an
2659 SBCL binary built from CLISP)
2660 * The system can now be dynamically linked on the MIPS platform,
2661 which enables dynamic loading of foreign code from Lisp. (thanks
2662 to Ralf Baechle for discussions on the MIPS ABI)
2663 * The system now records debugging information for its own source
2664 files in a filesystem-position-independent manner, relative to
2665 the "SYS" logical host.
2666 * fixed a compiler bug: MV-LET convertion did not check references
2667 to the "max args" entry point. (reported by Brian Downing)
2668 * tweaked disassembly notes to be less confident about proclaiming
2669 some instruction as an LRA. (thanks to Brian Downing)
2670 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2671 now each have their own history, command character, and other
2672 characteristics. (thanks to David Lichteblau)
2673 * fixed some bugs revealed by Paul Dietz' test suite:
2674 ** compiler failure in compiling LOGAND expressions including a
2676 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2677 shift greater than 32.
2678 ** FLUSH-DEST did not mark blocks for type check regeneration.
2679 ** HANDLER-CASE failed to accept declarations in handler clauses
2680 in some circumstances.
2682 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2683 * New code in contrib/sb-introspect (still probably not entirely
2684 stable yet) provides some support for smart Lisp development
2685 environments like SLIME.
2686 * The conditions signalled for errors occurring when loading .fasl
2687 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2688 in a way which should help ASDF recover gracefully.
2689 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2690 stuff has been cleaned up. If you code contrib/ stuff, this might
2691 affect you, and you can look at contrib/README, contrib/STANDARDS,
2692 and/or the 0.8.4.27 diff to check.
2693 * In full calls the compiler now does not generate checks for declared
2694 argument types for all arguments.
2695 * various threading fixes
2696 ** and some experimental patches which didn't make it into
2697 the main tree for this release, but which are shipped in
2698 contrib/experimental-thread.patch as a possible fix for some
2699 failures (deadlock, spinning...) in GC-intensive multithreaded
2701 * fixed PPC build problem (source code incompatibility of different
2702 library versions): added offsetof() hackery which attempts to divine
2703 where glibc maintainers put uc_mcontext today
2704 * fixed bug 282: compiler does not trust type assertions while passing
2705 arguments to a full call.
2706 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2707 &OPTIONAL VALUES type parameter.
2708 * fix bug 214: algorithm for noting rejected templates is now more
2709 similar to that of template seletion. (also reported by rydis on
2711 * fixed bug 141b: printing backquoted information readably and prettily
2712 inserts a space where necessary.
2713 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2714 (CHARACTER 10) are now reported as errors, rather than propagated
2715 as unknown types. (reported by piso on #lisp)
2716 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2717 with values NIL and :ERROR. (thanks to Milan Zamazal)
2718 * fixed bug 191c: CLOS now does proper keyword argument checking as
2719 described in CLHS 7.6.5 and 7.6.5.1.
2720 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2721 longer raise an error; further, using a list as a for-as-arithmetic
2722 counter now raises a meaningful error.
2723 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2724 now understood by sequence creation functions such as MAKE-SEQUENCE
2726 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2727 a STRING-INPUT-STREAM.
2728 * compiler enhancement: SIGNUM is now better able to derive the type
2730 * type declarations inside WITH-SLOTS are checked. (reported by
2732 * fixed some bugs revealed by Paul Dietz' test suite:
2733 ** incorrect optimization of TRUNCATE for a positive first
2734 argument and negative second.
2735 ** compiler failure in let-convertion during flushing dead code.
2736 ** compiler failure while deriving type of TRUNCATE on an
2737 interval, containing 0.
2738 ** ASH of a negative bignum by a negative bignum count now returns
2740 ** intersection of CONS types now canonicalizes properly, fixing
2741 inconsistencies in SUBTYPEP.
2743 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2744 * incompatible change: The --disable-debugger command line
2745 option now clobbers the debugger at a more fundamental
2746 level, by redefining #'INVOKE-DEBUGGER instead of by
2747 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2748 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2749 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2750 enter the debugger REPL and then suffer recursive errors
2751 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2752 while under the new system, BREAK in --disable-debugger mode
2753 terminates the system just as an unhandled error would.
2754 * fixed compiler performance when processing loops with a step >1;
2755 * bug fix: DOCUMENTATION now retrieves generic function
2756 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2757 support has been systematized, and now supports the methods
2758 specified by ANSI, along with a default method and a method for
2759 slot documentation. (reported by Nathan Froyd)
2760 * bug fix: effective methods associated with a generic function are
2761 no longer cached over a change of that generic function's method
2762 combination. (reported by Andreas Fuchs)
2763 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2764 names a non-existent directory. (thanks to Andreas Fuchs)
2765 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2766 the PPC platform, lead to stack corruption; this has been fixed.
2767 (reported by Rainer Joswig)
2768 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2769 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2770 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2771 arithmetic in general, is now much more reliable on the PPC
2773 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2774 platform now returns the right answer.
2775 * optimization: restored some effective method precomputation in
2776 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2777 precomputation is now tunable.
2778 * optimization: compiler-internal data structure use has been
2779 reviewed, and changes have been made that should improve the
2780 performance of the compiler by about 20%.
2781 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2782 simple-base-strings and simple-bit-vectors is improved.
2783 * optimization: the optimization of 32-bit logical and arithmetic
2784 functions introduced in version 0.8.3 on the x86 has been
2785 implemented on the mips, ppc and sparc platforms; an
2786 implementation of the same facility, but for 64-bit arithmetic,
2787 has been added for the alpha.
2788 * microoptimization: the compiler is better able to make use of the
2789 x86 LEA instruction for multiplication by constants.
2790 * bug fix: in some situations compiler did not report usage of
2791 generic arithmetic in (SPEED 3) policy.
2792 * bug 145b fix: compiler used wrong type specifier while converting
2793 MEMBER-types to numeric.
2794 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2796 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2797 index argument now works properly on the Alpha platform.
2798 * bug fix: floating point exception treatment on the Alpha platform
2800 * bug fix: FILE-POSITION works much better on string input and
2801 output streams. (thanks to Nikodemus Siivola)
2802 * bug fix: many threading/garbage collection symptoms sorted.
2803 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2804 might be pseudo-atomic.
2805 * internal change: Stopping for GC is now done with signals not
2806 ptrace. GC is now done in whichever thread wanted it, instead of
2808 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2810 * bug fix: result form in DO is not contained in the implicit
2812 * incompatible change: ICR structure is changed; the value part of
2813 CONTINUATION is now called LVAR; corresponding functions are
2814 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2815 * added type deriver for ISQRT (thanks to Robert E. Brown).
2816 * bug fix: better support for loading from the command line when an
2817 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2819 * fixed some bugs revealed by Paul Dietz' test suite:
2820 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2821 ** ROUND and FROUND now give the right answer when given very
2822 small float arguments.
2823 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2825 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2826 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2827 ** LCM should return a non-negative integer.
2828 ** PARSE-INTEGER returned the index of a terminator instead of the
2829 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2830 ** PARSE-INTEGER returned an incorrect index being applied to a
2832 ** LCM with two arguments of 0 returns 0 rather than signalling
2834 ** unsigned addition of a 32-bit constant with the high bit set no
2835 longer causes an internal compiler error on the x86.
2836 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2837 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2838 with negative last argument.
2839 ** byte specifiers with zero size and position no longer cause
2840 an error during type derivation.
2841 ** bignum multiplication on the Alpha platform now returns the
2843 * porting: The system now builds on SuSE AMD64, although it still
2844 generates a 32-bit binary.
2845 * .fasl file incompatibility: The fasl file version number has
2846 been incremented (because of the changes to internal compiler
2847 data structures referred to above).
2849 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2850 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2851 more accurately, on the Darwin kernel running on PowerPC hardware.
2852 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2853 * Compiler code deletion notes now signal a condition of type
2854 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2855 an associated MUFFLE-WARNING restart.
2856 * The compiler now performs limited argument count validation of
2857 constant format strings in FORMAT, and where appropriate in ERROR,
2858 CERROR and WARN. (thanks to Gerd Moellmann)
2859 * New ASDF-INSTALL contrib can be used for automatic download and
2860 installation of third-party Lisp code from CCLAN or other sites
2862 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2863 which forces another thread to execute a function supplied by the
2865 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2866 now accept and act upon their :ELEMENT-TYPE keyword argument.
2867 (reported by Martin Atzmueller, Edi Weitz)
2868 * bug fix: FILE-POSITION now accepts position designators up to
2869 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2870 is the greater. (thanks to Patrik Nordebo)
2871 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2872 Kalle Olavi Niemitalo)
2873 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2874 printable. (reported by Eric Marsden)
2875 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2876 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2877 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2878 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2879 (thanks to Stig Sandoe)
2880 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2881 UNREAD-CHAR now correctly includes the unread character in the
2882 target sequence. (thanks to Gerd Moellmann)
2883 * bug fix in threaded builds: the system can now be suspended and
2884 resumed by shell job control with minimal disruption.
2885 * bug fixes in times and timezones >2038AD
2886 * better handling of "where is GNU make?" problem in build scripts
2887 (thanks to Nikodemus Siivola)
2888 * new optimization: inside a named function any reference to a
2889 function with the same name is considered to be a self-reference;
2890 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2891 optimization quality.
2892 * new optimization on x86: logical functions and + now have
2893 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2894 used when the result is truncated to 32 bits.
2895 * VALUES declaration is partially enabled.
2896 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2897 * fixed some bugs revealed by Paul Dietz' test suite:
2898 ** The system now obeys the constraint imposed by
2899 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2900 types form a lattice under type intersection.
2901 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2902 ** ASSOC now ignores NIL elements in an alist.
2903 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2904 and (1+ MOST-POSITIVE-FIXNUM) answers.
2905 ** The addition of a method with invalid qualifiers to a generic
2906 function does not cause an error to be signalled immediately;
2907 a warning is signalled, and the error is generated only on
2908 calling the generic function.
2909 * changed .fasl file version number, in part to add type codes for
2910 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2911 obscure ANSI requirements
2913 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2914 * fixed bug 148: failure to inline-expand a local function left
2915 garbage, confusing the compiler.
2916 * fixed bugs 3cd: structure slot readers perform type check if the
2917 slot can have an invalid value (i.e. it is either not initialized
2918 or can be written with a less specific slot writer).
2919 * bug fix: the compiler now traps array references to elements off
2920 the end of an array; previously, the bounds checking in some
2921 circumstances could go off-by-one.
2922 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2924 * type declarations for array element types now obey the description
2925 on the CLHS page "Declaration TYPE", as per discussions on
2926 sbcl-help around 2003-05-08. This means that a declaration
2927 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2928 declaration, all references to BAR will be asserted or assumed
2929 (with THE, so dependent on compiler policy) to involve objects of
2930 type FOO. Note that no such declaration is implied in
2931 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2932 * declared types of functions from the "Conditions"
2933 chapter. (reported by Paul Dietz)
2934 * bug fix: CERROR accepts a function as its first argument.
2935 * bug fix: NTH an NTHCDR accept a bignum as index
2936 arguments. (reported by Adam Warner)
2937 * optimization: character compare routines now optimize comparing
2938 against a constant character. (reported by Gilbert Baumann)
2939 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2940 argument now works properly on the MIPS platform.
2941 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2942 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2943 declared to have a functional type, does not cause an error
2945 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2946 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2947 * fixed bug: if last continuation of a deleted block has a
2948 destination, this destination should be deleted too. (reported by
2950 * fixed a bug in the bootstrap process: the host compiler's values
2951 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2952 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2953 test case from Patrik Nordebo)
2954 * improved the ability of the disassembler on the PPC platform to
2955 provide helpful disassembly notes.
2956 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2957 the class in more cases than previously.
2958 * bug fix: FILE-POSITION now understands :START and :END for
2959 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2960 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2961 8) data. (thanks to David Lichteblau)
2962 * bug fix: it is possible to add a method to a generic function
2963 without lambda list.
2964 * bug fix: reader failed to signal END-OF-FILE inside an
2965 object representation. (reported by Nikodemus Siivola)
2966 * fixed some bugs revealed by Paul Dietz' test suite:
2967 ** LAST and [N]BUTLAST should accept a bignum.
2968 ** condition slot accessors are methods.
2969 ** (VECTOR NIL) is a subtype of STRING.
2971 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2972 * minor incompatible change: some nonsensical specialized lambda
2973 lists (used in DEFMETHOD) which were previously ignored now signal
2975 * minor incompatible change: the system is now aware of the types of
2976 variables in the COMMON-LISP package, and will signal errors for
2977 most violations of these type constraints (where previously they
2978 were silently accepted).
2979 * minor incompatible change: COMPILE-FILE now uses the freedom
2980 afforded (ANSI 3.2.2.3) to use derived function types for
2981 functions defined in the same file. This also permits the system
2982 to warn on static type mismatches and function
2983 redefinition. (Currently it does not work with high DEBUG level.)
2984 * minor incompatible change: VALUES declaration is disabled.
2985 * When issuing notes, the compiler now signals a condition of type
2986 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2987 restart for use in user handlers. It is expected that the
2988 COMPILER-NOTE condition will eventually become a condition
2989 supertype to a hierarchy of note types, which will then be
2990 handleable in a similar fashion. However, at the moment, no such
2991 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2992 but it's an implementation detail, not a classification for the
2994 * Changes in type checking closed the following bugs:
2995 ** type checking of unused values (192b, 194d, 203);
2996 ** template selection based on unsafe type assertions (192c, 236);
2997 ** type checking in branches (194bc).
2998 * A short form of VALUES type specifier has ANSI meaning (it has
2999 increased the number of situations when SBCL cannot perform type
3001 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3002 keywords or constants is permissible.
3003 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3004 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3005 classes. (thanks to Antonio Martinez)
3006 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3007 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3008 argument to a pretty-print stream. Instead, it leaves any such
3009 operations to DESCRIBE-OBJECT methods.
3010 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3012 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3013 lambda lists. (thanks to David Lichteblau)
3014 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3015 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3016 (thanks to Gerd Moellmann)
3017 * fixed bug 246: increased compilation speed of long
3018 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3020 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3022 * DEFCONSTANT now throws a condition of type
3023 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3024 constant to a non-EQL value; CONTINUE and ABORT restarts
3025 respectively change and preserve the value.
3026 * fixed bug 63: The code walker, part of the implementation of CLOS,
3027 is now better at handling symbol macros.
3028 * bug fix: There is no longer an internal implementation type named
3029 CL:LENGTH. (reported by Raymond Toy)
3030 * bug fix: In macro-like defining macros/special operators the
3031 implicit block does not enclose the lambda list.
3032 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3033 enclosed in parentheses) are not suitable as type specifiers, and
3034 their use properly signals an error now.
3035 * bug fix: An argument count mismatch for a type specifier in code
3036 being compiled no longer causes an unhandled error at compile
3037 time, but signals a compile-time warning.
3038 * fixed simple vector readable printing
3039 * bug fix: DESCRIBE takes more care over whether the class
3040 precedence list slot of a class is bound before accessing it.
3041 (reported by Markus Krummenacker)
3042 * bug fix: FORMATTER can successfully compile pretty-printer format
3043 strings which use variants of the ~* directive inside.
3044 * bug fix: SEARCH now applies its TEST predicate to the elements of
3045 the arguments in the correct order. (thanks to Wolfhard Buss)
3046 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3047 in type inference. (thanks to Robert E. Brown)
3048 * bug fix: Reading in symbols with an explicit package name of ""
3049 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3050 * fixed some bugs revealed by Paul Dietz' test suite:
3051 ** NIL is now allowed as a structure slot name.
3052 ** Arbitrary numbers, not just REALs, are allowed in certain
3053 circumstances in LOOP for-as-arithmetic clauses.
3054 ** Multiple class redefinitions before slot access no longer
3055 causes a type error.
3056 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3057 association between the name and a class.
3058 ** Generic functions with non-standard method-combination and over
3059 five methods all of which return constants no longer return NIL
3060 after the first few invocations. (thanks to Gerd Moellmann)
3061 ** CALL-NEXT-METHOD with no arguments now passes the original
3062 values of the arguments, even in the presence of assignment.
3063 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3065 ** DISASSEMBLE works with closures and funcallable instances.
3066 ** ADD-METHOD now returns the generic function, not the new method.
3067 ** FIND-METHOD signals an error if the lengths of the specializers
3068 is incompatible with the generic function, even if the ERRORP
3070 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3071 which its argument is a member.
3072 ** DEFCLASS only redefines the class named by its class-name
3073 argument if that name is the proper name of the class;
3074 otherwise, it creates a new class.
3075 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3076 of the UNBOUND-SLOT condition to the name of the slot.
3077 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3078 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3079 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3080 SLOT-MAKUNBOUND in the specified fashion.
3082 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3083 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3084 cross-compilation host. As a consequence, we can now bootstrap our
3085 way up to SBCL starting with a bare gcc toolchain and human-readable
3086 source code (first the source to CLISP, then the source to SBCL).
3087 * A contributed module containing a partial implementation of the
3088 simple-streams interface has been included. (thanks to Rudi
3090 * A contributed module implementing the RFC1321 Message Digest
3091 Algorithm, known as MD5, has been included.
3092 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3093 no longer has any effect, as the code controlled by this feature
3094 has been deleted. (As far as we know, no-one has ever built using
3095 this feature, and its semantics were confused in any case).
3096 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3097 work on conditions (as required by the ANSI specification),
3098 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3099 expected behaviour on conditions. Users should note, however,
3100 that such behaviour is not required by the ANSI specification,
3101 and so use of this behaviour may render their code unportable.
3102 * fixed some bugs revealed by Paul Dietz' test suite:
3103 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3105 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3106 ** SXHASH on condition objects no longer returns NIL.
3107 ** :ALLOCATION :CLASS slots are better treated; their values are
3108 updated on class redefinition, and initforms inherited from
3109 superclasses are applied.
3110 ** REMOVE-METHOD returns its generic function argument even when
3111 no method was removed.
3112 ** SHARED-INITIALIZE now initializes the values of the requested
3113 slots, including those with :ALLOCATION :CLASS.
3114 ** ALLOCATE-INSTANCE now works on structure classes defined via
3115 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3117 ** SLOT-EXISTS-P now works on conditions, as well as structures
3119 ** MAKE-LOAD-FORM now has the required methods on
3120 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3121 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3122 keyword, and now implements the SLOT-NAMES argument.
3123 ** methods with &OPTIONAL arguments no longer allow too many
3124 arguments to be passed in the call without error.
3125 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3126 option is consistent with the required arguments of the generic
3127 function lambda list.
3128 * bug fix: REQUIRE accepts a string designator. (Thanks to
3130 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3131 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3132 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3133 * bug fix: The compiler now checks for duplicated variables in macro
3135 * bug fix: SETQ on globals returns the correct value.
3136 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3137 not a COMPILER-ERROR (followed by some other strange error on
3138 choosing the CONTINUE restart).
3139 * bug fix: make.sh and friends are now more consistent in the way that
3140 they look for GNU "make".
3142 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3143 * experimental native threads support (on x86 Linux >=2.4 only).
3144 This is not compiled in by default: you need to add :SB-THREAD to
3145 the target features. See the "Beyond ANSI" chapter of the manual
3147 * fix for longstanding nonANSIism: The old distinction between
3148 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3149 The return value from CL:FIND-CLASS is now a CLOS class, and
3150 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3151 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3152 * An interface to the MetaObject Protocol, as described in Kiczales,
3153 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3154 MIT Press, 1991, is available from the SB-MOP package.
3155 * incompatible change: the SB-PCL package should now be considered
3156 a private implementation detail, and no longer a semi-private MOP
3158 * minor incompatible change: due to rearrangement for threads, the
3159 control stack and binding stack are now allocated at arbitrary
3160 addresses instead of being hardcoded per-port. Users affected by
3161 this probably have to be doing advanced things with shared
3162 libraries, and will know who they are.
3163 * minor incompatible change: Previously, all --eval forms used were
3164 processed with READ before any of them were processed with EVAL.
3165 Now each --eval form is processed with both READ and EVAL before
3166 the next --eval form is processed. (Thus package operations like
3167 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3168 work as the user might reasonably expect.)
3169 * minor incompatible change: *STANDARD-INPUT* is now only an
3170 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3172 * minor incompatible change: Y-OR-N-P is now character-oriented,
3173 not line oriented. Also, YES-OR-NO-P now works without errors.
3174 (thanks to Antonio Martinez)
3175 * sb-aclrepl module improvements: an integrated inspector, added
3176 repl features, and a bug fix to :trace command.
3177 * Known functions, which cannot be open coded by the backend, are
3178 considered to be able to check types of their arguments. (fixing
3179 a bug report by Nathan J. Froyd)
3180 * fixed a bug in computing method discriminating functions: It is
3181 now possible to define methods specialized on classes which have
3182 forward-referenced superclasses. (thanks to Gerd Moellmann)
3183 * fixed evaluation order in optional entries (reported by Gilbert
3185 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3186 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3187 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3188 required-by-AMOP NAME argument, as well as CLASS and
3189 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3190 * fixed bug 20: DEFMETHOD can define methods using names that are
3191 not the proper names of classes to designate class specializers.
3192 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3193 Unix file descriptors, instead of blowing up. (thanks to Antonio
3195 * Garbage collection refactoring: user-visible change is that a
3196 call to the GC function during WITHOUT-GCING will not do garbage
3197 collection until the end of the WITHOUT-GCING. If you were doing
3198 this you were probably losing anyway.
3199 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3200 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3201 TYPEP the latter but not the former.
3202 * The compiler issues a full WARNING for calls to undefined functions
3203 with names from the CL package.
3204 * MAP-INTO for a vector destination is open coded. (reported by
3205 Brian Downing on c.l.l)
3206 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3207 documentation string.
3208 * fixed some bugs revealed by Paul Dietz' test suite:
3209 ** COPY-ALIST now signals an error if its argument is a dotted
3211 ** Condition slots are now accessed more correctly in the presence
3212 of multiple initargs for a given slot.
3213 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3214 exclude restarts of the same name associated with a different
3216 ** DEFCLASS of forward-referenced classes with another
3217 forward-referenced class in the superclasses list no longer
3219 ** Condition slots are now initialized once each, not multiple
3220 times. (thanks to Gerd Moellmann)
3221 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3223 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3224 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3225 arguments contain duplicated elements.
3226 ** RESTART-CASE understands local macros.
3227 ** RESTART-CASE associates exactly its own restarts with a condition.
3228 ** ENDP in safe mode checks its argument to be of type LIST.
3229 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3230 in question is unbound.
3231 ** Optimized MAKE-INSTANCE functions no longer cause internal
3232 assertion failures in the presence of duplicate initargs.
3233 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3234 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3235 propagated even in the ctor optimized implementation of
3237 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3239 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3240 * a better implementation of SXHASH on (simple) bit vectors,
3241 measured both in execution speed and in distribution of results
3242 over the positive fixnums, has been installed. Likewise, a better
3243 implementation of EQUAL for simple bit vectors is now available.
3244 * fixed CEILING optimization for a divisor of form 2^k.
3245 * fixed bug 240 (emitting extra style warnings "using the lexical
3246 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3247 by Antonio Martinez)
3248 * fixed SXHASH, giving different results for NIL depending on type
3249 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3250 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3251 variable. (found by Rolf Wester)
3252 * SBCL does not ignore type declarations for special
3253 variables. (reported by rif on c.l.l 2003-03-05)
3254 * some bug fixes in contrib/sb-aclrepl/
3255 * fixed some bugs revealed by Paul Dietz' test suite:
3256 ** a bug in the CONS type specifier, whereby the CAR and CDR
3257 types got intertwined, has been fixed;
3258 ** the type system is now able to reason about the interaction
3259 between INTEGER and RATIO types more completely;
3260 ** APPEND, [N]REVERSE and NRECONC check that those their
3261 arguments, which must be proper lists, are really so;
3262 ** An array specialized to be unable to hold elements has been
3263 implemented, as required -- yes, really -- by ANSI;
3264 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3265 on malformed property lists;
3267 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3268 * incompatible packaging change: in line with Unix convention,
3269 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3270 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3271 * REQUIRE and PROVIDE are now optionally capable of doing something
3272 useful. See the documentation string for REQUIRE.
3273 * infrastructure for a managed SBCL contrib system: contributed
3274 modules in this release include:
3275 ** the ASDF system definition facility;
3276 ** an interface to the BSD Sockets API;
3277 ** an ACL-like convenience interface to the repl;
3278 (thanks to Kevin Rosenberg)
3279 ** an implementation of ROTATE-BYTE, with efficient implementation
3281 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3282 gives a closer approximation to the right answer than previously.
3283 (thanks to Raymond Toy)
3284 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3285 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3286 optional environment arguments, as required by ANSI.
3287 * fixed bugs in other functions taking environment objects, allowing
3288 calls with an explicit NIL environment argument to be compiled
3290 * fixed bug 228: primary return values from
3291 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3292 COMPILE or FUNCTION.
3293 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3294 :TYPE will no longer signal errors on innocuous objects.
3295 * fixed bug 231b: SETQ is better at respecting type declarations in
3296 the lexical environment.
3297 * fixed a bug in DEFCLASS: classes named by symbols with no or
3298 unprintable packages can now be defined.
3299 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3300 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3301 * fixed compiler failure related to checking types of functions.
3302 (reported by Robert E. Brown)
3303 * the compiler is now much more consistent in its error-checking
3304 treatment of bounding index arguments to sequence functions: in
3305 (SAFETY 3) code, errors will be signalled in almost all cases if
3306 invalid sequence bounding indices are passed to functions defined
3307 by ANSI to operate on sequences.
3308 * fixed a bug in the build procedure: documentation of SBCL-specific
3309 packages is now preserved and available in the final Lisp image.
3310 * lifted FDEFINITION lookup out of loops in the implementation of
3311 many list operations. (thanks to Robert E. Brown)
3312 * fixed a bug in the reader: the #n# reader macro now works for
3313 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3314 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3315 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3316 E. Brown and Tony Martinez respectively)
3317 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3318 SLOT-BOUNDP: in particular, it is now able to optimize them much
3319 better, and is now not vulnerable to having packages renamed.
3320 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3321 always called when a slot is not present in an instance. (thanks
3323 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3324 longer signal an error on unreading a character following EOF on
3325 the previous constituent stream. (thanks to Tony Martinez)
3326 * fixed some bugs revealed by Paul Dietz' test suite:
3327 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3328 not just nonnegative fixnums;
3329 ** the logical bit-array operators such as BIT-AND now accept an
3330 explicit NIL for their "opt-arg" argument (to indicate a
3331 freshly-consed result bit-array);
3332 ** ELT now signals an error on an invalid sequence index in safe
3334 ** the type system is now cleverer about negations of numeric
3335 types, and consequently understands the BIGNUM and RATIO types
3337 ** the type system is now cleverer about the interaction between
3338 INTEGER and RATIO types: while bugs still remain, many more
3339 cases are accurately computed;
3340 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3341 if it is in the last clause;
3342 ** CONSTANTLY now correctly returns a side-effect-free function in
3344 ** DECLARE is no longer treated as a special-operator; in
3345 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3346 * incremented fasl file version number due to the change in the
3347 DEFSTRUCT-SLOT-DESCRIPTION structure.
3349 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3350 * minor incompatible change: code processed by the "interpreter" or
3351 EVAL now has a compilation optimization policy of (DEBUG 2)
3352 (changed from (DEBUG 1)) to improve debuggability of interactive
3353 development, and to allow the use of the debug RETURN command in
3355 * an experimental implementation of the RETURN command for the
3356 debugger has been included. (thanks to Frederik Kuivinen)
3357 * fixed bug 62: constraints were not propagated into a loop.
3358 * fixed bug in embedded calls of SORT (reported and investigated by
3360 * fixed some bugs revealed by Paul Dietz' test suite:
3361 ** printing and reading of arrays with some dimensions having
3362 length 0 (thanks to Gerd Moellmann);
3363 ** BOA constructor with &AUX argument without a default value does
3364 not cause a type error;
3365 ** CONSTANTP now returns true for all self-evaluating objects.
3367 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3368 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3369 accessors that are related by inheritance, as specified in the
3370 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3371 Valtteri Vuorikoski)
3372 * The compiler is now able to inline functions that were defined in
3373 a complex lexical environment (e.g. inside a MACROLET).
3374 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3376 * Support for the upcoming FreeBSD-5.0 release has been included.
3377 (thanks to Dag-Erling Smorgrav)
3378 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3379 effect when it is not in a toplevel context.
3380 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3381 stand a better chance of being correct. (thanks to Gerd
3383 * fixed bug in COERCE, which now signals an error on coercing a
3384 rational to a bounded real type which excludes the expected
3386 * The compiler is now able to derive types more accurately from the
3387 COERCE and COMPILE functions.
3388 * fixed bug 223: functional binding is considered to be constant
3389 only for symbols in the CL package.
3390 * fixed bug 231: SETQ did not check the type of a variable being set
3391 (reported by Robert E. Brown)
3392 * A new optimization for MAKE-INSTANCE has been included, fixing
3393 various bugs (including relating to :ALLOCATION :CLASS slots and
3394 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3396 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3397 ** As required by ANSI, LOOP now disallows anonymous collection
3398 clauses such as COLLECT I in conjunction with aggregate boolean
3399 clauses such as THEREIS (= I 1);
3400 ** LOOP now signals an error when any variable is reused in the
3401 same loop (including the potentially useful construct analogous
3402 to WITH A = 1 WITH A = (1+ A);
3403 ** IT is only a special loop symbol within the first clause of a
3404 conditional loop clause;
3405 ** LOOP with a typed iteration variable over a hashtable now
3406 signals a type error iff it should.
3407 * fixed some other bugs revealed by Paul Dietz' test suite:
3408 ** FILE-STREAM now names the class previously known as FD-STREAM;
3409 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3410 argument) no longer signals an error;
3411 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3412 of the slot symbol, rather than using the current package
3413 ((:CONC-NAME "") continues to intern the slot's name in the
3415 * incremented fasl file version number, because of the incompatible
3416 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3417 of the new implementation of DEFINE-COMPILER-MACRO.
3419 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3420 * Support for building SBCL for MIPS platforms running in
3421 little-endian mode has now been checked in, and basic
3422 functionality on said platforms verified.
3423 * minor incompatible change: PCL now records the pathname of a file
3424 in which methods and the like are defined, rather than its
3426 * minor incompatible change: TRUENAME now considers the truename of
3427 a file naming a directory to be the pathname with :DIRECTORY
3428 component indicating that directory.
3429 * minor incompatible change: a NAMED clause in the extended form of
3430 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3431 reason for the previous behaviour is unclear.
3432 * more systematization and improvement of CLOS and MOP conformance
3433 in PCL (thanks to Gerd Moellman and Pierre Mai):
3434 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3436 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3437 primary methods with no specializers;
3438 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3440 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3441 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3442 FORWARD-REFERENCED-CLASSes; error reporting on
3443 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3445 ** SXHASH on CLOS instances now uses a slot internal to the
3446 instance to return different numbers on distinct instances,
3447 while preserving the same return value through invocations of
3449 ** DEFMETHOD signals errors when methods with longer incongruent
3450 lambda lists are added to generic functions;
3451 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3452 CLASS, as specified in AMOP;
3453 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3454 on the order returned by the primary method for classes of
3455 class STANDARD-CLASS;
3456 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3457 * fixed some bugs shown by Paul Dietz' test suite:
3458 ** DOLIST puts its body in TAGBODY;
3459 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3461 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3462 value producing form;
3463 ** if more variables are given to PROGV than values, extra
3464 variables are bound and made to have no value;
3465 ** NSUBSTITUTE on list arguments gets the right answer with
3467 ** ELT signals an error of type TYPE-ERROR when the index argument
3468 is not a valid sequence index;
3469 ** LOOP signals (at macroexpansion time) an error of type
3470 PROGRAM-ERROR when duplicate variable names are found;
3471 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3472 ** FUNCALL of special-operators now cause an error of type
3474 ** PSETQ now works as required in the presence of side-effecting
3475 symbol-macro places;
3476 ** NCONC accepts any object as its last argument;
3477 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3479 ** loop-for-as-package does not require a package to be explicitely
3481 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3483 * fixed bug 166: compiler preserves "there is a way to go"
3484 invariant when deleting code.
3485 * fixed bug 172: macro lambda lists with required arguments after
3486 &REST arguments now cause an error to be signalled. (thanks to
3488 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3489 bug. (thanks to Matthew Danish)
3490 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3492 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3493 arguments are lexically rebound. (thanks to Gerd Moellmann and
3495 * fixed bug 194: error messages are now more informative when there
3496 is no primary method applicable in a call to a generic
3497 function. (thanks to Gerd Moellmann)
3498 * fixed bug in command line argument checking (thanks to Julian
3500 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3501 (thanks to Matthew Danish)
3502 * incremented fasl file version number, because of the
3503 SXHASH-related changes in the layout of CLOS data structures
3505 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3506 * minor incompatible change: The runtime (the Unix executable named
3507 "sbcl") is now much pickier about the .core files it will load.
3508 Essentially it now requires .core files to descend from the same
3509 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3510 as the runtime does. (The intent is to prevent the crashes which
3511 can occur, and which can even be reported as mysterious failures,
3512 when people patch the sources or change the build parameters
3513 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3514 sbcl and .core files.)
3515 * fixed bug: VALUES-LIST is no longer optimized away.
3516 * fixed bug 142: The FFI conversion of C string values to Lisp
3517 string values no longer conses excessively. (thanks to Nathan
3518 Froyd porting Raymond Toy's fix to CMU CL)
3519 * began to systematize and improve MOP conformance in PCL (thanks to
3520 Nathan Froyd, Gerd Moellman and Pierre Mai):
3521 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3523 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3524 ** FINALIZE-INHERITANCE is now called on class finalization;
3525 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3526 argument precedence order.
3527 * fixed bug 202: The compiler no longer fails on functions whose
3528 derived types contradict their declared type.
3529 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3530 so it can be non-toplevel.
3531 * The fasl file version number has changed (because of the new
3532 implementation of DEFMACRO).
3533 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3534 safe code, that any length requirement by their type-specifier
3535 argument is valid. The exceptions to this are described in bug
3537 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3538 functions now check the types of their inputs as required by ANSI.
3539 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3540 introduced symbol is DECLAREd to be SPECIAL.
3541 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3542 * fixed bug: :COUNT argument to sequence functions may be negative
3543 * fixed bug: body of DO-SYMBOLS may contain declarations
3544 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3545 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3546 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3547 symbol macro only once
3548 * fixed printing of call frame when argument list is unavailable
3549 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3550 * compiler no longer signals WARNING on unknown keyword
3553 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3554 * A beta-quality port to the mips architecture running Linux,
3555 based on the old CMUCL backend, has been made. It has been tested
3556 on a big-endian kernel, and works sufficiently well to be able to
3557 rebuild itself; it has not been tested in little-endian mode.
3558 * fixed an inconsistency between gencgc.c and purify.c which made
3559 dumping/loading .core files unreliable
3560 * fixed bug 120a: The compiler now deals correctly with IFs where
3561 the consequent is the same as the alternative, instead of
3562 misderiving the return type. (thanks to Alexey Dejneka)
3563 * fixed bug 113: Logical pathnames are now dumpable (the logical
3564 host is resolved at load-time, throwing an error if it is not
3566 * fixed bug 174: FORMAT's error message is slightly clearer when a
3567 non-printing character is used in a format directive.
3568 * fixed several bugs in compiler checking of type declarations, i.e.
3569 violations of the Python "declarations are assertions" principle
3570 (thanks to Alexey Dejneka)
3571 * fixed several bugs in PCL's error checking (thanks to Gerd
3573 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3575 * fixed bug in compilation of functions as first class values
3576 (thanks to Antonio Martinez-Shotton)
3577 * The compiler's handling TYPE-ERRORs which it can prove will
3578 inevitably happen at runtime has been cleaned up and corrected
3579 in several ways. (thanks to Alexey Dejneka)
3580 * improved argument type checking for various basic arithmetic
3581 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3582 lavished on them in the past that they can be compiled in many
3583 ways in different special cases
3584 * fixed bug 181: compiler checks validity of user supplied type
3586 * cleaned up code flushing in optimization: Function calls which
3587 should signal errors for safety purposes (e.g. which ANSI says
3588 should signal errors when their arguments are of incorrect type)
3589 are no longer optimized away.
3590 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3591 * incremented fasl file version number, because changes in the
3592 implementation of sequence functions like COERCE caused
3593 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3594 in old inline expansions) to become undefined. (Actually these
3595 changes were later undone, so we might very well be binary
3596 compatible with 0.7.7 after all, but leaving the version number
3597 incremented seemed like the simplest and most conservative
3600 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3601 * An alpha-quality port to the parisc architecture running Linux,
3602 based on the old CMUCL backend, has been made. This, even more so
3603 than the other backends, should be considered still a work in
3604 progress; known problems include that the Linux kernel in 64-bit
3605 mode does not propagate the correct sigcontext structure to
3606 userspace, and consequently SBCL on a parisc64 kernel will not
3608 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3609 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3610 suppressed.) Also now that the compiler is looking at declarations
3611 in the environment, it checks optimization declarations as well,
3612 and suppresses inlining when (> DEBUG SPEED).
3613 * More fixes have been made to treatment of floating point exception
3614 treatment and other Unix signals. In particular, floating point
3615 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3616 * The detection and handling of control stack exhaustion (infinite
3617 or very deeply nested recursion) has changed. Stack exhaustion
3618 detection is now done by write-protecting pages at the OS level
3619 and applies at all optimization settings; when found, a
3620 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3621 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3622 be caught using IGNORE-ERRORS.
3623 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3624 correctly when transforming recursive LABELS functions to LETs.
3625 (thanks to Alexey Dejneka)
3626 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3627 that are names of constants or global variables.
3628 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3629 alien routines with docstrings.
3630 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3631 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3633 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3634 (thanks to Eric Marsden)
3635 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3636 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3637 * Bug fix: The SPARC backend can now compile functions involving
3638 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3639 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3640 to a functions expecting a PATHNAME-DESIGNATOR.
3641 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3642 lambda lists. (thanks to Alexey Dejneka)
3643 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3644 functions so that EXPORTing the name of the class doesn't cause
3645 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3646 to Antonio Martinez for reporting this)
3647 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3648 constant-folded without failing an assertion. (thanks to Einar
3649 Floystad Dorum for reporting this)
3650 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3651 types are now dealt with more correctly by the compiler.
3652 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3653 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3654 behaviour analogous to RENAME-FILE. This puts its behaviour more
3655 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3657 * The fasl file version number has changed again. (because of the
3658 bug fix involving the names of PCL MAKE-INSTANCE functions)
3660 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3661 * bug fix: Floating point exceptions are treated much more
3662 consistently on the x86/Linux and PPC/Linux platforms.
3663 * Array initialization with :INITIAL-ELEMENT is now much faster for
3664 cases when the compiler cannot open code the array creation, but
3665 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3666 array accesses have also seen a speed increase.
3667 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3668 specified. (This was at the root of some bad interactions between
3669 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3671 * bug fix: Internal error arguments for undefined functions are now
3672 computed correctly on the PPC/Linux platform.
3673 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3674 Raymond Toy's patch for CMU CL)
3675 * Support for the Solaris 9 operating environment has been included
3676 (thanks to Daniel Merritt)
3677 * A very ugly but hopefully complete draft of the missing FFI chapter
3678 of the manual has been created by reformatting the corresponding
3679 CMU CL manual chapter into (currently very ugly and incoherent)
3680 DocBook and bringing it up to date for SBCL behavior. Thus, the
3681 manual is now essentially complete, at least by my extreme
3682 once-and-only-once standards, whereby it's acceptable to refer to
3683 the doc strings of SB-EXT functions as the primary documentation.
3684 * The fasl file version number has changed again, due to cleanup of
3685 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3686 is no longer a static symbol.)
3688 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3689 * SBCL now builds with OpenMCL (version 0.12) as the
3690 cross-compilation host; also, more progress has been made toward
3691 bootstrapping under CLISP.
3692 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3694 * bug 158 fixed: The compiler can now deal with integer loop
3695 increments different from 1; fixing this turned out also to fix
3697 * bug 169 fixed: no more bogus warnings about using lexical bindings
3698 despite the presence of perfectly good SPECIAL declarations (thanks
3699 to David Lichteblau)
3700 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3701 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3702 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3704 * bug 180 fixed: Method combination specifications no longer ignore
3705 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3706 * bug fix: Structure type predicate functions now check their argument
3707 count as they should.
3708 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3709 correctly. (thanks to Pierre Mai)
3710 * minor incompatible change: The --noprogrammer option is deprecated
3711 in favor of the new --disable-debugger option, which is very similar.
3712 (The major difference is that it takes effect at a slightly different
3713 time at startup, causing handling of errors in --sysinit and
3714 --userinit files will be affected differently.) The
3715 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3716 been added to allow this functionality to be controlled from ordinary
3717 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3718 maintainers, who might want to run non-interactive scripts to
3719 build SBCL cores which will later be used interactively.)
3720 * minor incompatible change: The LOAD function no longer, when given
3721 a wild pathname to load, loads all files matching that pathname.
3722 Instead, an error of type FILE-ERROR is signalled.
3724 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3725 * bug 147 fixed: The compiler preserves its block link/count
3726 invariants more correctly now so that it doesn't crash. (thanks
3728 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3730 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3731 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3732 no-value-for-this-character as NIL (as specified by ANSI).
3733 * HOST-NAMESTRING on physical pathnames now returns a string that is
3734 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3735 (thanks to Christophe Rhodes)
3736 * The Alpha port handles icache flushing more correctly. (thanks to
3738 * More progress has been made toward bootstrapping under CLISP. (thanks
3739 to Christophe Rhodes)
3740 * The fasl file format has changed again, because dynamic loading
3741 on OpenBSD (which has non-ELF object files) motivated some cleanups
3742 in the way that foreign symbols are transformed and passed around.
3743 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3744 is no longer treated as whitespace by the reader, but instead as
3745 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3746 |A
\7fB|, instead of A as it used to.
3748 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3749 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3750 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3751 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3752 this as of 0.7.1.45, but was left out of the previous news section
3753 (thanks to Dan Barlow)
3754 * SBCL now runs on the Solaris operating system on SPARC architectures
3755 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3756 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3757 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3758 * SPARC backend cleanups, allowing builds of cores optimized for V8
3759 and V9 SPARCS, and also emission of code targeted to a particular
3760 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3762 * SBCL is closer to bootstrapping under CLISP, thanks to various
3763 fixes by Christophe Rhodes.
3764 * The fasl file format has changed again, to allow the compiler's
3765 INFO database to support symbol macros.
3766 * The user manual (in doc/) is formatted into HTML more nicely.
3767 (thanks to coreythomas)
3768 * The system is smarter about SUBTYPEP relationships, especially
3769 those involving NOT types (including types such as ATOM which are
3770 represented internally using NOT types). Thus SUBTYPEP is less
3771 likely to return (VALUES NIL NIL) in general, and in particular
3772 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3774 * The fasl file format has changed again, because the internal
3775 representation of types now includes a new slot to support the new
3776 SUBTYPEP-of-NOT-types logic.
3777 * (not a change in the main branch of SBCL, but a related prototype
3778 which can hopefully be merged into the main branch of SBCL in the
3779 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3780 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3781 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3782 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3783 dispatch macro character combinations. (thanks to Alexey Dejneka)
3784 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3785 that superficially look like logical namestrings correctly.
3786 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3788 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3789 default supplied-p) for &optional and &key arguments. (thanks to
3792 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3793 * incompatible change: The compiler is now less aggressive about
3794 tail call optimization, doing it only when (> SPACE DEBUG) or
3795 (> SPEED DEBUG). (This is an incompatible change because there are
3796 programs which relied on the old CMU-CL-style behavior to optimize
3797 away their unbounded recursion which will now die of stack overflow.)
3798 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3799 for non-GENCGC systems has been increased to 20M (since that
3800 seems much closer to the likely performance optimum for modern
3801 systems than the old 4M value was)
3802 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3803 and *DEBUG-PRINT-LEVEL*
3804 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3805 of CMU CL's support for SPARC, and various endianness and other
3806 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3807 * new syntactic sugar for the Unix command line: --load foo.bar is now
3808 an alternate notation for --eval '(load "foo.bar")'.
3810 ** The system now detects stack overflow and handles it gracefully,
3811 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3812 optimization settings. (This is a good thing in general, and
3813 its introduction in this version should be particularly timely
3814 for anyone whose code fails because of suppression of tail
3816 ** The system now hunts for the C variable "environ" in a more
3817 devious way, to avoid segfaults when the C library version differs
3818 between compile time and run time. (thanks to Christophe Rhodes)
3819 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3820 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3821 ** The compiler no longer issues bogus style warnings for undefined
3822 classes in the same source file as the DEFCLASSes which defined
3823 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3825 ** fixes in CONDITION class precedence list for undefined function
3826 errors (thanks to Alexei Dejneka)
3827 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3828 correctly. (thanks to Dan Barlow)
3829 ** portability fixes aiming at bootstrapping under CLISP (thanks
3830 to Dave McDonald and Christophe Rhodes)
3831 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3832 ** fixes in type translation and and type inference (thanks to
3834 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3835 ** various fixes in the new ports (thanks to Dan Barlow)
3836 * several changes related to debugging:
3837 ** suppression of tail recursion, as noted above
3838 ** stack overflow detection, as noted above
3839 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3840 is now the default. (For some time encapsulation has been more
3841 reliable than the breakpoint-based :ENCAPSULATE NIL
3842 implementation, at least on X86 systems; and I just noticed that
3843 encapsulation also seems closer to the spirit of the ANSI
3846 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3848 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3849 up properly again. (There was a packaging bug in 0.7.0 which
3850 left their definitions in SB-SYS::LOAD-FOREIGN and
3851 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3852 vital for most things which interface to C-level interfaces,
3853 like extensions working with sockets or databases or
3854 Perl-compatible regexes or whatever, and the need to fix
3855 this bug is the main reason that 0.7.1 was released so
3857 ** DEFGENERIC is now choosier about the methods it redefines, so that
3858 reLOADing a previously-LOADed file containing DEFGENERICs does
3859 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3860 cycle now works as it should. (thanks to Alexey Dejneka)
3861 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3862 fixed by Christophe Rhodes. (He actually submitted this patch
3863 months ago, and I delayed until after 0.7.0.)
3864 ** Bug 111 (internal compiler confusion about runtime checks on
3865 FUNCTION types) was fixed by Alexey Dejneka.
3866 * Some internal cleanups (getting rid of variables which aren't
3867 needed now that the byte interpreter is gone) caused the fasl
3868 file format number to change again.
3870 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3871 * major incompatible change: The default fasl file extension, i.e. the
3872 default extension for files produced by COMPILE-FILE, has changed
3873 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3875 ** There are many changes in the implementation of the compiler.
3876 SBCL is now essentially a compiler-only implementation of ANSI
3877 Common Lisp. EVAL still "interprets" a few special cases, but
3878 almost all the interesting cases are handled by creating
3879 a LAMBDA expression, calling COMPILE on it, then calling
3880 FUNCALL on the result.
3881 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3882 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3883 Since the code is newer, there might still be some new bugs
3884 (though not as many as before Martin Atzmueller's fixes:-). But
3885 the new code is substantially simpler and clearer, and hopefully
3886 any remaining bugs will be simpler, less fundamental, and more
3887 fixable then the bugs in the old code.
3888 ** The revised compiler is still a little unsteady on its feet.
3890 *** The debugging information it produces (particularly the names
3891 of FUNCTION objects) is sometimes much less useful than what
3892 the old compiler produced.
3893 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3894 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3895 within a MACROLET) has been temporarily weakened.
3896 ** There are new compiler optimizations for various functions:
3897 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3898 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3899 *** the math functions TRUNCATE, FLOOR, and CEILING
3900 *** the function-of-all-trades COERCE
3901 Mostly these should be transparent, but there's one
3902 potentially-annoying problem (bug 117): when the compiler
3903 inline-expands a function and does type analysis on the result,
3904 it can create control paths which have type mismatches, and
3905 when it can't prove that those control paths aren't taken,
3906 it will issue WARNINGs about the type mismatches. This is
3907 a particular problem in practice for the new sequence functions.
3908 It's not clear how this should be fixed, and for now, a
3909 workaround is given in the entry for 117 in the BUGS file.
3910 ** (Because of the interaction between the two previous items --
3911 occasional inlining problems and new inline expansions -- some
3912 of the new sequence function optimizations won't really kick in
3913 completely until debugging information, and then inlining, are
3914 straightened out in some future version.)
3915 * minor incompatible changes:
3916 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3917 DIRECTORY no longer implicitly promotes NIL slots of its
3918 pathname argument to :WILD. In particular, when you ask for the
3919 contents of a directory (which you used to be able to do without
3920 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3921 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3922 ** changes in behavior that ANSI explicitly defines to be
3923 implementation dependent:
3924 *** The new compiler-only implementation still conforms with ANSI,
3925 but acts a little different than before. Besides the obvious
3926 changes in performance tradeoffs (that the cost per form passed
3927 to EVAL has gone up, and the cost per form executed by EVAL
3928 has gone down), the behavior of the system changes a little
3929 because there are no longer any interpreted function objects.
3930 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3931 e.g. doing COMPILE on the output of interactive DEFUN is
3933 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3935 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3936 has changed from (:CL) to NIL.
3937 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3938 e.g. #\Newline and #\Tab, don't have names specified in the
3939 ANSI Common Lisp standard, is now based on their ASCII symbolic
3940 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3941 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3942 are no longer used for output.
3943 ** changes in internal implementation constants:
3944 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3945 4 million. (If your application spends a lot of time GCing and
3946 you have a lot of RAM, you might want to experiment with
3947 increasing it even more.)
3948 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3949 However, almost all old code should still continue to work without
3950 immediate update, as SB-C-CALL is now a (deprecated) nickname
3952 ** Old operator names in the style DEF-FOO are now deprecated in
3953 favor of new corresponding names DEFINE-FOO, for consistency with
3954 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3955 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3956 internal symbols, but a few supported extensions like
3957 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3958 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3959 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3960 etc. as you get deeper into recursive calls to the debugger
3961 command loop, instead of the old "5]", "5]]", "5]]]"
3962 sequence. (I was motivated to do this when squabbles between
3963 ILISP and SBCL left me very deeply nested in the debugger. In the
3964 short term, this change will probably provoke more ILISP/SBCL
3965 squabbles, but hopefully it will be an improvement in the long run.)
3966 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3967 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3968 function, since declaiming return types involving aliens is
3969 (1) annoyingly messy to do by hand and (2) vital to efficient
3970 compilation of code which calls such functions.
3971 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3972 longer reexported by the SB-EXT package. They're solely useful
3973 for alien code, so it seems more logical that you should get
3974 them from the SB-ALIEN package, not in SB-EXT.
3975 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3976 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3977 features. Instead, the code that they used to control is always
3978 built into the system.
3979 * many other bug fixes
3980 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3981 advantage of the new EVAL-WHEN stuff and to clean them up in
3982 general, and they are now more ANSI-compliant in a number of
3983 ways. Martin Atzmueller is responsible for a lot of this.
3984 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3986 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3987 *** correct ERROR type for various file operations
3988 *** some fixes for Lisp streams
3989 *** DEFMETHOD syntax checking
3990 *** changing old weird representation of debug information as
3991 strings (which, among their other deficiencies, don't transform
3992 correctly when you rename packages, and don't change their
3993 print representation when you change things like *PACKAGE*
3994 and *PRINT-LENGTH*) to symbols and lists of symbols
3995 He also made several improvements and fixed several bugs in DESCRIBE.
3996 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3998 *** misbehavior of WRITE-STRING/WRITE-LINE
3999 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4000 and several other LOOP problems as well
4001 *** DIRECTORY when similar filenames are present
4002 *** DEFGENERIC with :METHOD options
4003 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4004 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4005 *** argument ordering in FIND with :TEST option
4006 *** mishandled package designator argument in APROPOS-LIST
4007 *** various problems in the backquote readmacro
4008 *** a bug in APROPOS
4009 *** probably some others that I'm not describing very well here,
4010 since the CVS log documents them by reference to sbcl-devel
4011 messages, and the SourceForge archives aren't working well.:-(
4012 ** Dan Barlow improved the Alpha port (and is making progress on the
4013 PPC port, for those of you who think different).
4014 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4015 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4016 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4017 continues to work on the SPARC port (for those of you in a position
4018 to look down upon our little PC-compatible boxes from a great height).
4019 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4020 and uses that copy, rather than the current dynamic value, when
4021 it's trying to decide whether to truncate output. Thus e.g.
4022 (let ((*print-lines* 50))
4023 (pprint-logical-block (stream nil)
4025 (let ((*print-lines* 8))
4026 (print (aref possiblybigthings i) stream)))))
4027 should now truncate the logical block only at 50 lines, instead of
4028 often truncating it at 8 lines, as it did before.
4029 * The doc/cmucl/ directory, containing old CMU CL documentation
4030 from the time of the fork, is no longer part of the base system.
4031 SourceForge has shut down its anonymous FTP service, and with it
4032 my original plan for distributing the old CMU CL documentation
4033 there. For now, if you need these files you can download an old
4034 SBCL source release and extract them from it.
4035 * The fasl file version number changed again, for dozens of reasons,
4036 some of which are apparent above.
4038 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4039 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4040 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4041 patch, so that DISASSEMBLE output is much nicer.
4042 * The code in the SB-PROFILE package now seems reasonably stable.
4043 I still haven't decided what the final interface should look like
4044 (I'd like PROFILE to interact cleanly with TRACE, since both
4045 facilities use function encapsulation) but if you have a need
4046 for profiling now, you can probably use it successfully with
4047 the current CMU-CL-style interface.
4048 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4049 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4050 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4051 link now returns the dangling link itself, and for similar
4052 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4053 link itself. (In these cases the old code signalled an error and
4054 looped endlessly, respectively.) Thus, DIRECTORY now works even
4055 in the presence of dangling and cyclic symbolic links.
4056 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4057 is now a supported extension again, since the consensus on
4058 sbcl-devel was that it can be useful for ordinary development
4059 work, not just for debugging SBCL itself.
4060 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4061 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4062 that currently-defined functions might be redefined later with
4063 different return types.
4064 * Hash tables can be printed readably, as inspired by CMU CL code
4065 of Eric Marsden and SBCL code of Martin Atzmueller.
4066 * better error handling in CLOS method combination, thanks to
4067 Martin Atzmueller porting Pierre Mai's CMU CL patches
4068 * more overflow fixes for >16Mbyte I/O buffers
4069 * A bug in READ has been fixed, so that now a single Ctrl-D
4070 character suffices to cause end-of-file on character streams.
4071 In particular, now you only need one Ctrl-D at the command
4072 line (not two) to exit SBCL.
4073 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4075 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4076 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4077 does the right thing.
4078 * The implementation of some type tests, especially for CONDITION
4079 types, is now tidier and maybe faster, due to CMU CL code
4080 originally by Douglas Crosher, ported by Martin Atzmueller.
4081 * Some math functions have been fixed, and there are new
4082 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4083 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4084 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4085 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4086 likely to be built into sbcl-0.7.0.)
4087 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4088 CL, ported to SBCL by Martin Atzmueller)
4089 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4090 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4091 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4092 * There's a new slam.sh hack to shorten the edit/compile/debug
4093 cycle for low-level changes to SBCL itself, and a new
4094 :SB-AFTER-XC-CORE target feature to control the generation of
4095 the after-xc.core file needed by slam.sh.
4096 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4097 COMPILE-FILE is no longer supported, so that now every function
4098 gets an entry point, so that block compilation looks a little
4099 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4100 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4101 returns the number of bytes consed since the system started,
4102 rather than the number consed since the first time the function
4103 was called. (The new definition parallels ANSI functions like
4104 CL:GET-INTERNAL-RUN-TIME.)
4105 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4106 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4107 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4108 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4109 * Fasl file version numbers are now independent of the target CPU,
4110 since historically most system changes which required version
4111 number changes have affected all CPUs equally. Similarly,
4112 the byte fasl file version is now equal to the ordinary
4115 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4116 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4117 is no longer recognized. I apologize for this, because it was
4118 listed in SB-EXT as a supported extension, but I found that
4119 its existing behavior was poorly specified, as well as incorrectly
4120 specified, and it looked like too much of a mess to straighten it
4121 out. I have enough on my hands trying to get ANSI stuff to work..
4122 * many patches ported from CMU CL by Martin Atzmueller, with
4123 half a dozen bug fixes in pretty-printing and the debugger, and
4124 half a dozen others elsewhere
4125 * fixed bug 13: Floating point infinities are now supported again.
4126 They might still be a little bit flaky, but thanks to bug reports
4127 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4128 as flaky as they were.
4129 * The --noprogrammer command line option is now supported. (Its
4130 behavior is slightly different in detail from what the old man
4131 page claimed it would do, but it's still appropriate under the
4132 same circumstances that the man page talks about.)
4133 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4134 are now supported, and enabled by default. Thus, the compiler can
4135 handle many floating point and complex operations much less
4136 inefficiently. (Thus e.g. you can implement a complex FFT
4138 * The compiler now detects type mismatches between DECLAIM FTYPE
4139 and DEFUN better, and implements CHECK-TYPE more correctly, and
4140 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4141 thanks to patches from Martin Atzmueller.
4142 * various fixes to make the cross-compiler more portable to
4143 ANSI-conforming-but-different cross-compilation hosts (notably
4144 Lispworks for Windows, following bug reports from Arthur Lemmens)
4145 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4146 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4147 fixed thanks to Pierre Mai's CMU CL patches.
4148 * fixes to keep the system from overflowing internal counters when
4149 it tries to use i/o buffers larger than 16M bytes
4150 * fixed bug 45a: Various internal functions required to support
4151 complex special functions have been merged from CMU CL sources.
4152 (When I was first setting up SBCL, I misunderstood a compile-time
4153 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4154 * improved support for type intersection and union, fixing bug 12
4155 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4156 more obscure bugs as well
4157 * some steps toward byte-compiling non-performance-critical
4158 parts of the system, courtesy of patches from Martin Atzmueller
4159 * Christophe Rhodes has made some debian packages of sbcl at
4160 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4161 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4162 stable, but are nonetheless usable. When he's ready, I'd be happy
4163 to add them to the SourceForge "File Releases" section. (And if
4164 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4165 * new fasl file format version number (because of changes in
4166 internal representation of (OR ..) types to accommodate the new
4167 support for (AND ..) types, among other things)
4169 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4170 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4172 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4174 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4175 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4176 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4177 * DESCRIBE now works on structure objects again.
4178 * Most function call argument type mismatches are now handled as
4179 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4180 know whether the function will be redefined before the call is
4181 executed. (The compiler could flag local calls with full WARNINGs,
4182 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4183 it doesn't keep track of enough information to know whether calls
4184 are local in this sense.)
4185 * Compiler output is now more verbose, with messages truncated
4186 later than before. (There should be some supported way for users
4187 to override the default verbosity, but I haven't decided how to
4188 provide it yet, so this behavior is still controlled by the internal
4189 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4190 src/compiler/ir1util.lisp.)
4191 * Fasl file format version numbers have increased again, because
4192 support for the Gray streams extension changes the layout of the
4193 system's STREAM objects.
4194 * The Gray subclassable streams extension now works, thanks to a
4195 patch from Martin Atzmueller.
4196 * The full LOAD-FOREIGN extension (not just the primitive
4197 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4198 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4199 but like most other programs, it defaults to copying the Unix
4200 environment from the original process instead of starting the
4201 new process in an empty environment.
4202 * Extensions which manipulate the Unix environment now support
4203 an :ENVIRONMENT keyword option which doesn't smash case or
4204 do other bad things. The CMU-CL-style :ENV option is retained
4205 for porting convenience.
4206 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4207 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4209 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4211 * A patch from Martin Atzmueller seems to have solved the SIGINT
4212 problem, and as far as we know, signal-handling now works cleanly.
4213 (If you find any new bugs, please report them!)
4214 * The system no longer defaults Lisp source file names to types
4215 ".l", ".cl", or ".lsp", but only to ".lisp".
4216 * The compiler no longer uses special default file extensions for
4217 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4218 seems to expect a single default extension for all compiled code,
4219 and there's no compelling reason to try to stretch the standard
4220 to allow two different extensions.) Instead, byte-compiled files
4221 default to the same extension as native-compiled files.
4222 * Fasl file format version numbers have increased again, because
4223 a rearrangement of internal implementation packages made some
4224 dumped symbols in old fasl files unreadable in new cores.
4225 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4226 many fewer weird special cases.
4227 * Bug #17 (differing COMPILE-FILE behavior between logical and
4228 physical pathnames) has been fixed, and some related misbehavior too,
4229 thanks to a patch from Martin Atzmueller.
4230 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4231 by Tim Moore, ported to SBCL by Martin Atzmueller.
4232 * Martin Atzmueller fixed several filesystem-related problems,
4233 including bug #36, in part by porting CMU CL patches, which were
4234 written in part by Paul Werkowski.
4235 * More compiler warnings in src/runtime/ are gone, thanks to
4236 more patches from Martin Atzmueller.
4237 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4240 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4242 * DESCRIBE now works on CONDITION objects.
4243 * The debugger now handles errors which arise when trying to print
4244 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4246 * The build system now uses an additional file, customize-target-features.lisp,
4247 to allow local modifications to the target *FEATURES* list. (The point of
4248 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4249 debugging features enabled, without having to worry about propagating it
4250 into everyone's system when I do a "cvs update".) When no
4251 customize-target-features.lisp file exists, the target *FEATURES* list
4252 should be constructed the same way as before.
4253 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4254 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4255 warn about behavior which is undefined under the ANSI spec. Note: This
4256 is specified by ANSI, but it's not very popular with programmers.
4257 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4258 macro in the SBCL sources for an example of a workaround which you
4259 might use to make portable ANSI-standard code which does what you want.
4260 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4261 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4262 when it's used as a non-toplevel form. (This is required in order
4263 to implement the DEFCONSTANT-EQX macro.)
4264 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4265 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4266 funny things, due to bugs in the implementation of EVAL-WHEN
4267 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4268 non-toplevel DEFCONSTANTs are uncommon.)
4269 * The core file version number and fasl file version number have been
4270 incremented, because the old noncompliant DEFCONSTANT behavior involved
4271 calling functions which no longer exist, and because I also took the
4272 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4273 * fixed bug 1 (error handling before read-eval-print loop starts), and
4274 redid debugger restarts and related debugger commands somewhat while
4276 ** The QUIT debugger command is gone, since it did something
4277 rather different than the SB-EXT:QUIT command, and since it never
4278 worked properly outside the main toplevel read/eval/print loop.
4279 Invoking the new TOPLEVEL restart provides the same functionality.
4280 ** The GO debugger command is also gone, since you can just invoke
4281 the CONTINUE restart directly instead.
4282 ** The TOP debugger command is also gone, since it's redundant with the
4283 FRAME 0 command, and since it interfered with abbreviations for the
4285 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4287 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4288 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4289 some porting work by Martin Atzmueller)
4290 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4291 with a boatload of patches which he ported to SBCL. Now that those
4293 ** The system tries to make sure that its low-priority messages
4294 are prefixed by semicolons, to help people who like to use
4295 syntax highlighting in their ILISP buffer. (This patch
4296 was originally due to Raymond Toy.)
4297 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4298 patches originally written by Raymond Toy.
4299 ** The compiler understands coercion between single-value and
4300 multiple-VALUES type expressions better, getting rid of some very
4301 weird behavior, thanks to patches originally by Robert MacLachlan
4302 and Douglas Crosher.
4303 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4304 lambda lists, thanks to a patch originally by Pierre Mai.
4305 ** The system no longer bogusly warns about "abbreviated type
4307 ** The compiler gets less confused by inlining and RETURN-FROM,
4308 thanks to some patches originally by Tim Moore.
4309 ** The system no longer hangs when dumping circular lists to fasl
4310 files, thanks to a patch originally from Douglas Crosher.
4311 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4312 undefined function error.
4313 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4314 in structure slot initforms) for now, documented workaround instead:-|
4315 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4316 * fixed bug 5: added stubs for various Gray stream functions called
4317 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4318 installed, at least appropriate type errors are generated
4319 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4320 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4321 consistently in DEFMETHOD forms.
4322 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4323 it doesn't seem to affect SBCL after all
4324 * The C runtime system now builds with better optimization and many
4325 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4327 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4329 * The system is now under CVS at SourceForge (instead of the
4330 CVS repository on my home machine).
4331 * The new signal handling code has been tweaked to treat register
4332 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4333 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4334 downstream, e.g. in debug-int.lisp, has implicit dependencies
4335 on the unsignedness of integer representation of machine words,
4336 and that caused the system to bomb out with infinite regress
4337 when trying to recover from type errors involving signed values,
4338 e.g. (BUTLAST '(1 2 3) -1).)
4339 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4340 (This was one of the bugs Peter Van Eynde reported back in July.)
4341 * The system now uses code inspired by Colin Walters' O(N)
4342 implementation of MAP (from the cmucl-imp@cons.org mailing
4343 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4344 inline the MAP operation, and there is more than one
4345 sequence argument to the MAP call (so that it can't just
4346 do ETYPECASE once and for all based on the type of the
4347 single sequence argument). (The old non-inline implementation
4348 of the general M-argument sequence-of-length-N case required
4349 O(M*N*N) time when any of the sequence arguments were LISTs.)
4350 * The QUIT :UNIX-CODE keyword argument has been renamed to
4351 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4352 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4354 * Raymond Wiker's patches to port dynamic loading from Linux to
4355 FreeBSD have been added.
4356 * The BUGS file is now more nearly up to date, thanks in large part
4357 to Martin Atzmueller's review of it.
4358 * The debugger now flushes standard output streams before it begins
4359 its output ("debugger invoked" and so forth).
4360 * The core version number and fasl file version number have both
4361 been incremented, because of incompatible changes in the layout
4363 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4364 used to not be called for a saved Lisp image.)
4365 * Martin Atzmueller's version of a patch to fix a compiler crash,
4366 as posted on sbcl-devel 13 September 2000, has been installed.
4367 * Instead of installing Martin Atzmueller's patch for the
4368 compiler transform for SUBSEQ, I deleted the compiler transform,
4369 and transforms for some similar consing operations.
4370 * A bug in signal handling which kept TRACE from working on OpenBSD
4372 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4373 away by constant folding
4374 * The system now defines its address space constants in one place
4375 (in the Lisp sources), and propagates them automatically elsewhere
4376 (through GENESIS and the sbcl.h file). Therefore, patching the
4377 address map is less unnecessarily tedious and error-prone. The
4378 Lisp names of address space constants have also been systematized.
4379 * CVS tags like dollar-Header-dollar have been removed from
4380 the sources, because they have never saved me trouble and
4381 they've been source of trouble working with patches and other
4382 diff-related operations.
4383 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4384 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4386 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4388 * The system has been ported to OpenBSD.
4389 * The system now compiles with a simple "sh make.sh" on the systems
4390 that it's supported on. I.e., now you no longer need to tweak
4391 text in the target-features.lisp-expr and symlinks in src/runtime/
4392 by hand, the make.sh takes care of it for you.
4393 * The system is no longer so grossly inefficient when compiling code
4394 involving vectors implemented as general (not simple) vectors (VECTOR T),
4395 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4396 which dares to use the various sequence functions on non-simple
4397 vectors, takes less of a performance hit.
4398 * There is now a primitive type predicate VECTOR-T-P
4399 to test for the (VECTOR T) type, so that e.g.
4400 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4401 can now be compiled with some semblance of efficiency. (The old code
4402 turned the type declaration into a full call to %TYPEP at runtime!)
4403 * AREF on (VECTOR T) is still not fast, since it's still compiled
4404 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4405 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4406 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4407 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4408 without making full calls to SUBTYPEP at runtime.
4409 (Some analogous efficiency issues for non-simple vectors specialized to
4410 element types other than T, or for non-simple multidimensional arrays,
4411 have not been addressed. They could almost certainly be handled the
4412 same way if anyone is motivated to do so.)
4413 * The changes in array handling break binary compatibility, so
4414 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4415 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4416 instead of (VALUES T T).
4417 * By following the instructions that Dan Barlow posted to sbcl-devel
4418 on 2 July 2000, I was able to enable primitive dynamic object
4419 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4420 functionality is not implemented (since it calls ld to resolve
4421 library references automatically, requiring RUN-PROGRAM for its
4422 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4423 try to resolve library references) is now supported.
4424 * The system now flushes the standard output streams when it terminates,
4425 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4426 them at several other probably-convenient times, e.g. in each pass of
4427 the toplevel read-eval-print loop, and after evaluating a form given
4428 as an "--eval" command-line option. (These changes were motivated by a
4429 discussion of stream flushing issues on cmucl-imp in August 2000.)
4430 * The source transform for TYPEP of array types no longer assumes
4431 that an array whose element type is a not-yet-defined type
4432 is implemented as an array of T, but instead punts, so that the
4433 type will be interpreted at runtime.
4434 * There is now some support for cross-compiling in make.sh: each of
4435 the phases of make.sh has its own script. (This should be transparent
4436 to people doing ordinary, non-cross-compile builds.)
4437 * Since my laptop doesn't have hundreds of megabytes of memory like
4438 my desktop machine, I became more motivated to do some items on
4439 my to-do list in order to reduce the size of the system a little:
4440 ** Arrange for various needed-only-at-cold-init things to be
4441 uninterned after cold init. To support this, those things have
4442 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4443 symbols with such names are now uninterned after cold init).
4444 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4445 fasl files for cold load.
4446 ** Remove the old compiler structure pooling code (which used to
4447 be conditional on the target feature :SB-ALLOC) completely.
4448 ** Redo the representation of some data in cold init to be more compact.
4449 (I also looked into supporting byte compiled code at bootstrap time,
4450 which would probably reduce the size of the system a lot, but that
4451 looked too complicated, so I punted for now.)
4452 * The maximum signal nesting depth in the src/runtime/ support code has
4453 been reduced from 4096 to 256. (I don't know any reason for the very
4454 large old value. If the new smaller value turns out to break something,
4455 I'll probably just bump it back up.)
4456 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4458 * Many, many bugs reported by Peter Van Eynde have been added to
4459 the BUGS list; some have even been fixed.
4460 * While enabling dynamic object file loading, I tried to make the
4461 code easier to understand, renaming various functions and variables
4462 with less ambiguous names, and changing some function calling
4463 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4464 * While trying to figure out how to do the OpenBSD port, I tried to
4465 clean up some of the code in src/runtime/. In particular, I dropped
4466 support for non-POSIX signal handling, added various comments,
4467 tweaked the code to reduce the number of compilation warnings, and
4468 renamed some files to increase consistency.
4469 * To support the new automatic configuration functionality in make.sh,
4470 the source file target-features.lisp-expr has been replaced with the
4471 source file base-target-features.lisp-expr and the machine-generated
4472 file local-target-features.lisp-expr.
4473 * fixed a stupid quoting error in make.sh so that using CMU CL
4474 "lisp -batch" as cross-compilation host works again
4476 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4478 * DESCRIBE no longer tries to call itself recursively to describe
4479 bound/fbound values, so that it no longer fails on symbols which are
4480 bound to themselves (like keywords, T, and NIL).
4481 * DESCRIBE now works on generic functions.
4482 * The printer now prints less-screwed-up representations of closures
4483 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4484 * A private symbol is used instead of the :EMPTY keyword previously
4485 used to mark empty slots in hash tables. Thus
4486 (DEFVAR *HT* (MAKE-HASH-TABLE))
4487 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4488 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4489 now does what ANSI says that it should. (You can still get
4490 similar noncompliant behavior if bang on the hash table
4491 implementation with all the symbols you get back from
4492 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4493 This breaks binary compatibility, since tests for equality to
4494 :EMPTY are wired into things like the macroexpansion of
4495 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4497 * There's now a minimal placeholder implementation for CL:STEP,
4498 as required by ANSI.
4499 * An obscure bug in the interaction of the normal compiler, the byte
4500 compiler, inlining, and structure predicates has been patched
4501 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4502 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4504 * Missing ordinary arguments in a macro call are now detected even
4505 when the macro lambda list contains &KEY or &REST.
4506 * The debugger no longer complains about encountering the top of the
4507 stack when you type "FRAME 0" to explicitly instruct it to go to
4508 the top of the stack. And it now prints the frame you request even
4509 if it's the current frame (instead of saying "You are here.").
4510 * As specified by ANSI, the system now always prints keywords
4511 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4513 * The default initial SIZE of HASH-TABLEs is now smaller.
4514 * Type information from CLOS class dispatch is now propagated
4515 into DEFMETHOD bodies, so that e.g.
4516 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4518 is now basically equivalent to
4519 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4520 (DECLARE (TYPE SINGLE-FLOAT X))
4522 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4523 operation, without having to do run-time type dispatch.
4524 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4525 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4526 * Since I don't seem to be making big file reorganizations very often
4527 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4528 is overflowing my ability to conveniently back them up), I've finally
4529 checked the system into CVS. (The CVS repository is on my home system,
4530 not at SourceForge -- putting it on SourceForge might come later.)
4531 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4532 high-level GC-NOTIFY-FOO functions send their output. (There's
4533 still very little control of where low-level verbose GC functions
4534 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4535 less than it used to -- the GC-NOTIFY-FOO functions are now under
4536 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4537 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4538 in only one place in the source code, and propagates it automatically
4539 everywhere that it's needed. Thus e.g. when I bump the version from
4540 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4541 * The C source files now include boilerplate legalese and documentation
4542 at the head of each file (just as the Lisp source files already did).
4543 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4544 to his page will be replaced with a link to his new CLiki service.
4546 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4548 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4549 * The build process now looks for GNU make under the default name "gmake",
4550 instead of "make" as it used to. If GNU make is not available as "gmake"
4551 on your system, you can change this default behavior by setting the
4552 GNUMAKE environment variable.
4553 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4554 can build without error under CMU CL.
4556 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4558 * There is now a partial SBCL user manual (with some new text and some
4559 text cribbed from the CMU CL manual).
4560 * The beginnings of a profiler have been added (starting with the
4561 CMU CL profiler and simplifying and cleaning up). Eventually the
4562 main interface should be through the TRACE macro, but for now,
4563 it's still accessed through vaguely CMU-CL-style functions and macros
4564 exported from the package SB-PROFILE.
4565 * Some problems left over from porting CMU CL to the new
4566 cross-compilation bootstrap process have been cleaned up:
4567 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4568 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4569 part when getting rid of the compiler *BACKEND* stuff.)
4570 ** The value of *NULL-TYPE* was screwed up, because it was
4571 being initialized before the type system knew the final
4572 definition of the 'NULL type. This screwed up several key
4573 optimizations in the compiler, causing inefficiency in all sorts
4574 of places. (I found it because I wanted to understand why
4575 GET-INTERNAL-RUN-TIME was consing.)
4576 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4577 PROCLAIM FTYPE information. Unfortunately this broke binary
4578 compatibility again, since now the forms output by DEFGENERIC
4579 to refer to functions which didn't exist in 0.6.3.
4580 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4581 can use the new (as of 0.6.3) transform for SOME into MAP into
4583 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4584 values to (INTEGER 0 1000000), so that the time code will no longer
4585 occasionally get blown up by Linux returning 1000000 microseconds
4586 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4587 its output conform to the ANSI spec. (Alas, this makes its output
4588 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4589 * A full call to MAP NIL with a single sequence argument no longer conses.
4590 * fixes to problems pointed out by Martin Atzmueller:
4591 * The manual page no longer talks about multiprocessing as though
4592 it were currently supported.
4593 * The ILISP support patches have been removed from the distribution,
4594 because as of version 5.10.1, ILISP now supports SBCL without us
4595 having to maintain patches.
4596 * added a modified version of Raymond Toy's recent CMU CL patch for
4597 EQUALP comparison of HASH-TABLE
4599 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4601 * The system still can't cross-compile itself with
4602 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4603 entails), but at least it can compile more of itself that way
4604 than it used to be able to, and various buglets which were uncovered
4605 by trying to cross-compile itself that way have now been fixed.
4606 * This release breaks binary compatibility again. This time
4607 at least I've incremented the FASL file format version to 2, so that the
4608 problem can be detected reliably instead of just causing weird errors.
4609 * various new style warnings:
4610 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4611 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4612 ** using the lexical binding of a variable named in the *FOO* style
4613 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4614 as specified by ANSI.
4615 * *RANDOM-STATE* is no longer automatically initialized from
4616 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4617 default behavior of the system is to repeat its behavior every time
4618 it's run. If you'd like to change this behavior, you can always
4619 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4620 old convention there was no comparably easy way to get the system to
4621 repeat its behavior every time it was run.
4622 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4623 FTYPE declarations has been removed, in favor of their ANSI
4624 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4625 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4626 the types of their sequence arguments can be determined at compile time.
4627 This is done through a new open code expansion for MAP which eliminates
4628 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4629 argument types can be determined at compile time.
4630 * The optimizer now transforms COERCE into an identity operation when it
4631 can prove that the coerced object is already of the correct type. (This
4632 can be a win for machine generated code, including the output of other
4633 optimization transforms, such as the MAP transform above.)
4634 * Credit information has been moved from source file headers into CREDITS.
4635 * Source file headers have been made more standard.
4636 * The CASE macro now compiles without complaining even when it has
4639 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4641 * (Note that the way that the PCL macroexpansions were rewritten
4642 to accommodate the change in DEFGENERIC below breaks binary
4643 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4644 not run under sbcl-0.6.2. Once we get out of alpha releases,
4645 i.e. hit release 1.0.0, we'll probably try to maintain binary
4646 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4647 and sbcl-1.4.4. Until then, however, it might be fairly common
4648 for maintenance releases to break binary compatibility.)
4649 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4651 * The reporting of unhandled signals has been changed to print some
4652 explanatory text as well as the report form. (Previously only
4653 the report form was printed.)
4654 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4655 it defines, so that the compiler no longer issues undefined function
4656 warnings for compiled-but-not-yet-loaded generic functions.
4657 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4658 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4659 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4660 section of the ANSI spec are supported.
4661 * The "" nickname for the "KEYWORD" package has been removed.
4662 The reader still handles symbol tokens which begin with a package marker
4663 as keywords, but it doesn't expose its mechanism for doing so in the
4664 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4665 * The system now issues STYLE-WARNINGs for contradictory TYPE
4666 proclamations. (Warnings for contradictory FTYPE proclamations would
4667 be nice too, but those can't be done usefully unless the type system
4668 is made smarter about FUNCTION types.)
4669 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4670 systematized, so that "*target-*.lisp is supposed to exist only on the
4671 target and imply that there's a related file which exists on the
4672 host, and *host-*.lisp is supposed to exist only on the host and imply
4673 that there's a related file which exists on the target. This involves a
4674 lot of renaming. Hopefully the acute confusion caused by the renaming
4675 will be justified by the reduction in chronic confusion..
4676 ** runtime-type.lisp -> early-target-type.lisp
4677 ** target-type.lisp -> late-target-type.lisp
4678 ** early-host-format.lisp -> early-format.lisp
4679 ** late-host-format.lisp -> late-format.lisp
4680 ** host-error.lisp -> misc-error.lisp
4681 ** early-error.lisp -> early-target-error.lisp
4682 ** late-error.lisp -> late-target-error.lisp
4683 ** host-defboot.lisp -> early-defboot.lisp
4684 ** code/misc.lisp -> code/target-misc.lisp
4685 ** code/host-misc.lisp -> code/misc.lisp
4686 ** code/numbers.lisp -> code/target-numbers.lisp
4687 ** code/early-numbers.lisp -> numbers.lisp
4688 ** early-host-type.lisp -> early-type.lisp
4689 ** late-host-type.lisp -> late-type.lisp
4690 ** host-typep.lisp -> typep.lisp
4691 ** load.lisp -> target-load.lisp
4692 ** host-load.lisp -> load.lisp
4693 ** host-disassem.lisp -> disassem.lisp
4694 ** host-insts.lisp -> insts.lisp
4695 ** byte-comp.lisp -> target-byte-comp.lisp
4696 ** host-byte-comp.lisp -> byte-comp.lisp
4697 ** host-signal.lisp -> signal.lisp
4698 ** host-defstruct.lisp -> defstruct.lisp
4699 ** late-target-type.lisp -> deftypes-for-target.lisp
4700 Furthermore, several other previously target-only files foo.lisp (e.g.
4701 hash-table.lisp and random.lisp) have been split into a target-and-host
4702 foo.lisp file and a target-only target-foo.lisp file, with their key type
4703 definitions in the target-and-host part, so that the cross-compiler will
4704 know more about target types.
4705 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4706 gone away. In their place are various *BACKEND-FOO* variables
4707 corresponding to the slots of the old structure.
4708 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4709 into the target SBCL has been fixed.
4710 * The chill.lisp system for loading cold code into a running SBCL
4712 * Support for the CMU CL "scavenger hook" extension has been removed.
4713 (It was undocumented and unused in the CMU CL sources that SBCL was
4714 derived from, and stale in sbcl-0.6.1.)
4715 * Various errors in the cross-compiler type system were detected
4716 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4717 (enabling various consistency checks). Many of them were fixed,
4718 but some hard problems remain, so the compiler is back to
4719 running without *TYPE-SYSTEM-INITIALIZED* for now.
4720 * As part of the cross-compiler type system cleanup, I implemented
4721 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4722 * I have started adding UNCROSS calls throughout the type system
4723 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4724 on cross-compiler input files will be able to go away, and various
4726 * CONSTANTP now returns true for quoted forms (as explicitly required
4729 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4731 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4732 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4733 (MAKE-LIST -1). In the longer term, it ought to become true
4734 that declarations are assertions even at SAFETY 1. For now, it's not
4735 quite true even at SAFETY 3, but it's at least more nearly true..
4736 (Note that this change seems to increases the size of the system by
4737 O(5%) and to decrease the speed of the compiler by 20% or more.)
4738 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4739 for the problem of printing dozens of lines of distracting information
4740 about low-level system machinery as part of the top stack frame
4741 on entry to the debugger when an undefined function was called.
4742 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4743 is not reset to COMMON-LISP-USER.
4744 * Compilation of stuff related to dyncount.lisp has been made conditional
4745 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4746 smaller. The various dyncount-related symbols have been moved into
4747 a new "SB-DYNCOUNT" package.
4748 * tty-inspect.lisp has been renamed to inspect.lisp.
4749 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4750 feature has gone away. (When we eventually port to other flavors of
4751 libc and/or Unix, we'll try to make the differences between flavors
4752 invisible at the user level.)
4753 * Various other *FEATURES* tags, and/or their associated conditionals,
4754 have been removed if obsolescent, or given better documentation, or
4755 sometimes given more-mnemonic names.
4757 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4759 * tidied up "make.sh" script
4760 * tidied up system directory structure
4761 * better "clean.sh" behavior
4762 * added doc/FOR-CMUCL-DEVELOPERS
4763 * many many small tweaks to output format, e.g. removing possibly-confusing
4764 trailing #\. character in DESCRIBE-INSTANCE
4765 * (EQUALP #\A 'A) no longer signals an error.
4766 * new hashing code, including EQUALP hashing
4767 * tidied up Lisp initialization and toplevel
4768 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4769 * command line argument processing
4770 * added POSIX-GETENV function to deal with Unix-ish environment variables
4771 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4772 terminating SBCL on EOF
4773 * non-verbose GC by default
4774 * There is no more "sbcl" shell script; the sbcl file is now the C
4775 runtime executable (just like CMU CL).
4776 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4778 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4779 symbol and package information as Lisp native symbol and package objects
4780 instead of strings naming symbols and strings naming packages. This way,
4781 whenever packages are renamed (as in warm init), debug information is
4782 transformed along with everything else.
4783 * tweaked the optimization policy declarations which control the building
4784 of SBCL itself. Now, among other things, the system no longer saves
4785 source location debugging information. (This helps two problems at once
4786 by reducing SBCL size and by keeping SBCL from trying to look for its
4787 sources -- which may not exist -- when reporting errors.)
4788 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4789 debugging and testing purposes
4790 * cleaned up printing, making the printer call PRINT-OBJECT for
4791 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4792 methods, giving nearly-ANSI behavior
4793 * converted almost all special variables to use *FOO* naming convention
4794 * deleted PARSE-TIME functionality, since it can be done portably
4795 * moved some files out of cold init into warm init
4796 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4798 * regularized formatting of source files
4799 * added an install.sh script
4800 * fixed ridiculous memory usage of cross-compiler by making
4801 compiler/alloc.lisp not try to do pooling unless it can hook
4802 itself into the GC of the cross-compilation host. Now the system
4803 builds nicely on my old laptop.
4804 * added :SB-ALLOC in target-features.lisp-expr
4805 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4806 was not implemented)
4807 * re-did condition handling and note reporting in the compiler. Notes
4808 are no longer handled by signalling conditions. Style warnings
4809 and warnings are handled more correctly and reported in such a way
4810 that it's easy to find one or the other in your output (so that you
4811 can e.g. figure out which of many problems caused COMPILE-FILE to
4813 * changed the severity of several compiler warnings from full WARNING
4814 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4815 compiler note reporting so that it doesn't use the condition system
4816 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4817 COMPILE-FILE command)
4818 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4819 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4820 no longer supported, since their implementation was deeply intertwingled
4821 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4822 * removed bogus "support" for compiler macros named (SETF FOO), and
4823 removed the compiler macro for SETF INFO (but only after making a fool
4824 of myself on the cmucl-imp mailing list by posting a bogus patch for
4825 DEFINE-COMPILER-MACRO..)
4826 * Compiled files containing forms which have side effects on the Lisp
4827 reader (such as DEFPACKAGE forms) are now handled more correctly.
4828 (Compiler queuing of top level lambdas has been suppressed by setting
4829 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4830 * deleted various currently-unused source files, e.g. gengc.lisp. They
4831 may be added back at some point e.g. when porting to other architectures,
4832 but until they are it's distracting to distribute them and to try to
4834 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4835 there were so many of them they're just distractions, and UNCROSS is
4836 known to be able to handle the current sources
4837 * moved PROFILE functionality into TRACE, so that it will be clear
4838 how the wrapping and unwrapping of functions when you profile them
4839 interacts with the wrapping and unwrapping of functions when you
4840 trace them. (Actually, the functionality isn't there yet, but at least
4841 the interface specification is there. Hopefully, the functionality will
4842 arrive with some future maintenance release.)
4843 * removed host-oops.lisp
4844 * changed signature of QUIT function to allow UNIX-CODE argument
4845 * fixed READ-SEQUENCE bug
4846 * tweaked verbose GC output so that it looks more like the progress
4847 output that ANSI specifies for functions like LOAD
4848 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4849 * added <http://sbcl.sourceforge.com> to the banner information printed by