1 ;;;; -*- coding: utf-8; -*-
2 * reduced conservativism on GENCGC platforms: on average 45% less
3 pages pinned (measured from SBCL self build).
5 changes in sbcl-1.0.20 relative to 1.0.19:
6 * minor incompatible change: OPTIMIZE qualities
7 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
8 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
9 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
10 * documentation: some slot access efficiency guidelines have been
11 added to the user manual.
12 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
13 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
15 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
16 to more efficient EQ-comparison versions more often.
17 * optimization: enhanced derivation of DOLIST iteration variable type
19 * optimization: constant folding of simple (LIST ...) forms as DOLIST
21 * optimization: runtime lookup of function definitions can be
22 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
23 * optimization: compiler is able to derive the return type of
24 (AREF (THE STRING X) Y) as being CHARACTER.
25 * optimization: CLRHASH on empty hash-tables no longer does pointless
26 work. (thanks to Alec Berryman)
27 * optimization: REPLACE deftransforms don't punt when :START1 or
29 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
30 no longer emits a code deletion note. (reported by Andrew
32 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
33 functions with an odd number of &OPTIONAL arguments, a &REST
34 argument, and one or more &KEY arguments at the call site.
35 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
36 arguments appeared at call sites as well.
37 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
38 READ-LINE &co can no longer read from them afterwards. (reported
40 * bug fix: fixed #427: unused local aliens no longer cause compiler
41 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
43 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
44 alien-stack leakage. (reported by Andy Hefner)
45 * bug fix: PROGV signals an error when an attempt to violate declared
46 type of a variable or bind a constant is made.
47 * bug fix: SET signals an error when an attempt to violate declared
48 type of a variable is made.
49 * bug fix: restart computation during the execution of a restart
50 test function no longer causes infinite recursion. (reported by
52 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
53 signals an error. (thanks to Pierre Mai)
54 * bug fix: EXPORT left symbol unexported in conflict situations.
55 (thanks to Michael Weber)
56 * bug fix: correctly handle name conflicts involving CL:NIL.
57 (thanks to Michael Weber)
58 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
59 conflicts arising from USEing package with conflicting symbols
60 correctly. (thanks to Michael Weber)
62 changes in sbcl-1.0.19 relative to 1.0.18:
63 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
64 warnings that go otherwise unhandled will be muffled if they are
65 of the type that's the value of this variable.
66 * optimization: stack allocation is slightly more efficient on x86
68 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
69 where a raw slot always is initialized using the initform whose
70 type is not know sufficiently well a compile-time are now compiled
71 correctly. (reported by John Morrison)
72 * bug fix: compiler no longer makes erronous assumptions in the
73 presense of non-foldable SATISFIES types.
74 * bug fix: stack analysis missed cleanups of dynamic-extent
75 arguments in non-let-converted calls to local functions.
76 * improvements to the Windows port:
77 ** adjusted address spaces for building on both Win32 and
78 Win64. (thanks for John Connors)
79 * fixed some bugs revealed by Paul Dietz' test suite:
80 ** interval arithmetic during type derivation used inexact integer
81 to single-float coercions.
82 ** arithmetic operations involving large integers and single
83 floats give the same results in compiled and interpreted code.
84 ** deriving the result type of COERCE no longer signals an error
85 if the derived type of the second argument is a MEMBER type
86 containing invalid type specifiers.
87 ** ADJOIN with constant NIL as second argument works correctly.
88 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
90 changes in sbcl-1.0.18 relative to 1.0.17:
91 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
92 profiles only the current thread.
93 * minor incompatible change: changes to SYMBOL-VALUE of constants
94 defined with DEFCONSTANT now signal an error.
95 * enhancement: SB-SPROF now has support for wallclock profiling,
96 and is also able to profile specific threads. REPORT output
97 has also additional sorting options.
98 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
100 * optimization: structure allocation has been improved
101 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
102 ** out of line constructors are ~10% faster.
103 ** inline constructors are ~15% faster.
104 ** inline constructors are capable of dynamic extent allocation
105 (generally on x86 and x86-64, in some cases on other platforms
107 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
109 * optimization: file compiler is now able to coalesce non-circular
110 lists, non-base strings, and bit-vectors. Additionally, constants
111 are never referenced through SYMBOL-VALUE at runtime.
112 * optimization: code defining methods on PRINT-OBJECT (and other
113 generic functions in the COMMON-LISP package) now loads faster.
114 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
116 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
117 fasl is now deleted. (reported by Attila Lendvai)
118 * bug fix: READ-LINE always returned NIL for the last line in files.
119 (reported by Yoshinori Tahara)
120 * bug fix: more accurate disassembly annotations of foreign function
121 calls. (thanks to Andy Hefner)
122 * bug fix: trimming non-simple strings and non-string string
123 designators when the there is nothing to trim works properly.
124 (thanks to James Knight)
125 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
126 (thanks to Travis Cross)
127 * fixed some bugs revealed by Paul Dietz' test suite:
128 ** NIL is a valid function name (regression at 1.0.13.38)
129 ** FILL on lists was missing its return value (regression at 1.0.12.27)
130 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
131 fill pointers properly (regression at 1.0.12.23)
132 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
133 displacement indices properly (regression at 1.0.12.23)
135 changes in sbcl-1.0.17 relative to 1.0.16:
136 * temporary regression: user code can no longer allocate closure
137 variable storage on stack, due to bug 419 without explicitly
138 requesting it. Please consult sbcl-devel for advice if you need to
139 use this feature in the meanwhile.
140 * new feature: runtime argument --control-stack-size can be used to
141 adjust thread default control stack size.
142 * enhancement: improved TIME output
143 ** all times are reported using the measured accuracy (milliseconds
144 for real and GC times, microseconds for everything else.)
145 ** processor cycle counts on x86 and x86-64.
146 ** interpreted forms are counted for both evaluator modes.
147 ** number of lambdas converted by the compiler is reported.
148 ** CPU percentage report (computed from real and total run time.)
149 ** more comprehensive run time reporting, using a condenced format
150 ** interperted form, lambda, and page fault counts are omitted
152 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
154 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
155 * optimization: two argument forms of LAST are upto ~10% faster
156 in normal SPEED policies.
157 * optimization: NCONC no longer needs to heap cons its &REST list
158 in normal SPEED policies.
159 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
161 * bug fix: UNION and NUNION work with :TEST-NOT once more,
162 regression since 1.0.9.1. (thanks to Eric Marsden)
163 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
164 since 1.0.15.36. (thanks to Paul Khuong)
165 * bug fix: LAST when always returned the whole list when given a bignum
166 as the second argument.
167 * bug fix: dynamic extent allocation of nested lists and vectors
168 could leak to otherwise accessible parts.
169 * bug fix: invalid optimization of heap-allocated alien variable
171 * bug fix: fasl header checking is less vulnerable to different
172 platform word lengths.
173 * bug fix: more correct assembler syntax for GNU binutils
174 2.18.50.0.4 support. (thanks to Marijn Schouten)
175 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
177 * bug fix: the fopcompiler can handle LOCALLY forms (with no
178 declarations) successfully. (reported by Attila Lendvai)
180 changes in sbcl-1.0.16 relative to 1.0.15:
181 * minor incompatible change: revert the changes to sb-posix's error
182 signaling added in 1.0.14.
183 * minor incompatible change: change PROBE-FILE back to returning
184 NIL whenever we can't get a truename, as was the case before 1.0.14.
185 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
186 inherits from ERROR instead of just CONDITION.
187 * new feature: SB-INTROSPECT can provide source locations for instances
188 as well. (thanks to Tobian Ritterweiler)
189 * optimization: binding special variables now generates smaller code
190 on threaded platforms.
191 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
192 and cases where no :TEST is given but the compiler can infer that
193 the element to search is of type (OR FIXNUM (NOT NUMBER)).
194 * optimization: better LOGNOT on fixnums.
195 * optimization: modular arithmetic for a particular requested width
196 is implemented using a tagged representation unless a better
197 representation is available.
198 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
199 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
200 for wellformedness and returns a specialized vector. (reported by
201 Francois-Rene Rideau)
202 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
203 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
204 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
205 unscheduling timer at the same time as another timer fires.
206 * bug fix: don't reschedule timers for dead threads.
207 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
208 * bug fix: copying output from RUN-PROGRAM to a stream signalled
209 bogus errors if select() was interrupted.
210 * enhancement: add support for fcntl's struct flock to SB-POSIX.
212 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
213 * enhancement: cleaner backtraces for interactive interrupts, as
214 well as other cases where the interesting frames used to be
215 obscured by interrupt handling frames.
216 * enhancement: untracing a whole package using (UNTRACE "FOO") is
217 now supported, and tracing a whole package using (TRACE "FOO") now
218 traces SETF-functions as well.
219 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
220 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
221 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
222 * unreadably printed representation of hash-tables now includes
224 * bug fix: partially fixed #188: type propagation from assignments
225 is now more efficient.
226 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
227 'DOUBLE-FLOAT) are not flushable.
228 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
229 full address of the object, and none of the tag bits.
230 * bug fix: readably printing hash-tables now respects other printer
231 control variables. (reported by Cedric St-Jean)
232 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
233 keyword with MAKE-HASH-TABLE.
234 * bug fix: export SB-POSIX:MKSTEMP.
235 * bug fix: SORT was not interrupt safe.
236 * bug fix: XREF accounts for the last node of each basic-block as
238 * bug fix: MAKE-INSTANCE optimizations interacted badly with
239 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
240 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
241 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
242 create obsolete instances in the case of redefinition or
243 obsoletion of a superclass. (thanks to Andy Hefner)
244 * bug fix: Support for the Alpha architecture has been revived; it had
245 suffered somewhat from lack of maintenance since sbcl-1.0.
246 * improvements to the Windows port:
247 ** The system detects the codepage to use at startup, instead of
248 using the value from the saved core. (thanks to Kei Suzuki)
250 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
251 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
252 (see documentation for details.)
253 * revived support for OpenBSD (contributed by Josh Elsasser)
254 * partially fixed bug #108: ROOM no longer suffers from occasional
255 (AVER (SAP= CURRENT END)) failures.
256 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
257 bodies no longer cause a WARNING to be signalled. (reported by
259 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
260 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
261 the bug in TRUNCATE.) (reported by Michael Weber)
262 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
263 no samples. (reported by Andy Hefner)
264 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
265 no longer appear as (NIL ...) frames in backtraces.
266 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
267 is now actually exported from SB-EXT as documented. (reported by
269 * bug fix: sb-aclrepl now correctly understands how to inspect
270 single-floats on 64-bit platforms where single-floats are not boxed.
271 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
272 yet been finalized. (reported by Levente Meszaros)
273 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
274 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
275 SBCL-specific optimize qualities.
277 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
278 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
279 an executable in the search path, and does so in the child
280 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
281 been removed; it can be found in the CVS history, for somebody who
282 needs that search behavior (see the manual).
283 * minor incompatible change: compiler policy re. weakening type
284 checks has changed: now type checks are weakened only if SAFETY < 2
286 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
287 unparsing of directory pathnames as files. Analogously,
288 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
289 filename to parse into a directory pathname.
290 * enhancement: implicit generic function creation now signals a
291 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
292 users can bind handlers for and muffle around calls to LOAD.
293 * enhancement: RUN-PROGRAM allows unicode arguments and environments
294 to be used (using the default stream external format), and allows
295 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
296 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
297 strings and vectors whose element-type or simplicity is not fully
298 known at compile-time.
299 * optimization: STRING-TRIM and related functions no longer allocate
300 a new string when no trimming needs to be performed. These functions
301 are also faster than before when the input string has been declared
303 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
304 * optimization: READ-LINE is significantly faster for files containing
306 * optimization: non-open coded uses of character comparison operators
307 (e.g. char=) no longer cons when called with more than one parameter
308 on platforms supporting dynamic-extent allocation.
309 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
310 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
311 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
313 * bug fix: some sequence functions elided bounds checking when
315 * bug fix: too liberal weakening of union-type checks when SPEED >
317 * bug fix: more bogus fixnum declarations in ROOM implementation
320 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
321 * new feature: MAKE-HASH-TABLE now experimentally accepts a
322 :SYNCHRONIZED argument, which makes the hash-table safe for
323 concurrent accesses (but not iteration.) See also:
324 SB-EXT:WITH-LOCKED-HASH-TABLE, and
325 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
326 * optimization: CONCATENATE on strings is an order of magnitude faster
327 in code compiled with (> SPEED SPACE).
328 * optimization: SUBSEQ is ~50% faster on lists.
329 * bug fix: bug 417 fixed -- source location reporting is now more robust.
330 * bug fix: SUBSEQ on a list will now correctly signal an error if if
331 END is smaller then START.
332 * bug fix: SB-PROFILE will no longer report extra consing for nested
333 calls to profiled functions.
334 * bug fix: ROOM implementation had bogus fixnum declarations which
335 could cause type-errors when calling ROOM on large images.
336 * bug fix: if file compilation is aborted, the partial fasl is now
337 deleted, and COMPILE-FILE returns NIL as the primary value.
338 * bug fix: number of thread safety issues relating to SBCL's internal
339 hash-table usage have been fixed.
340 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
341 be returned from its body when the values were being returned
342 using unknown-values return convection and the W-P-O was wrapped
343 inside an UNWIND-PROTECT.
344 * bug fix: sb-posix should now compile again under Windows, enabling
347 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
348 * incompatible change: hash-table accessor functions are no longer
349 automatically protected by locks. Concurrent accesses on the same hash-table
350 from multiple threads can give inconsistent results or even corrupt the
351 hash-table completely. Multi-threaded applications should do their own
352 locking at the correct granularity. In the current implementation it is
353 still safe to have multiple readers access the same table, but it's not
354 guaranteed that this property will be maintained in future releases.
355 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
356 and will signal an error at runtime.
357 * enhancement: SB-THREAD package now exports a semaphore interface.
358 * enhancement: CONS can now stack-allocate on x86 and
359 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
361 * enhancement: nested lists can now be stack allocated on
362 platforms providing stack allocation support.
363 * enhancement: dynamic-extent support has been extended to support
364 cases where there are multiple possible sources for the stack
366 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
367 if the mutex is uncontested on Linux.
368 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
369 as specified by the CLHS. (thanks to Tobias Rittweiler)
370 * bug fix: NaN comparison now works on x86-64.
371 * bug fix: CLOSE :ABORT T on a stream with pending output now
373 * bug fix: instances of non-standard metaclasses using standard
374 instance structure protocol sometimes missed the slot type checks
376 * bug fix: known functions can be inlined in high-debug code.
378 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
379 * minor incompatible change: the MSI installer on Windows no longer
380 associates .lisp and .fasl files with the installed SBCL.
381 * minor incompatible change: :UNIX is no longer present in *FEATURES*
382 on Windows. (thanks to Luis Oliviera)
383 * new feature: SB-CLTL2 contrib module now implements
384 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
385 * optimization: scavenging weak pointers is now more efficient,
386 requiring O(1) instead of O(N) per weak pointer to identify
387 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
388 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
389 method bodies using constant slot-names when the first argument is
390 a specializer parameter for the method.
391 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
392 names, when the class of the instance is a direct instance
393 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
394 as slow as the constant slot-name case.)
395 * optimization: member type construction is now O(N) instead
397 * optimization: UNION and NUNION are now O(N+M) for large
398 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
400 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
401 now more readable in environments like Slime which display it.
402 (thanks to Tobias C. Rittweiler)
403 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
404 was able to infer the correct argument type for the object on which
405 the CAS operation was being performed.
406 * bug fix: copy propagation interfered with parallel assignment
407 semantics in local calls. (reported by Paul Khuong)
408 * bug fix: the signed modular fixnum shift compiled to wrong code on
409 x86 and x86-64. (spotted by a slight modification to some of
412 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
413 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
414 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
415 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
417 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
418 and generic function now print the object identity for anonymous
419 instances. (thanks to Attila Lendvai)
420 * enhancement: as an extension to MOP, SBCL now supports SETF of
421 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
422 (thanks to Attila Lendvai)
423 * enhancement: sb-cover has an alternate annotation mode which puts
424 the form annotations at the car of the form, instead of annotating
425 it completely. The benefit of this mode is that it shows better
426 which forms were instrumented by the compiler.
427 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
428 by printer control variables. (reported by Dan Corkill)
429 * bug fix: system leaked memory when delayed output was performed by
430 the OS in smaller chunks then expected. (thanks to David Smith)
431 * bug fix: system leaked memory when file streams were not closed
433 * bug fix: large objects written to slow streams that were modified
434 after the write could end up with the modified state written to
435 the underlying file descriptor.
436 * bug fix: multiple threads operating in parallel on the same stream
437 could cause buffer-overflows.
438 * bug fix: source location information is stored correctly
439 (broken since 1.0.6). This bug would generally show up as the
440 Slime debugger highlighting the wrong form when the sldb-show-source
442 * bug fix: Branch forms are again annotated as branches in the sb-cover
444 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
445 on x86-64. (reported by Gregory Vanuxem)
446 * bug fix: EQUALP could return wrong results for structures with raw
447 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
448 word). (reported by Vjacheslav Fyodorov)
449 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
452 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
453 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
454 atomic compare-and-swap operations on threaded platforms.
455 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
456 allows assigning a global minimum value to optimization qualities
457 (overriding proclamations and declarations).
458 * enhancement: closed over variables can be stack-allocated on x86
460 * performance bug fix: GETHASH and (SETF GETHASH) are once again
462 * optimization: slot definition lookup is now O(1). This speeds up
463 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
464 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
465 * optimization: ASSOC and MEMBER can now be open-coded for all
466 combinations of keyword arguments when second argument is constant
467 and SPEED >= SPACE. In other cases a specialized version is
469 * bug fix: using obsoleted structure instances with TYPEP and
470 generic functions now signals a sensible error.
471 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
472 (reported by Kristoffer Kvello)
473 * bug fix: backtrace construction is now more careful when making
474 lisp-objects from pointers on the stack, to avoid creating bogus
475 objects that can be seen by the GC.
476 * bug fix: defaulting of values in contexts expecting more than 7
477 variables now works on x86-64. (reported by Christopher Laux)
478 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
480 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
481 as the property-list of a symbol.
482 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
483 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
484 situations requested, are once again file-compileable. (reported
487 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
488 * MOP improvement: support for user-defined subclasses of
489 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
490 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
491 the expansion of DEFMETHOD, is responsible for generating a form
492 which creates a list of specializers when evaluated. Additional
493 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
494 debugging and introspective support.
495 * minor incompatible change: the (unsupported) spinlock interface
496 has changed: free spinlock now has the value NIL, and a held spinlock
497 has the owning thread as its value.
498 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
499 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
501 * enhancement: name of a socket-stream is now "a socket" instead of
503 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
504 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
505 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
506 for generic functions. (Thanks to Tobias C. Rittweiler)
507 * optimization: bignum printing speed has been improved by 20-40%
508 (depending on the bignum size.)
509 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
511 * bug fix: the cache used by the CLOS to store precomputed effective
512 methods, slot offsets, and constant return values is now thread and
514 * bug fix: generic function dispatch function updating is now thread
515 and interrupt safe (in the sense that the known issues have been
517 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
518 * bug fix: interning EQL-specializers is now thread and interrupt safe.
519 * bug fix: asdf systems with dependencies to the SB-POSIX or
520 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
521 * bug fix: interrupt safety of applicable method computation has been
524 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
525 * new contrib: sb-cover, an experimental code coverage tool, is included
527 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
528 significantly faster.
529 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
530 produce more efficient code on x86-64 if the compiler can determine
531 that (+ INDEX <constant>) does not require a bounds check and FOO
532 has an element type at least 8 bits wide.
533 * enhancement: a new, experimental synchronous timeout facility is
534 provided. Refer to SB-SYS:WITH-DEADLINE for details.
535 * enhancement: when a symbol name conflict error arises, the
536 conflicting symbols are always printed with a package prefix.
537 (thanks to Kevin Reid)
538 * enhancement: stepping is now once again supported on the SPARC and
539 MIPS platforms. (It is also now more likely to work on CheneyGC
541 * enhancement: sb-sprof can now also track and report accurate call
543 * bug fixes: the treatment of non-standard subclasses of
544 SB-MOP:SPECIALIZER is more correct.
545 * incompatible change: PURIFY no longer copies the data from the
546 dynamic space into the static and read-only spaces on platforms
547 that use the generational garbage collector
548 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
550 * bug fix: GC race condition occasionally resulting in crashes with
551 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
553 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
555 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
556 a STYLE-WARNING to be signalled (regression from 1.0.4.)
557 * bug fix: an asynchronous interrupt could previously leave the
558 system running with GC inhibited.
559 * bug fix: a DECLARE form evaluated at top-level now causes an error
560 rather than silently (or verbosely) returning NIL.
561 * bug fix: trying to dissassemble functions compiled at high DEBUG could
562 result in a error being signalled due to source form lookup errors.
563 (reported by Peter Graves)
565 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
566 * incompatible change: removed writer methods for host-ent-name,
567 host-ent-addresses -- changing the values did not update the DNS
569 * minor incompatible change: changed experimental JOIN-THREAD interface
570 * documentation: the manual now lists reader and writer methods
571 in class slot documentation sections. (thanks to Richard M Kreuter)
572 * documentation: unwinding from asyncronous events has been
573 documented as unsafe.
574 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
575 in multithreaded application code.
576 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
577 platforms. (thanks to James Anderson for the optimization hint)
578 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
580 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
581 variants no longer cons.
582 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
583 their NOT- variants no longer cons.
584 * optimization: EQUAL hash tables no longer use SXHASH for objects
585 of all data types, but instead use an EQL hash for types for which
586 EQUAL is the same as EQL.
587 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
588 are significantly faster.
589 * optimization: new STRING-OUTPUT-STREAM implementation is much
590 faster for large outputs and conses 30% less on average.
591 * enhancement: XREF information is now collected for references made
592 to global variables using SYMBOL-VALUE and a constant argument.
593 * enhancement: SIGINT now causes a specific condition
594 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
595 * bug fix: FILE-POSITION used to signal an error in some cases where
596 ANSI requires it to return NIL.
597 * bug fix: ADJUST-ARRAY is now interrupt-safe.
598 * bug fix: adding and removing fd-handlers is now interrupt-safe.
599 * bug fix: inlined calls to C now ensure 16byte stack alignment on
601 * bug fix: bad type declaration in the CLOS implementation has
602 been fixed. (reported by James Anderson)
603 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
604 (thanks to Lutz Euler)
605 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
606 required by the CLHS. (reported by Samium Gromoff)
607 * bug fix: dead unbound variable references now signal an error.
608 (reported by Marco Monteiro)
609 * bug fix: / with an unused value was being deleted in safe code.
610 (thanks to Marco Monteiro and Kevin Reid)
611 * bug fix: number of characters that can be written onto a single
612 line in a file is unlimited.
613 * bug fix: some GC deadlocks caused by asynchronous interrupts have
614 been fixed by inhibiting interrupts for when GC is disabled.
615 * bug fix: some interrupt safety issues with GETHASH, (SETF
616 GETHASH), CLRHASH and REMHASH have been fixed.
617 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
618 specifier no longer causes infinite recursion.
619 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
620 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
621 * bug fix: an &environment argument with the correct variable information
622 is passed to macros that are expanded during byte compilation.
623 (reported by Samium Gromoff)
624 * bug fix: a logic error was causing the hash function for bignums to
625 have excessive amounts of collisions. (reported by Faré Rideau)
626 * bug fix: modifying the contents of an array could change the return
627 value of SXHASH on that array, which is only allowed for strings
628 and bit vectors. (bug introduced in 0.9.16)
629 * improvement: the x86-64/darwin port now passes all tests (except
630 for the debugger tests) but should still be considered
631 experimental until this is fixed.
632 * improvement: a style-warning is signaled for CASE (etc) clauses with
633 duplicate keys. (thanks to Kevin Reid)
634 * improvement: macos/x86 and macos/x86-64 now use mach exception
635 handlers for dealing with illegal instructions (for trapping and
636 error handling) and memory protection violations (for GC).
638 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
639 * new platform: experimental support for x86-64/darwin (MacOS).
640 * incompatible change: the thread-safe (on most platforms) getaddrinfo
641 and getnameinfo sockets functions are used instead of gethostbyaddr
642 and gethostbyname, on platforms where the newer functions are available.
643 As a result, the ALIASES field of HOST-ENT will always be NIL on these
645 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
646 don't print the startup banner, but behave as if --noinform was passed
647 as a command line argument. (thanks to Kevin Reid)
648 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
649 * optimization: code using alien values with undeclared types is much faster.
650 * optimization: the compiler is now able to open code SEARCH in more cases.
651 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
652 * bug fix: using standardized COMMON-LISP special variables as loop
653 variables no longer signals bogus package lock violations. (reported
655 * bug fix: declaring local loop variables to be of a range-limited type
656 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
657 (reported by Andras Simon)
658 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
659 bugs remain on x86-64.)
660 * bug fix: the #= and ## reader macros now interact reasonably with
661 funcallable instances.
662 * bug fix: type-checks for function arguments were compiled using the
663 compiler policy settings of the wrong lexical environment (in 1.0.2
665 * bug fix: SHADOW accepts characters as string designators, as required
666 by the spec (thanks to Eric Marsden)
667 * bug fix: fixed GC safety issues when foreign functions are called with
668 non-base strings as arguments
669 * bug fix: more consistent error output for fatal-compiler-warnings, like
671 * bug fix: functions with proper names were showing up as NIL in
672 backtraces (reported by Edi Weitz, regression in 1.0.2)
674 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
675 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
676 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
677 produce more efficient code on the x86 if the compiler can determine
678 that (+ INDEX <constant>) does not require a bounds check and FOO
679 has an element type at least 8 bits wide.
680 * bug fix: references to undefined variables are handled the same way
681 in toplevel forms as in the normal compiler
682 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
684 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
685 (thanks to Jon Buller)
686 * improvement: faster compilation times for complex functions
687 * improvement: added readlink support to SB-POSIX (thanks to Richard
690 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
691 * improvement: experimental support for mach exception handling on
692 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
694 * improvement: support for GBK external format.
695 (thanks to Chun Tian (binghe))
696 * improvement: the debugger now displays variables that have been closed
697 over, in code compiled with (DEBUG 2) or higher
698 * improvement: support for executable cores on NetBSD (thanks to
700 * new feature: added a RESTART-FRAME debugger command
701 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
702 be used to provide an implementation for FILE-POSITION on Gray streams
703 (thanks to Eric Marsden)
704 * new feature: add syslog, openlog and closelog support to SB-POSIX
705 (thanks to Richard Kreuter)
706 * optimization: the function call overhead in code compiled with
707 a high DEBUG optimization setting is significantly
708 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
709 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
710 (thanks to Tony Martinez)
711 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
712 evaluated once (reported by Kevin Reid)
713 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
714 works again (thanks to Kilian Sprotte)
715 * bug fix: an error is signaled for tagbodies with duplicate tags
716 (thanks to Stephen Wilson)
717 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
718 * bug fix: Win32 port can now handle foreign code unwinding Lisp
719 stack frames from alien callbacks.
720 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
721 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
722 65k pages (thanks to David Woodhouse)
723 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
725 changes in sbcl-1.0.1 relative to sbcl-1.0:
726 * new platform: FreeBSD/x86-64, including support for threading.
727 * new feature: the compiler stores cross-referencing information
728 about function calls (who-calls), macroexpansion (who-macroexpands)
729 and special variables (who-binds, who-sets, who-references) for code
730 compiled with (< SPACE 3). This information is available through the
731 sb-introspect contrib.
732 * new feature: users may subclass SEQUENCE, and have instances of
733 these classes interoperate with standard Common Lisp functions if
734 a number of methods are defined. (This feature is experimental
735 and the interface subject to change based on feedback from SBCL
736 users and the general community)
737 * improvement: sb-sprof traces call stacks to an arbitrary depth on
738 x86/x86-64, rather than the previous fixed depth of 8
739 * improvement: another pthread back-end of mutex "pthread-futex".
740 * bug fix: non-ascii command-line arguments are processed correctly
741 (thanks to Yaroslav Kavenchuk)
742 * bug fix: non-required arguments were not passed correctly when a method
743 defined using DEFMETHOD was called from a mop-generated method using
744 CALL-NEXT-METHOD (reported by Pascal Costanza)
745 * bug fix: recursion is now permitted in accessors through
746 SLOT-UNBOUND. (reported by Pascal Costanza)
747 * bug fix: an error was signaled at startup if the HOME environment
748 variable was defined, but had an empty value (reported by Peter Van Eynde)
749 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
750 signal a PROGRAM-ERROR, not a BUG.
751 * bug fix: SB-POSIX:READDIR works when built with large file support.
752 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
754 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
755 for better automatic indentation support. (thanks to Matt Pillsbury)
756 * optimization: loading generic functions no longer takes O(n^2) time,
757 proportional to the amount of methods in the generic function
758 (reported by Todd Sabin and Jeremy Brown)
759 * optimization: the FIND and POSITION family of sequence functions
760 are significantly faster on arrays whose element types have been
762 * improvements to the Windows port:
763 ** Intermittent heap corruption problems have been fixed. (thanks
764 to Alastair Bridgewater)
765 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
767 ** Lisp is able to unwind foreign exception frames from alien
768 callbacks. (thanks to Alastair Bridgewater)
770 changes in sbcl-1.0 relative to sbcl-0.9.18:
771 * improvement: experimental support for threading on FreeBSD/x86.
772 (thanks to NIIMI Satoshi)
773 * improvement: runtime option --dynamic-space-size can be used
774 to set the size of the dynamic space reserved on startup.
775 * improvement: floating point modes in effect are now saved in
776 core, and restored on startup.
777 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
778 startup, not time since first call to GET-INTERNAL-REAL-TIME.
779 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
780 threads are not running after *SAVE-HOOKS* have run.
781 * improvement: writes to CLOS instance slots are type-checked in code
782 compiled with (SAFETY 3)
783 * improvement: floating-point exception handling on FreeBSD (thanks to
785 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
786 (thanks to Zach Beane)
787 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
789 * improvement: added support for the Shift-JIS external format.
790 (contributed by NIIMI Satoshi)
791 * improvement: callbacks are supported on Linux/PPC. (thanks to
793 * bug fix: compiler bug triggered by a (non-standard) VALUES
794 declaration in a LET* was fixed. (reported by Kaersten Poeck)
795 * bug fix: file compiler no longer confuses validated and already
796 dumped structurres. (reported by Kaersten Poeck)
797 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
798 fill-pointer signals a type-error as required. (thanks to
800 * bug fix: disassemly of funcallable instances works.
801 * bug fix: single stepping on PPC.
802 * bug fix: fix thread-safety problems in the type system (generally
803 manifesting as nonsensical errors like "STRING is a bad type specifier
804 for sequences" or "The value 1 is not of type FIXNUM")
805 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
806 (reported by Josip Gracin)
807 * bug fix: an error is signaled for attempts to displace arrays with
808 incompatible element types (thanks to Mario Mommer)
809 * bug fix: more correct handling of wide characters in debug info
810 (bug reported by Attila Lendvai and fixed by Juho Snellman)
811 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
812 and don't cause extra consing
813 * optimization: MAP and MAP-INTO are significantly faster on vectors
814 whose elements types have been declared.
815 * Improvements to SB-SPROF:
816 ** Support for allocation profiling
817 ** Reduced profiling overhead, especially for long profiling runs
818 * Improvements to the Windows port:
819 ** floating point exceptions are now reported correctly.
820 ** stack exhaustion detection works partially.
821 ** more accurate GET-INTERNAL-REAL-TIME.
822 ** better RUN-PROGRAM behaviour in case of errors in the child
824 ** PROBE-FILE now simplifies pathnames correctly.
825 ** DIRECTORY now works correctly with :WILD-INFERIORS.
826 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
827 1st 1970 (although time timezone data will be unreliable).
828 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
830 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
831 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
832 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
834 * bug fix: two potential GC deadlocks affecting threaded builds.
835 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
836 returns T (reported by Anton Kazennikov)
837 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
838 with non-variable places
839 * bug fix: remove a race condition in the setting of
840 funcallable-instance functions, this should make threaded CLOS
841 code more stable against memory faults.
842 * bug fix: corruption of specials when unbinding is interrupted by an
843 asynchronous unwind (reported by Hannu Koivisto)
844 * improvement: the debugger will now also display local variables that
845 are only used once, for code compiled with a DEBUG optimization quality
848 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
849 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
850 * incompatible change: External-format support for FFI calls. The
851 SB-ALIEN:C-STRING no longer implies an ASCII
852 external-format. Instead the string is subject to external-format
853 conversion. Additionally return values of type SB-ALIEN:C-STRING
854 are by default of type (SIMPLE-ARRAY CHARACTER), not
855 SIMPLE-BASE-STRING. For an alien type that behaves like the old
856 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
857 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
858 * incompatible change: SB-EXT package no longer contains the
859 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
860 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
862 * incompatible change: the single-stepper is no longer available
863 on Alpha, Mips and Sparc platforms.
864 * minor incompatible change: the direct superclasses of
865 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
866 not (STANDARD-OBJECT FUNCTION). This makes the
867 class-precedence-lists of GENERIC-FUNCTION and
868 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
870 * minor incompatible change: the default stream external format on
871 non-unicode SBCL is no longer detected from the locale, but is always
873 * new feature: Add a version of evaluator that uses an interpreter instead
874 of the compiler. EVAL still uses the compiler by default, to switch it
875 to use the interpreter, set the value of the variable
876 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
877 * minor incompatible change: the single-stepper REPL has been merged
878 with the normal debugger (see the "Stepping" heading of the debugger help
879 for more details). The debugger command STEP will no longer switch
880 to the single-stepper REPL.
881 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
882 for a type now works.
883 * bug fix: loading of default sysinit file works. (thanks to Leonid
885 * bug fix: better detection of circularities in the file-compiler.
886 (reported by Marco Monteiro)
887 * bug fix: the CL pathname functions now work with files that have
888 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
889 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
890 systems with Unix98 pty semantics.
891 * bug fix: ASDF-INSTALL will now work with bsd tar.
892 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
894 * bug fix: timers expiring in dead threads no longer cause a
895 type-error (reported by Paul "Nonny Mouse").
896 * bug fix: thanks to more lightweight single-stepper instrumentation,
897 code compiled with (DEBUG 3) will compile and execute significantly faster,
898 and will have more accurate type-inferencing than before
899 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
900 whose bindings are modified
901 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
902 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
903 ** replace dummy implementations of CL:MACHINE-INSTANCE and
904 CL:SOFTWARE-VERSION with proper version
906 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
907 * feature: implemented the READER-METHOD-CLASS and
908 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
909 as specified by AMOP.
910 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
912 * optimization: faster LOGCOUNT implementation on x86 and x86-64
913 (thanks to Lutz Euler)
914 * optimization: hashing of general arrays and vectors has been
915 improved. (reported by Any Fingerhut)
916 * enhancement: SB-INTROSPECT is now able to find definitions of
917 profiled functions. (thanks to Troels Henriksen)
918 * enhancement: compiler-macro expansion applies now to FUNCALL forms
919 as well, allowing compiler-macros for SETF-functions to expand.
920 * enhancement: step-instrumentation no longer wraps known
921 single-value functions with multiple-value context, allowing
922 better type inference.
923 * fixed bug #337: use of MAKE-METHOD in method combination now works
924 even in the presence of user-defined method classes. (reported by
925 Bruno Haible and Pascal Costanza)
926 * fixed bug #339(c): if there are applicable methods not part of any
927 long-form method-combination group, call INVALID-METHOD-ERROR.
928 (reported by Bruno Haible)
929 * fixed bug #361: the :FUNCTION initarg in the protocol for
930 initialization of methods can now be used to override
931 internally-produced optimized functions. (reported by Bruno
933 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
934 system-provided lambda expression no longer cause warnings about
935 unbound #:|pv-table| symbols.
936 * bug fix: improved the handling of type declarations and the
937 detection of violations for keyword arguments with non-constant
939 * bug fix: potentially erronous calls to PATHNAME and
940 MERGE-PATHNAMES were being flushed in some cases.
941 (reported by Richard Kreuter)
942 * bug fix: compiled calls to TYPEP were mishandling obsolete
943 instances. (reported by James Bielman and Attila Lendvai)
944 * bug fix: format strings with ~> without matching ~< no longer
945 trigger an AVER, but signal an understandable error instead.
946 (reported by Antonio Martinez)
947 * bug fix: specifying an output-file without a pathname-type for
948 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
949 of input-file instead of "fasl". (reported by Robert Dodier)
950 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
951 (reported by James Y Knight).
952 * bug fix: compiler-macros-function did not consider the environment
953 argument for shadowing by local functions.
954 * bug fix: compiler-macros expansion was inhibited by local INLINE
956 * bug fix: inline expansions of known functions were subject to
957 step-instrumentation in high debug policies, leading to problems
959 * bug fix: compiler failed to differentiate between different CONS
961 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
962 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
963 * bug fix: #368: incorrect use of expressed vs. upgraded array
965 * bug fix: #306a: more precise unions of array types.
966 * thread-safety improvements:
967 ** CONDITION-WAIT could return early on Linux, if the thread was
968 interrupted and subsequently continued with SIGCONT.
969 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
971 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
972 * added support for the ucs-2 external format. (contributed by Ivan
974 * minor incompatible change: pretty printing of objects of type
975 (cons symbol) is, in the default pprint-dispatch-table, now
976 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
978 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
979 called later in a class's lifetime, possibly as late as when the
980 first instance of the class is created. Previously,
981 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
982 class became finalizeable.
983 * fixed bug: FILE-POSITION sometimes returned inconsistent results
984 for multibyte external-format streams. (thanks to "vbzoli")
985 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
986 with :ALLOCATION :CLASS inherited from superclasses of the
988 * fixed bug: anonymous classes can now be created using the :NAME
989 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
990 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
991 * fixed bug: core-files saved with :EXECUTABLE T can again be
992 executed when SBCL_HOME isn't set. (reported by James Knight)
993 * fixed bug: toplevel LOCALLY forms with declarations could
994 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
995 * fixed bug: printing from several different threads using different
996 values of *print-case* could cause invalid output, due to
997 some internal special variables of the printer not being bound
998 thread-locally (reported by Max Mikhanosha)
999 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1000 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1001 (reported by Pascal Costanza)
1002 * fixed bug: FINALIZE-INHERITANCE is called from
1003 REINITIALIZE-INSTANCE on classes when the class has previously
1004 been finalized, as required by AMOP.
1005 * minor code generation optimizations:
1006 ** better register allocation in CLOS dispatching functions
1007 ** overflow detection when coercing signed bytes to fixnums on x86-64
1008 ** is now implemented with one IMUL instruction instead of three shifts
1009 ** more efficient bit-vector access on x86 and x86-64
1010 ** more efficient access to raw structure slots on x86 and x86-64
1011 * fixed some bugs revealed by Paul Dietz' test suite:
1012 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1013 return its argument.
1015 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1016 * feature: thread support on Solaris/x86, and experimental thread support
1018 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1020 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1021 default method for SLOT-UNBOUND.
1022 * minor incompatible change: prevent the user from specializing the
1023 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1024 somewhat counter to the intent of the protocol, I (CSR) think, and
1025 additionally it just doesn't work in SBCL as currently
1026 implemented, thanks to optimizations (that are always valid for
1027 the other three SLOT-VALUEish functions, but not for the setter).
1028 * minor incompatible change: the :SB-LDB feature is now enabled by
1029 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1030 the low-level debugger.
1031 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1032 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1033 down to an SB-SYS:FD-STREAM, allowing direct communication in
1034 more cases. Temporary files and pipes are still used as a fallback
1036 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1037 on the non-reentrant C functions localtime(3) and gmtime(3).
1038 * bug fix: global optimization policy was not visible in LOCALLY and
1040 * bug fix: class objects can be used as specializers in methods.
1041 (reported by Pascal Costanza)
1042 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1043 with a type error. (reported by blitz_ on #lisp)
1044 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1045 failed with a type error. (reported by Pascal Bourguignon)
1046 * bug fix: merging pathnames against defaults with :DIRECTORY
1047 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1048 (reported by James Y Knight)
1049 * bug fix: saving large (>2GB) cores on x86-64 now works
1050 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1051 constant index and a (simple-array (signed-byte 32)) array
1052 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1053 error (patch by Robert J. Macomber)
1054 * feature: TIME now displays the amount of run-time spent in GC
1055 * bug fix: The debugger now does a better job of respecting
1056 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1057 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1058 workaround for bug 403.)
1059 * bug fix: floating point exception handling now works on FreeBSD
1060 (thanks to NIIMI Satoshi)
1061 * fixed some bugs revealed by Paul Dietz' test suite:
1062 ** MISC.641: LET-conversion were not supposed to work in late
1064 * improvements to the Win32/x86 port:
1065 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1066 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1068 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1069 * new feature: source path information is generated for macro-expansion
1070 errors for use in IDE's like Slime (thanks to Helmut Eller)
1071 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1072 * bug fix: compiler does not loop forever on an invalid type in
1074 * improvement: compilation of most CLOS applications is significantly
1076 * optimization: added a limited bytecode compiler for simple toplevel
1077 forms, speeding up compilation and FASL loading
1078 * bug fix: the statistical profiler now properly distinguishes anonymous
1081 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1082 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1083 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1084 into subclasses of the finalized class. Now user calls to
1085 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1086 system return before any subclasses are finalized.
1087 * minor incompatible change: The reader no longer ignores errors
1088 regarding non-existent packages in #+ and #- feature tests.
1089 * new feature: command line options --no-sysinit, --no-userinit to
1090 inhibit loading the corresponding init files
1091 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1092 for sending data through UDP sockets (thanks to François-René Rideau)
1093 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1094 error is signalled instead (thanks to François-René Rideau)
1095 * minor incompatible change: Error signalling behaviour of lexical
1096 operations violating package locks has changed slightly. Refer to
1097 documentation on package locks for details.
1098 * enhancement: EVAL can process IF-expressions without resorting to the
1100 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1101 constant-folded properly due to missing out-of-line functions.
1102 (reported by Mika Pihlajamäki)
1103 * bug fix: LISTEN sometimes returned T even in cases where no data was
1104 immediately available from the stream
1105 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1106 were reversed. (reported by Levente Mészáros)
1107 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1108 when given a vector displaced to an adjustable array. (reported
1110 * bug fix: derivation of float boundaries from numbers outside the
1111 appropriate float range (reported by John Wiseman)
1112 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1114 * bug fix: precomputing cache entries for generic functions with
1115 some subclasses of specializers as yet invalid does not attempt to
1116 fill a cache line with a negative offset. (reported by Levente
1118 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1119 allow condition class objects as arguments to DOCUMENTATION and
1120 (SETF DOCUMENTATION); only find and set documentation for
1121 structure names for the STRUCTURE doc-type. (suggested by Gary
1123 * improvements to the Win32/x86 port:
1124 ** (user-homedir-pathname) and default initialization file
1125 locations now know about the user's "Documents and Settings"
1126 directory (thanks to Yaroslav Kavenchuk)
1127 ** run-program is implemented (thanks to Mike Thomas)
1128 ** sockets support (thanks to Timothy Ritchey)
1129 ** better backtrace support (thanks to Alastair Bridgewater)
1130 ** sb-grovel supported
1131 ** asdf-install and sb-posix work somewhat
1132 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1133 * improvements to the Solaris/x86 port:
1134 ** works on Solaris 11/Solaris Express
1135 ** floating-point exception handling support
1136 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1137 * fixed some bugs revealed by Paul Dietz' test suite:
1138 ** REMOVE-METHOD returns its generic function argument even when
1139 the method is not one of the generic functions' methods.
1140 ** objects declared as MEMBER types can be admissible arguments to
1141 structure accessors.
1142 ** printing characters should simply be printed by the FORMAT ~:C
1144 ** compiler failure when compiling functions with hairy constant
1145 defaults for optional parameters.
1146 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1147 function, which is already optimized.
1149 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1150 * new platform: experimental support for SBCL x86/Darwin, including
1151 MacOS X 10.4.5 on Intel.
1152 * new feature: Unicode character names are now known to the system
1153 (through CHAR-NAME and NAME-CHAR).
1154 * new feature: the filesystem location of SBCL's core file is
1155 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1156 * minor incompatible change: the contrib modules SB-POSIX and
1157 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1158 this change is to make it easier to distribute
1159 location-independent binaries.
1160 * bug fix: as implied by AMOP, standardized classes no longer have
1161 slots named by external symbols of public packages. (reported by
1163 * optimization: calling functions via constant symbols -- as in
1164 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1165 via the function object as in (FUNCALL #'FOO).
1166 * optimization: on x86 and x86-64, the calling convention for the
1167 case of dealing with an unknown number of values has been altered
1168 to be friendlier to the prediction heuristics implemented,
1169 particularly with reference to CALL and RET pairing. (thanks to
1170 Alastair Bridgewater)
1171 * optimization: on x86, the code for access to thread-local symbol
1172 values and binding/unbinding of thread-local symbols is smaller.
1173 (thanks to Alastair Bridgewater)
1174 * enhancement: CONSTANTP is now able to determine constantness of
1175 more complex forms, including calls to constant-foldable standardized
1176 functions and some special forms beyond QUOTE.
1177 * fixed bug: occasional GC crashes on Solaris/x86
1178 * optimization: x86-64 supports stack allocation of results of simple
1179 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1180 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1181 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1182 (thanks to James Knight)
1183 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1184 by the standard (thanks to Stephen Compall)
1186 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1187 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1188 be used for bundling the runtime and the core file into one
1189 executable binary. This feature is not currently supported on all SBCL
1190 platforms. (thanks to James Bielman and NIIMI Satoshi)
1191 * new feature: a generational or ephemeral garbage collector is now
1192 the default on the PowerPC platform (both Linux and Darwin). The
1193 old Cheney (stop and copy) collector is a build-time option.
1194 (thanks to Cyrus Harmon, after Raymond Toy)
1195 * minor incompatible change: the method by which SBCL finds its
1196 contributed modules has changed; it no longer relies on symbolic
1197 links from an $SBCL_HOME/systems directory, but searches directly
1198 in the subdirectories of $SBCL_HOME.
1199 * enhancement: the dynamic heap size on the Linux/PPC platform is
1200 markedly larger, even using the older Cheney garbage collector.
1201 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1202 certain complicated string types. (reported by Gary King)
1203 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1204 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1205 * fixed bug: DOCUMENTATION on structures no longer signals an
1206 error if no structure type of the right name exists. (reported by
1208 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1209 applicable non-standard primary or :AROUND method on
1210 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1211 no longer cause unbound variable errors. (reported by Kevin Reid)
1212 * fixed bug: metacircle resolution in cases where methods have slots
1213 added before the slots from STANDARD-METHOD. (reported by Jean
1215 * fixed bug: the Power PC assembly code for calling into lisp
1216 sometimes computed its return address wrongly (depending on the
1217 15th bit of the address, fixed at link time). This used to
1218 manifest itself in a segmentation violation while building PCL.
1219 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1220 many others over the years)
1221 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1222 the stack pointer (treated as a signed integer) is negative.
1223 (thanks to Peter van Eynde)
1225 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1226 * new platform: experimental support for the Windows operating
1227 system has been added. (thanks to Alastair Bridgewater)
1228 * fixed several bugs in and robustified the PPC FFI (including
1229 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1230 * bug fix: allow non-simple string symbol names (reported by Paul
1232 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1234 * bug fix: saving a core corrupted callbacks on x86/x86-64
1235 * bug fix: closed a loophole in metacircularity detection and
1236 grounding in the PCL implementation of CLOS.
1237 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1238 (reported by tomppa on #lisp)
1239 * bug fix: clear the direction flag on Lisp -> C transitions, as
1240 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1241 (reported by Andrej Grozin and Hendrik Maryns)
1242 * optimization: major improvements to GC efficiency on GENCGC platforms
1243 * optimization: faster implementation of EQUAL
1244 * optimization: emit more efficient opcodes for some common
1245 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1246 * optimization: several other minor code-generation improvements on x86-64
1247 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1249 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1250 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1251 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1252 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1253 by ANSI as a generic function, and for consistency (SETF
1254 GENERIC-FUNCTION-NAME) is treated likewise.
1255 * fixed bug #233.b: make constraint propagation notice when a variable
1256 value is changed after it is referenced but before it is used
1257 * fixed bug #296: no more arbitrary behaviour from filenames with
1258 odd characters as --load arguments. (reported by Adam Warner)
1259 * fixed bug #390: :CHARACTER-SET pathname components now work as
1260 expected. (reported by Tim Daly Jr)
1261 * fixed bug #391: complicated :TYPE intersections in slot
1262 definitions no longer cause an error in PCL internals.
1263 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1264 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1265 returning the number of octets which would be written to the
1266 file-stream. (thanks to Robert J. Macomber)
1267 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1268 arguments to the internal stream functions. (thanks to David
1270 * bug fix: stack exhaustion is now handled on the x86/SunOS
1271 platform. (thanks to Daisuke Homma)
1272 * bug fix: more accurate ROOM results on GENCGC platforms
1273 * optimization: improved type inference for arithmetic-for
1274 index variables in LOOP
1275 * optimization: faster floating-point SQRT on x86-64
1276 * fixed some bugs revealed by Paul Dietz' test suite:
1277 ** DOCUMENTATION returns NIL instead of "" for method combinations
1278 that don't have a docstring
1280 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1281 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1282 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1283 therefore are not customizeable by user code (as seems to be at
1284 least permitted and maybe required by AMOP). As a consolation,
1285 however, the SBCL implementation of these functions now calls
1286 REINITIALIZE-INSTANCE as specified by AMOP.
1287 * bug fix: slot-definition documentation is propagated as per ANSI
1288 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1289 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1290 argument, as do eslotd initialization methods. (from Pascal
1291 Costanza's "Closer" project)
1292 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1293 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1295 * bug fix: it is now possible to have more than one subclass of
1296 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1297 (reported by Bruno Haible, Pascal Costanza and others)
1298 * bug fix: the dependent update protocol now works for generic
1299 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1300 and Pascal Costanza)
1301 * bug fix: condition-class instances corresponding to
1302 DEFINE-CONDITION forms are now created eagerly. (reported by
1303 Kalle Olavi Niemitalo on comp.lang.lisp)
1304 * bug fix: floating point printing is more accurate in some
1305 circumstances. (thanks to Simon Alexander)
1306 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1307 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1308 * bug fix: callbacks on OS X now preserve stack-alignment.
1309 (thanks to Cyrus Harmon)
1310 * enhancement: the x86-64 disassembler is much better at
1311 disassembling SSE instructions. (thanks to Lutz Euler)
1312 * enhancement: CHECK-TYPE on a variable now allows type-inference
1313 (thanks to Vincent Arkesteijn)
1314 * optimization: improved performance of EUC-JP external format.
1315 (thanks to NIIMI Satoshi)
1316 * optimization: performance improvements to IO on file streams of
1317 :ELEMENT-TYPE CHARACTER
1318 * optimization: much faster memory allocation on x86-64
1319 * optimization: faster garbage collections (latency and throughput)
1321 * optimization: faster bignum subtraction and division on x86 and x86-64
1322 (thanks to Lutz Euler)
1323 * optimization: more accurate type-inference for loops that use a
1324 floating point index variable or a negative step.
1326 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1327 * bug fix: add a workaround to SBCL looping infinitely at startup on
1328 Linux kernels with apparently buggy implementations of personality().
1329 (thanks to Svein Ove Aas)
1330 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1331 (thanks to David Lichteblau)
1332 * bug fix: local bindings shadow global symbol macros for
1333 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1334 * optimization: non-open coded uses of numeric comparison operators
1335 (e.g. >) no longer cons when called with more than one parameter
1336 on platforms supporting dynamic-extent allocation.
1337 * enhancement: saving cores with foreign code loaded is now supported
1338 on MIPS/Linux in addition to the previously supported platforms.
1339 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1340 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1341 * bug fix: division by zero in sb-sprof when no samples were collected
1342 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1343 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1344 (thanks to Alexander Kjeldaas)
1345 * new feature: ignoring the return values of destructive functions
1346 like NREVERSE signals a compile-time style-warning.
1347 (thanks to Kevin Reid)
1348 * new feature: the alignment of alien structure fields can be
1349 explicitly specified. (thanks to Cyrus Harmon)
1350 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1351 and dump core on SIGQUIT
1353 ** incompatible change: threads do not inherit values of specials
1354 from their parents (see manual)
1355 ** bug fix: threads stacks belonging to dead threads are freed by the
1356 next exiting thread, no need to gc to collect thread stacks anymore
1357 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1358 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1359 ** bug fix: dynamic variable and thread start related gc lossage
1360 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1362 * fixed some bugs revealed by Paul Dietz' test suite:
1363 ** SUBTYPEP is slightly more accurate on heinously complicated
1364 CONS types where some of the members have uncertain (in the
1365 NIL, NIL sense) type relationships to each other.
1367 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1368 no longer purified unless :PURIFY T is explicitly specified.
1369 ** Non-purified cores are significantly smaller than before
1371 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1372 * new feature: timers based on Zach Beane's excellent timer package
1373 * added support for the following external formats: koi8-u,
1374 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1375 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1376 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1377 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1378 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1379 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1380 * incompatible change: a threaded SBCL will no longer revert to
1381 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1382 * bug fix: interrupts are disabled until startup is complete; no
1383 more sigsegvs when receiving a signal to soon
1384 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1386 * bug fix: add a workaround for the memory randomization features in
1387 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1388 workaround will only be in effect on systems with the proc filesystem
1390 * bug fix: printing objects of type HASH-TABLE signals a
1391 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1393 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1394 have been running for over 50 days. (reported by Gilbert Baumann)
1395 * bug fix: the logic for getting names of functions gets less
1396 confused when confronded with alternate-metaclass
1397 funcallable-instances. (reported by Cyrus Harmon)
1398 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1399 consistent, even on internal alternate-metaclass objects.
1400 * bug fix: flush closure information collected by physical
1401 environment analysis prepass before the main pass. (bug reported
1403 * bug fix: compiler pack phase does not modify a hashtable, which is
1404 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1405 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1406 may be expanded. (reported by Alexander <malishev> on c.l.l)
1407 * bug fix: no more style-warnings when compiling code using LAST
1408 under high-SPEED low-SPACE settings. (reported by David Wallin
1410 * fixed some incompatibilities between SBCL's MOP and the MOP
1412 ** the METAOBJECT class is now implemented;
1413 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1414 STANDARD-OBJECT, as required;
1415 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1416 now compatible (as required by VALIDATE-SUPERCLASS); there
1417 remains a consistency requirement over the presence or absence
1418 of the FUNCTION class in the superclass list of finalized
1419 classes; see the manual for more details;
1420 ** the :AROUND method for COMPUTE-SLOTS on
1421 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1422 requested slot ordering.
1424 ** bug fix: parent thread now can be gc'ed even with a live
1426 ** bug fix: binding a special with PROGV to no value is not
1427 the same as that symbol not having been bound (thanks to
1429 ** bug fix: binding specials is thread safe (thanks to
1431 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1432 and SB-SPROF do not die with 'no handler for signal XX in
1433 interrupt_handle_now(..)' anymore
1434 ** bug fix: WITH-TIMEOUT works with multiple running threads
1435 * fixed some bugs revealed by Paul Dietz' test suite:
1436 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1437 the :method-class keyword argument.
1439 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1440 * new port: the Solaris operating system on x86 processors is now
1441 mostly supported, though some rough edges in the environment
1442 remain. (thanks to Daisuke Homma)
1443 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1444 space for its heap. (thanks to Thiemo Seufer)
1445 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1446 closures. (thanks to Thiemo Seufer)
1447 * minor incompatible change: eof selects abort in the debugger.
1448 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1449 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1451 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1452 deprecated, and will go away in a future revision of SBCL.
1453 * minor incompatible change: GC-ON and GC-OFF are no longer
1454 implemented with a counter, it does not matter how many times gc
1455 is switched on or off
1456 * bug fix: discriminating functions for generic function classes
1457 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1458 make invalid assumptions about method precedence order. (reported
1460 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1461 platforms. (thanks to Thiemo Seufer)
1462 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1463 * bug fix: correct pathname printing with printer escaping is on.
1464 (thanks to Kevin Reid)
1465 * bug fix: complex VOP definitions in "user-space" no longer trigger
1466 package locks. (reported by Zach Beane)
1467 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1468 for funcallable-instance functions, and is no different from
1470 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1471 characters in the current readtable. (reported by Nicholas Neuss)
1472 * bug fix: SUBTYPEP on various CONS types returns more a more
1473 accurate acknowledgment of its certainty.
1474 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1475 lists in some cases. This partially fixes bug 384.
1476 * flush all standard streams before prompting in the REPL and the
1478 * bug fix: signal handling and triggering gc do not conflict
1479 directly anymore, in particular a high frequency sb-sprof does
1480 not prevent gc from running
1481 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1482 approximation for timezone and DST information between the
1483 universal time epoch and the smallest negative 32-bit time_t.
1484 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1485 year 1899 so long as the encoded time is non-negative.
1486 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1487 the correct answer for non-constant rotate amounts when performing
1488 an inline 32-bit rotation.
1490 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1491 there is only one thread in the session
1492 ** bug fix: memory leak for streams created in one thread and
1493 written to in another
1494 ** bug fix: lockup when compiled with gcc4
1495 ** bug fix: race that allows the gc to be triggered when gc is
1497 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1498 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1500 ** bug fix: finished threads are gc'ed properly
1501 * fixed some bugs revealed by Paul Dietz' test suite:
1502 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1503 has arguments with a different set of applicable methods from
1504 the orignal arguments.
1505 ** The type error thrown by MAP now has a correct expected-type
1507 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1508 name a compiled function.
1509 ** Three MISC tests where a large bignum was improperly coerced to
1510 a float that couldn't represent that bignum during type
1511 derivation were fixed.
1512 ** SUBTYPEP can now handle types involving the negation of a
1513 list-form FUNCTION type.
1514 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1515 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1516 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1518 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1519 * New feature: Experimental support for bivalent streams: streams
1520 opened with :element-type :default now allow character and binary
1521 (unsigned-byte 8) I/O
1522 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1523 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1525 * bug fix: no more highly sporadic "couldn't check whether ~S is
1526 readable" when reading a stream and an interrupt hits in the middle
1527 of a select system call
1528 * compiler better recognizes complex arrays (reported by Eduardo
1530 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1531 for variables on SBCL built with linkage-tables. (reported by Luis
1533 * various error reporting improvements.
1534 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1535 (Thanks to Thiemo Seufer)
1536 * bug fix: nested reader invokations work correctly
1537 * bug fix: it is possible to have simultaneous references to foreign
1538 code and foreign data with the same name.
1540 ** added x86-64 support
1541 ** incompatible change: the threading api now works with thread
1542 objects instead of thread ids
1543 ** bug fix: threads are protected from signals and interruption when
1544 starting up or going down
1545 ** bug fix: a race where an exiting thread could lose its stack to gc
1546 ** bug fix: don't halt on infinite error in threads if possible
1547 ** fixed numerous gc deadlocks introduced in the pthread merge
1548 ** bug fix: fixed thread safety issues in read and print
1549 ** bug fix: debugger doesn't hang on session lock if interrupted at
1550 an inappropriate moment
1551 ** bug fix: run-program is now thread safe(r)
1552 ** bug fix: inner with-recursive-lock no longer releases the mutex
1553 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1554 * fixed some bugs revealed by Paul Dietz' test suite:
1555 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1556 EXPECTED-TYPE slots filled.
1557 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1559 ** MAKE-CONDITION accepts classes as type-designators.
1560 ** COMPILE may never return NIL.
1561 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1562 range before calling Unix time functions
1564 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1565 * numerous signal handling fixes to increase stability
1566 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1567 * minor incompatible change: we now correctly canonize default
1568 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1569 as per the MOP, rather than the historical (INITARG INITFUNCTION
1570 INITFORM). (reported by Bruno Haible)
1571 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1572 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1573 for more information.
1574 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1575 pathname is a directory pathname.
1576 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1577 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1579 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1580 :METACLASS argument in addition to a class name. (reported by
1581 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1582 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1583 (reported by Bruno Haible; patch from Gerd Moellmann)
1584 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1586 * bug fix: more cleanups to the floating point exception handling on
1587 x86-64 (thanks to James Knight)
1588 * bug fix: the compiler does not try to constant fold calls of
1589 COERCE and BIT-* functions when they return freshly constructed
1590 objects. (reported by Nikodemus Siivola and Paul Dietz)
1591 * optimization: improved the MIPS versions of generic (in the
1592 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1593 * optimization: direct conversion of (unsigned-byte 32) to floats on
1594 the PowerPC platform.
1595 * optimization: structure instances with raw slots now use less
1596 memory, and probably show better memory locality. (thanks to
1598 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1599 are treated as requests for stack allocation on the x86-64,
1600 Alpha-32, PPC and SPARC platforms.
1601 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1602 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1603 (thanks to Tony Martinez)
1605 ** gcing a dead thread can no longer lead to lockups
1606 ** threads block signals until they are set up properly
1607 ** errno is no longer shared by threads
1608 ** interrupt-thread restores the eflags register on x86
1609 ** fixed some lockups due to gc/thread interaction
1610 * fixed some bugs revealed by Paul Dietz' test suite:
1611 ** invalid dotted lists no longer raise a read error when
1612 *READ-SUPPRESS* is T
1613 ** signal an error if a symbol that names a declaration is used
1614 as the name of a type, or vice versa
1615 ** allow using the (declare (typespec var*)) abbreviation for
1616 (declare (type typespec var*)) with all type specifiers
1617 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1618 ** always return NIL from PPRINT-POP when OBJECT is NIL
1619 ** don't signal errors when pretty-printing malformed LABELS,
1620 FLET or MACROLET forms
1621 ** declarations in a DOLIST body are also in scope for the
1623 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1625 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1628 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1629 * fixed cross-compiler leakages that prevented building a 32-bit
1630 target with a 64-bit host compiler.
1631 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1632 opened with :IF-EXISTS :OVERWRITE.
1633 * fixed bug 281: error for an invalid qualifier in a short-form method
1634 combination method is not signalled until the faulty method is called.
1635 * bug fix: iteration variable type inferrer failed to deal with open
1636 intervals. (reported by Alan Shields)
1637 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1638 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1639 or more runtime options were provided to the sbcl binary.
1640 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1641 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1642 * optimization: there's now a fast-path for fixnum arguments in the
1643 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1644 * optimization: the code generated on x86-64 is more compact thanks
1645 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1646 * bug fix: floating-point exception handling now partly works on
1647 x86-64. (Thanks to James Knight)
1648 * improvement to the MIPS backend from Thiemo Seufer: C-style
1649 64-bit long long arguments and return values to alien functions
1651 * fixed some bugs revealed by Paul Dietz' test suite:
1652 ** the type-error signalled from WARN has a filled-in DATUM slot.
1653 ** the type-error required when a stream is not associated with
1654 a file has the stream as its datum.
1655 ** type-errors on single-floats on x86-64 no longer have
1656 :INVALID-OBJECT as the datum
1657 ** the type-errors signalled for invalid function names now have
1658 a correct expected type
1659 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1660 for structures defined with a :TYPE. Documentation strings for
1661 typed structures are no longer immediately discarded
1662 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1663 broadcast streams and synonym streams. FILE-LENGTH now also works
1664 on broadcast streams.
1666 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1667 * incompatible change: the --noprogrammer option, deprecated since
1668 version 0.7.5, has been removed. Please use the equivalent
1669 --disable-debugger option instead.
1670 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1672 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1673 inoperational for a while now) has been completely removed.
1674 * null lexical environments are now printed as #<NULL-LEXENV>,
1675 significantly reducing the amount of clutter in typical backtraces.
1676 * documentation on weak pointers, finalization, and after GC hooks
1677 has been added to the manual.
1678 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1679 as other specialized array types, is much faster. SUBSEQ and
1680 COPY-SEQ on such arrays have also been sped up.
1681 * optimization: EQL is now more efficient when at least other argument
1682 is known to be of type (OR FIXNUM (NOT NUMBER)).
1683 * fixed bug: the runtime is now less vulnerable to changes in the
1684 size of the SBCL object on OS X, and virtual memory is reserved for
1685 all spaces that need to be at a fixed address.
1686 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1687 * fixed bug: finalizers and after GC hooks that cause consing are now
1689 * fixed bug: compiler error messages and summaries are now printed to
1690 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1691 * fixed inference of the upper bound of an iteration variable.
1692 (reported by Rajat Datta).
1693 * fixed bug 211e: calling local functions with duplicated constant
1694 keyword argument no longer causes a bogus style warning about an
1696 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1697 ftype declaration to be disregarded. (reported by Dave Roberts)
1698 * fixed bug 373: caused by erronous compilation of references to alien
1699 variables in the runtime on ppc/darwin.
1700 * fixed bug 376: CONJUGATE type deriver.
1701 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1702 (reported by Baughn on #lisp)
1703 * a cross-compiler bug on non-x86 platforms has been identified and
1704 fixed. (thanks to Bruno Haible)
1705 * improvements to the MIPS runtime code for increased stability.
1706 (thanks to Thiemo Seufer)
1707 * increased the maximimum compact environment size to allow
1708 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1709 * partial workaround for bug 135: don't name the function we're
1710 calling for hairy cases of EVAL, so as not to accumulate one environment
1711 entry per call to EVAL. (reported by Kevin Reid)
1712 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1713 * optimization: type testing for non-vector arrays should be faster.
1714 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1716 * bug fix: setting 31st element of a bit vector to zero did not work
1718 * bug fix: redefining a class definition which failed due to a
1719 previous accessor / function clash now works (but see BUGS entry
1720 #380 for more problems in this area). (thanks to Zach Beane)
1721 * the long form of DEFINE-METHOD-COMBINATION disables method group
1722 checking when given a single method group with pattern *. (thanks
1724 * on x86 compiler supports stack allocation of results of simple
1725 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1726 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1727 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1728 * fixed some bugs related to Unicode integration:
1729 ** the restarts for recovering from input and output encoding
1730 errors only appear when there is in fact such an error to
1732 * fixed some bugs revealed by Paul Dietz' test suite:
1733 ** MISC.549 and similar: late transformation of unsafe type
1734 assertions into derived types caused unexpected code
1736 ** SCALE-FLOAT type deriver is less wrong.
1737 ** type derivers for EXP, LOG and similar functions compute result
1738 types for complex arguments better.
1739 ** (MISC.563) CONJUGATE type deriver works for very restricted
1741 ** out-of-line type testers for character strings are available.
1742 ** EQUAL compiler transform understands specialness of objects
1744 ** accessing double-floats stored on the stack now works on x86-64.
1745 ** debugger internals could sometimes create invalid lispobjs,
1746 resulting in GC crashes.
1747 ** MISC.548: type check weakening can convert required type into
1749 ** initialization forms for bindings are not in scope of free special
1752 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1753 * incompatible change: thread support for non-NPTL systems has
1754 been removed - locking is buggy and unreliable. A threaded
1755 SBCL build will now warn at startup and refuse to create
1756 new threads, unless futex support is detected in the kernel
1757 * incompatible change: the top level REPL now has only an ABORT
1758 restart associated with it, not TOPLEVEL and ABORT as it used to.
1759 TOP and TOPLEVEL are now available as debugger commands for
1760 returning to the top level.
1761 * incompatible change: forms evaluated in the REPL now use the
1762 global optimization policy.
1763 * incompatible change: user- and system-initialization files are
1764 no longer processed with LOAD, but by READ and EVAL; hence the
1765 global optimization policy, startup package, readtable, etc,
1767 * The .fasl file format number has been incremented because of
1768 various incompatible changes.
1769 * internal entry point details and argument counts no longer appear
1770 in backtraces unless explicitly requested by setting
1771 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1772 * built-in and standard functions no longer have names like "top
1773 level local call to FOO".
1774 * fixed bug 32: functions defined in non-null lexical environments
1775 now have more legible printed representation
1776 * fixed bug 33: functions defined in non-null lexical environemnts
1777 are now more amenable to inspection by INSPECT.
1778 * workaround for bug 354: XEPs no longer appear in backtraces unless
1779 explicitly requested.
1780 * fixed bug: receiving the signal which results from attempting to
1781 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1782 SIGBUS on other *BSDs) on architectures where the C stack is also
1783 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1784 to debugger confusion or wild instability. (reported by Cheuksan
1786 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1787 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1788 (reported by Lutz Euler)
1789 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1790 compiled code. (reported by Paul Dietz)
1791 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1792 specialized parameter to a method to something that is not TYPEP
1793 the specializer is now possible.
1794 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1795 face of package deletion.
1796 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1797 pathname. (reported by David Tolpin on c.l.l)
1798 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1799 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1800 STRING-STREAM and FILE-STREAM, respectively.
1801 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1802 than previously; in addition, its test suite is now run on build.
1803 (thanks to Luke Gorrie)
1804 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1805 * optimization: added a immediate representation for single-floats
1807 * fixed some bugs revealed by Paul Dietz' test suite:
1808 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1809 ** attempting to create a package with a colliding nickname causes
1810 correctable errors to be signalled.
1811 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1812 ** several x86-64 backend bugs related to sign-extension of immediate
1815 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1816 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1817 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1819 * fixed loading of multiply forward-referenced layouts.
1820 (thanks to Cheuksan Wang)
1821 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1822 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1823 (see COMPILE-FILE documentation for details of :PRINT option)
1824 * fixed bugs 19 and 317: fixed-format floating point printing is
1825 more accurate. This also fixes a bug reported by Adam Warner
1826 related to the ~@F format directive.
1827 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1829 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1830 dispatch table if the from-char is a dispatch macro character.
1831 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1832 by the wordsize no longer ignore the last word. (reported by Lutz
1834 * fixed bug in type checking of dynamic-extent variables. (reported
1836 * optimization: sequence traversal functions use their freedom to
1837 coerce function designators to functions.
1838 * optimization: code with many calls to CLOS methods specialized on
1839 CLOS classes has had redundant type checks removed.
1840 * optimization: added declarations to speed up operations that access
1841 the internal character database (for example STRING-UPCASE)
1842 * optimization: comparison operations between floats and sufficiently small
1843 fixnums no longer create extra rationals
1844 * fixed some bugs related to Unicode integration:
1845 ** portions of multibyte characters at the end of buffers for
1846 character-based file input are correctly transferred to the
1847 start of the buffer at the next read.
1848 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1849 passing it through to OPEN.
1850 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1851 argument given to internal calls to OPEN.
1852 ** the built-in comment readers (introduced by character sequences
1853 ";" and "#|") are more forgiving to encoding errors; they will
1854 STYLE-WARN and then attempt to resync the stream at a character
1855 boundary. (thanks to Teemu Kalvas)
1856 * fixed some bugs in the x86-64 port:
1857 ** Negative short int return values from c-calls are sign-extended
1859 ** The stack is aligned to 16-bytes for c-calls, as required by
1860 the ABI. (thanks to Cheuksan Wang)
1861 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1862 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1863 for enums. (thanks to Vincent Arkesteijn)
1864 ** Multiple small optimizations and bugfixes for floating point
1866 * fixed some bugs revealed by Paul Dietz' test suite:
1867 ** Space, Tab, Linefeed, Return and Page have the invalid
1868 secondary constituent character trait.
1869 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1871 ** WITH-INPUT-FROM-STRING should only update the index place on
1873 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1874 works more reliably.
1875 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1876 with one-dimensional arrays with fill pointers.
1877 ** TYPE-OF failed on a complex with an integer realpart and a
1879 ** compiler failure during type inference for the code of form
1880 (IF (EQL X (THE ...)) ...) (MISC.535).
1882 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1883 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1884 platform. The port supports 61-bit fixnums, large memory spaces
1885 and reloading shared object files.
1886 * enhancement: saving cores with foreign code loaded is now
1887 supported on ppc/Darwin in addition to the previously supported
1889 * enhancement: the statistical profiler now walks deeper into the
1890 call stack for more meaningful call-graphs and accrued time
1891 reports (x86/x86-64 only). It also now reports time spent in
1893 * enhancement: it is now possible to trace most individual methods
1894 of a generic function in addition to tracing the generic function
1896 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1898 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1899 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1900 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1901 produces an error. (thanks to Vincent Arkesteijn)
1902 * bug fix: NAMESTRING on pathnames with :WILD components in their
1903 directories works correctly. (thanks to Artem V. Andreev)
1904 * fixed bug 125: compiler preserves identity of closures. (reported
1906 * bug fixed: functions with &REST arguments sometimes failed with
1907 "Undefined function" when compiled with (DEBUG 3). (reported by
1909 * bug fix: overflow during compiling of setting element of a bit
1910 vector with constant index and value. (reported by Timmy Douglas)
1911 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1912 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1913 * on x86 compiler partially supports stack allocation of dynamic-extent
1915 * GO and RETURN-FROM do not check the extent of their exit points
1916 when compiled with SAFETY 0.
1917 * fixed some bugs related to Unicode integration:
1918 ** encoding and decoding errors are now much more robustly
1919 handled; it should now be possible to recover even from invalid
1920 input or output to the terminal. (thanks to Teemu Kalvas)
1921 ** provided a first cut at implementing STRING-TO-OCTETS and
1922 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1923 ** altered the SB-MD5 contributed module slightly, changing the
1924 interface just enough for it to be supportable for builds where
1925 lisp characters are not eight bits.
1926 * fixed some bugs revealed by Paul Dietz' test suite:
1927 ** the FORMATTER-generated functions for ~V[ conditionals require
1928 the correct number of arguments.
1929 ** READ-FROM-STRING returns the mandated second value when applied
1930 to displaced strings.
1931 ** the #\Rubout and #\Backspace characters are treated as invalid
1932 constituent characters by the tokenizer.
1934 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1935 * new feature: reloading changed shared object files with
1936 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1937 * new feature: references to foreign variables and functions
1938 can now be compiled and loaded before the corresponding shared
1939 object file is loaded, as long as the foreign definitions are
1940 available at runtime.
1941 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1942 supported. (thanks to Dan Debertin)
1943 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1944 just keywords. (thanks to Vincent Arkesteijn)
1945 * fixed bug #331: structure-class instances corresponding to
1946 DEFSTRUCT forms are now created eagerly.
1947 * fixed bug #345: backtraces from calls to undefined functions work
1948 on x86 as well. Related bug #61 is now also partially fixed on x86
1949 (backtraces from throws to unknown catch tags.)
1950 * bug fix: lambda-list parsing is now stricter vrt. order and number
1951 of lambda-list keywords.
1952 * bug fix: as specified by AMOP, an error is signalled if a
1953 class-option appears multiple times in a DEFCLASS form. (reported
1955 * bug fix: RANDOM can be compiled when the compiler derives the type
1956 of its numeric argument as a disjoint set of small integers.
1957 (reported by Paul Dietz)
1958 * bug fix: starting a core saved with shared objects loaded when
1959 those objects are not available no longer causes threaded SBCL to
1960 hang. (reported by Sean Ross)
1961 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1963 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1964 argument, which must always be NIL. (reported by Kalle Niemitalo)
1965 * bug fix: printing 1.0d+23 no longer results in an error.
1966 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1967 * bug fix: structure slot setters preserve evaluation order. (thanks
1969 * fixed some bugs related to Unicode integration:
1970 ** RUN-PROGRAM can allow its child to take input from a Lisp
1971 stream. (reported by Stefan Scholl)
1972 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1973 locales. (reported by Ken Causey)
1974 * fixed some bugs revealed by Paul Dietz' test suite:
1975 ** INCF, DECF and REMF evaluate their place form as specified in
1977 ** FORMATTER expands ~{ ~} iteration directives with V or #
1978 parameters correctly.
1979 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1980 consequent uses no arguments correctly.
1981 ** the system has a partial understanding of the (COMPLEX RATIO)
1983 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1986 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1987 * new feature: a build-time option (controlled by the :SB-UNICODE
1988 keyword feature, enabled by default) for building the system with
1989 support for the entire 21-bit character space defined by the
1991 * new feature: the system now has rudimentary external-format
1992 support; the primary user-visible change at this time is that
1993 characters with the high bit set (such as lower-case-e-acute) will
1994 print correctly to a terminal in a UTF-8 environment.
1995 * minor incompatible change: BASE-CHAR no longer names a class;
1996 however, CHARACTER continues to do so, as required by ANSI.
1997 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1998 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1999 has been moved to the SB-EXT package (temporarily re-exported from
2001 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2002 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2004 * fixed bug #351: better error-handlind and reporting for malformed
2006 * fixed bug #350: bignum-printing is now more memory-efficient,
2007 allowing printing of very large bignums, eg. (expt 2 10000000).
2008 (reported by Bruno Haible)
2009 * fixed bug #302: better primitive-type selection for intersection
2011 * fixed bug #308: non-graphic characters now all have names, as
2012 required. (reported by Bruno Haible)
2013 * bug fix: redefining a class with different superclasses now correctly
2014 removes it from the direct-subclasses of its previous superclasses.
2015 (reported by David Morse)
2016 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2017 new value now works. (reported by Bruno Haible)
2018 * bug fix: correct canonicalization of multiple non-standard slot
2019 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2020 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2021 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2022 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2023 now exists, an signals an error.
2024 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2025 during macroexpansion for non-iteration variables as well. (reported
2026 by Bruno Haible for CMUCL)
2027 * bug fix: Cyclic structures and unprintable objects in compiler
2028 messages no longer cause errors. (reported by Bruno Haible)
2029 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2030 and READ-FROM-STRING all now return a primary value of NIL if
2031 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2032 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2033 (reported by Bruno Haible for CMUCL)
2034 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2035 do not designate a condition. (reported by Bruno Haible for
2037 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2038 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2039 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2040 permitting resolution in favour of any of the conflicting symbols.
2041 (reported by Bruno Haible for CMUCL)
2042 * bug fix: EQUAL compiler optimizations is less aggressive on
2043 strings which can potentially compare true despite having distinct
2044 specialized array element types.
2045 * bug fix: unit enumerations can be defined without dividing by
2046 zero. (thanks to Vincent Arkesteijn)
2047 * FORMAT compile-time argument count checking has been enhanced.
2048 (report from Bruno Haible for CMUCL)
2049 * a partial workaround for the bug 262: the compiler does not try to
2050 inline-expand a local function doing RETURN-FROM from a deleted
2051 BLOCK. (thanks to Peter Denno for the bug report and to David
2052 Wragg for the simple test case)
2053 * fixed some bugs revealed by Paul Dietz' test suite:
2054 ** RENAME-PACKAGE allows all package designators as new package
2056 ** constraint propagation and lambda variable substitution are
2057 more cautious in dealing with partially deleted code.
2058 ** compiler sometimes lost reoptimization passes.
2059 ** CERROR, given a condition as condition designator, passes its
2060 remaining arguments to the continue format control without
2062 ** Case-altering FORMAT directives work correctly on non-ASCII
2064 ** The REAL type specifier handles bounds outside the floating
2065 point ranges without signalling FLOATING-POINT-OVERFLOW.
2066 ** Functions with IR1-transformations can create intercomponent
2067 references to global functions.
2068 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2070 ** FORMAT ~R treats a nil value for its first parameter correctly.
2072 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2073 * enhancement: saving cores with foreign code loaded is now
2074 supported on x86/NetBSD and sparc/Linux in addition to the previously
2075 supported platforms.
2076 * bug fix: on some platforms repeated installations caused multiple
2077 copies of HTML documentation to be installed -- should not happen
2078 any more. (reported by Stefan Scholl)
2079 * bug fix: parsing self-recursive alien record types multiple times
2080 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2081 original patch by Helmut Eller for CMUCL)
2082 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2083 (thanks to Richard Kreuter)
2084 * bug fix: defining classes whose accessors are methods on existing
2085 generic functions in other (locked) packages no longer signals
2086 bogus package lock violations. (reported by François-René Rideau)
2087 * bug fix: special variables as DEFMETHOD parameters no longer have
2088 associated bogus type declarations. (reported by David Wragg and
2090 * bug fix: read-write consistency on streams of element-type
2091 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2092 * bug fix: redefiniton of the only method of a generic function with
2093 no DEFGENERIC no longer emits a full WARNING. In addition,
2094 redefinition of generic functions with no DEFGENERIC to an
2095 incompatible lambda list now signals an error. (thanks to Zach
2097 * bug fix: DEFGENERIC now works even when there's a function of the
2098 same name in an enclosing lexical environment. (thanks to Zach
2100 * fixed compiler failure, caused by instrumenting code during
2101 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2102 * optimization: added loop analysis and improved register allocation
2103 * fixed some bugs revealed by Paul Dietz' test suite:
2104 ** POSITION on displaced vectors with non-zero displacement
2105 returns the right answer.
2106 ** (SIMPLE-STRING) is a valid type specifier for sequence
2108 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2110 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2111 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2113 ** The justification version of the FORMAT ~< directive treats
2114 non-zero minpad parameter correctly.
2116 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2117 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2118 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2119 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2120 the supported interface.
2121 * new feature: Single-stepping of code compiled with DEBUG 2 or
2122 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2123 * new feature: saving cores with foreign code loaded is now
2124 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2125 Timothy Moore's work for CMUCL)
2126 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2127 parameters to * instead of NIL if no initform is supplied.
2128 (reported by Johan Bockgaard)
2129 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2130 correctly when FUNCALL appears as the car of the form. Note:
2131 despite this FUNCALL forms are not currently subject to
2132 compiler-macro expansion. (port of Raymond Toy's fix for the
2133 same from CMUCL, reported by Johan Bockgaard)
2134 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2135 (thanks for Teemu Kalvas)
2136 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2137 case instead of terminating the iteration (thanks for Julian
2138 Squires, Sean Champ and Raymond Toy)
2139 * bug fix: incorrect expansion of defgeneric that caused a style
2140 warning. (thanks for Zach Beane)
2141 * bug fix: slot accessor effective method computation works properly
2142 for classes with multiple non-standard applicable methods on
2143 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2144 * on x86 compiler supports stack allocation of results of LIST and
2145 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2146 CMUCL implementation by Gerd Moellmann)
2147 * fixed some bugs revealed by Paul Dietz' test suite:
2148 ** FORMAT strings with both the ~<~:;~> form of the justification
2149 directive and pretty-printing directives cause an error.
2151 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2152 * incompatible change: the internal functions
2153 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2154 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2155 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2156 instead of the old functions.
2157 * new feature: on platforms where "dladdr" is available foreign
2158 function names now appear in backtraces. (based on Helmut Eller's
2160 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2161 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2163 * documentation: ASDF manual is now distributed with SBCL.
2164 * bug fix: SBCL can now load its contributed modules with REQUIRE
2165 even if the system-provided entries have been removed from
2166 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2168 * bug fix: backtraces involving undefined functions or assembly
2169 routines are more informative. (thanks to Brian Downing)
2170 * bug fix: mutually referent alien structures now work correctly.
2171 (reported by Rick Taube)
2172 * bug fix: structures defined by WITH-ALIEN can be referred to
2173 within other definitions in the same WITH-ALIEN.
2174 * bug fix: division operators (MOD, TRUNCATE and the like) with
2175 constant zero divisors and integer dividends no longer generate
2177 * bug fix: provide default methods for INPUT-STREAM-P and
2178 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2179 * bug fix: improve the detection and resolution of MOP metacycles
2180 (where effective method computation depends on the generic function
2181 having its effective method computed). (reported by Bruno Haible)
2182 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2183 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2184 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2185 * bug fix: pathnames with relative directory components can be
2186 represented relative to default pathnames.
2187 * optimization: in taking the GCD of bignums, reduce the two bignums
2188 to approximately the same size (using Euclid's algorithm) before
2189 applying the more sophisticated binary GCD. (thanks to Juho
2191 * optimization: COUNT on bitvectors now operates word-at-a-time.
2192 * optimization: ASH with a positive, but not necessarily constant,
2193 (leftwards) shift, when in a modular context, is compiled to a
2195 * fixed some bugs revealed by Paul Dietz' test suite:
2196 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2197 if the corresponding argument is NIL.
2198 ** FORMAT directives accepting a minpad parameter treat negative
2200 ** literal commas as character parameters to FORMAT directives are
2202 ** literal spaces directly after ~<Newline> directives within a
2203 format-logical-block (~:< ~@:>) do not induce :FILL-style
2204 conditional newlines.
2205 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2206 as specified: it no longer includes conditional newlines).
2207 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2209 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2210 :PER-LINE-PREFIX argument does not evaluate to a string.
2212 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2213 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2214 the manual for details; for now, package locks can be disabled by
2215 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2216 if no major problems are found then it is likely that they will be
2217 compiled in unconditionally.
2218 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2219 unsupported operators on all platforms. To load a shared library
2220 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2221 object file, link it into a shared library outside of SBCL and
2222 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2224 * bug workaround: ROOM T and the GC were not working together
2225 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2226 reporting facility aren't preserved. That reporting has been
2227 disabled, so now until and unless someone figures out how to make it
2228 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2229 * minor incompatible change: as threatened around sbcl-0.8.0, the
2230 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2231 The symbols are also exported from SB-PCL for backwards
2232 compatibility, but more so than before SB-PCL should be treated as
2233 an implementation-internal package.
2234 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2236 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2237 It is known as of this release not to work on the Alpha, however.
2238 * fixed bug #167: errors signalled due to illegal syntax in method
2239 bodies are now more legible.
2240 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2241 designators and can hence be used with TYPEP.
2242 * fixed bug #333: CHECK-TYPE now ensures that the type error
2243 signalled, if any, has the right object to be accessed by
2244 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2245 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2246 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2248 * fixed bug #334: programmatic addition of slots using specialized
2249 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2250 and :ALLOCATION :CLASS slots.
2251 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2252 float exponents. (rereported by Peter Seibel)
2253 * fixed a bug: #\Space (and other whitespace characters) are no
2254 longer considered to be macro characters in standard syntax by
2255 GET-MACRO-CHARACTER.
2256 * fixed bug: initialization of condition class metaobjects no longer
2257 causes an instance of the condition to be created. (reported by
2259 * fixed bug: it is now possible to have slots such that
2260 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2261 description is neither :INSTANCE nor :CLASS.
2262 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2263 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2264 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2265 non-standard SLOT-DEFINITION-ALLOCATION values.
2266 * fixed bugs: various race conditions handling exiting threads.
2267 CL-PPCRE's thread torture test now passes.
2268 * fixed arguably-a-bug: GC time proportional to number of threads in
2269 system even when most of them are idle
2270 * optimization: improved performance of BIT and SBIT on bit-vectors.
2271 * .fasl file incompatibility: The fasl file version number has
2272 been incremented because of changes associated with package locks.
2274 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2275 * minor incompatible change: the system no longer provides
2276 optimization hints (conditions which are TYPEP
2277 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2278 mode; these hints are still emitted when higher SPEED optimization
2280 * new contrib module: a sampling profiler (profiling by statistical
2281 sampling, rather than by instrumenting functions) is available as
2282 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2283 * the behaviour of the standard function ED is now customizeable by
2284 third parties through a hook variable: see ED's documentation
2285 string for information on the protocol.
2286 * the compiler no longer emits efficiency notes for (FUNCALL X)
2287 when the type of X is uncertain under default optimization
2289 * fixed bug 276: mutating a binding of a specialized parameter to a
2290 method to something that is not TYPEP the specializer is now
2292 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2293 exhibit the required behaviour on the x86 platform. (thanks to
2294 Peter van Eynde, Eric Marsden and Bruno Haible)
2295 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2296 even for difficult arguments. (reported by Peter Graves)
2297 * fixed bug 141a: the backquote printer now descends quoted
2299 * fixed another bug in backquote printing: no more destructive
2300 modification of the form's list structure. (reported by Brian
2302 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2303 move between its address being taken and the call to
2304 interrupt_thread, fixing a crashing race condition.
2305 * the SB-POSIX contrib implementation has been adjusted so that it
2306 no longer exhibits ridiculously poor performance when constructing
2307 instances corresponding to C structs.
2309 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2310 * minor incompatible change: the sb-grovel contrib now treats C
2311 structures as alien (in the sense of SB-ALIEN) objects rather than
2312 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2313 has implications for memory management of client code
2314 (sb-grovel-returned objects must now be manually managed) and for
2315 type safety (alien objects now have full types).
2316 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2317 used to control emission of compiler diagnostics, rather than the
2318 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2319 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2320 quality should be considered deprecated.
2321 * install.sh now installs the user manual as well
2322 * (not quite a new documentable feature, but worth considering in
2323 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2324 beginnings of a semantically meaningful condition hierarchy is
2325 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2327 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2328 designator as the defaults argument.
2329 * fixed bug: Displaced arrays whose displaced-to array has become
2330 too small now cause ARRAY-DIMENSION to signal an error, providing
2331 for safer bounds-checking. (reported by Bruno Haible)
2332 * fixed bug: DEFCLASS slot definitions with identical :READER and
2333 :WRITER names now signal a reasonable error. (reported by Thomas
2335 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2337 * fixed bug: Invalid slot specification errors now print correctly.
2338 (thanks to Zach Beane)
2339 * fixed bug 320: Shared to local slot value transfers in class
2340 redefinitions now happen corrently from superclasses as
2341 well. (reported by Bruno Haible)
2342 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2344 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2345 improper lists correctly. (reported by Bruno Haible)
2346 * fixed bug 313: source-transform for <fun-name> was erroneously
2347 applied to a call of a value of a variable with name <fun-name>.
2348 (reported by Antonio Menezes Leitao)
2349 * fixed bug 307: The obsolete instance protocol ensures that
2350 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2351 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2352 remove dead unknown-values globs from the middle of the stack.
2353 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2354 easier to resume long computations after using *BREAK-ON-SIGNALS*
2355 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2356 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2357 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2358 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2359 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2360 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2362 * fixed a (fairly theoretical) bug in string printing: if
2363 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2364 does not have array-element-type equal to the most general string
2366 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2367 function name. (thanks to Nikodemus Siivola)
2368 * fixed bug: compile-time format string checker failed on
2369 non-closed ~{. (reported by Thomas F Burdick)
2370 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2371 don't warn on legal loop code involving more than one aggregate
2372 boolean. (thanks to Nikodemus Siivola)
2373 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2374 in keyword position to MAKE-PACKAGE was not being reported as
2376 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2377 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2379 * fixed bugs 280 and 312: the checking for multiple definitions in a
2380 file is less likely to become confused by uses of inline
2382 * fixed bug: the #S reader macro performs the keyword coercion
2383 specified for slot names. (reported by Kalle Niemitalo)
2384 * fixed bug: lambda lists may contain symbols whose names start with
2385 & but are not lambda-list-keywords; their occurrence triggers a
2387 * fixed bug 321: define-method-combination argument lambda lists do
2388 not cause invalid code to be generated when &optional or &aux
2389 variables have default values. (reported by Bruno Haible)
2390 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2391 have CLOS classes; this appears also to have fixed reported
2392 instability in stack exhaustion detection.
2393 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2394 MUFFLE-WARNING when no associated restart is present is now
2396 * optimization: rearranged the expansion of various defining macros
2397 so that each expands into only one top-level form in a
2398 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2400 * optimization: used a previously little-used slot in symbols to
2401 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2403 * fixed some bugs revealed by Paul Dietz' test suite:
2404 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2406 ** Fixed an optimization bug related to inheritance of initforms
2407 from local to shared slots.
2408 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2409 ** CAST optimizer forgot to flush argument derived type.
2410 ** print/read consistency on floats is now orders of magnitude
2411 more likely. (thanks also to Bruno Haible for a similar report
2413 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2414 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2415 for objects of type REAL. Make it so.
2416 ** SXHASH is specified (infelicitously) to respect similarity,
2417 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2418 it so. (thanks to Markus Ziegler)
2419 ** on the Alpha, the compiler succeeds in compiling functions
2420 returning a known number of arguments greater than 63.
2421 ** fixed handling of invalid NIL arguments in keyword position
2423 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2425 ** provided a readably-printable representation for RANDOM-STATE
2427 ** ensured that pathnames, where they have a namestring, always
2428 print using #P"..." syntax.
2430 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2431 * Support for the forthcoming 2.0 version of the NetBSD kernel
2432 running on x86 hardware has been added. (thanks to Perry
2433 E. Metzger most immediately, and others for their past work)
2434 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2435 current release version; SBCL's OpenBSD support had been broken
2436 since about the time of OpenBSD's switch to ELF binary format.)
2437 * [placeholder for DX summary]
2438 ** user code with &REST lists declared dynamic-extent, under high
2439 speed or space and low safety and debug optimization policy.
2440 * The manual has been converted to Texinfo format and the debugger
2441 chapter from the cmucl manual has been added.
2442 * A facility has been added to extract documentation strings from
2443 sbcl and store them as Texinfo-formatted snippets for inclusion in
2444 the manual (via Texinfo's @include directive)
2445 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2446 the test case to Dave Roberts)
2447 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2448 pointers. (reported by Sean Ross)
2449 * bug fix: PROFILE output is printed nicely even for large numerical
2450 values. (thanks to Zach Beane)
2451 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2452 greater than 32 handle EOF correctly.
2453 * bug fix: on X86 an immediate argument of the IMUL instruction is
2454 correctly printed in disassembly. (thanks to Lutz Euler)
2455 * bug fix: class slots in redefined classes preserve their old
2456 values. (thanks to Bruno Haible and Nikodemus Siivola)
2457 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2458 succeeds. (reported by Marco Baringer)
2459 * fixed some bugs revealed by Paul Dietz' test suite:
2460 ** READ-SEQUENCE now works on ECHO-STREAMs.
2461 ** RATIONALIZE works more according to its specification. (thanks
2464 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2465 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2466 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2467 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2468 should matter to you only if you rebind the printer control
2469 variables and then find you want different bindings in the
2470 debugger than in the ordinary execution of your program.)
2471 * The runtime build system has been tweaked to support building
2472 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2473 assembler and linker. (thanks to Nikodemus Siivola)
2474 * Unbound, undefined, undeclared variables now trigger full
2475 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2476 more often programmer error than deliberate exploitation of undefined
2478 * optimization: the hash algorithm for strings has changed to one
2479 that is less vulnerable to spurious collisions. (thanks to Juho
2481 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2482 needless bounds checking. (thanks to Juho Snellman)
2483 * optimization: implemented multiplication as a modular
2484 (UNSIGNED-BYTE 32) operation on the PPC backend.
2485 * fixed some bugs revealed by Paul Dietz' test suite:
2486 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2488 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2489 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2490 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2492 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2494 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2496 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2497 their output stream on EOF from read.
2498 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2499 have been read to end-of-file.
2500 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2502 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2503 description of determination of which consecutive characters
2505 ** Printing the "Space" character with escaping on now yields "#\\ ",
2506 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2507 ** Reading floating-point numbers with *READ-BASE* set to a number
2508 less than 10 works correctly.
2509 ** Reading floating-point numbers with *READ-BASE* set to a number
2510 more than 10 works correctly.
2511 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2512 the readtable currently in effect.
2514 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2515 * minor incompatible change: parsing of namestrings on a physical
2516 (Unix) host has changed; numbers after the final #\. in a
2517 namestring are no longer interpreted as a version field. This is
2518 intented to be largely invisible to the user, except that the
2519 meaning of the namestring "*.*.*" has changed: it now refers to a
2520 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2521 should usually be replaced by
2522 (make-pathname :name :wild :type :wild :version :wild)
2523 with the added benefit that this is more likely to be portable.
2524 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2525 option to OPEN now signals an error if the file being opened
2526 exists; this may have an impact on existing code.
2527 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2528 signals received, on the PowerPC platforms (both Linux and
2529 Darwin). (thanks to Pierre Mai for pointing out the location of
2531 * several fixes on OS X: The system now builds and runs cleanly on
2532 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2533 (thanks to Brian Mastenbrook)
2534 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2535 second-resolution: integer multiples of 1/3600 between -24 and 24.
2536 (thanks to Vincent Arkesteijn)
2537 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2538 type when called with 1 argument; PEEK-CHAR checked type of
2539 PEEK-TYPE only after having read first character from a
2540 stream. (reported by Peter Graves)
2541 * bug fix: the garbage collector now has much better locality
2542 behaviour, and in particular no longer treats all memory as being
2543 exhausted when many small objects point to each other in a deeply
2545 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2546 recognized as being TYPEP their class.
2547 * bug fix: the PUSHNEW documentation string has been corrected.
2548 (thanks to Vincent Arkesteijn)
2549 * bug fix: defaulting of the value for the last of an atypically
2550 large number of multiple values being bound was not being
2551 performed correctly on the Alpha or PPC platforms
2552 * optimization: implemented multiplication as a modular
2553 (UNSIGNED-BYTE 32) operation on the x86 backend.
2554 * optimization: SEARCH on simple-base-strings can now be open-coded.
2555 (see also contrib/compiler-extras.lisp for inspiration for
2556 teaching the compiler about the Boyer-Moore algorithm).
2557 * value, returned by MAX (and MIN) called with several EQUALP, but
2558 not EQL, arguments now does not depend on compiler settings.
2559 * fixed some bugs revealed by Paul Dietz' test suite:
2560 ** in stack analysis liveness information is propagated from
2561 non-local entry points.
2562 ** pathwise CAST removing failed when the CAST node did not start
2564 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2565 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2567 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2568 host is already defined.
2569 ** RENAME-FILE works on streams instead of signalling an internal
2571 ** PEEK-CHAR uses the current readtable when determining whether
2572 or not a character is whitespace.
2573 ** MERGE-PATHNAMES handles the case when the pathname does not
2574 specify a name while the default-pathname specifies a version.
2575 ** Pathnames now stand a better chance of respecting print/read
2577 ** Attempting to use standardized file system operators with a
2578 pathname with invalid :DIRECTORY components signals a
2580 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2582 ** DIRECTORY on logical pathnames is more correct.
2583 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2584 signal a TYPE-ERROR if their argument is not a stream.
2585 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2586 designator argument does not designate a stream.
2587 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2588 examining the synonym.
2589 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2591 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2592 element-type larger than ([UN]SIGNED-BYTE 32).
2594 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2595 * When built with the :SB-FUTEX feature, threaded builds now take
2596 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2597 for faster/more reliable mutex and condition variable support.
2598 * Incompatible change (but one you probably shouldn't have been using
2599 anyway): the interface and code for arbitrating between multiple
2600 threads in the same user session has been redesigned.
2601 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2602 error when called without an explicit environment argument.
2603 (thanks to Nikodemus Siivola)
2604 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2605 confused about their position. (thanks to Adam Warner and Gerd
2607 * bug fix: Pretty printing backquoted forms with unquotations in the
2608 argument list position of various code constructs such as LAMBDA
2609 now works correctly. (reported by Paul Dietz)
2610 * bug fix: Pretty printing unquotations no longer loses all
2611 stream position information.
2612 * optimization: performance of string output streams is now less
2613 poor for multiple small sequence writes.
2614 * optimization: performance of CSUBTYPEP in the presence of complex
2615 expressions involving CONS and NOT many times has been improved.
2616 (reported by Paul Dietz)
2617 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2619 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2620 streams for the REPL, for the debugger, and so on. (thanks to
2622 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2623 * fixed some bugs revealed by Paul Dietz' test suite:
2624 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2625 is ignored if it is too small, rather than propagating through
2626 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2627 ** extremely complex negations of CONS types were not being
2628 sufficiently canonicalized, leading to inconsistencies in
2630 ** VALUES tranformer lost derived type.
2632 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2633 * fixed a bootstrapping bug: the build process no longer assumes
2634 that the various BOOLE-related constants have the same value in
2635 host and target lisps. (noted by Paul Dietz' test suite on an
2636 SBCL binary built from CLISP)
2637 * The system can now be dynamically linked on the MIPS platform,
2638 which enables dynamic loading of foreign code from Lisp. (thanks
2639 to Ralf Baechle for discussions on the MIPS ABI)
2640 * The system now records debugging information for its own source
2641 files in a filesystem-position-independent manner, relative to
2642 the "SYS" logical host.
2643 * fixed a compiler bug: MV-LET convertion did not check references
2644 to the "max args" entry point. (reported by Brian Downing)
2645 * tweaked disassembly notes to be less confident about proclaiming
2646 some instruction as an LRA. (thanks to Brian Downing)
2647 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2648 now each have their own history, command character, and other
2649 characteristics. (thanks to David Lichteblau)
2650 * fixed some bugs revealed by Paul Dietz' test suite:
2651 ** compiler failure in compiling LOGAND expressions including a
2653 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2654 shift greater than 32.
2655 ** FLUSH-DEST did not mark blocks for type check regeneration.
2656 ** HANDLER-CASE failed to accept declarations in handler clauses
2657 in some circumstances.
2659 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2660 * New code in contrib/sb-introspect (still probably not entirely
2661 stable yet) provides some support for smart Lisp development
2662 environments like SLIME.
2663 * The conditions signalled for errors occurring when loading .fasl
2664 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2665 in a way which should help ASDF recover gracefully.
2666 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2667 stuff has been cleaned up. If you code contrib/ stuff, this might
2668 affect you, and you can look at contrib/README, contrib/STANDARDS,
2669 and/or the 0.8.4.27 diff to check.
2670 * In full calls the compiler now does not generate checks for declared
2671 argument types for all arguments.
2672 * various threading fixes
2673 ** and some experimental patches which didn't make it into
2674 the main tree for this release, but which are shipped in
2675 contrib/experimental-thread.patch as a possible fix for some
2676 failures (deadlock, spinning...) in GC-intensive multithreaded
2678 * fixed PPC build problem (source code incompatibility of different
2679 library versions): added offsetof() hackery which attempts to divine
2680 where glibc maintainers put uc_mcontext today
2681 * fixed bug 282: compiler does not trust type assertions while passing
2682 arguments to a full call.
2683 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2684 &OPTIONAL VALUES type parameter.
2685 * fix bug 214: algorithm for noting rejected templates is now more
2686 similar to that of template seletion. (also reported by rydis on
2688 * fixed bug 141b: printing backquoted information readably and prettily
2689 inserts a space where necessary.
2690 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2691 (CHARACTER 10) are now reported as errors, rather than propagated
2692 as unknown types. (reported by piso on #lisp)
2693 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2694 with values NIL and :ERROR. (thanks to Milan Zamazal)
2695 * fixed bug 191c: CLOS now does proper keyword argument checking as
2696 described in CLHS 7.6.5 and 7.6.5.1.
2697 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2698 longer raise an error; further, using a list as a for-as-arithmetic
2699 counter now raises a meaningful error.
2700 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2701 now understood by sequence creation functions such as MAKE-SEQUENCE
2703 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2704 a STRING-INPUT-STREAM.
2705 * compiler enhancement: SIGNUM is now better able to derive the type
2707 * type declarations inside WITH-SLOTS are checked. (reported by
2709 * fixed some bugs revealed by Paul Dietz' test suite:
2710 ** incorrect optimization of TRUNCATE for a positive first
2711 argument and negative second.
2712 ** compiler failure in let-convertion during flushing dead code.
2713 ** compiler failure while deriving type of TRUNCATE on an
2714 interval, containing 0.
2715 ** ASH of a negative bignum by a negative bignum count now returns
2717 ** intersection of CONS types now canonicalizes properly, fixing
2718 inconsistencies in SUBTYPEP.
2720 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2721 * incompatible change: The --disable-debugger command line
2722 option now clobbers the debugger at a more fundamental
2723 level, by redefining #'INVOKE-DEBUGGER instead of by
2724 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2725 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2726 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2727 enter the debugger REPL and then suffer recursive errors
2728 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2729 while under the new system, BREAK in --disable-debugger mode
2730 terminates the system just as an unhandled error would.
2731 * fixed compiler performance when processing loops with a step >1;
2732 * bug fix: DOCUMENTATION now retrieves generic function
2733 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2734 support has been systematized, and now supports the methods
2735 specified by ANSI, along with a default method and a method for
2736 slot documentation. (reported by Nathan Froyd)
2737 * bug fix: effective methods associated with a generic function are
2738 no longer cached over a change of that generic function's method
2739 combination. (reported by Andreas Fuchs)
2740 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2741 names a non-existent directory. (thanks to Andreas Fuchs)
2742 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2743 the PPC platform, lead to stack corruption; this has been fixed.
2744 (reported by Rainer Joswig)
2745 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2746 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2747 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2748 arithmetic in general, is now much more reliable on the PPC
2750 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2751 platform now returns the right answer.
2752 * optimization: restored some effective method precomputation in
2753 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2754 precomputation is now tunable.
2755 * optimization: compiler-internal data structure use has been
2756 reviewed, and changes have been made that should improve the
2757 performance of the compiler by about 20%.
2758 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2759 simple-base-strings and simple-bit-vectors is improved.
2760 * optimization: the optimization of 32-bit logical and arithmetic
2761 functions introduced in version 0.8.3 on the x86 has been
2762 implemented on the mips, ppc and sparc platforms; an
2763 implementation of the same facility, but for 64-bit arithmetic,
2764 has been added for the alpha.
2765 * microoptimization: the compiler is better able to make use of the
2766 x86 LEA instruction for multiplication by constants.
2767 * bug fix: in some situations compiler did not report usage of
2768 generic arithmetic in (SPEED 3) policy.
2769 * bug 145b fix: compiler used wrong type specifier while converting
2770 MEMBER-types to numeric.
2771 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2773 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2774 index argument now works properly on the Alpha platform.
2775 * bug fix: floating point exception treatment on the Alpha platform
2777 * bug fix: FILE-POSITION works much better on string input and
2778 output streams. (thanks to Nikodemus Siivola)
2779 * bug fix: many threading/garbage collection symptoms sorted.
2780 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2781 might be pseudo-atomic.
2782 * internal change: Stopping for GC is now done with signals not
2783 ptrace. GC is now done in whichever thread wanted it, instead of
2785 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2787 * bug fix: result form in DO is not contained in the implicit
2789 * incompatible change: ICR structure is changed; the value part of
2790 CONTINUATION is now called LVAR; corresponding functions are
2791 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2792 * added type deriver for ISQRT (thanks to Robert E. Brown).
2793 * bug fix: better support for loading from the command line when an
2794 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2796 * fixed some bugs revealed by Paul Dietz' test suite:
2797 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2798 ** ROUND and FROUND now give the right answer when given very
2799 small float arguments.
2800 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2802 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2803 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2804 ** LCM should return a non-negative integer.
2805 ** PARSE-INTEGER returned the index of a terminator instead of the
2806 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2807 ** PARSE-INTEGER returned an incorrect index being applied to a
2809 ** LCM with two arguments of 0 returns 0 rather than signalling
2811 ** unsigned addition of a 32-bit constant with the high bit set no
2812 longer causes an internal compiler error on the x86.
2813 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2814 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2815 with negative last argument.
2816 ** byte specifiers with zero size and position no longer cause
2817 an error during type derivation.
2818 ** bignum multiplication on the Alpha platform now returns the
2820 * porting: The system now builds on SuSE AMD64, although it still
2821 generates a 32-bit binary.
2822 * .fasl file incompatibility: The fasl file version number has
2823 been incremented (because of the changes to internal compiler
2824 data structures referred to above).
2826 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2827 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2828 more accurately, on the Darwin kernel running on PowerPC hardware.
2829 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2830 * Compiler code deletion notes now signal a condition of type
2831 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2832 an associated MUFFLE-WARNING restart.
2833 * The compiler now performs limited argument count validation of
2834 constant format strings in FORMAT, and where appropriate in ERROR,
2835 CERROR and WARN. (thanks to Gerd Moellmann)
2836 * New ASDF-INSTALL contrib can be used for automatic download and
2837 installation of third-party Lisp code from CCLAN or other sites
2839 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2840 which forces another thread to execute a function supplied by the
2842 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2843 now accept and act upon their :ELEMENT-TYPE keyword argument.
2844 (reported by Martin Atzmueller, Edi Weitz)
2845 * bug fix: FILE-POSITION now accepts position designators up to
2846 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2847 is the greater. (thanks to Patrik Nordebo)
2848 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2849 Kalle Olavi Niemitalo)
2850 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2851 printable. (reported by Eric Marsden)
2852 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2853 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2854 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2855 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2856 (thanks to Stig Sandoe)
2857 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2858 UNREAD-CHAR now correctly includes the unread character in the
2859 target sequence. (thanks to Gerd Moellmann)
2860 * bug fix in threaded builds: the system can now be suspended and
2861 resumed by shell job control with minimal disruption.
2862 * bug fixes in times and timezones >2038AD
2863 * better handling of "where is GNU make?" problem in build scripts
2864 (thanks to Nikodemus Siivola)
2865 * new optimization: inside a named function any reference to a
2866 function with the same name is considered to be a self-reference;
2867 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2868 optimization quality.
2869 * new optimization on x86: logical functions and + now have
2870 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2871 used when the result is truncated to 32 bits.
2872 * VALUES declaration is partially enabled.
2873 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2874 * fixed some bugs revealed by Paul Dietz' test suite:
2875 ** The system now obeys the constraint imposed by
2876 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2877 types form a lattice under type intersection.
2878 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2879 ** ASSOC now ignores NIL elements in an alist.
2880 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2881 and (1+ MOST-POSITIVE-FIXNUM) answers.
2882 ** The addition of a method with invalid qualifiers to a generic
2883 function does not cause an error to be signalled immediately;
2884 a warning is signalled, and the error is generated only on
2885 calling the generic function.
2886 * changed .fasl file version number, in part to add type codes for
2887 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2888 obscure ANSI requirements
2890 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2891 * fixed bug 148: failure to inline-expand a local function left
2892 garbage, confusing the compiler.
2893 * fixed bugs 3cd: structure slot readers perform type check if the
2894 slot can have an invalid value (i.e. it is either not initialized
2895 or can be written with a less specific slot writer).
2896 * bug fix: the compiler now traps array references to elements off
2897 the end of an array; previously, the bounds checking in some
2898 circumstances could go off-by-one.
2899 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2901 * type declarations for array element types now obey the description
2902 on the CLHS page "Declaration TYPE", as per discussions on
2903 sbcl-help around 2003-05-08. This means that a declaration
2904 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2905 declaration, all references to BAR will be asserted or assumed
2906 (with THE, so dependent on compiler policy) to involve objects of
2907 type FOO. Note that no such declaration is implied in
2908 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2909 * declared types of functions from the "Conditions"
2910 chapter. (reported by Paul Dietz)
2911 * bug fix: CERROR accepts a function as its first argument.
2912 * bug fix: NTH an NTHCDR accept a bignum as index
2913 arguments. (reported by Adam Warner)
2914 * optimization: character compare routines now optimize comparing
2915 against a constant character. (reported by Gilbert Baumann)
2916 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2917 argument now works properly on the MIPS platform.
2918 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2919 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2920 declared to have a functional type, does not cause an error
2922 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2923 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2924 * fixed bug: if last continuation of a deleted block has a
2925 destination, this destination should be deleted too. (reported by
2927 * fixed a bug in the bootstrap process: the host compiler's values
2928 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2929 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2930 test case from Patrik Nordebo)
2931 * improved the ability of the disassembler on the PPC platform to
2932 provide helpful disassembly notes.
2933 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2934 the class in more cases than previously.
2935 * bug fix: FILE-POSITION now understands :START and :END for
2936 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2937 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2938 8) data. (thanks to David Lichteblau)
2939 * bug fix: it is possible to add a method to a generic function
2940 without lambda list.
2941 * bug fix: reader failed to signal END-OF-FILE inside an
2942 object representation. (reported by Nikodemus Siivola)
2943 * fixed some bugs revealed by Paul Dietz' test suite:
2944 ** LAST and [N]BUTLAST should accept a bignum.
2945 ** condition slot accessors are methods.
2946 ** (VECTOR NIL) is a subtype of STRING.
2948 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2949 * minor incompatible change: some nonsensical specialized lambda
2950 lists (used in DEFMETHOD) which were previously ignored now signal
2952 * minor incompatible change: the system is now aware of the types of
2953 variables in the COMMON-LISP package, and will signal errors for
2954 most violations of these type constraints (where previously they
2955 were silently accepted).
2956 * minor incompatible change: COMPILE-FILE now uses the freedom
2957 afforded (ANSI 3.2.2.3) to use derived function types for
2958 functions defined in the same file. This also permits the system
2959 to warn on static type mismatches and function
2960 redefinition. (Currently it does not work with high DEBUG level.)
2961 * minor incompatible change: VALUES declaration is disabled.
2962 * When issuing notes, the compiler now signals a condition of type
2963 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2964 restart for use in user handlers. It is expected that the
2965 COMPILER-NOTE condition will eventually become a condition
2966 supertype to a hierarchy of note types, which will then be
2967 handleable in a similar fashion. However, at the moment, no such
2968 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2969 but it's an implementation detail, not a classification for the
2971 * Changes in type checking closed the following bugs:
2972 ** type checking of unused values (192b, 194d, 203);
2973 ** template selection based on unsafe type assertions (192c, 236);
2974 ** type checking in branches (194bc).
2975 * A short form of VALUES type specifier has ANSI meaning (it has
2976 increased the number of situations when SBCL cannot perform type
2978 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2979 keywords or constants is permissible.
2980 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2981 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2982 classes. (thanks to Antonio Martinez)
2983 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2984 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2985 argument to a pretty-print stream. Instead, it leaves any such
2986 operations to DESCRIBE-OBJECT methods.
2987 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2989 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2990 lambda lists. (thanks to David Lichteblau)
2991 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2992 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2993 (thanks to Gerd Moellmann)
2994 * fixed bug 246: increased compilation speed of long
2995 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2997 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2999 * DEFCONSTANT now throws a condition of type
3000 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3001 constant to a non-EQL value; CONTINUE and ABORT restarts
3002 respectively change and preserve the value.
3003 * fixed bug 63: The code walker, part of the implementation of CLOS,
3004 is now better at handling symbol macros.
3005 * bug fix: There is no longer an internal implementation type named
3006 CL:LENGTH. (reported by Raymond Toy)
3007 * bug fix: In macro-like defining macros/special operators the
3008 implicit block does not enclose the lambda list.
3009 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3010 enclosed in parentheses) are not suitable as type specifiers, and
3011 their use properly signals an error now.
3012 * bug fix: An argument count mismatch for a type specifier in code
3013 being compiled no longer causes an unhandled error at compile
3014 time, but signals a compile-time warning.
3015 * fixed simple vector readable printing
3016 * bug fix: DESCRIBE takes more care over whether the class
3017 precedence list slot of a class is bound before accessing it.
3018 (reported by Markus Krummenacker)
3019 * bug fix: FORMATTER can successfully compile pretty-printer format
3020 strings which use variants of the ~* directive inside.
3021 * bug fix: SEARCH now applies its TEST predicate to the elements of
3022 the arguments in the correct order. (thanks to Wolfhard Buss)
3023 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3024 in type inference. (thanks to Robert E. Brown)
3025 * bug fix: Reading in symbols with an explicit package name of ""
3026 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3027 * fixed some bugs revealed by Paul Dietz' test suite:
3028 ** NIL is now allowed as a structure slot name.
3029 ** Arbitrary numbers, not just REALs, are allowed in certain
3030 circumstances in LOOP for-as-arithmetic clauses.
3031 ** Multiple class redefinitions before slot access no longer
3032 causes a type error.
3033 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3034 association between the name and a class.
3035 ** Generic functions with non-standard method-combination and over
3036 five methods all of which return constants no longer return NIL
3037 after the first few invocations. (thanks to Gerd Moellmann)
3038 ** CALL-NEXT-METHOD with no arguments now passes the original
3039 values of the arguments, even in the presence of assignment.
3040 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3042 ** DISASSEMBLE works with closures and funcallable instances.
3043 ** ADD-METHOD now returns the generic function, not the new method.
3044 ** FIND-METHOD signals an error if the lengths of the specializers
3045 is incompatible with the generic function, even if the ERRORP
3047 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3048 which its argument is a member.
3049 ** DEFCLASS only redefines the class named by its class-name
3050 argument if that name is the proper name of the class;
3051 otherwise, it creates a new class.
3052 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3053 of the UNBOUND-SLOT condition to the name of the slot.
3054 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3055 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3056 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3057 SLOT-MAKUNBOUND in the specified fashion.
3059 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3060 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3061 cross-compilation host. As a consequence, we can now bootstrap our
3062 way up to SBCL starting with a bare gcc toolchain and human-readable
3063 source code (first the source to CLISP, then the source to SBCL).
3064 * A contributed module containing a partial implementation of the
3065 simple-streams interface has been included. (thanks to Rudi
3067 * A contributed module implementing the RFC1321 Message Digest
3068 Algorithm, known as MD5, has been included.
3069 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3070 no longer has any effect, as the code controlled by this feature
3071 has been deleted. (As far as we know, no-one has ever built using
3072 this feature, and its semantics were confused in any case).
3073 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3074 work on conditions (as required by the ANSI specification),
3075 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3076 expected behaviour on conditions. Users should note, however,
3077 that such behaviour is not required by the ANSI specification,
3078 and so use of this behaviour may render their code unportable.
3079 * fixed some bugs revealed by Paul Dietz' test suite:
3080 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3082 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3083 ** SXHASH on condition objects no longer returns NIL.
3084 ** :ALLOCATION :CLASS slots are better treated; their values are
3085 updated on class redefinition, and initforms inherited from
3086 superclasses are applied.
3087 ** REMOVE-METHOD returns its generic function argument even when
3088 no method was removed.
3089 ** SHARED-INITIALIZE now initializes the values of the requested
3090 slots, including those with :ALLOCATION :CLASS.
3091 ** ALLOCATE-INSTANCE now works on structure classes defined via
3092 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3094 ** SLOT-EXISTS-P now works on conditions, as well as structures
3096 ** MAKE-LOAD-FORM now has the required methods on
3097 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3098 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3099 keyword, and now implements the SLOT-NAMES argument.
3100 ** methods with &OPTIONAL arguments no longer allow too many
3101 arguments to be passed in the call without error.
3102 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3103 option is consistent with the required arguments of the generic
3104 function lambda list.
3105 * bug fix: REQUIRE accepts a string designator. (Thanks to
3107 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3108 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3109 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3110 * bug fix: The compiler now checks for duplicated variables in macro
3112 * bug fix: SETQ on globals returns the correct value.
3113 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3114 not a COMPILER-ERROR (followed by some other strange error on
3115 choosing the CONTINUE restart).
3116 * bug fix: make.sh and friends are now more consistent in the way that
3117 they look for GNU "make".
3119 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3120 * experimental native threads support (on x86 Linux >=2.4 only).
3121 This is not compiled in by default: you need to add :SB-THREAD to
3122 the target features. See the "Beyond ANSI" chapter of the manual
3124 * fix for longstanding nonANSIism: The old distinction between
3125 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3126 The return value from CL:FIND-CLASS is now a CLOS class, and
3127 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3128 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3129 * An interface to the MetaObject Protocol, as described in Kiczales,
3130 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3131 MIT Press, 1991, is available from the SB-MOP package.
3132 * incompatible change: the SB-PCL package should now be considered
3133 a private implementation detail, and no longer a semi-private MOP
3135 * minor incompatible change: due to rearrangement for threads, the
3136 control stack and binding stack are now allocated at arbitrary
3137 addresses instead of being hardcoded per-port. Users affected by
3138 this probably have to be doing advanced things with shared
3139 libraries, and will know who they are.
3140 * minor incompatible change: Previously, all --eval forms used were
3141 processed with READ before any of them were processed with EVAL.
3142 Now each --eval form is processed with both READ and EVAL before
3143 the next --eval form is processed. (Thus package operations like
3144 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3145 work as the user might reasonably expect.)
3146 * minor incompatible change: *STANDARD-INPUT* is now only an
3147 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3149 * minor incompatible change: Y-OR-N-P is now character-oriented,
3150 not line oriented. Also, YES-OR-NO-P now works without errors.
3151 (thanks to Antonio Martinez)
3152 * sb-aclrepl module improvements: an integrated inspector, added
3153 repl features, and a bug fix to :trace command.
3154 * Known functions, which cannot be open coded by the backend, are
3155 considered to be able to check types of their arguments. (fixing
3156 a bug report by Nathan J. Froyd)
3157 * fixed a bug in computing method discriminating functions: It is
3158 now possible to define methods specialized on classes which have
3159 forward-referenced superclasses. (thanks to Gerd Moellmann)
3160 * fixed evaluation order in optional entries (reported by Gilbert
3162 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3163 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3164 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3165 required-by-AMOP NAME argument, as well as CLASS and
3166 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3167 * fixed bug 20: DEFMETHOD can define methods using names that are
3168 not the proper names of classes to designate class specializers.
3169 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3170 Unix file descriptors, instead of blowing up. (thanks to Antonio
3172 * Garbage collection refactoring: user-visible change is that a
3173 call to the GC function during WITHOUT-GCING will not do garbage
3174 collection until the end of the WITHOUT-GCING. If you were doing
3175 this you were probably losing anyway.
3176 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3177 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3178 TYPEP the latter but not the former.
3179 * The compiler issues a full WARNING for calls to undefined functions
3180 with names from the CL package.
3181 * MAP-INTO for a vector destination is open coded. (reported by
3182 Brian Downing on c.l.l)
3183 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3184 documentation string.
3185 * fixed some bugs revealed by Paul Dietz' test suite:
3186 ** COPY-ALIST now signals an error if its argument is a dotted
3188 ** Condition slots are now accessed more correctly in the presence
3189 of multiple initargs for a given slot.
3190 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3191 exclude restarts of the same name associated with a different
3193 ** DEFCLASS of forward-referenced classes with another
3194 forward-referenced class in the superclasses list no longer
3196 ** Condition slots are now initialized once each, not multiple
3197 times. (thanks to Gerd Moellmann)
3198 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3200 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3201 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3202 arguments contain duplicated elements.
3203 ** RESTART-CASE understands local macros.
3204 ** RESTART-CASE associates exactly its own restarts with a condition.
3205 ** ENDP in safe mode checks its argument to be of type LIST.
3206 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3207 in question is unbound.
3208 ** Optimized MAKE-INSTANCE functions no longer cause internal
3209 assertion failures in the presence of duplicate initargs.
3210 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3211 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3212 propagated even in the ctor optimized implementation of
3214 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3216 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3217 * a better implementation of SXHASH on (simple) bit vectors,
3218 measured both in execution speed and in distribution of results
3219 over the positive fixnums, has been installed. Likewise, a better
3220 implementation of EQUAL for simple bit vectors is now available.
3221 * fixed CEILING optimization for a divisor of form 2^k.
3222 * fixed bug 240 (emitting extra style warnings "using the lexical
3223 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3224 by Antonio Martinez)
3225 * fixed SXHASH, giving different results for NIL depending on type
3226 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3227 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3228 variable. (found by Rolf Wester)
3229 * SBCL does not ignore type declarations for special
3230 variables. (reported by rif on c.l.l 2003-03-05)
3231 * some bug fixes in contrib/sb-aclrepl/
3232 * fixed some bugs revealed by Paul Dietz' test suite:
3233 ** a bug in the CONS type specifier, whereby the CAR and CDR
3234 types got intertwined, has been fixed;
3235 ** the type system is now able to reason about the interaction
3236 between INTEGER and RATIO types more completely;
3237 ** APPEND, [N]REVERSE and NRECONC check that those their
3238 arguments, which must be proper lists, are really so;
3239 ** An array specialized to be unable to hold elements has been
3240 implemented, as required -- yes, really -- by ANSI;
3241 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3242 on malformed property lists;
3244 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3245 * incompatible packaging change: in line with Unix convention,
3246 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3247 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3248 * REQUIRE and PROVIDE are now optionally capable of doing something
3249 useful. See the documentation string for REQUIRE.
3250 * infrastructure for a managed SBCL contrib system: contributed
3251 modules in this release include:
3252 ** the ASDF system definition facility;
3253 ** an interface to the BSD Sockets API;
3254 ** an ACL-like convenience interface to the repl;
3255 (thanks to Kevin Rosenberg)
3256 ** an implementation of ROTATE-BYTE, with efficient implementation
3258 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3259 gives a closer approximation to the right answer than previously.
3260 (thanks to Raymond Toy)
3261 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3262 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3263 optional environment arguments, as required by ANSI.
3264 * fixed bugs in other functions taking environment objects, allowing
3265 calls with an explicit NIL environment argument to be compiled
3267 * fixed bug 228: primary return values from
3268 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3269 COMPILE or FUNCTION.
3270 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3271 :TYPE will no longer signal errors on innocuous objects.
3272 * fixed bug 231b: SETQ is better at respecting type declarations in
3273 the lexical environment.
3274 * fixed a bug in DEFCLASS: classes named by symbols with no or
3275 unprintable packages can now be defined.
3276 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3277 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3278 * fixed compiler failure related to checking types of functions.
3279 (reported by Robert E. Brown)
3280 * the compiler is now much more consistent in its error-checking
3281 treatment of bounding index arguments to sequence functions: in
3282 (SAFETY 3) code, errors will be signalled in almost all cases if
3283 invalid sequence bounding indices are passed to functions defined
3284 by ANSI to operate on sequences.
3285 * fixed a bug in the build procedure: documentation of SBCL-specific
3286 packages is now preserved and available in the final Lisp image.
3287 * lifted FDEFINITION lookup out of loops in the implementation of
3288 many list operations. (thanks to Robert E. Brown)
3289 * fixed a bug in the reader: the #n# reader macro now works for
3290 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3291 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3292 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3293 E. Brown and Tony Martinez respectively)
3294 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3295 SLOT-BOUNDP: in particular, it is now able to optimize them much
3296 better, and is now not vulnerable to having packages renamed.
3297 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3298 always called when a slot is not present in an instance. (thanks
3300 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3301 longer signal an error on unreading a character following EOF on
3302 the previous constituent stream. (thanks to Tony Martinez)
3303 * fixed some bugs revealed by Paul Dietz' test suite:
3304 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3305 not just nonnegative fixnums;
3306 ** the logical bit-array operators such as BIT-AND now accept an
3307 explicit NIL for their "opt-arg" argument (to indicate a
3308 freshly-consed result bit-array);
3309 ** ELT now signals an error on an invalid sequence index in safe
3311 ** the type system is now cleverer about negations of numeric
3312 types, and consequently understands the BIGNUM and RATIO types
3314 ** the type system is now cleverer about the interaction between
3315 INTEGER and RATIO types: while bugs still remain, many more
3316 cases are accurately computed;
3317 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3318 if it is in the last clause;
3319 ** CONSTANTLY now correctly returns a side-effect-free function in
3321 ** DECLARE is no longer treated as a special-operator; in
3322 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3323 * incremented fasl file version number due to the change in the
3324 DEFSTRUCT-SLOT-DESCRIPTION structure.
3326 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3327 * minor incompatible change: code processed by the "interpreter" or
3328 EVAL now has a compilation optimization policy of (DEBUG 2)
3329 (changed from (DEBUG 1)) to improve debuggability of interactive
3330 development, and to allow the use of the debug RETURN command in
3332 * an experimental implementation of the RETURN command for the
3333 debugger has been included. (thanks to Frederik Kuivinen)
3334 * fixed bug 62: constraints were not propagated into a loop.
3335 * fixed bug in embedded calls of SORT (reported and investigated by
3337 * fixed some bugs revealed by Paul Dietz' test suite:
3338 ** printing and reading of arrays with some dimensions having
3339 length 0 (thanks to Gerd Moellmann);
3340 ** BOA constructor with &AUX argument without a default value does
3341 not cause a type error;
3342 ** CONSTANTP now returns true for all self-evaluating objects.
3344 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3345 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3346 accessors that are related by inheritance, as specified in the
3347 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3348 Valtteri Vuorikoski)
3349 * The compiler is now able to inline functions that were defined in
3350 a complex lexical environment (e.g. inside a MACROLET).
3351 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3353 * Support for the upcoming FreeBSD-5.0 release has been included.
3354 (thanks to Dag-Erling Smorgrav)
3355 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3356 effect when it is not in a toplevel context.
3357 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3358 stand a better chance of being correct. (thanks to Gerd
3360 * fixed bug in COERCE, which now signals an error on coercing a
3361 rational to a bounded real type which excludes the expected
3363 * The compiler is now able to derive types more accurately from the
3364 COERCE and COMPILE functions.
3365 * fixed bug 223: functional binding is considered to be constant
3366 only for symbols in the CL package.
3367 * fixed bug 231: SETQ did not check the type of a variable being set
3368 (reported by Robert E. Brown)
3369 * A new optimization for MAKE-INSTANCE has been included, fixing
3370 various bugs (including relating to :ALLOCATION :CLASS slots and
3371 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3373 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3374 ** As required by ANSI, LOOP now disallows anonymous collection
3375 clauses such as COLLECT I in conjunction with aggregate boolean
3376 clauses such as THEREIS (= I 1);
3377 ** LOOP now signals an error when any variable is reused in the
3378 same loop (including the potentially useful construct analogous
3379 to WITH A = 1 WITH A = (1+ A);
3380 ** IT is only a special loop symbol within the first clause of a
3381 conditional loop clause;
3382 ** LOOP with a typed iteration variable over a hashtable now
3383 signals a type error iff it should.
3384 * fixed some other bugs revealed by Paul Dietz' test suite:
3385 ** FILE-STREAM now names the class previously known as FD-STREAM;
3386 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3387 argument) no longer signals an error;
3388 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3389 of the slot symbol, rather than using the current package
3390 ((:CONC-NAME "") continues to intern the slot's name in the
3392 * incremented fasl file version number, because of the incompatible
3393 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3394 of the new implementation of DEFINE-COMPILER-MACRO.
3396 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3397 * Support for building SBCL for MIPS platforms running in
3398 little-endian mode has now been checked in, and basic
3399 functionality on said platforms verified.
3400 * minor incompatible change: PCL now records the pathname of a file
3401 in which methods and the like are defined, rather than its
3403 * minor incompatible change: TRUENAME now considers the truename of
3404 a file naming a directory to be the pathname with :DIRECTORY
3405 component indicating that directory.
3406 * minor incompatible change: a NAMED clause in the extended form of
3407 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3408 reason for the previous behaviour is unclear.
3409 * more systematization and improvement of CLOS and MOP conformance
3410 in PCL (thanks to Gerd Moellman and Pierre Mai):
3411 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3413 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3414 primary methods with no specializers;
3415 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3417 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3418 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3419 FORWARD-REFERENCED-CLASSes; error reporting on
3420 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3422 ** SXHASH on CLOS instances now uses a slot internal to the
3423 instance to return different numbers on distinct instances,
3424 while preserving the same return value through invocations of
3426 ** DEFMETHOD signals errors when methods with longer incongruent
3427 lambda lists are added to generic functions;
3428 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3429 CLASS, as specified in AMOP;
3430 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3431 on the order returned by the primary method for classes of
3432 class STANDARD-CLASS;
3433 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3434 * fixed some bugs shown by Paul Dietz' test suite:
3435 ** DOLIST puts its body in TAGBODY;
3436 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3438 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3439 value producing form;
3440 ** if more variables are given to PROGV than values, extra
3441 variables are bound and made to have no value;
3442 ** NSUBSTITUTE on list arguments gets the right answer with
3444 ** ELT signals an error of type TYPE-ERROR when the index argument
3445 is not a valid sequence index;
3446 ** LOOP signals (at macroexpansion time) an error of type
3447 PROGRAM-ERROR when duplicate variable names are found;
3448 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3449 ** FUNCALL of special-operators now cause an error of type
3451 ** PSETQ now works as required in the presence of side-effecting
3452 symbol-macro places;
3453 ** NCONC accepts any object as its last argument;
3454 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3456 ** loop-for-as-package does not require a package to be explicitely
3458 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3460 * fixed bug 166: compiler preserves "there is a way to go"
3461 invariant when deleting code.
3462 * fixed bug 172: macro lambda lists with required arguments after
3463 &REST arguments now cause an error to be signalled. (thanks to
3465 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3466 bug. (thanks to Matthew Danish)
3467 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3469 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3470 arguments are lexically rebound. (thanks to Gerd Moellmann and
3472 * fixed bug 194: error messages are now more informative when there
3473 is no primary method applicable in a call to a generic
3474 function. (thanks to Gerd Moellmann)
3475 * fixed bug in command line argument checking (thanks to Julian
3477 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3478 (thanks to Matthew Danish)
3479 * incremented fasl file version number, because of the
3480 SXHASH-related changes in the layout of CLOS data structures
3482 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3483 * minor incompatible change: The runtime (the Unix executable named
3484 "sbcl") is now much pickier about the .core files it will load.
3485 Essentially it now requires .core files to descend from the same
3486 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3487 as the runtime does. (The intent is to prevent the crashes which
3488 can occur, and which can even be reported as mysterious failures,
3489 when people patch the sources or change the build parameters
3490 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3491 sbcl and .core files.)
3492 * fixed bug: VALUES-LIST is no longer optimized away.
3493 * fixed bug 142: The FFI conversion of C string values to Lisp
3494 string values no longer conses excessively. (thanks to Nathan
3495 Froyd porting Raymond Toy's fix to CMU CL)
3496 * began to systematize and improve MOP conformance in PCL (thanks to
3497 Nathan Froyd, Gerd Moellman and Pierre Mai):
3498 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3500 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3501 ** FINALIZE-INHERITANCE is now called on class finalization;
3502 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3503 argument precedence order.
3504 * fixed bug 202: The compiler no longer fails on functions whose
3505 derived types contradict their declared type.
3506 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3507 so it can be non-toplevel.
3508 * The fasl file version number has changed (because of the new
3509 implementation of DEFMACRO).
3510 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3511 safe code, that any length requirement by their type-specifier
3512 argument is valid. The exceptions to this are described in bug
3514 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3515 functions now check the types of their inputs as required by ANSI.
3516 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3517 introduced symbol is DECLAREd to be SPECIAL.
3518 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3519 * fixed bug: :COUNT argument to sequence functions may be negative
3520 * fixed bug: body of DO-SYMBOLS may contain declarations
3521 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3522 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3523 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3524 symbol macro only once
3525 * fixed printing of call frame when argument list is unavailable
3526 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3527 * compiler no longer signals WARNING on unknown keyword
3530 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3531 * A beta-quality port to the mips architecture running Linux,
3532 based on the old CMUCL backend, has been made. It has been tested
3533 on a big-endian kernel, and works sufficiently well to be able to
3534 rebuild itself; it has not been tested in little-endian mode.
3535 * fixed an inconsistency between gencgc.c and purify.c which made
3536 dumping/loading .core files unreliable
3537 * fixed bug 120a: The compiler now deals correctly with IFs where
3538 the consequent is the same as the alternative, instead of
3539 misderiving the return type. (thanks to Alexey Dejneka)
3540 * fixed bug 113: Logical pathnames are now dumpable (the logical
3541 host is resolved at load-time, throwing an error if it is not
3543 * fixed bug 174: FORMAT's error message is slightly clearer when a
3544 non-printing character is used in a format directive.
3545 * fixed several bugs in compiler checking of type declarations, i.e.
3546 violations of the Python "declarations are assertions" principle
3547 (thanks to Alexey Dejneka)
3548 * fixed several bugs in PCL's error checking (thanks to Gerd
3550 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3552 * fixed bug in compilation of functions as first class values
3553 (thanks to Antonio Martinez-Shotton)
3554 * The compiler's handling TYPE-ERRORs which it can prove will
3555 inevitably happen at runtime has been cleaned up and corrected
3556 in several ways. (thanks to Alexey Dejneka)
3557 * improved argument type checking for various basic arithmetic
3558 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3559 lavished on them in the past that they can be compiled in many
3560 ways in different special cases
3561 * fixed bug 181: compiler checks validity of user supplied type
3563 * cleaned up code flushing in optimization: Function calls which
3564 should signal errors for safety purposes (e.g. which ANSI says
3565 should signal errors when their arguments are of incorrect type)
3566 are no longer optimized away.
3567 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3568 * incremented fasl file version number, because changes in the
3569 implementation of sequence functions like COERCE caused
3570 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3571 in old inline expansions) to become undefined. (Actually these
3572 changes were later undone, so we might very well be binary
3573 compatible with 0.7.7 after all, but leaving the version number
3574 incremented seemed like the simplest and most conservative
3577 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3578 * An alpha-quality port to the parisc architecture running Linux,
3579 based on the old CMUCL backend, has been made. This, even more so
3580 than the other backends, should be considered still a work in
3581 progress; known problems include that the Linux kernel in 64-bit
3582 mode does not propagate the correct sigcontext structure to
3583 userspace, and consequently SBCL on a parisc64 kernel will not
3585 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3586 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3587 suppressed.) Also now that the compiler is looking at declarations
3588 in the environment, it checks optimization declarations as well,
3589 and suppresses inlining when (> DEBUG SPEED).
3590 * More fixes have been made to treatment of floating point exception
3591 treatment and other Unix signals. In particular, floating point
3592 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3593 * The detection and handling of control stack exhaustion (infinite
3594 or very deeply nested recursion) has changed. Stack exhaustion
3595 detection is now done by write-protecting pages at the OS level
3596 and applies at all optimization settings; when found, a
3597 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3598 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3599 be caught using IGNORE-ERRORS.
3600 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3601 correctly when transforming recursive LABELS functions to LETs.
3602 (thanks to Alexey Dejneka)
3603 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3604 that are names of constants or global variables.
3605 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3606 alien routines with docstrings.
3607 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3608 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3610 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3611 (thanks to Eric Marsden)
3612 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3613 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3614 * Bug fix: The SPARC backend can now compile functions involving
3615 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3616 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3617 to a functions expecting a PATHNAME-DESIGNATOR.
3618 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3619 lambda lists. (thanks to Alexey Dejneka)
3620 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3621 functions so that EXPORTing the name of the class doesn't cause
3622 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3623 to Antonio Martinez for reporting this)
3624 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3625 constant-folded without failing an assertion. (thanks to Einar
3626 Floystad Dorum for reporting this)
3627 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3628 types are now dealt with more correctly by the compiler.
3629 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3630 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3631 behaviour analogous to RENAME-FILE. This puts its behaviour more
3632 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3634 * The fasl file version number has changed again. (because of the
3635 bug fix involving the names of PCL MAKE-INSTANCE functions)
3637 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3638 * bug fix: Floating point exceptions are treated much more
3639 consistently on the x86/Linux and PPC/Linux platforms.
3640 * Array initialization with :INITIAL-ELEMENT is now much faster for
3641 cases when the compiler cannot open code the array creation, but
3642 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3643 array accesses have also seen a speed increase.
3644 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3645 specified. (This was at the root of some bad interactions between
3646 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3648 * bug fix: Internal error arguments for undefined functions are now
3649 computed correctly on the PPC/Linux platform.
3650 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3651 Raymond Toy's patch for CMU CL)
3652 * Support for the Solaris 9 operating environment has been included
3653 (thanks to Daniel Merritt)
3654 * A very ugly but hopefully complete draft of the missing FFI chapter
3655 of the manual has been created by reformatting the corresponding
3656 CMU CL manual chapter into (currently very ugly and incoherent)
3657 DocBook and bringing it up to date for SBCL behavior. Thus, the
3658 manual is now essentially complete, at least by my extreme
3659 once-and-only-once standards, whereby it's acceptable to refer to
3660 the doc strings of SB-EXT functions as the primary documentation.
3661 * The fasl file version number has changed again, due to cleanup of
3662 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3663 is no longer a static symbol.)
3665 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3666 * SBCL now builds with OpenMCL (version 0.12) as the
3667 cross-compilation host; also, more progress has been made toward
3668 bootstrapping under CLISP.
3669 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3671 * bug 158 fixed: The compiler can now deal with integer loop
3672 increments different from 1; fixing this turned out also to fix
3674 * bug 169 fixed: no more bogus warnings about using lexical bindings
3675 despite the presence of perfectly good SPECIAL declarations (thanks
3676 to David Lichteblau)
3677 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3678 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3679 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3681 * bug 180 fixed: Method combination specifications no longer ignore
3682 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3683 * bug fix: Structure type predicate functions now check their argument
3684 count as they should.
3685 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3686 correctly. (thanks to Pierre Mai)
3687 * minor incompatible change: The --noprogrammer option is deprecated
3688 in favor of the new --disable-debugger option, which is very similar.
3689 (The major difference is that it takes effect at a slightly different
3690 time at startup, causing handling of errors in --sysinit and
3691 --userinit files will be affected differently.) The
3692 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3693 been added to allow this functionality to be controlled from ordinary
3694 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3695 maintainers, who might want to run non-interactive scripts to
3696 build SBCL cores which will later be used interactively.)
3697 * minor incompatible change: The LOAD function no longer, when given
3698 a wild pathname to load, loads all files matching that pathname.
3699 Instead, an error of type FILE-ERROR is signalled.
3701 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3702 * bug 147 fixed: The compiler preserves its block link/count
3703 invariants more correctly now so that it doesn't crash. (thanks
3705 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3707 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3708 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3709 no-value-for-this-character as NIL (as specified by ANSI).
3710 * HOST-NAMESTRING on physical pathnames now returns a string that is
3711 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3712 (thanks to Christophe Rhodes)
3713 * The Alpha port handles icache flushing more correctly. (thanks to
3715 * More progress has been made toward bootstrapping under CLISP. (thanks
3716 to Christophe Rhodes)
3717 * The fasl file format has changed again, because dynamic loading
3718 on OpenBSD (which has non-ELF object files) motivated some cleanups
3719 in the way that foreign symbols are transformed and passed around.
3720 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3721 is no longer treated as whitespace by the reader, but instead as
3722 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3723 |A
\x7fB|, instead of A as it used to.
3725 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3726 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3727 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3728 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3729 this as of 0.7.1.45, but was left out of the previous news section
3730 (thanks to Dan Barlow)
3731 * SBCL now runs on the Solaris operating system on SPARC architectures
3732 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3733 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3734 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3735 * SPARC backend cleanups, allowing builds of cores optimized for V8
3736 and V9 SPARCS, and also emission of code targeted to a particular
3737 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3739 * SBCL is closer to bootstrapping under CLISP, thanks to various
3740 fixes by Christophe Rhodes.
3741 * The fasl file format has changed again, to allow the compiler's
3742 INFO database to support symbol macros.
3743 * The user manual (in doc/) is formatted into HTML more nicely.
3744 (thanks to coreythomas)
3745 * The system is smarter about SUBTYPEP relationships, especially
3746 those involving NOT types (including types such as ATOM which are
3747 represented internally using NOT types). Thus SUBTYPEP is less
3748 likely to return (VALUES NIL NIL) in general, and in particular
3749 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3751 * The fasl file format has changed again, because the internal
3752 representation of types now includes a new slot to support the new
3753 SUBTYPEP-of-NOT-types logic.
3754 * (not a change in the main branch of SBCL, but a related prototype
3755 which can hopefully be merged into the main branch of SBCL in the
3756 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3757 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3758 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3759 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3760 dispatch macro character combinations. (thanks to Alexey Dejneka)
3761 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3762 that superficially look like logical namestrings correctly.
3763 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3765 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3766 default supplied-p) for &optional and &key arguments. (thanks to
3769 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3770 * incompatible change: The compiler is now less aggressive about
3771 tail call optimization, doing it only when (> SPACE DEBUG) or
3772 (> SPEED DEBUG). (This is an incompatible change because there are
3773 programs which relied on the old CMU-CL-style behavior to optimize
3774 away their unbounded recursion which will now die of stack overflow.)
3775 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3776 for non-GENCGC systems has been increased to 20M (since that
3777 seems much closer to the likely performance optimum for modern
3778 systems than the old 4M value was)
3779 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3780 and *DEBUG-PRINT-LEVEL*
3781 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3782 of CMU CL's support for SPARC, and various endianness and other
3783 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3784 * new syntactic sugar for the Unix command line: --load foo.bar is now
3785 an alternate notation for --eval '(load "foo.bar")'.
3787 ** The system now detects stack overflow and handles it gracefully,
3788 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3789 optimization settings. (This is a good thing in general, and
3790 its introduction in this version should be particularly timely
3791 for anyone whose code fails because of suppression of tail
3793 ** The system now hunts for the C variable "environ" in a more
3794 devious way, to avoid segfaults when the C library version differs
3795 between compile time and run time. (thanks to Christophe Rhodes)
3796 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3797 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3798 ** The compiler no longer issues bogus style warnings for undefined
3799 classes in the same source file as the DEFCLASSes which defined
3800 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3802 ** fixes in CONDITION class precedence list for undefined function
3803 errors (thanks to Alexei Dejneka)
3804 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3805 correctly. (thanks to Dan Barlow)
3806 ** portability fixes aiming at bootstrapping under CLISP (thanks
3807 to Dave McDonald and Christophe Rhodes)
3808 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3809 ** fixes in type translation and and type inference (thanks to
3811 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3812 ** various fixes in the new ports (thanks to Dan Barlow)
3813 * several changes related to debugging:
3814 ** suppression of tail recursion, as noted above
3815 ** stack overflow detection, as noted above
3816 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3817 is now the default. (For some time encapsulation has been more
3818 reliable than the breakpoint-based :ENCAPSULATE NIL
3819 implementation, at least on X86 systems; and I just noticed that
3820 encapsulation also seems closer to the spirit of the ANSI
3823 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3825 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3826 up properly again. (There was a packaging bug in 0.7.0 which
3827 left their definitions in SB-SYS::LOAD-FOREIGN and
3828 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3829 vital for most things which interface to C-level interfaces,
3830 like extensions working with sockets or databases or
3831 Perl-compatible regexes or whatever, and the need to fix
3832 this bug is the main reason that 0.7.1 was released so
3834 ** DEFGENERIC is now choosier about the methods it redefines, so that
3835 reLOADing a previously-LOADed file containing DEFGENERICs does
3836 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3837 cycle now works as it should. (thanks to Alexey Dejneka)
3838 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3839 fixed by Christophe Rhodes. (He actually submitted this patch
3840 months ago, and I delayed until after 0.7.0.)
3841 ** Bug 111 (internal compiler confusion about runtime checks on
3842 FUNCTION types) was fixed by Alexey Dejneka.
3843 * Some internal cleanups (getting rid of variables which aren't
3844 needed now that the byte interpreter is gone) caused the fasl
3845 file format number to change again.
3847 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3848 * major incompatible change: The default fasl file extension, i.e. the
3849 default extension for files produced by COMPILE-FILE, has changed
3850 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3852 ** There are many changes in the implementation of the compiler.
3853 SBCL is now essentially a compiler-only implementation of ANSI
3854 Common Lisp. EVAL still "interprets" a few special cases, but
3855 almost all the interesting cases are handled by creating
3856 a LAMBDA expression, calling COMPILE on it, then calling
3857 FUNCALL on the result.
3858 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3859 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3860 Since the code is newer, there might still be some new bugs
3861 (though not as many as before Martin Atzmueller's fixes:-). But
3862 the new code is substantially simpler and clearer, and hopefully
3863 any remaining bugs will be simpler, less fundamental, and more
3864 fixable then the bugs in the old code.
3865 ** The revised compiler is still a little unsteady on its feet.
3867 *** The debugging information it produces (particularly the names
3868 of FUNCTION objects) is sometimes much less useful than what
3869 the old compiler produced.
3870 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3871 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3872 within a MACROLET) has been temporarily weakened.
3873 ** There are new compiler optimizations for various functions:
3874 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3875 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3876 *** the math functions TRUNCATE, FLOOR, and CEILING
3877 *** the function-of-all-trades COERCE
3878 Mostly these should be transparent, but there's one
3879 potentially-annoying problem (bug 117): when the compiler
3880 inline-expands a function and does type analysis on the result,
3881 it can create control paths which have type mismatches, and
3882 when it can't prove that those control paths aren't taken,
3883 it will issue WARNINGs about the type mismatches. This is
3884 a particular problem in practice for the new sequence functions.
3885 It's not clear how this should be fixed, and for now, a
3886 workaround is given in the entry for 117 in the BUGS file.
3887 ** (Because of the interaction between the two previous items --
3888 occasional inlining problems and new inline expansions -- some
3889 of the new sequence function optimizations won't really kick in
3890 completely until debugging information, and then inlining, are
3891 straightened out in some future version.)
3892 * minor incompatible changes:
3893 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3894 DIRECTORY no longer implicitly promotes NIL slots of its
3895 pathname argument to :WILD. In particular, when you ask for the
3896 contents of a directory (which you used to be able to do without
3897 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3898 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3899 ** changes in behavior that ANSI explicitly defines to be
3900 implementation dependent:
3901 *** The new compiler-only implementation still conforms with ANSI,
3902 but acts a little different than before. Besides the obvious
3903 changes in performance tradeoffs (that the cost per form passed
3904 to EVAL has gone up, and the cost per form executed by EVAL
3905 has gone down), the behavior of the system changes a little
3906 because there are no longer any interpreted function objects.
3907 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3908 e.g. doing COMPILE on the output of interactive DEFUN is
3910 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3912 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3913 has changed from (:CL) to NIL.
3914 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3915 e.g. #\Newline and #\Tab, don't have names specified in the
3916 ANSI Common Lisp standard, is now based on their ASCII symbolic
3917 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3918 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3919 are no longer used for output.
3920 ** changes in internal implementation constants:
3921 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3922 4 million. (If your application spends a lot of time GCing and
3923 you have a lot of RAM, you might want to experiment with
3924 increasing it even more.)
3925 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3926 However, almost all old code should still continue to work without
3927 immediate update, as SB-C-CALL is now a (deprecated) nickname
3929 ** Old operator names in the style DEF-FOO are now deprecated in
3930 favor of new corresponding names DEFINE-FOO, for consistency with
3931 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3932 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3933 internal symbols, but a few supported extensions like
3934 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3935 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3936 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3937 etc. as you get deeper into recursive calls to the debugger
3938 command loop, instead of the old "5]", "5]]", "5]]]"
3939 sequence. (I was motivated to do this when squabbles between
3940 ILISP and SBCL left me very deeply nested in the debugger. In the
3941 short term, this change will probably provoke more ILISP/SBCL
3942 squabbles, but hopefully it will be an improvement in the long run.)
3943 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3944 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3945 function, since declaiming return types involving aliens is
3946 (1) annoyingly messy to do by hand and (2) vital to efficient
3947 compilation of code which calls such functions.
3948 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3949 longer reexported by the SB-EXT package. They're solely useful
3950 for alien code, so it seems more logical that you should get
3951 them from the SB-ALIEN package, not in SB-EXT.
3952 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3953 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3954 features. Instead, the code that they used to control is always
3955 built into the system.
3956 * many other bug fixes
3957 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3958 advantage of the new EVAL-WHEN stuff and to clean them up in
3959 general, and they are now more ANSI-compliant in a number of
3960 ways. Martin Atzmueller is responsible for a lot of this.
3961 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3963 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3964 *** correct ERROR type for various file operations
3965 *** some fixes for Lisp streams
3966 *** DEFMETHOD syntax checking
3967 *** changing old weird representation of debug information as
3968 strings (which, among their other deficiencies, don't transform
3969 correctly when you rename packages, and don't change their
3970 print representation when you change things like *PACKAGE*
3971 and *PRINT-LENGTH*) to symbols and lists of symbols
3972 He also made several improvements and fixed several bugs in DESCRIBE.
3973 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3975 *** misbehavior of WRITE-STRING/WRITE-LINE
3976 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3977 and several other LOOP problems as well
3978 *** DIRECTORY when similar filenames are present
3979 *** DEFGENERIC with :METHOD options
3980 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3981 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3982 *** argument ordering in FIND with :TEST option
3983 *** mishandled package designator argument in APROPOS-LIST
3984 *** various problems in the backquote readmacro
3985 *** a bug in APROPOS
3986 *** probably some others that I'm not describing very well here,
3987 since the CVS log documents them by reference to sbcl-devel
3988 messages, and the SourceForge archives aren't working well.:-(
3989 ** Dan Barlow improved the Alpha port (and is making progress on the
3990 PPC port, for those of you who think different).
3991 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3992 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3993 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3994 continues to work on the SPARC port (for those of you in a position
3995 to look down upon our little PC-compatible boxes from a great height).
3996 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3997 and uses that copy, rather than the current dynamic value, when
3998 it's trying to decide whether to truncate output. Thus e.g.
3999 (let ((*print-lines* 50))
4000 (pprint-logical-block (stream nil)
4002 (let ((*print-lines* 8))
4003 (print (aref possiblybigthings i) stream)))))
4004 should now truncate the logical block only at 50 lines, instead of
4005 often truncating it at 8 lines, as it did before.
4006 * The doc/cmucl/ directory, containing old CMU CL documentation
4007 from the time of the fork, is no longer part of the base system.
4008 SourceForge has shut down its anonymous FTP service, and with it
4009 my original plan for distributing the old CMU CL documentation
4010 there. For now, if you need these files you can download an old
4011 SBCL source release and extract them from it.
4012 * The fasl file version number changed again, for dozens of reasons,
4013 some of which are apparent above.
4015 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4016 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4017 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4018 patch, so that DISASSEMBLE output is much nicer.
4019 * The code in the SB-PROFILE package now seems reasonably stable.
4020 I still haven't decided what the final interface should look like
4021 (I'd like PROFILE to interact cleanly with TRACE, since both
4022 facilities use function encapsulation) but if you have a need
4023 for profiling now, you can probably use it successfully with
4024 the current CMU-CL-style interface.
4025 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4026 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4027 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4028 link now returns the dangling link itself, and for similar
4029 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4030 link itself. (In these cases the old code signalled an error and
4031 looped endlessly, respectively.) Thus, DIRECTORY now works even
4032 in the presence of dangling and cyclic symbolic links.
4033 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4034 is now a supported extension again, since the consensus on
4035 sbcl-devel was that it can be useful for ordinary development
4036 work, not just for debugging SBCL itself.
4037 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4038 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4039 that currently-defined functions might be redefined later with
4040 different return types.
4041 * Hash tables can be printed readably, as inspired by CMU CL code
4042 of Eric Marsden and SBCL code of Martin Atzmueller.
4043 * better error handling in CLOS method combination, thanks to
4044 Martin Atzmueller porting Pierre Mai's CMU CL patches
4045 * more overflow fixes for >16Mbyte I/O buffers
4046 * A bug in READ has been fixed, so that now a single Ctrl-D
4047 character suffices to cause end-of-file on character streams.
4048 In particular, now you only need one Ctrl-D at the command
4049 line (not two) to exit SBCL.
4050 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4052 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4053 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4054 does the right thing.
4055 * The implementation of some type tests, especially for CONDITION
4056 types, is now tidier and maybe faster, due to CMU CL code
4057 originally by Douglas Crosher, ported by Martin Atzmueller.
4058 * Some math functions have been fixed, and there are new
4059 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4060 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4061 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4062 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4063 likely to be built into sbcl-0.7.0.)
4064 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4065 CL, ported to SBCL by Martin Atzmueller)
4066 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4067 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4068 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4069 * There's a new slam.sh hack to shorten the edit/compile/debug
4070 cycle for low-level changes to SBCL itself, and a new
4071 :SB-AFTER-XC-CORE target feature to control the generation of
4072 the after-xc.core file needed by slam.sh.
4073 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4074 COMPILE-FILE is no longer supported, so that now every function
4075 gets an entry point, so that block compilation looks a little
4076 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4077 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4078 returns the number of bytes consed since the system started,
4079 rather than the number consed since the first time the function
4080 was called. (The new definition parallels ANSI functions like
4081 CL:GET-INTERNAL-RUN-TIME.)
4082 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4083 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4084 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4085 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4086 * Fasl file version numbers are now independent of the target CPU,
4087 since historically most system changes which required version
4088 number changes have affected all CPUs equally. Similarly,
4089 the byte fasl file version is now equal to the ordinary
4092 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4093 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4094 is no longer recognized. I apologize for this, because it was
4095 listed in SB-EXT as a supported extension, but I found that
4096 its existing behavior was poorly specified, as well as incorrectly
4097 specified, and it looked like too much of a mess to straighten it
4098 out. I have enough on my hands trying to get ANSI stuff to work..
4099 * many patches ported from CMU CL by Martin Atzmueller, with
4100 half a dozen bug fixes in pretty-printing and the debugger, and
4101 half a dozen others elsewhere
4102 * fixed bug 13: Floating point infinities are now supported again.
4103 They might still be a little bit flaky, but thanks to bug reports
4104 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4105 as flaky as they were.
4106 * The --noprogrammer command line option is now supported. (Its
4107 behavior is slightly different in detail from what the old man
4108 page claimed it would do, but it's still appropriate under the
4109 same circumstances that the man page talks about.)
4110 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4111 are now supported, and enabled by default. Thus, the compiler can
4112 handle many floating point and complex operations much less
4113 inefficiently. (Thus e.g. you can implement a complex FFT
4115 * The compiler now detects type mismatches between DECLAIM FTYPE
4116 and DEFUN better, and implements CHECK-TYPE more correctly, and
4117 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4118 thanks to patches from Martin Atzmueller.
4119 * various fixes to make the cross-compiler more portable to
4120 ANSI-conforming-but-different cross-compilation hosts (notably
4121 Lispworks for Windows, following bug reports from Arthur Lemmens)
4122 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4123 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4124 fixed thanks to Pierre Mai's CMU CL patches.
4125 * fixes to keep the system from overflowing internal counters when
4126 it tries to use i/o buffers larger than 16M bytes
4127 * fixed bug 45a: Various internal functions required to support
4128 complex special functions have been merged from CMU CL sources.
4129 (When I was first setting up SBCL, I misunderstood a compile-time
4130 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4131 * improved support for type intersection and union, fixing bug 12
4132 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4133 more obscure bugs as well
4134 * some steps toward byte-compiling non-performance-critical
4135 parts of the system, courtesy of patches from Martin Atzmueller
4136 * Christophe Rhodes has made some debian packages of sbcl at
4137 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4138 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4139 stable, but are nonetheless usable. When he's ready, I'd be happy
4140 to add them to the SourceForge "File Releases" section. (And if
4141 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4142 * new fasl file format version number (because of changes in
4143 internal representation of (OR ..) types to accommodate the new
4144 support for (AND ..) types, among other things)
4146 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4147 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4149 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4151 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4152 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4153 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4154 * DESCRIBE now works on structure objects again.
4155 * Most function call argument type mismatches are now handled as
4156 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4157 know whether the function will be redefined before the call is
4158 executed. (The compiler could flag local calls with full WARNINGs,
4159 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4160 it doesn't keep track of enough information to know whether calls
4161 are local in this sense.)
4162 * Compiler output is now more verbose, with messages truncated
4163 later than before. (There should be some supported way for users
4164 to override the default verbosity, but I haven't decided how to
4165 provide it yet, so this behavior is still controlled by the internal
4166 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4167 src/compiler/ir1util.lisp.)
4168 * Fasl file format version numbers have increased again, because
4169 support for the Gray streams extension changes the layout of the
4170 system's STREAM objects.
4171 * The Gray subclassable streams extension now works, thanks to a
4172 patch from Martin Atzmueller.
4173 * The full LOAD-FOREIGN extension (not just the primitive
4174 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4175 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4176 but like most other programs, it defaults to copying the Unix
4177 environment from the original process instead of starting the
4178 new process in an empty environment.
4179 * Extensions which manipulate the Unix environment now support
4180 an :ENVIRONMENT keyword option which doesn't smash case or
4181 do other bad things. The CMU-CL-style :ENV option is retained
4182 for porting convenience.
4183 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4184 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4186 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4188 * A patch from Martin Atzmueller seems to have solved the SIGINT
4189 problem, and as far as we know, signal-handling now works cleanly.
4190 (If you find any new bugs, please report them!)
4191 * The system no longer defaults Lisp source file names to types
4192 ".l", ".cl", or ".lsp", but only to ".lisp".
4193 * The compiler no longer uses special default file extensions for
4194 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4195 seems to expect a single default extension for all compiled code,
4196 and there's no compelling reason to try to stretch the standard
4197 to allow two different extensions.) Instead, byte-compiled files
4198 default to the same extension as native-compiled files.
4199 * Fasl file format version numbers have increased again, because
4200 a rearrangement of internal implementation packages made some
4201 dumped symbols in old fasl files unreadable in new cores.
4202 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4203 many fewer weird special cases.
4204 * Bug #17 (differing COMPILE-FILE behavior between logical and
4205 physical pathnames) has been fixed, and some related misbehavior too,
4206 thanks to a patch from Martin Atzmueller.
4207 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4208 by Tim Moore, ported to SBCL by Martin Atzmueller.
4209 * Martin Atzmueller fixed several filesystem-related problems,
4210 including bug #36, in part by porting CMU CL patches, which were
4211 written in part by Paul Werkowski.
4212 * More compiler warnings in src/runtime/ are gone, thanks to
4213 more patches from Martin Atzmueller.
4214 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4217 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4219 * DESCRIBE now works on CONDITION objects.
4220 * The debugger now handles errors which arise when trying to print
4221 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4223 * The build system now uses an additional file, customize-target-features.lisp,
4224 to allow local modifications to the target *FEATURES* list. (The point of
4225 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4226 debugging features enabled, without having to worry about propagating it
4227 into everyone's system when I do a "cvs update".) When no
4228 customize-target-features.lisp file exists, the target *FEATURES* list
4229 should be constructed the same way as before.
4230 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4231 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4232 warn about behavior which is undefined under the ANSI spec. Note: This
4233 is specified by ANSI, but it's not very popular with programmers.
4234 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4235 macro in the SBCL sources for an example of a workaround which you
4236 might use to make portable ANSI-standard code which does what you want.
4237 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4238 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4239 when it's used as a non-toplevel form. (This is required in order
4240 to implement the DEFCONSTANT-EQX macro.)
4241 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4242 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4243 funny things, due to bugs in the implementation of EVAL-WHEN
4244 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4245 non-toplevel DEFCONSTANTs are uncommon.)
4246 * The core file version number and fasl file version number have been
4247 incremented, because the old noncompliant DEFCONSTANT behavior involved
4248 calling functions which no longer exist, and because I also took the
4249 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4250 * fixed bug 1 (error handling before read-eval-print loop starts), and
4251 redid debugger restarts and related debugger commands somewhat while
4253 ** The QUIT debugger command is gone, since it did something
4254 rather different than the SB-EXT:QUIT command, and since it never
4255 worked properly outside the main toplevel read/eval/print loop.
4256 Invoking the new TOPLEVEL restart provides the same functionality.
4257 ** The GO debugger command is also gone, since you can just invoke
4258 the CONTINUE restart directly instead.
4259 ** The TOP debugger command is also gone, since it's redundant with the
4260 FRAME 0 command, and since it interfered with abbreviations for the
4262 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4264 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4265 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4266 some porting work by Martin Atzmueller)
4267 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4268 with a boatload of patches which he ported to SBCL. Now that those
4270 ** The system tries to make sure that its low-priority messages
4271 are prefixed by semicolons, to help people who like to use
4272 syntax highlighting in their ILISP buffer. (This patch
4273 was originally due to Raymond Toy.)
4274 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4275 patches originally written by Raymond Toy.
4276 ** The compiler understands coercion between single-value and
4277 multiple-VALUES type expressions better, getting rid of some very
4278 weird behavior, thanks to patches originally by Robert MacLachlan
4279 and Douglas Crosher.
4280 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4281 lambda lists, thanks to a patch originally by Pierre Mai.
4282 ** The system no longer bogusly warns about "abbreviated type
4284 ** The compiler gets less confused by inlining and RETURN-FROM,
4285 thanks to some patches originally by Tim Moore.
4286 ** The system no longer hangs when dumping circular lists to fasl
4287 files, thanks to a patch originally from Douglas Crosher.
4288 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4289 undefined function error.
4290 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4291 in structure slot initforms) for now, documented workaround instead:-|
4292 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4293 * fixed bug 5: added stubs for various Gray stream functions called
4294 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4295 installed, at least appropriate type errors are generated
4296 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4297 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4298 consistently in DEFMETHOD forms.
4299 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4300 it doesn't seem to affect SBCL after all
4301 * The C runtime system now builds with better optimization and many
4302 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4304 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4306 * The system is now under CVS at SourceForge (instead of the
4307 CVS repository on my home machine).
4308 * The new signal handling code has been tweaked to treat register
4309 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4310 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4311 downstream, e.g. in debug-int.lisp, has implicit dependencies
4312 on the unsignedness of integer representation of machine words,
4313 and that caused the system to bomb out with infinite regress
4314 when trying to recover from type errors involving signed values,
4315 e.g. (BUTLAST '(1 2 3) -1).)
4316 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4317 (This was one of the bugs Peter Van Eynde reported back in July.)
4318 * The system now uses code inspired by Colin Walters' O(N)
4319 implementation of MAP (from the cmucl-imp@cons.org mailing
4320 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4321 inline the MAP operation, and there is more than one
4322 sequence argument to the MAP call (so that it can't just
4323 do ETYPECASE once and for all based on the type of the
4324 single sequence argument). (The old non-inline implementation
4325 of the general M-argument sequence-of-length-N case required
4326 O(M*N*N) time when any of the sequence arguments were LISTs.)
4327 * The QUIT :UNIX-CODE keyword argument has been renamed to
4328 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4329 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4331 * Raymond Wiker's patches to port dynamic loading from Linux to
4332 FreeBSD have been added.
4333 * The BUGS file is now more nearly up to date, thanks in large part
4334 to Martin Atzmueller's review of it.
4335 * The debugger now flushes standard output streams before it begins
4336 its output ("debugger invoked" and so forth).
4337 * The core version number and fasl file version number have both
4338 been incremented, because of incompatible changes in the layout
4340 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4341 used to not be called for a saved Lisp image.)
4342 * Martin Atzmueller's version of a patch to fix a compiler crash,
4343 as posted on sbcl-devel 13 September 2000, has been installed.
4344 * Instead of installing Martin Atzmueller's patch for the
4345 compiler transform for SUBSEQ, I deleted the compiler transform,
4346 and transforms for some similar consing operations.
4347 * A bug in signal handling which kept TRACE from working on OpenBSD
4349 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4350 away by constant folding
4351 * The system now defines its address space constants in one place
4352 (in the Lisp sources), and propagates them automatically elsewhere
4353 (through GENESIS and the sbcl.h file). Therefore, patching the
4354 address map is less unnecessarily tedious and error-prone. The
4355 Lisp names of address space constants have also been systematized.
4356 * CVS tags like dollar-Header-dollar have been removed from
4357 the sources, because they have never saved me trouble and
4358 they've been source of trouble working with patches and other
4359 diff-related operations.
4360 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4361 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4363 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4365 * The system has been ported to OpenBSD.
4366 * The system now compiles with a simple "sh make.sh" on the systems
4367 that it's supported on. I.e., now you no longer need to tweak
4368 text in the target-features.lisp-expr and symlinks in src/runtime/
4369 by hand, the make.sh takes care of it for you.
4370 * The system is no longer so grossly inefficient when compiling code
4371 involving vectors implemented as general (not simple) vectors (VECTOR T),
4372 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4373 which dares to use the various sequence functions on non-simple
4374 vectors, takes less of a performance hit.
4375 * There is now a primitive type predicate VECTOR-T-P
4376 to test for the (VECTOR T) type, so that e.g.
4377 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4378 can now be compiled with some semblance of efficiency. (The old code
4379 turned the type declaration into a full call to %TYPEP at runtime!)
4380 * AREF on (VECTOR T) is still not fast, since it's still compiled
4381 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4382 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4383 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4384 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4385 without making full calls to SUBTYPEP at runtime.
4386 (Some analogous efficiency issues for non-simple vectors specialized to
4387 element types other than T, or for non-simple multidimensional arrays,
4388 have not been addressed. They could almost certainly be handled the
4389 same way if anyone is motivated to do so.)
4390 * The changes in array handling break binary compatibility, so
4391 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4392 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4393 instead of (VALUES T T).
4394 * By following the instructions that Dan Barlow posted to sbcl-devel
4395 on 2 July 2000, I was able to enable primitive dynamic object
4396 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4397 functionality is not implemented (since it calls ld to resolve
4398 library references automatically, requiring RUN-PROGRAM for its
4399 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4400 try to resolve library references) is now supported.
4401 * The system now flushes the standard output streams when it terminates,
4402 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4403 them at several other probably-convenient times, e.g. in each pass of
4404 the toplevel read-eval-print loop, and after evaluating a form given
4405 as an "--eval" command-line option. (These changes were motivated by a
4406 discussion of stream flushing issues on cmucl-imp in August 2000.)
4407 * The source transform for TYPEP of array types no longer assumes
4408 that an array whose element type is a not-yet-defined type
4409 is implemented as an array of T, but instead punts, so that the
4410 type will be interpreted at runtime.
4411 * There is now some support for cross-compiling in make.sh: each of
4412 the phases of make.sh has its own script. (This should be transparent
4413 to people doing ordinary, non-cross-compile builds.)
4414 * Since my laptop doesn't have hundreds of megabytes of memory like
4415 my desktop machine, I became more motivated to do some items on
4416 my to-do list in order to reduce the size of the system a little:
4417 ** Arrange for various needed-only-at-cold-init things to be
4418 uninterned after cold init. To support this, those things have
4419 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4420 symbols with such names are now uninterned after cold init).
4421 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4422 fasl files for cold load.
4423 ** Remove the old compiler structure pooling code (which used to
4424 be conditional on the target feature :SB-ALLOC) completely.
4425 ** Redo the representation of some data in cold init to be more compact.
4426 (I also looked into supporting byte compiled code at bootstrap time,
4427 which would probably reduce the size of the system a lot, but that
4428 looked too complicated, so I punted for now.)
4429 * The maximum signal nesting depth in the src/runtime/ support code has
4430 been reduced from 4096 to 256. (I don't know any reason for the very
4431 large old value. If the new smaller value turns out to break something,
4432 I'll probably just bump it back up.)
4433 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4435 * Many, many bugs reported by Peter Van Eynde have been added to
4436 the BUGS list; some have even been fixed.
4437 * While enabling dynamic object file loading, I tried to make the
4438 code easier to understand, renaming various functions and variables
4439 with less ambiguous names, and changing some function calling
4440 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4441 * While trying to figure out how to do the OpenBSD port, I tried to
4442 clean up some of the code in src/runtime/. In particular, I dropped
4443 support for non-POSIX signal handling, added various comments,
4444 tweaked the code to reduce the number of compilation warnings, and
4445 renamed some files to increase consistency.
4446 * To support the new automatic configuration functionality in make.sh,
4447 the source file target-features.lisp-expr has been replaced with the
4448 source file base-target-features.lisp-expr and the machine-generated
4449 file local-target-features.lisp-expr.
4450 * fixed a stupid quoting error in make.sh so that using CMU CL
4451 "lisp -batch" as cross-compilation host works again
4453 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4455 * DESCRIBE no longer tries to call itself recursively to describe
4456 bound/fbound values, so that it no longer fails on symbols which are
4457 bound to themselves (like keywords, T, and NIL).
4458 * DESCRIBE now works on generic functions.
4459 * The printer now prints less-screwed-up representations of closures
4460 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4461 * A private symbol is used instead of the :EMPTY keyword previously
4462 used to mark empty slots in hash tables. Thus
4463 (DEFVAR *HT* (MAKE-HASH-TABLE))
4464 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4465 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4466 now does what ANSI says that it should. (You can still get
4467 similar noncompliant behavior if bang on the hash table
4468 implementation with all the symbols you get back from
4469 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4470 This breaks binary compatibility, since tests for equality to
4471 :EMPTY are wired into things like the macroexpansion of
4472 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4474 * There's now a minimal placeholder implementation for CL:STEP,
4475 as required by ANSI.
4476 * An obscure bug in the interaction of the normal compiler, the byte
4477 compiler, inlining, and structure predicates has been patched
4478 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4479 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4481 * Missing ordinary arguments in a macro call are now detected even
4482 when the macro lambda list contains &KEY or &REST.
4483 * The debugger no longer complains about encountering the top of the
4484 stack when you type "FRAME 0" to explicitly instruct it to go to
4485 the top of the stack. And it now prints the frame you request even
4486 if it's the current frame (instead of saying "You are here.").
4487 * As specified by ANSI, the system now always prints keywords
4488 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4490 * The default initial SIZE of HASH-TABLEs is now smaller.
4491 * Type information from CLOS class dispatch is now propagated
4492 into DEFMETHOD bodies, so that e.g.
4493 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4495 is now basically equivalent to
4496 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4497 (DECLARE (TYPE SINGLE-FLOAT X))
4499 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4500 operation, without having to do run-time type dispatch.
4501 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4502 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4503 * Since I don't seem to be making big file reorganizations very often
4504 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4505 is overflowing my ability to conveniently back them up), I've finally
4506 checked the system into CVS. (The CVS repository is on my home system,
4507 not at SourceForge -- putting it on SourceForge might come later.)
4508 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4509 high-level GC-NOTIFY-FOO functions send their output. (There's
4510 still very little control of where low-level verbose GC functions
4511 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4512 less than it used to -- the GC-NOTIFY-FOO functions are now under
4513 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4514 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4515 in only one place in the source code, and propagates it automatically
4516 everywhere that it's needed. Thus e.g. when I bump the version from
4517 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4518 * The C source files now include boilerplate legalese and documentation
4519 at the head of each file (just as the Lisp source files already did).
4520 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4521 to his page will be replaced with a link to his new CLiki service.
4523 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4525 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4526 * The build process now looks for GNU make under the default name "gmake",
4527 instead of "make" as it used to. If GNU make is not available as "gmake"
4528 on your system, you can change this default behavior by setting the
4529 GNUMAKE environment variable.
4530 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4531 can build without error under CMU CL.
4533 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4535 * There is now a partial SBCL user manual (with some new text and some
4536 text cribbed from the CMU CL manual).
4537 * The beginnings of a profiler have been added (starting with the
4538 CMU CL profiler and simplifying and cleaning up). Eventually the
4539 main interface should be through the TRACE macro, but for now,
4540 it's still accessed through vaguely CMU-CL-style functions and macros
4541 exported from the package SB-PROFILE.
4542 * Some problems left over from porting CMU CL to the new
4543 cross-compilation bootstrap process have been cleaned up:
4544 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4545 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4546 part when getting rid of the compiler *BACKEND* stuff.)
4547 ** The value of *NULL-TYPE* was screwed up, because it was
4548 being initialized before the type system knew the final
4549 definition of the 'NULL type. This screwed up several key
4550 optimizations in the compiler, causing inefficiency in all sorts
4551 of places. (I found it because I wanted to understand why
4552 GET-INTERNAL-RUN-TIME was consing.)
4553 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4554 PROCLAIM FTYPE information. Unfortunately this broke binary
4555 compatibility again, since now the forms output by DEFGENERIC
4556 to refer to functions which didn't exist in 0.6.3.
4557 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4558 can use the new (as of 0.6.3) transform for SOME into MAP into
4560 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4561 values to (INTEGER 0 1000000), so that the time code will no longer
4562 occasionally get blown up by Linux returning 1000000 microseconds
4563 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4564 its output conform to the ANSI spec. (Alas, this makes its output
4565 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4566 * A full call to MAP NIL with a single sequence argument no longer conses.
4567 * fixes to problems pointed out by Martin Atzmueller:
4568 * The manual page no longer talks about multiprocessing as though
4569 it were currently supported.
4570 * The ILISP support patches have been removed from the distribution,
4571 because as of version 5.10.1, ILISP now supports SBCL without us
4572 having to maintain patches.
4573 * added a modified version of Raymond Toy's recent CMU CL patch for
4574 EQUALP comparison of HASH-TABLE
4576 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4578 * The system still can't cross-compile itself with
4579 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4580 entails), but at least it can compile more of itself that way
4581 than it used to be able to, and various buglets which were uncovered
4582 by trying to cross-compile itself that way have now been fixed.
4583 * This release breaks binary compatibility again. This time
4584 at least I've incremented the FASL file format version to 2, so that the
4585 problem can be detected reliably instead of just causing weird errors.
4586 * various new style warnings:
4587 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4588 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4589 ** using the lexical binding of a variable named in the *FOO* style
4590 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4591 as specified by ANSI.
4592 * *RANDOM-STATE* is no longer automatically initialized from
4593 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4594 default behavior of the system is to repeat its behavior every time
4595 it's run. If you'd like to change this behavior, you can always
4596 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4597 old convention there was no comparably easy way to get the system to
4598 repeat its behavior every time it was run.
4599 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4600 FTYPE declarations has been removed, in favor of their ANSI
4601 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4602 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4603 the types of their sequence arguments can be determined at compile time.
4604 This is done through a new open code expansion for MAP which eliminates
4605 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4606 argument types can be determined at compile time.
4607 * The optimizer now transforms COERCE into an identity operation when it
4608 can prove that the coerced object is already of the correct type. (This
4609 can be a win for machine generated code, including the output of other
4610 optimization transforms, such as the MAP transform above.)
4611 * Credit information has been moved from source file headers into CREDITS.
4612 * Source file headers have been made more standard.
4613 * The CASE macro now compiles without complaining even when it has
4616 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4618 * (Note that the way that the PCL macroexpansions were rewritten
4619 to accommodate the change in DEFGENERIC below breaks binary
4620 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4621 not run under sbcl-0.6.2. Once we get out of alpha releases,
4622 i.e. hit release 1.0.0, we'll probably try to maintain binary
4623 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4624 and sbcl-1.4.4. Until then, however, it might be fairly common
4625 for maintenance releases to break binary compatibility.)
4626 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4628 * The reporting of unhandled signals has been changed to print some
4629 explanatory text as well as the report form. (Previously only
4630 the report form was printed.)
4631 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4632 it defines, so that the compiler no longer issues undefined function
4633 warnings for compiled-but-not-yet-loaded generic functions.
4634 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4635 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4636 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4637 section of the ANSI spec are supported.
4638 * The "" nickname for the "KEYWORD" package has been removed.
4639 The reader still handles symbol tokens which begin with a package marker
4640 as keywords, but it doesn't expose its mechanism for doing so in the
4641 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4642 * The system now issues STYLE-WARNINGs for contradictory TYPE
4643 proclamations. (Warnings for contradictory FTYPE proclamations would
4644 be nice too, but those can't be done usefully unless the type system
4645 is made smarter about FUNCTION types.)
4646 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4647 systematized, so that "*target-*.lisp is supposed to exist only on the
4648 target and imply that there's a related file which exists on the
4649 host, and *host-*.lisp is supposed to exist only on the host and imply
4650 that there's a related file which exists on the target. This involves a
4651 lot of renaming. Hopefully the acute confusion caused by the renaming
4652 will be justified by the reduction in chronic confusion..
4653 ** runtime-type.lisp -> early-target-type.lisp
4654 ** target-type.lisp -> late-target-type.lisp
4655 ** early-host-format.lisp -> early-format.lisp
4656 ** late-host-format.lisp -> late-format.lisp
4657 ** host-error.lisp -> misc-error.lisp
4658 ** early-error.lisp -> early-target-error.lisp
4659 ** late-error.lisp -> late-target-error.lisp
4660 ** host-defboot.lisp -> early-defboot.lisp
4661 ** code/misc.lisp -> code/target-misc.lisp
4662 ** code/host-misc.lisp -> code/misc.lisp
4663 ** code/numbers.lisp -> code/target-numbers.lisp
4664 ** code/early-numbers.lisp -> numbers.lisp
4665 ** early-host-type.lisp -> early-type.lisp
4666 ** late-host-type.lisp -> late-type.lisp
4667 ** host-typep.lisp -> typep.lisp
4668 ** load.lisp -> target-load.lisp
4669 ** host-load.lisp -> load.lisp
4670 ** host-disassem.lisp -> disassem.lisp
4671 ** host-insts.lisp -> insts.lisp
4672 ** byte-comp.lisp -> target-byte-comp.lisp
4673 ** host-byte-comp.lisp -> byte-comp.lisp
4674 ** host-signal.lisp -> signal.lisp
4675 ** host-defstruct.lisp -> defstruct.lisp
4676 ** late-target-type.lisp -> deftypes-for-target.lisp
4677 Furthermore, several other previously target-only files foo.lisp (e.g.
4678 hash-table.lisp and random.lisp) have been split into a target-and-host
4679 foo.lisp file and a target-only target-foo.lisp file, with their key type
4680 definitions in the target-and-host part, so that the cross-compiler will
4681 know more about target types.
4682 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4683 gone away. In their place are various *BACKEND-FOO* variables
4684 corresponding to the slots of the old structure.
4685 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4686 into the target SBCL has been fixed.
4687 * The chill.lisp system for loading cold code into a running SBCL
4689 * Support for the CMU CL "scavenger hook" extension has been removed.
4690 (It was undocumented and unused in the CMU CL sources that SBCL was
4691 derived from, and stale in sbcl-0.6.1.)
4692 * Various errors in the cross-compiler type system were detected
4693 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4694 (enabling various consistency checks). Many of them were fixed,
4695 but some hard problems remain, so the compiler is back to
4696 running without *TYPE-SYSTEM-INITIALIZED* for now.
4697 * As part of the cross-compiler type system cleanup, I implemented
4698 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4699 * I have started adding UNCROSS calls throughout the type system
4700 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4701 on cross-compiler input files will be able to go away, and various
4703 * CONSTANTP now returns true for quoted forms (as explicitly required
4706 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4708 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4709 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4710 (MAKE-LIST -1). In the longer term, it ought to become true
4711 that declarations are assertions even at SAFETY 1. For now, it's not
4712 quite true even at SAFETY 3, but it's at least more nearly true..
4713 (Note that this change seems to increases the size of the system by
4714 O(5%) and to decrease the speed of the compiler by 20% or more.)
4715 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4716 for the problem of printing dozens of lines of distracting information
4717 about low-level system machinery as part of the top stack frame
4718 on entry to the debugger when an undefined function was called.
4719 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4720 is not reset to COMMON-LISP-USER.
4721 * Compilation of stuff related to dyncount.lisp has been made conditional
4722 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4723 smaller. The various dyncount-related symbols have been moved into
4724 a new "SB-DYNCOUNT" package.
4725 * tty-inspect.lisp has been renamed to inspect.lisp.
4726 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4727 feature has gone away. (When we eventually port to other flavors of
4728 libc and/or Unix, we'll try to make the differences between flavors
4729 invisible at the user level.)
4730 * Various other *FEATURES* tags, and/or their associated conditionals,
4731 have been removed if obsolescent, or given better documentation, or
4732 sometimes given more-mnemonic names.
4734 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4736 * tidied up "make.sh" script
4737 * tidied up system directory structure
4738 * better "clean.sh" behavior
4739 * added doc/FOR-CMUCL-DEVELOPERS
4740 * many many small tweaks to output format, e.g. removing possibly-confusing
4741 trailing #\. character in DESCRIBE-INSTANCE
4742 * (EQUALP #\A 'A) no longer signals an error.
4743 * new hashing code, including EQUALP hashing
4744 * tidied up Lisp initialization and toplevel
4745 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4746 * command line argument processing
4747 * added POSIX-GETENV function to deal with Unix-ish environment variables
4748 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4749 terminating SBCL on EOF
4750 * non-verbose GC by default
4751 * There is no more "sbcl" shell script; the sbcl file is now the C
4752 runtime executable (just like CMU CL).
4753 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4755 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4756 symbol and package information as Lisp native symbol and package objects
4757 instead of strings naming symbols and strings naming packages. This way,
4758 whenever packages are renamed (as in warm init), debug information is
4759 transformed along with everything else.
4760 * tweaked the optimization policy declarations which control the building
4761 of SBCL itself. Now, among other things, the system no longer saves
4762 source location debugging information. (This helps two problems at once
4763 by reducing SBCL size and by keeping SBCL from trying to look for its
4764 sources -- which may not exist -- when reporting errors.)
4765 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4766 debugging and testing purposes
4767 * cleaned up printing, making the printer call PRINT-OBJECT for
4768 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4769 methods, giving nearly-ANSI behavior
4770 * converted almost all special variables to use *FOO* naming convention
4771 * deleted PARSE-TIME functionality, since it can be done portably
4772 * moved some files out of cold init into warm init
4773 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4775 * regularized formatting of source files
4776 * added an install.sh script
4777 * fixed ridiculous memory usage of cross-compiler by making
4778 compiler/alloc.lisp not try to do pooling unless it can hook
4779 itself into the GC of the cross-compilation host. Now the system
4780 builds nicely on my old laptop.
4781 * added :SB-ALLOC in target-features.lisp-expr
4782 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4783 was not implemented)
4784 * re-did condition handling and note reporting in the compiler. Notes
4785 are no longer handled by signalling conditions. Style warnings
4786 and warnings are handled more correctly and reported in such a way
4787 that it's easy to find one or the other in your output (so that you
4788 can e.g. figure out which of many problems caused COMPILE-FILE to
4790 * changed the severity of several compiler warnings from full WARNING
4791 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4792 compiler note reporting so that it doesn't use the condition system
4793 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4794 COMPILE-FILE command)
4795 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4796 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4797 no longer supported, since their implementation was deeply intertwingled
4798 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4799 * removed bogus "support" for compiler macros named (SETF FOO), and
4800 removed the compiler macro for SETF INFO (but only after making a fool
4801 of myself on the cmucl-imp mailing list by posting a bogus patch for
4802 DEFINE-COMPILER-MACRO..)
4803 * Compiled files containing forms which have side effects on the Lisp
4804 reader (such as DEFPACKAGE forms) are now handled more correctly.
4805 (Compiler queuing of top level lambdas has been suppressed by setting
4806 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4807 * deleted various currently-unused source files, e.g. gengc.lisp. They
4808 may be added back at some point e.g. when porting to other architectures,
4809 but until they are it's distracting to distribute them and to try to
4811 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4812 there were so many of them they're just distractions, and UNCROSS is
4813 known to be able to handle the current sources
4814 * moved PROFILE functionality into TRACE, so that it will be clear
4815 how the wrapping and unwrapping of functions when you profile them
4816 interacts with the wrapping and unwrapping of functions when you
4817 trace them. (Actually, the functionality isn't there yet, but at least
4818 the interface specification is there. Hopefully, the functionality will
4819 arrive with some future maintenance release.)
4820 * removed host-oops.lisp
4821 * changed signature of QUIT function to allow UNIX-CODE argument
4822 * fixed READ-SEQUENCE bug
4823 * tweaked verbose GC output so that it looks more like the progress
4824 output that ANSI specifies for functions like LOAD
4825 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4826 * added <http://sbcl.sourceforge.com> to the banner information printed by