1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.20 relative to 1.0.19:
3 * minor incompatible change: OPTIMIZE qualities
4 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
5 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
6 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
7 * documentation: some slot access efficiency guidelines have been
8 added to the user manual.
9 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
10 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
12 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
13 to more efficient EQ-comparison versions more often.
14 * optimization: enhanced derivation of DOLIST iteration variable type
16 * optimization: constant folding of simple (LIST ...) forms as DOLIST
18 * optimization: runtime lookup of function definitions can be
19 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
20 * optimization: compiler is able to derive the return type of
21 (AREF (THE STRING X) Y) as being CHARACTER.
22 * optimization: CLRHASH on empty hash-tables no longer does pointless
23 work. (thanks to Alec Berryman)
24 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
25 READ-LINE &co can no longer read from them afterwards. (reported
27 * bug fix: fixed #427: unused local aliens no longer cause compiler
28 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
30 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
31 alien-stack leakage. (reported by Andy Hefner)
32 * bug fix: PROGV signals an error when an attempt to violate declared
33 type of a variable or bind a constant is made.
34 * bug fix: SET signals an error when an attempt to violate declared
35 type of a variable is made.
36 * bug fix: restart computation during the execution of a restart
37 test function no longer causes infinite recursion. (reported by
39 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
40 signals an error. (thanks to Pierre Mai)
41 * bug fix: EXPORT left symbol unexported in conflict situations.
42 (thanks to Michael Weber)
43 * bug fix: correctly handle name conflicts involving CL:NIL.
44 (thanks to Michael Weber)
45 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
46 conflicts arising from USEing package with conflicting symbols
47 correctly. (thanks to Michael Weber)
49 changes in sbcl-1.0.19 relative to 1.0.18:
50 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
51 warnings that go otherwise unhandled will be muffled if they are
52 of the type that's the value of this variable.
53 * optimization: stack allocation is slightly more efficient on x86
55 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
56 where a raw slot always is initialized using the initform whose
57 type is not know sufficiently well a compile-time are now compiled
58 correctly. (reported by John Morrison)
59 * bug fix: compiler no longer makes erronous assumptions in the
60 presense of non-foldable SATISFIES types.
61 * bug fix: stack analysis missed cleanups of dynamic-extent
62 arguments in non-let-converted calls to local functions.
63 * improvements to the Windows port:
64 ** adjusted address spaces for building on both Win32 and
65 Win64. (thanks for John Connors)
66 * fixed some bugs revealed by Paul Dietz' test suite:
67 ** interval arithmetic during type derivation used inexact integer
68 to single-float coercions.
69 ** arithmetic operations involving large integers and single
70 floats give the same results in compiled and interpreted code.
71 ** deriving the result type of COERCE no longer signals an error
72 if the derived type of the second argument is a MEMBER type
73 containing invalid type specifiers.
74 ** ADJOIN with constant NIL as second argument works correctly.
75 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
77 changes in sbcl-1.0.18 relative to 1.0.17:
78 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
79 profiles only the current thread.
80 * minor incompatible change: changes to SYMBOL-VALUE of constants
81 defined with DEFCONSTANT now signal an error.
82 * enhancement: SB-SPROF now has support for wallclock profiling,
83 and is also able to profile specific threads. REPORT output
84 has also additional sorting options.
85 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
87 * optimization: structure allocation has been improved
88 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
89 ** out of line constructors are ~10% faster.
90 ** inline constructors are ~15% faster.
91 ** inline constructors are capable of dynamic extent allocation
92 (generally on x86 and x86-64, in some cases on other platforms
94 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
96 * optimization: file compiler is now able to coalesce non-circular
97 lists, non-base strings, and bit-vectors. Additionally, constants
98 are never referenced through SYMBOL-VALUE at runtime.
99 * optimization: code defining methods on PRINT-OBJECT (and other
100 generic functions in the COMMON-LISP package) now loads faster.
101 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
103 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
104 fasl is now deleted. (reported by Attila Lendvai)
105 * bug fix: READ-LINE always returned NIL for the last line in files.
106 (reported by Yoshinori Tahara)
107 * bug fix: more accurate disassembly annotations of foreign function
108 calls. (thanks to Andy Hefner)
109 * bug fix: trimming non-simple strings and non-string string
110 designators when the there is nothing to trim works properly.
111 (thanks to James Knight)
112 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
113 (thanks to Travis Cross)
114 * fixed some bugs revealed by Paul Dietz' test suite:
115 ** NIL is a valid function name (regression at 1.0.13.38)
116 ** FILL on lists was missing its return value (regression at 1.0.12.27)
117 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
118 fill pointers properly (regression at 1.0.12.23)
119 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
120 displacement indices properly (regression at 1.0.12.23)
122 changes in sbcl-1.0.17 relative to 1.0.16:
123 * temporary regression: user code can no longer allocate closure
124 variable storage on stack, due to bug 419 without explicitly
125 requesting it. Please consult sbcl-devel for advice if you need to
126 use this feature in the meanwhile.
127 * new feature: runtime argument --control-stack-size can be used to
128 adjust thread default control stack size.
129 * enhancement: improved TIME output
130 ** all times are reported using the measured accuracy (milliseconds
131 for real and GC times, microseconds for everything else.)
132 ** processor cycle counts on x86 and x86-64.
133 ** interpreted forms are counted for both evaluator modes.
134 ** number of lambdas converted by the compiler is reported.
135 ** CPU percentage report (computed from real and total run time.)
136 ** more comprehensive run time reporting, using a condenced format
137 ** interperted form, lambda, and page fault counts are omitted
139 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
141 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
142 * optimization: two argument forms of LAST are upto ~10% faster
143 in normal SPEED policies.
144 * optimization: NCONC no longer needs to heap cons its &REST list
145 in normal SPEED policies.
146 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
148 * bug fix: UNION and NUNION work with :TEST-NOT once more,
149 regression since 1.0.9.1. (thanks to Eric Marsden)
150 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
151 since 1.0.15.36. (thanks to Paul Khuong)
152 * bug fix: LAST when always returned the whole list when given a bignum
153 as the second argument.
154 * bug fix: dynamic extent allocation of nested lists and vectors
155 could leak to otherwise accessible parts.
156 * bug fix: invalid optimization of heap-allocated alien variable
158 * bug fix: fasl header checking is less vulnerable to different
159 platform word lengths.
160 * bug fix: more correct assembler syntax for GNU binutils
161 2.18.50.0.4 support. (thanks to Marijn Schouten)
162 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
164 * bug fix: the fopcompiler can handle LOCALLY forms (with no
165 declarations) successfully. (reported by Attila Lendvai)
167 changes in sbcl-1.0.16 relative to 1.0.15:
168 * minor incompatible change: revert the changes to sb-posix's error
169 signaling added in 1.0.14.
170 * minor incompatible change: change PROBE-FILE back to returning
171 NIL whenever we can't get a truename, as was the case before 1.0.14.
172 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
173 inherits from ERROR instead of just CONDITION.
174 * new feature: SB-INTROSPECT can provide source locations for instances
175 as well. (thanks to Tobian Ritterweiler)
176 * optimization: binding special variables now generates smaller code
177 on threaded platforms.
178 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
179 and cases where no :TEST is given but the compiler can infer that
180 the element to search is of type (OR FIXNUM (NOT NUMBER)).
181 * optimization: better LOGNOT on fixnums.
182 * optimization: modular arithmetic for a particular requested width
183 is implemented using a tagged representation unless a better
184 representation is available.
185 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
186 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
187 for wellformedness and returns a specialized vector. (reported by
188 Francois-Rene Rideau)
189 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
190 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
191 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
192 unscheduling timer at the same time as another timer fires.
193 * bug fix: don't reschedule timers for dead threads.
194 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
195 * bug fix: copying output from RUN-PROGRAM to a stream signalled
196 bogus errors if select() was interrupted.
197 * enhancement: add support for fcntl's struct flock to SB-POSIX.
199 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
200 * enhancement: cleaner backtraces for interactive interrupts, as
201 well as other cases where the interesting frames used to be
202 obscured by interrupt handling frames.
203 * enhancement: untracing a whole package using (UNTRACE "FOO") is
204 now supported, and tracing a whole package using (TRACE "FOO") now
205 traces SETF-functions as well.
206 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
207 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
208 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
209 * unreadably printed representation of hash-tables now includes
211 * bug fix: partially fixed #188: type propagation from assignments
212 is now more efficient.
213 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
214 'DOUBLE-FLOAT) are not flushable.
215 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
216 full address of the object, and none of the tag bits.
217 * bug fix: readably printing hash-tables now respects other printer
218 control variables. (reported by Cedric St-Jean)
219 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
220 keyword with MAKE-HASH-TABLE.
221 * bug fix: export SB-POSIX:MKSTEMP.
222 * bug fix: SORT was not interrupt safe.
223 * bug fix: XREF accounts for the last node of each basic-block as
225 * bug fix: MAKE-INSTANCE optimizations interacted badly with
226 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
227 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
228 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
229 create obsolete instances in the case of redefinition or
230 obsoletion of a superclass. (thanks to Andy Hefner)
231 * bug fix: Support for the Alpha architecture has been revived; it had
232 suffered somewhat from lack of maintenance since sbcl-1.0.
233 * improvements to the Windows port:
234 ** The system detects the codepage to use at startup, instead of
235 using the value from the saved core. (thanks to Kei Suzuki)
237 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
238 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
239 (see documentation for details.)
240 * revived support for OpenBSD (contributed by Josh Elsasser)
241 * partially fixed bug #108: ROOM no longer suffers from occasional
242 (AVER (SAP= CURRENT END)) failures.
243 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
244 bodies no longer cause a WARNING to be signalled. (reported by
246 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
247 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
248 the bug in TRUNCATE.) (reported by Michael Weber)
249 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
250 no samples. (reported by Andy Hefner)
251 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
252 no longer appear as (NIL ...) frames in backtraces.
253 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
254 is now actually exported from SB-EXT as documented. (reported by
256 * bug fix: sb-aclrepl now correctly understands how to inspect
257 single-floats on 64-bit platforms where single-floats are not boxed.
258 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
259 yet been finalized. (reported by Levente Meszaros)
260 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
261 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
262 SBCL-specific optimize qualities.
264 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
265 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
266 an executable in the search path, and does so in the child
267 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
268 been removed; it can be found in the CVS history, for somebody who
269 needs that search behavior (see the manual).
270 * minor incompatible change: compiler policy re. weakening type
271 checks has changed: now type checks are weakened only if SAFETY < 2
273 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
274 unparsing of directory pathnames as files. Analogously,
275 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
276 filename to parse into a directory pathname.
277 * enhancement: implicit generic function creation now signals a
278 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
279 users can bind handlers for and muffle around calls to LOAD.
280 * enhancement: RUN-PROGRAM allows unicode arguments and environments
281 to be used (using the default stream external format), and allows
282 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
283 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
284 strings and vectors whose element-type or simplicity is not fully
285 known at compile-time.
286 * optimization: STRING-TRIM and related functions no longer allocate
287 a new string when no trimming needs to be performed. These functions
288 are also faster than before when the input string has been declared
290 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
291 * optimization: READ-LINE is significantly faster for files containing
293 * optimization: non-open coded uses of character comparison operators
294 (e.g. char=) no longer cons when called with more than one parameter
295 on platforms supporting dynamic-extent allocation.
296 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
297 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
298 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
300 * bug fix: some sequence functions elided bounds checking when
302 * bug fix: too liberal weakening of union-type checks when SPEED >
304 * bug fix: more bogus fixnum declarations in ROOM implementation
307 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
308 * new feature: MAKE-HASH-TABLE now experimentally accepts a
309 :SYNCHRONIZED argument, which makes the hash-table safe for
310 concurrent accesses (but not iteration.) See also:
311 SB-EXT:WITH-LOCKED-HASH-TABLE, and
312 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
313 * optimization: CONCATENATE on strings is an order of magnitude faster
314 in code compiled with (> SPEED SPACE).
315 * optimization: SUBSEQ is ~50% faster on lists.
316 * bug fix: bug 417 fixed -- source location reporting is now more robust.
317 * bug fix: SUBSEQ on a list will now correctly signal an error if if
318 END is smaller then START.
319 * bug fix: SB-PROFILE will no longer report extra consing for nested
320 calls to profiled functions.
321 * bug fix: ROOM implementation had bogus fixnum declarations which
322 could cause type-errors when calling ROOM on large images.
323 * bug fix: if file compilation is aborted, the partial fasl is now
324 deleted, and COMPILE-FILE returns NIL as the primary value.
325 * bug fix: number of thread safety issues relating to SBCL's internal
326 hash-table usage have been fixed.
327 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
328 be returned from its body when the values were being returned
329 using unknown-values return convection and the W-P-O was wrapped
330 inside an UNWIND-PROTECT.
331 * bug fix: sb-posix should now compile again under Windows, enabling
334 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
335 * incompatible change: hash-table accessor functions are no longer
336 automatically protected by locks. Concurrent accesses on the same hash-table
337 from multiple threads can give inconsistent results or even corrupt the
338 hash-table completely. Multi-threaded applications should do their own
339 locking at the correct granularity. In the current implementation it is
340 still safe to have multiple readers access the same table, but it's not
341 guaranteed that this property will be maintained in future releases.
342 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
343 and will signal an error at runtime.
344 * enhancement: SB-THREAD package now exports a semaphore interface.
345 * enhancement: CONS can now stack-allocate on x86 and
346 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
348 * enhancement: nested lists can now be stack allocated on
349 platforms providing stack allocation support.
350 * enhancement: dynamic-extent support has been extended to support
351 cases where there are multiple possible sources for the stack
353 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
354 if the mutex is uncontested on Linux.
355 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
356 as specified by the CLHS. (thanks to Tobias Rittweiler)
357 * bug fix: NaN comparison now works on x86-64.
358 * bug fix: CLOSE :ABORT T on a stream with pending output now
360 * bug fix: instances of non-standard metaclasses using standard
361 instance structure protocol sometimes missed the slot type checks
363 * bug fix: known functions can be inlined in high-debug code.
365 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
366 * minor incompatible change: the MSI installer on Windows no longer
367 associates .lisp and .fasl files with the installed SBCL.
368 * minor incompatible change: :UNIX is no longer present in *FEATURES*
369 on Windows. (thanks to Luis Oliviera)
370 * new feature: SB-CLTL2 contrib module now implements
371 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
372 * optimization: scavenging weak pointers is now more efficient,
373 requiring O(1) instead of O(N) per weak pointer to identify
374 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
375 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
376 method bodies using constant slot-names when the first argument is
377 a specializer parameter for the method.
378 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
379 names, when the class of the instance is a direct instance
380 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
381 as slow as the constant slot-name case.)
382 * optimization: member type construction is now O(N) instead
384 * optimization: UNION and NUNION are now O(N+M) for large
385 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
387 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
388 now more readable in environments like Slime which display it.
389 (thanks to Tobias C. Rittweiler)
390 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
391 was able to infer the correct argument type for the object on which
392 the CAS operation was being performed.
393 * bug fix: copy propagation interfered with parallel assignment
394 semantics in local calls. (reported by Paul Khuong)
395 * bug fix: the signed modular fixnum shift compiled to wrong code on
396 x86 and x86-64. (spotted by a slight modification to some of
399 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
400 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
401 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
402 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
404 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
405 and generic function now print the object identity for anonymous
406 instances. (thanks to Attila Lendvai)
407 * enhancement: as an extension to MOP, SBCL now supports SETF of
408 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
409 (thanks to Attila Lendvai)
410 * enhancement: sb-cover has an alternate annotation mode which puts
411 the form annotations at the car of the form, instead of annotating
412 it completely. The benefit of this mode is that it shows better
413 which forms were instrumented by the compiler.
414 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
415 by printer control variables. (reported by Dan Corkill)
416 * bug fix: system leaked memory when delayed output was performed by
417 the OS in smaller chunks then expected. (thanks to David Smith)
418 * bug fix: system leaked memory when file streams were not closed
420 * bug fix: large objects written to slow streams that were modified
421 after the write could end up with the modified state written to
422 the underlying file descriptor.
423 * bug fix: multiple threads operating in parallel on the same stream
424 could cause buffer-overflows.
425 * bug fix: source location information is stored correctly
426 (broken since 1.0.6). This bug would generally show up as the
427 Slime debugger highlighting the wrong form when the sldb-show-source
429 * bug fix: Branch forms are again annotated as branches in the sb-cover
431 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
432 on x86-64. (reported by Gregory Vanuxem)
433 * bug fix: EQUALP could return wrong results for structures with raw
434 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
435 word). (reported by Vjacheslav Fyodorov)
436 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
439 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
440 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
441 atomic compare-and-swap operations on threaded platforms.
442 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
443 allows assigning a global minimum value to optimization qualities
444 (overriding proclamations and declarations).
445 * enhancement: closed over variables can be stack-allocated on x86
447 * performance bug fix: GETHASH and (SETF GETHASH) are once again
449 * optimization: slot definition lookup is now O(1). This speeds up
450 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
451 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
452 * optimization: ASSOC and MEMBER can now be open-coded for all
453 combinations of keyword arguments when second argument is constant
454 and SPEED >= SPACE. In other cases a specialized version is
456 * bug fix: using obsoleted structure instances with TYPEP and
457 generic functions now signals a sensible error.
458 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
459 (reported by Kristoffer Kvello)
460 * bug fix: backtrace construction is now more careful when making
461 lisp-objects from pointers on the stack, to avoid creating bogus
462 objects that can be seen by the GC.
463 * bug fix: defaulting of values in contexts expecting more than 7
464 variables now works on x86-64. (reported by Christopher Laux)
465 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
467 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
468 as the property-list of a symbol.
469 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
470 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
471 situations requested, are once again file-compileable. (reported
474 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
475 * MOP improvement: support for user-defined subclasses of
476 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
477 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
478 the expansion of DEFMETHOD, is responsible for generating a form
479 which creates a list of specializers when evaluated. Additional
480 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
481 debugging and introspective support.
482 * minor incompatible change: the (unsupported) spinlock interface
483 has changed: free spinlock now has the value NIL, and a held spinlock
484 has the owning thread as its value.
485 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
486 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
488 * enhancement: name of a socket-stream is now "a socket" instead of
490 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
491 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
492 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
493 for generic functions. (Thanks to Tobias C. Rittweiler)
494 * optimization: bignum printing speed has been improved by 20-40%
495 (depending on the bignum size.)
496 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
498 * bug fix: the cache used by the CLOS to store precomputed effective
499 methods, slot offsets, and constant return values is now thread and
501 * bug fix: generic function dispatch function updating is now thread
502 and interrupt safe (in the sense that the known issues have been
504 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
505 * bug fix: interning EQL-specializers is now thread and interrupt safe.
506 * bug fix: asdf systems with dependencies to the SB-POSIX or
507 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
508 * bug fix: interrupt safety of applicable method computation has been
511 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
512 * new contrib: sb-cover, an experimental code coverage tool, is included
514 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
515 significantly faster.
516 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
517 produce more efficient code on x86-64 if the compiler can determine
518 that (+ INDEX <constant>) does not require a bounds check and FOO
519 has an element type at least 8 bits wide.
520 * enhancement: a new, experimental synchronous timeout facility is
521 provided. Refer to SB-SYS:WITH-DEADLINE for details.
522 * enhancement: when a symbol name conflict error arises, the
523 conflicting symbols are always printed with a package prefix.
524 (thanks to Kevin Reid)
525 * enhancement: stepping is now once again supported on the SPARC and
526 MIPS platforms. (It is also now more likely to work on CheneyGC
528 * enhancement: sb-sprof can now also track and report accurate call
530 * bug fixes: the treatment of non-standard subclasses of
531 SB-MOP:SPECIALIZER is more correct.
532 * incompatible change: PURIFY no longer copies the data from the
533 dynamic space into the static and read-only spaces on platforms
534 that use the generational garbage collector
535 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
537 * bug fix: GC race condition occasionally resulting in crashes with
538 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
540 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
542 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
543 a STYLE-WARNING to be signalled (regression from 1.0.4.)
544 * bug fix: an asynchronous interrupt could previously leave the
545 system running with GC inhibited.
546 * bug fix: a DECLARE form evaluated at top-level now causes an error
547 rather than silently (or verbosely) returning NIL.
548 * bug fix: trying to dissassemble functions compiled at high DEBUG could
549 result in a error being signalled due to source form lookup errors.
550 (reported by Peter Graves)
552 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
553 * incompatible change: removed writer methods for host-ent-name,
554 host-ent-addresses -- changing the values did not update the DNS
556 * minor incompatible change: changed experimental JOIN-THREAD interface
557 * documentation: the manual now lists reader and writer methods
558 in class slot documentation sections. (thanks to Richard M Kreuter)
559 * documentation: unwinding from asyncronous events has been
560 documented as unsafe.
561 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
562 in multithreaded application code.
563 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
564 platforms. (thanks to James Anderson for the optimization hint)
565 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
567 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
568 variants no longer cons.
569 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
570 their NOT- variants no longer cons.
571 * optimization: EQUAL hash tables no longer use SXHASH for objects
572 of all data types, but instead use an EQL hash for types for which
573 EQUAL is the same as EQL.
574 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
575 are significantly faster.
576 * optimization: new STRING-OUTPUT-STREAM implementation is much
577 faster for large outputs and conses 30% less on average.
578 * enhancement: XREF information is now collected for references made
579 to global variables using SYMBOL-VALUE and a constant argument.
580 * enhancement: SIGINT now causes a specific condition
581 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
582 * bug fix: FILE-POSITION used to signal an error in some cases where
583 ANSI requires it to return NIL.
584 * bug fix: ADJUST-ARRAY is now interrupt-safe.
585 * bug fix: adding and removing fd-handlers is now interrupt-safe.
586 * bug fix: inlined calls to C now ensure 16byte stack alignment on
588 * bug fix: bad type declaration in the CLOS implementation has
589 been fixed. (reported by James Anderson)
590 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
591 (thanks to Lutz Euler)
592 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
593 required by the CLHS. (reported by Samium Gromoff)
594 * bug fix: dead unbound variable references now signal an error.
595 (reported by Marco Monteiro)
596 * bug fix: / with an unused value was being deleted in safe code.
597 (thanks to Marco Monteiro and Kevin Reid)
598 * bug fix: number of characters that can be written onto a single
599 line in a file is unlimited.
600 * bug fix: some GC deadlocks caused by asynchronous interrupts have
601 been fixed by inhibiting interrupts for when GC is disabled.
602 * bug fix: some interrupt safety issues with GETHASH, (SETF
603 GETHASH), CLRHASH and REMHASH have been fixed.
604 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
605 specifier no longer causes infinite recursion.
606 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
607 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
608 * bug fix: an &environment argument with the correct variable information
609 is passed to macros that are expanded during byte compilation.
610 (reported by Samium Gromoff)
611 * bug fix: a logic error was causing the hash function for bignums to
612 have excessive amounts of collisions. (reported by Faré Rideau)
613 * bug fix: modifying the contents of an array could change the return
614 value of SXHASH on that array, which is only allowed for strings
615 and bit vectors. (bug introduced in 0.9.16)
616 * improvement: the x86-64/darwin port now passes all tests (except
617 for the debugger tests) but should still be considered
618 experimental until this is fixed.
619 * improvement: a style-warning is signaled for CASE (etc) clauses with
620 duplicate keys. (thanks to Kevin Reid)
621 * improvement: macos/x86 and macos/x86-64 now use mach exception
622 handlers for dealing with illegal instructions (for trapping and
623 error handling) and memory protection violations (for GC).
625 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
626 * new platform: experimental support for x86-64/darwin (MacOS).
627 * incompatible change: the thread-safe (on most platforms) getaddrinfo
628 and getnameinfo sockets functions are used instead of gethostbyaddr
629 and gethostbyname, on platforms where the newer functions are available.
630 As a result, the ALIASES field of HOST-ENT will always be NIL on these
632 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
633 don't print the startup banner, but behave as if --noinform was passed
634 as a command line argument. (thanks to Kevin Reid)
635 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
636 * optimization: code using alien values with undeclared types is much faster.
637 * optimization: the compiler is now able to open code SEARCH in more cases.
638 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
639 * bug fix: using standardized COMMON-LISP special variables as loop
640 variables no longer signals bogus package lock violations. (reported
642 * bug fix: declaring local loop variables to be of a range-limited type
643 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
644 (reported by Andras Simon)
645 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
646 bugs remain on x86-64.)
647 * bug fix: the #= and ## reader macros now interact reasonably with
648 funcallable instances.
649 * bug fix: type-checks for function arguments were compiled using the
650 compiler policy settings of the wrong lexical environment (in 1.0.2
652 * bug fix: SHADOW accepts characters as string designators, as required
653 by the spec (thanks to Eric Marsden)
654 * bug fix: fixed GC safety issues when foreign functions are called with
655 non-base strings as arguments
656 * bug fix: more consistent error output for fatal-compiler-warnings, like
658 * bug fix: functions with proper names were showing up as NIL in
659 backtraces (reported by Edi Weitz, regression in 1.0.2)
661 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
662 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
663 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
664 produce more efficient code on the x86 if the compiler can determine
665 that (+ INDEX <constant>) does not require a bounds check and FOO
666 has an element type at least 8 bits wide.
667 * bug fix: references to undefined variables are handled the same way
668 in toplevel forms as in the normal compiler
669 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
671 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
672 (thanks to Jon Buller)
673 * improvement: faster compilation times for complex functions
674 * improvement: added readlink support to SB-POSIX (thanks to Richard
677 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
678 * improvement: experimental support for mach exception handling on
679 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
681 * improvement: support for GBK external format.
682 (thanks to Chun Tian (binghe))
683 * improvement: the debugger now displays variables that have been closed
684 over, in code compiled with (DEBUG 2) or higher
685 * improvement: support for executable cores on NetBSD (thanks to
687 * new feature: added a RESTART-FRAME debugger command
688 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
689 be used to provide an implementation for FILE-POSITION on Gray streams
690 (thanks to Eric Marsden)
691 * new feature: add syslog, openlog and closelog support to SB-POSIX
692 (thanks to Richard Kreuter)
693 * optimization: the function call overhead in code compiled with
694 a high DEBUG optimization setting is significantly
695 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
696 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
697 (thanks to Tony Martinez)
698 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
699 evaluated once (reported by Kevin Reid)
700 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
701 works again (thanks to Kilian Sprotte)
702 * bug fix: an error is signaled for tagbodies with duplicate tags
703 (thanks to Stephen Wilson)
704 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
705 * bug fix: Win32 port can now handle foreign code unwinding Lisp
706 stack frames from alien callbacks.
707 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
708 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
709 65k pages (thanks to David Woodhouse)
710 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
712 changes in sbcl-1.0.1 relative to sbcl-1.0:
713 * new platform: FreeBSD/x86-64, including support for threading.
714 * new feature: the compiler stores cross-referencing information
715 about function calls (who-calls), macroexpansion (who-macroexpands)
716 and special variables (who-binds, who-sets, who-references) for code
717 compiled with (< SPACE 3). This information is available through the
718 sb-introspect contrib.
719 * new feature: users may subclass SEQUENCE, and have instances of
720 these classes interoperate with standard Common Lisp functions if
721 a number of methods are defined. (This feature is experimental
722 and the interface subject to change based on feedback from SBCL
723 users and the general community)
724 * improvement: sb-sprof traces call stacks to an arbitrary depth on
725 x86/x86-64, rather than the previous fixed depth of 8
726 * improvement: another pthread back-end of mutex "pthread-futex".
727 * bug fix: non-ascii command-line arguments are processed correctly
728 (thanks to Yaroslav Kavenchuk)
729 * bug fix: non-required arguments were not passed correctly when a method
730 defined using DEFMETHOD was called from a mop-generated method using
731 CALL-NEXT-METHOD (reported by Pascal Costanza)
732 * bug fix: recursion is now permitted in accessors through
733 SLOT-UNBOUND. (reported by Pascal Costanza)
734 * bug fix: an error was signaled at startup if the HOME environment
735 variable was defined, but had an empty value (reported by Peter Van Eynde)
736 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
737 signal a PROGRAM-ERROR, not a BUG.
738 * bug fix: SB-POSIX:READDIR works when built with large file support.
739 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
741 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
742 for better automatic indentation support. (thanks to Matt Pillsbury)
743 * optimization: loading generic functions no longer takes O(n^2) time,
744 proportional to the amount of methods in the generic function
745 (reported by Todd Sabin and Jeremy Brown)
746 * optimization: the FIND and POSITION family of sequence functions
747 are significantly faster on arrays whose element types have been
749 * improvements to the Windows port:
750 ** Intermittent heap corruption problems have been fixed. (thanks
751 to Alastair Bridgewater)
752 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
754 ** Lisp is able to unwind foreign exception frames from alien
755 callbacks. (thanks to Alastair Bridgewater)
757 changes in sbcl-1.0 relative to sbcl-0.9.18:
758 * improvement: experimental support for threading on FreeBSD/x86.
759 (thanks to NIIMI Satoshi)
760 * improvement: runtime option --dynamic-space-size can be used
761 to set the size of the dynamic space reserved on startup.
762 * improvement: floating point modes in effect are now saved in
763 core, and restored on startup.
764 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
765 startup, not time since first call to GET-INTERNAL-REAL-TIME.
766 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
767 threads are not running after *SAVE-HOOKS* have run.
768 * improvement: writes to CLOS instance slots are type-checked in code
769 compiled with (SAFETY 3)
770 * improvement: floating-point exception handling on FreeBSD (thanks to
772 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
773 (thanks to Zach Beane)
774 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
776 * improvement: added support for the Shift-JIS external format.
777 (contributed by NIIMI Satoshi)
778 * improvement: callbacks are supported on Linux/PPC. (thanks to
780 * bug fix: compiler bug triggered by a (non-standard) VALUES
781 declaration in a LET* was fixed. (reported by Kaersten Poeck)
782 * bug fix: file compiler no longer confuses validated and already
783 dumped structurres. (reported by Kaersten Poeck)
784 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
785 fill-pointer signals a type-error as required. (thanks to
787 * bug fix: disassemly of funcallable instances works.
788 * bug fix: single stepping on PPC.
789 * bug fix: fix thread-safety problems in the type system (generally
790 manifesting as nonsensical errors like "STRING is a bad type specifier
791 for sequences" or "The value 1 is not of type FIXNUM")
792 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
793 (reported by Josip Gracin)
794 * bug fix: an error is signaled for attempts to displace arrays with
795 incompatible element types (thanks to Mario Mommer)
796 * bug fix: more correct handling of wide characters in debug info
797 (bug reported by Attila Lendvai and fixed by Juho Snellman)
798 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
799 and don't cause extra consing
800 * optimization: MAP and MAP-INTO are significantly faster on vectors
801 whose elements types have been declared.
802 * Improvements to SB-SPROF:
803 ** Support for allocation profiling
804 ** Reduced profiling overhead, especially for long profiling runs
805 * Improvements to the Windows port:
806 ** floating point exceptions are now reported correctly.
807 ** stack exhaustion detection works partially.
808 ** more accurate GET-INTERNAL-REAL-TIME.
809 ** better RUN-PROGRAM behaviour in case of errors in the child
811 ** PROBE-FILE now simplifies pathnames correctly.
812 ** DIRECTORY now works correctly with :WILD-INFERIORS.
813 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
814 1st 1970 (although time timezone data will be unreliable).
815 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
817 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
818 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
819 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
821 * bug fix: two potential GC deadlocks affecting threaded builds.
822 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
823 returns T (reported by Anton Kazennikov)
824 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
825 with non-variable places
826 * bug fix: remove a race condition in the setting of
827 funcallable-instance functions, this should make threaded CLOS
828 code more stable against memory faults.
829 * bug fix: corruption of specials when unbinding is interrupted by an
830 asynchronous unwind (reported by Hannu Koivisto)
831 * improvement: the debugger will now also display local variables that
832 are only used once, for code compiled with a DEBUG optimization quality
835 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
836 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
837 * incompatible change: External-format support for FFI calls. The
838 SB-ALIEN:C-STRING no longer implies an ASCII
839 external-format. Instead the string is subject to external-format
840 conversion. Additionally return values of type SB-ALIEN:C-STRING
841 are by default of type (SIMPLE-ARRAY CHARACTER), not
842 SIMPLE-BASE-STRING. For an alien type that behaves like the old
843 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
844 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
845 * incompatible change: SB-EXT package no longer contains the
846 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
847 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
849 * incompatible change: the single-stepper is no longer available
850 on Alpha, Mips and Sparc platforms.
851 * minor incompatible change: the direct superclasses of
852 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
853 not (STANDARD-OBJECT FUNCTION). This makes the
854 class-precedence-lists of GENERIC-FUNCTION and
855 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
857 * minor incompatible change: the default stream external format on
858 non-unicode SBCL is no longer detected from the locale, but is always
860 * new feature: Add a version of evaluator that uses an interpreter instead
861 of the compiler. EVAL still uses the compiler by default, to switch it
862 to use the interpreter, set the value of the variable
863 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
864 * minor incompatible change: the single-stepper REPL has been merged
865 with the normal debugger (see the "Stepping" heading of the debugger help
866 for more details). The debugger command STEP will no longer switch
867 to the single-stepper REPL.
868 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
869 for a type now works.
870 * bug fix: loading of default sysinit file works. (thanks to Leonid
872 * bug fix: better detection of circularities in the file-compiler.
873 (reported by Marco Monteiro)
874 * bug fix: the CL pathname functions now work with files that have
875 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
876 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
877 systems with Unix98 pty semantics.
878 * bug fix: ASDF-INSTALL will now work with bsd tar.
879 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
881 * bug fix: timers expiring in dead threads no longer cause a
882 type-error (reported by Paul "Nonny Mouse").
883 * bug fix: thanks to more lightweight single-stepper instrumentation,
884 code compiled with (DEBUG 3) will compile and execute significantly faster,
885 and will have more accurate type-inferencing than before
886 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
887 whose bindings are modified
888 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
889 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
890 ** replace dummy implementations of CL:MACHINE-INSTANCE and
891 CL:SOFTWARE-VERSION with proper version
893 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
894 * feature: implemented the READER-METHOD-CLASS and
895 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
896 as specified by AMOP.
897 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
899 * optimization: faster LOGCOUNT implementation on x86 and x86-64
900 (thanks to Lutz Euler)
901 * optimization: hashing of general arrays and vectors has been
902 improved. (reported by Any Fingerhut)
903 * enhancement: SB-INTROSPECT is now able to find definitions of
904 profiled functions. (thanks to Troels Henriksen)
905 * enhancement: compiler-macro expansion applies now to FUNCALL forms
906 as well, allowing compiler-macros for SETF-functions to expand.
907 * enhancement: step-instrumentation no longer wraps known
908 single-value functions with multiple-value context, allowing
909 better type inference.
910 * fixed bug #337: use of MAKE-METHOD in method combination now works
911 even in the presence of user-defined method classes. (reported by
912 Bruno Haible and Pascal Costanza)
913 * fixed bug #339(c): if there are applicable methods not part of any
914 long-form method-combination group, call INVALID-METHOD-ERROR.
915 (reported by Bruno Haible)
916 * fixed bug #361: the :FUNCTION initarg in the protocol for
917 initialization of methods can now be used to override
918 internally-produced optimized functions. (reported by Bruno
920 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
921 system-provided lambda expression no longer cause warnings about
922 unbound #:|pv-table| symbols.
923 * bug fix: improved the handling of type declarations and the
924 detection of violations for keyword arguments with non-constant
926 * bug fix: potentially erronous calls to PATHNAME and
927 MERGE-PATHNAMES were being flushed in some cases.
928 (reported by Richard Kreuter)
929 * bug fix: compiled calls to TYPEP were mishandling obsolete
930 instances. (reported by James Bielman and Attila Lendvai)
931 * bug fix: format strings with ~> without matching ~< no longer
932 trigger an AVER, but signal an understandable error instead.
933 (reported by Antonio Martinez)
934 * bug fix: specifying an output-file without a pathname-type for
935 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
936 of input-file instead of "fasl". (reported by Robert Dodier)
937 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
938 (reported by James Y Knight).
939 * bug fix: compiler-macros-function did not consider the environment
940 argument for shadowing by local functions.
941 * bug fix: compiler-macros expansion was inhibited by local INLINE
943 * bug fix: inline expansions of known functions were subject to
944 step-instrumentation in high debug policies, leading to problems
946 * bug fix: compiler failed to differentiate between different CONS
948 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
949 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
950 * bug fix: #368: incorrect use of expressed vs. upgraded array
952 * bug fix: #306a: more precise unions of array types.
953 * thread-safety improvements:
954 ** CONDITION-WAIT could return early on Linux, if the thread was
955 interrupted and subsequently continued with SIGCONT.
956 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
958 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
959 * added support for the ucs-2 external format. (contributed by Ivan
961 * minor incompatible change: pretty printing of objects of type
962 (cons symbol) is, in the default pprint-dispatch-table, now
963 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
965 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
966 called later in a class's lifetime, possibly as late as when the
967 first instance of the class is created. Previously,
968 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
969 class became finalizeable.
970 * fixed bug: FILE-POSITION sometimes returned inconsistent results
971 for multibyte external-format streams. (thanks to "vbzoli")
972 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
973 with :ALLOCATION :CLASS inherited from superclasses of the
975 * fixed bug: anonymous classes can now be created using the :NAME
976 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
977 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
978 * fixed bug: core-files saved with :EXECUTABLE T can again be
979 executed when SBCL_HOME isn't set. (reported by James Knight)
980 * fixed bug: toplevel LOCALLY forms with declarations could
981 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
982 * fixed bug: printing from several different threads using different
983 values of *print-case* could cause invalid output, due to
984 some internal special variables of the printer not being bound
985 thread-locally (reported by Max Mikhanosha)
986 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
987 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
988 (reported by Pascal Costanza)
989 * fixed bug: FINALIZE-INHERITANCE is called from
990 REINITIALIZE-INSTANCE on classes when the class has previously
991 been finalized, as required by AMOP.
992 * minor code generation optimizations:
993 ** better register allocation in CLOS dispatching functions
994 ** overflow detection when coercing signed bytes to fixnums on x86-64
995 ** is now implemented with one IMUL instruction instead of three shifts
996 ** more efficient bit-vector access on x86 and x86-64
997 ** more efficient access to raw structure slots on x86 and x86-64
998 * fixed some bugs revealed by Paul Dietz' test suite:
999 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1000 return its argument.
1002 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1003 * feature: thread support on Solaris/x86, and experimental thread support
1005 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1007 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1008 default method for SLOT-UNBOUND.
1009 * minor incompatible change: prevent the user from specializing the
1010 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1011 somewhat counter to the intent of the protocol, I (CSR) think, and
1012 additionally it just doesn't work in SBCL as currently
1013 implemented, thanks to optimizations (that are always valid for
1014 the other three SLOT-VALUEish functions, but not for the setter).
1015 * minor incompatible change: the :SB-LDB feature is now enabled by
1016 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1017 the low-level debugger.
1018 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1019 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1020 down to an SB-SYS:FD-STREAM, allowing direct communication in
1021 more cases. Temporary files and pipes are still used as a fallback
1023 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1024 on the non-reentrant C functions localtime(3) and gmtime(3).
1025 * bug fix: global optimization policy was not visible in LOCALLY and
1027 * bug fix: class objects can be used as specializers in methods.
1028 (reported by Pascal Costanza)
1029 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1030 with a type error. (reported by blitz_ on #lisp)
1031 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1032 failed with a type error. (reported by Pascal Bourguignon)
1033 * bug fix: merging pathnames against defaults with :DIRECTORY
1034 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1035 (reported by James Y Knight)
1036 * bug fix: saving large (>2GB) cores on x86-64 now works
1037 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1038 constant index and a (simple-array (signed-byte 32)) array
1039 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1040 error (patch by Robert J. Macomber)
1041 * feature: TIME now displays the amount of run-time spent in GC
1042 * bug fix: The debugger now does a better job of respecting
1043 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1044 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1045 workaround for bug 403.)
1046 * bug fix: floating point exception handling now works on FreeBSD
1047 (thanks to NIIMI Satoshi)
1048 * fixed some bugs revealed by Paul Dietz' test suite:
1049 ** MISC.641: LET-conversion were not supposed to work in late
1051 * improvements to the Win32/x86 port:
1052 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1053 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1055 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1056 * new feature: source path information is generated for macro-expansion
1057 errors for use in IDE's like Slime (thanks to Helmut Eller)
1058 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1059 * bug fix: compiler does not loop forever on an invalid type in
1061 * improvement: compilation of most CLOS applications is significantly
1063 * optimization: added a limited bytecode compiler for simple toplevel
1064 forms, speeding up compilation and FASL loading
1065 * bug fix: the statistical profiler now properly distinguishes anonymous
1068 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1069 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1070 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1071 into subclasses of the finalized class. Now user calls to
1072 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1073 system return before any subclasses are finalized.
1074 * minor incompatible change: The reader no longer ignores errors
1075 regarding non-existent packages in #+ and #- feature tests.
1076 * new feature: command line options --no-sysinit, --no-userinit to
1077 inhibit loading the corresponding init files
1078 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1079 for sending data through UDP sockets (thanks to François-René Rideau)
1080 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1081 error is signalled instead (thanks to François-René Rideau)
1082 * minor incompatible change: Error signalling behaviour of lexical
1083 operations violating package locks has changed slightly. Refer to
1084 documentation on package locks for details.
1085 * enhancement: EVAL can process IF-expressions without resorting to the
1087 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1088 constant-folded properly due to missing out-of-line functions.
1089 (reported by Mika Pihlajamäki)
1090 * bug fix: LISTEN sometimes returned T even in cases where no data was
1091 immediately available from the stream
1092 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1093 were reversed. (reported by Levente Mészáros)
1094 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1095 when given a vector displaced to an adjustable array. (reported
1097 * bug fix: derivation of float boundaries from numbers outside the
1098 appropriate float range (reported by John Wiseman)
1099 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1101 * bug fix: precomputing cache entries for generic functions with
1102 some subclasses of specializers as yet invalid does not attempt to
1103 fill a cache line with a negative offset. (reported by Levente
1105 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1106 allow condition class objects as arguments to DOCUMENTATION and
1107 (SETF DOCUMENTATION); only find and set documentation for
1108 structure names for the STRUCTURE doc-type. (suggested by Gary
1110 * improvements to the Win32/x86 port:
1111 ** (user-homedir-pathname) and default initialization file
1112 locations now know about the user's "Documents and Settings"
1113 directory (thanks to Yaroslav Kavenchuk)
1114 ** run-program is implemented (thanks to Mike Thomas)
1115 ** sockets support (thanks to Timothy Ritchey)
1116 ** better backtrace support (thanks to Alastair Bridgewater)
1117 ** sb-grovel supported
1118 ** asdf-install and sb-posix work somewhat
1119 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1120 * improvements to the Solaris/x86 port:
1121 ** works on Solaris 11/Solaris Express
1122 ** floating-point exception handling support
1123 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1124 * fixed some bugs revealed by Paul Dietz' test suite:
1125 ** REMOVE-METHOD returns its generic function argument even when
1126 the method is not one of the generic functions' methods.
1127 ** objects declared as MEMBER types can be admissible arguments to
1128 structure accessors.
1129 ** printing characters should simply be printed by the FORMAT ~:C
1131 ** compiler failure when compiling functions with hairy constant
1132 defaults for optional parameters.
1133 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1134 function, which is already optimized.
1136 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1137 * new platform: experimental support for SBCL x86/Darwin, including
1138 MacOS X 10.4.5 on Intel.
1139 * new feature: Unicode character names are now known to the system
1140 (through CHAR-NAME and NAME-CHAR).
1141 * new feature: the filesystem location of SBCL's core file is
1142 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1143 * minor incompatible change: the contrib modules SB-POSIX and
1144 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1145 this change is to make it easier to distribute
1146 location-independent binaries.
1147 * bug fix: as implied by AMOP, standardized classes no longer have
1148 slots named by external symbols of public packages. (reported by
1150 * optimization: calling functions via constant symbols -- as in
1151 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1152 via the function object as in (FUNCALL #'FOO).
1153 * optimization: on x86 and x86-64, the calling convention for the
1154 case of dealing with an unknown number of values has been altered
1155 to be friendlier to the prediction heuristics implemented,
1156 particularly with reference to CALL and RET pairing. (thanks to
1157 Alastair Bridgewater)
1158 * optimization: on x86, the code for access to thread-local symbol
1159 values and binding/unbinding of thread-local symbols is smaller.
1160 (thanks to Alastair Bridgewater)
1161 * enhancement: CONSTANTP is now able to determine constantness of
1162 more complex forms, including calls to constant-foldable standardized
1163 functions and some special forms beyond QUOTE.
1164 * fixed bug: occasional GC crashes on Solaris/x86
1165 * optimization: x86-64 supports stack allocation of results of simple
1166 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1167 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1168 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1169 (thanks to James Knight)
1170 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1171 by the standard (thanks to Stephen Compall)
1173 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1174 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1175 be used for bundling the runtime and the core file into one
1176 executable binary. This feature is not currently supported on all SBCL
1177 platforms. (thanks to James Bielman and NIIMI Satoshi)
1178 * new feature: a generational or ephemeral garbage collector is now
1179 the default on the PowerPC platform (both Linux and Darwin). The
1180 old Cheney (stop and copy) collector is a build-time option.
1181 (thanks to Cyrus Harmon, after Raymond Toy)
1182 * minor incompatible change: the method by which SBCL finds its
1183 contributed modules has changed; it no longer relies on symbolic
1184 links from an $SBCL_HOME/systems directory, but searches directly
1185 in the subdirectories of $SBCL_HOME.
1186 * enhancement: the dynamic heap size on the Linux/PPC platform is
1187 markedly larger, even using the older Cheney garbage collector.
1188 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1189 certain complicated string types. (reported by Gary King)
1190 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1191 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1192 * fixed bug: DOCUMENTATION on structures no longer signals an
1193 error if no structure type of the right name exists. (reported by
1195 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1196 applicable non-standard primary or :AROUND method on
1197 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1198 no longer cause unbound variable errors. (reported by Kevin Reid)
1199 * fixed bug: metacircle resolution in cases where methods have slots
1200 added before the slots from STANDARD-METHOD. (reported by Jean
1202 * fixed bug: the Power PC assembly code for calling into lisp
1203 sometimes computed its return address wrongly (depending on the
1204 15th bit of the address, fixed at link time). This used to
1205 manifest itself in a segmentation violation while building PCL.
1206 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1207 many others over the years)
1208 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1209 the stack pointer (treated as a signed integer) is negative.
1210 (thanks to Peter van Eynde)
1212 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1213 * new platform: experimental support for the Windows operating
1214 system has been added. (thanks to Alastair Bridgewater)
1215 * fixed several bugs in and robustified the PPC FFI (including
1216 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1217 * bug fix: allow non-simple string symbol names (reported by Paul
1219 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1221 * bug fix: saving a core corrupted callbacks on x86/x86-64
1222 * bug fix: closed a loophole in metacircularity detection and
1223 grounding in the PCL implementation of CLOS.
1224 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1225 (reported by tomppa on #lisp)
1226 * bug fix: clear the direction flag on Lisp -> C transitions, as
1227 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1228 (reported by Andrej Grozin and Hendrik Maryns)
1229 * optimization: major improvements to GC efficiency on GENCGC platforms
1230 * optimization: faster implementation of EQUAL
1231 * optimization: emit more efficient opcodes for some common
1232 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1233 * optimization: several other minor code-generation improvements on x86-64
1234 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1236 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1237 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1238 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1239 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1240 by ANSI as a generic function, and for consistency (SETF
1241 GENERIC-FUNCTION-NAME) is treated likewise.
1242 * fixed bug #233.b: make constraint propagation notice when a variable
1243 value is changed after it is referenced but before it is used
1244 * fixed bug #296: no more arbitrary behaviour from filenames with
1245 odd characters as --load arguments. (reported by Adam Warner)
1246 * fixed bug #390: :CHARACTER-SET pathname components now work as
1247 expected. (reported by Tim Daly Jr)
1248 * fixed bug #391: complicated :TYPE intersections in slot
1249 definitions no longer cause an error in PCL internals.
1250 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1251 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1252 returning the number of octets which would be written to the
1253 file-stream. (thanks to Robert J. Macomber)
1254 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1255 arguments to the internal stream functions. (thanks to David
1257 * bug fix: stack exhaustion is now handled on the x86/SunOS
1258 platform. (thanks to Daisuke Homma)
1259 * bug fix: more accurate ROOM results on GENCGC platforms
1260 * optimization: improved type inference for arithmetic-for
1261 index variables in LOOP
1262 * optimization: faster floating-point SQRT on x86-64
1263 * fixed some bugs revealed by Paul Dietz' test suite:
1264 ** DOCUMENTATION returns NIL instead of "" for method combinations
1265 that don't have a docstring
1267 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1268 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1269 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1270 therefore are not customizeable by user code (as seems to be at
1271 least permitted and maybe required by AMOP). As a consolation,
1272 however, the SBCL implementation of these functions now calls
1273 REINITIALIZE-INSTANCE as specified by AMOP.
1274 * bug fix: slot-definition documentation is propagated as per ANSI
1275 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1276 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1277 argument, as do eslotd initialization methods. (from Pascal
1278 Costanza's "Closer" project)
1279 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1280 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1282 * bug fix: it is now possible to have more than one subclass of
1283 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1284 (reported by Bruno Haible, Pascal Costanza and others)
1285 * bug fix: the dependent update protocol now works for generic
1286 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1287 and Pascal Costanza)
1288 * bug fix: condition-class instances corresponding to
1289 DEFINE-CONDITION forms are now created eagerly. (reported by
1290 Kalle Olavi Niemitalo on comp.lang.lisp)
1291 * bug fix: floating point printing is more accurate in some
1292 circumstances. (thanks to Simon Alexander)
1293 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1294 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1295 * bug fix: callbacks on OS X now preserve stack-alignment.
1296 (thanks to Cyrus Harmon)
1297 * enhancement: the x86-64 disassembler is much better at
1298 disassembling SSE instructions. (thanks to Lutz Euler)
1299 * enhancement: CHECK-TYPE on a variable now allows type-inference
1300 (thanks to Vincent Arkesteijn)
1301 * optimization: improved performance of EUC-JP external format.
1302 (thanks to NIIMI Satoshi)
1303 * optimization: performance improvements to IO on file streams of
1304 :ELEMENT-TYPE CHARACTER
1305 * optimization: much faster memory allocation on x86-64
1306 * optimization: faster garbage collections (latency and throughput)
1308 * optimization: faster bignum subtraction and division on x86 and x86-64
1309 (thanks to Lutz Euler)
1310 * optimization: more accurate type-inference for loops that use a
1311 floating point index variable or a negative step.
1313 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1314 * bug fix: add a workaround to SBCL looping infinitely at startup on
1315 Linux kernels with apparently buggy implementations of personality().
1316 (thanks to Svein Ove Aas)
1317 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1318 (thanks to David Lichteblau)
1319 * bug fix: local bindings shadow global symbol macros for
1320 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1321 * optimization: non-open coded uses of numeric comparison operators
1322 (e.g. >) no longer cons when called with more than one parameter
1323 on platforms supporting dynamic-extent allocation.
1324 * enhancement: saving cores with foreign code loaded is now supported
1325 on MIPS/Linux in addition to the previously supported platforms.
1326 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1327 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1328 * bug fix: division by zero in sb-sprof when no samples were collected
1329 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1330 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1331 (thanks to Alexander Kjeldaas)
1332 * new feature: ignoring the return values of destructive functions
1333 like NREVERSE signals a compile-time style-warning.
1334 (thanks to Kevin Reid)
1335 * new feature: the alignment of alien structure fields can be
1336 explicitly specified. (thanks to Cyrus Harmon)
1337 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1338 and dump core on SIGQUIT
1340 ** incompatible change: threads do not inherit values of specials
1341 from their parents (see manual)
1342 ** bug fix: threads stacks belonging to dead threads are freed by the
1343 next exiting thread, no need to gc to collect thread stacks anymore
1344 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1345 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1346 ** bug fix: dynamic variable and thread start related gc lossage
1347 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1349 * fixed some bugs revealed by Paul Dietz' test suite:
1350 ** SUBTYPEP is slightly more accurate on heinously complicated
1351 CONS types where some of the members have uncertain (in the
1352 NIL, NIL sense) type relationships to each other.
1354 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1355 no longer purified unless :PURIFY T is explicitly specified.
1356 ** Non-purified cores are significantly smaller than before
1358 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1359 * new feature: timers based on Zach Beane's excellent timer package
1360 * added support for the following external formats: koi8-u,
1361 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1362 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1363 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1364 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1365 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1366 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1367 * incompatible change: a threaded SBCL will no longer revert to
1368 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1369 * bug fix: interrupts are disabled until startup is complete; no
1370 more sigsegvs when receiving a signal to soon
1371 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1373 * bug fix: add a workaround for the memory randomization features in
1374 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1375 workaround will only be in effect on systems with the proc filesystem
1377 * bug fix: printing objects of type HASH-TABLE signals a
1378 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1380 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1381 have been running for over 50 days. (reported by Gilbert Baumann)
1382 * bug fix: the logic for getting names of functions gets less
1383 confused when confronded with alternate-metaclass
1384 funcallable-instances. (reported by Cyrus Harmon)
1385 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1386 consistent, even on internal alternate-metaclass objects.
1387 * bug fix: flush closure information collected by physical
1388 environment analysis prepass before the main pass. (bug reported
1390 * bug fix: compiler pack phase does not modify a hashtable, which is
1391 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1392 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1393 may be expanded. (reported by Alexander <malishev> on c.l.l)
1394 * bug fix: no more style-warnings when compiling code using LAST
1395 under high-SPEED low-SPACE settings. (reported by David Wallin
1397 * fixed some incompatibilities between SBCL's MOP and the MOP
1399 ** the METAOBJECT class is now implemented;
1400 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1401 STANDARD-OBJECT, as required;
1402 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1403 now compatible (as required by VALIDATE-SUPERCLASS); there
1404 remains a consistency requirement over the presence or absence
1405 of the FUNCTION class in the superclass list of finalized
1406 classes; see the manual for more details;
1407 ** the :AROUND method for COMPUTE-SLOTS on
1408 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1409 requested slot ordering.
1411 ** bug fix: parent thread now can be gc'ed even with a live
1413 ** bug fix: binding a special with PROGV to no value is not
1414 the same as that symbol not having been bound (thanks to
1416 ** bug fix: binding specials is thread safe (thanks to
1418 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1419 and SB-SPROF do not die with 'no handler for signal XX in
1420 interrupt_handle_now(..)' anymore
1421 ** bug fix: WITH-TIMEOUT works with multiple running threads
1422 * fixed some bugs revealed by Paul Dietz' test suite:
1423 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1424 the :method-class keyword argument.
1426 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1427 * new port: the Solaris operating system on x86 processors is now
1428 mostly supported, though some rough edges in the environment
1429 remain. (thanks to Daisuke Homma)
1430 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1431 space for its heap. (thanks to Thiemo Seufer)
1432 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1433 closures. (thanks to Thiemo Seufer)
1434 * minor incompatible change: eof selects abort in the debugger.
1435 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1436 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1438 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1439 deprecated, and will go away in a future revision of SBCL.
1440 * minor incompatible change: GC-ON and GC-OFF are no longer
1441 implemented with a counter, it does not matter how many times gc
1442 is switched on or off
1443 * bug fix: discriminating functions for generic function classes
1444 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1445 make invalid assumptions about method precedence order. (reported
1447 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1448 platforms. (thanks to Thiemo Seufer)
1449 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1450 * bug fix: correct pathname printing with printer escaping is on.
1451 (thanks to Kevin Reid)
1452 * bug fix: complex VOP definitions in "user-space" no longer trigger
1453 package locks. (reported by Zach Beane)
1454 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1455 for funcallable-instance functions, and is no different from
1457 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1458 characters in the current readtable. (reported by Nicholas Neuss)
1459 * bug fix: SUBTYPEP on various CONS types returns more a more
1460 accurate acknowledgment of its certainty.
1461 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1462 lists in some cases. This partially fixes bug 384.
1463 * flush all standard streams before prompting in the REPL and the
1465 * bug fix: signal handling and triggering gc do not conflict
1466 directly anymore, in particular a high frequency sb-sprof does
1467 not prevent gc from running
1468 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1469 approximation for timezone and DST information between the
1470 universal time epoch and the smallest negative 32-bit time_t.
1471 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1472 year 1899 so long as the encoded time is non-negative.
1473 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1474 the correct answer for non-constant rotate amounts when performing
1475 an inline 32-bit rotation.
1477 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1478 there is only one thread in the session
1479 ** bug fix: memory leak for streams created in one thread and
1480 written to in another
1481 ** bug fix: lockup when compiled with gcc4
1482 ** bug fix: race that allows the gc to be triggered when gc is
1484 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1485 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1487 ** bug fix: finished threads are gc'ed properly
1488 * fixed some bugs revealed by Paul Dietz' test suite:
1489 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1490 has arguments with a different set of applicable methods from
1491 the orignal arguments.
1492 ** The type error thrown by MAP now has a correct expected-type
1494 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1495 name a compiled function.
1496 ** Three MISC tests where a large bignum was improperly coerced to
1497 a float that couldn't represent that bignum during type
1498 derivation were fixed.
1499 ** SUBTYPEP can now handle types involving the negation of a
1500 list-form FUNCTION type.
1501 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1502 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1503 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1505 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1506 * New feature: Experimental support for bivalent streams: streams
1507 opened with :element-type :default now allow character and binary
1508 (unsigned-byte 8) I/O
1509 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1510 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1512 * bug fix: no more highly sporadic "couldn't check whether ~S is
1513 readable" when reading a stream and an interrupt hits in the middle
1514 of a select system call
1515 * compiler better recognizes complex arrays (reported by Eduardo
1517 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1518 for variables on SBCL built with linkage-tables. (reported by Luis
1520 * various error reporting improvements.
1521 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1522 (Thanks to Thiemo Seufer)
1523 * bug fix: nested reader invokations work correctly
1524 * bug fix: it is possible to have simultaneous references to foreign
1525 code and foreign data with the same name.
1527 ** added x86-64 support
1528 ** incompatible change: the threading api now works with thread
1529 objects instead of thread ids
1530 ** bug fix: threads are protected from signals and interruption when
1531 starting up or going down
1532 ** bug fix: a race where an exiting thread could lose its stack to gc
1533 ** bug fix: don't halt on infinite error in threads if possible
1534 ** fixed numerous gc deadlocks introduced in the pthread merge
1535 ** bug fix: fixed thread safety issues in read and print
1536 ** bug fix: debugger doesn't hang on session lock if interrupted at
1537 an inappropriate moment
1538 ** bug fix: run-program is now thread safe(r)
1539 ** bug fix: inner with-recursive-lock no longer releases the mutex
1540 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1541 * fixed some bugs revealed by Paul Dietz' test suite:
1542 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1543 EXPECTED-TYPE slots filled.
1544 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1546 ** MAKE-CONDITION accepts classes as type-designators.
1547 ** COMPILE may never return NIL.
1548 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1549 range before calling Unix time functions
1551 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1552 * numerous signal handling fixes to increase stability
1553 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1554 * minor incompatible change: we now correctly canonize default
1555 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1556 as per the MOP, rather than the historical (INITARG INITFUNCTION
1557 INITFORM). (reported by Bruno Haible)
1558 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1559 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1560 for more information.
1561 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1562 pathname is a directory pathname.
1563 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1564 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1566 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1567 :METACLASS argument in addition to a class name. (reported by
1568 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1569 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1570 (reported by Bruno Haible; patch from Gerd Moellmann)
1571 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1573 * bug fix: more cleanups to the floating point exception handling on
1574 x86-64 (thanks to James Knight)
1575 * bug fix: the compiler does not try to constant fold calls of
1576 COERCE and BIT-* functions when they return freshly constructed
1577 objects. (reported by Nikodemus Siivola and Paul Dietz)
1578 * optimization: improved the MIPS versions of generic (in the
1579 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1580 * optimization: direct conversion of (unsigned-byte 32) to floats on
1581 the PowerPC platform.
1582 * optimization: structure instances with raw slots now use less
1583 memory, and probably show better memory locality. (thanks to
1585 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1586 are treated as requests for stack allocation on the x86-64,
1587 Alpha-32, PPC and SPARC platforms.
1588 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1589 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1590 (thanks to Tony Martinez)
1592 ** gcing a dead thread can no longer lead to lockups
1593 ** threads block signals until they are set up properly
1594 ** errno is no longer shared by threads
1595 ** interrupt-thread restores the eflags register on x86
1596 ** fixed some lockups due to gc/thread interaction
1597 * fixed some bugs revealed by Paul Dietz' test suite:
1598 ** invalid dotted lists no longer raise a read error when
1599 *READ-SUPPRESS* is T
1600 ** signal an error if a symbol that names a declaration is used
1601 as the name of a type, or vice versa
1602 ** allow using the (declare (typespec var*)) abbreviation for
1603 (declare (type typespec var*)) with all type specifiers
1604 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1605 ** always return NIL from PPRINT-POP when OBJECT is NIL
1606 ** don't signal errors when pretty-printing malformed LABELS,
1607 FLET or MACROLET forms
1608 ** declarations in a DOLIST body are also in scope for the
1610 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1612 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1615 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1616 * fixed cross-compiler leakages that prevented building a 32-bit
1617 target with a 64-bit host compiler.
1618 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1619 opened with :IF-EXISTS :OVERWRITE.
1620 * fixed bug 281: error for an invalid qualifier in a short-form method
1621 combination method is not signalled until the faulty method is called.
1622 * bug fix: iteration variable type inferrer failed to deal with open
1623 intervals. (reported by Alan Shields)
1624 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1625 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1626 or more runtime options were provided to the sbcl binary.
1627 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1628 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1629 * optimization: there's now a fast-path for fixnum arguments in the
1630 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1631 * optimization: the code generated on x86-64 is more compact thanks
1632 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1633 * bug fix: floating-point exception handling now partly works on
1634 x86-64. (Thanks to James Knight)
1635 * improvement to the MIPS backend from Thiemo Seufer: C-style
1636 64-bit long long arguments and return values to alien functions
1638 * fixed some bugs revealed by Paul Dietz' test suite:
1639 ** the type-error signalled from WARN has a filled-in DATUM slot.
1640 ** the type-error required when a stream is not associated with
1641 a file has the stream as its datum.
1642 ** type-errors on single-floats on x86-64 no longer have
1643 :INVALID-OBJECT as the datum
1644 ** the type-errors signalled for invalid function names now have
1645 a correct expected type
1646 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1647 for structures defined with a :TYPE. Documentation strings for
1648 typed structures are no longer immediately discarded
1649 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1650 broadcast streams and synonym streams. FILE-LENGTH now also works
1651 on broadcast streams.
1653 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1654 * incompatible change: the --noprogrammer option, deprecated since
1655 version 0.7.5, has been removed. Please use the equivalent
1656 --disable-debugger option instead.
1657 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1659 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1660 inoperational for a while now) has been completely removed.
1661 * null lexical environments are now printed as #<NULL-LEXENV>,
1662 significantly reducing the amount of clutter in typical backtraces.
1663 * documentation on weak pointers, finalization, and after GC hooks
1664 has been added to the manual.
1665 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1666 as other specialized array types, is much faster. SUBSEQ and
1667 COPY-SEQ on such arrays have also been sped up.
1668 * optimization: EQL is now more efficient when at least other argument
1669 is known to be of type (OR FIXNUM (NOT NUMBER)).
1670 * fixed bug: the runtime is now less vulnerable to changes in the
1671 size of the SBCL object on OS X, and virtual memory is reserved for
1672 all spaces that need to be at a fixed address.
1673 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1674 * fixed bug: finalizers and after GC hooks that cause consing are now
1676 * fixed bug: compiler error messages and summaries are now printed to
1677 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1678 * fixed inference of the upper bound of an iteration variable.
1679 (reported by Rajat Datta).
1680 * fixed bug 211e: calling local functions with duplicated constant
1681 keyword argument no longer causes a bogus style warning about an
1683 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1684 ftype declaration to be disregarded. (reported by Dave Roberts)
1685 * fixed bug 373: caused by erronous compilation of references to alien
1686 variables in the runtime on ppc/darwin.
1687 * fixed bug 376: CONJUGATE type deriver.
1688 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1689 (reported by Baughn on #lisp)
1690 * a cross-compiler bug on non-x86 platforms has been identified and
1691 fixed. (thanks to Bruno Haible)
1692 * improvements to the MIPS runtime code for increased stability.
1693 (thanks to Thiemo Seufer)
1694 * increased the maximimum compact environment size to allow
1695 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1696 * partial workaround for bug 135: don't name the function we're
1697 calling for hairy cases of EVAL, so as not to accumulate one environment
1698 entry per call to EVAL. (reported by Kevin Reid)
1699 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1700 * optimization: type testing for non-vector arrays should be faster.
1701 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1703 * bug fix: setting 31st element of a bit vector to zero did not work
1705 * bug fix: redefining a class definition which failed due to a
1706 previous accessor / function clash now works (but see BUGS entry
1707 #380 for more problems in this area). (thanks to Zach Beane)
1708 * the long form of DEFINE-METHOD-COMBINATION disables method group
1709 checking when given a single method group with pattern *. (thanks
1711 * on x86 compiler supports stack allocation of results of simple
1712 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1713 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1714 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1715 * fixed some bugs related to Unicode integration:
1716 ** the restarts for recovering from input and output encoding
1717 errors only appear when there is in fact such an error to
1719 * fixed some bugs revealed by Paul Dietz' test suite:
1720 ** MISC.549 and similar: late transformation of unsafe type
1721 assertions into derived types caused unexpected code
1723 ** SCALE-FLOAT type deriver is less wrong.
1724 ** type derivers for EXP, LOG and similar functions compute result
1725 types for complex arguments better.
1726 ** (MISC.563) CONJUGATE type deriver works for very restricted
1728 ** out-of-line type testers for character strings are available.
1729 ** EQUAL compiler transform understands specialness of objects
1731 ** accessing double-floats stored on the stack now works on x86-64.
1732 ** debugger internals could sometimes create invalid lispobjs,
1733 resulting in GC crashes.
1734 ** MISC.548: type check weakening can convert required type into
1736 ** initialization forms for bindings are not in scope of free special
1739 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1740 * incompatible change: thread support for non-NPTL systems has
1741 been removed - locking is buggy and unreliable. A threaded
1742 SBCL build will now warn at startup and refuse to create
1743 new threads, unless futex support is detected in the kernel
1744 * incompatible change: the top level REPL now has only an ABORT
1745 restart associated with it, not TOPLEVEL and ABORT as it used to.
1746 TOP and TOPLEVEL are now available as debugger commands for
1747 returning to the top level.
1748 * incompatible change: forms evaluated in the REPL now use the
1749 global optimization policy.
1750 * incompatible change: user- and system-initialization files are
1751 no longer processed with LOAD, but by READ and EVAL; hence the
1752 global optimization policy, startup package, readtable, etc,
1754 * The .fasl file format number has been incremented because of
1755 various incompatible changes.
1756 * internal entry point details and argument counts no longer appear
1757 in backtraces unless explicitly requested by setting
1758 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1759 * built-in and standard functions no longer have names like "top
1760 level local call to FOO".
1761 * fixed bug 32: functions defined in non-null lexical environments
1762 now have more legible printed representation
1763 * fixed bug 33: functions defined in non-null lexical environemnts
1764 are now more amenable to inspection by INSPECT.
1765 * workaround for bug 354: XEPs no longer appear in backtraces unless
1766 explicitly requested.
1767 * fixed bug: receiving the signal which results from attempting to
1768 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1769 SIGBUS on other *BSDs) on architectures where the C stack is also
1770 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1771 to debugger confusion or wild instability. (reported by Cheuksan
1773 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1774 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1775 (reported by Lutz Euler)
1776 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1777 compiled code. (reported by Paul Dietz)
1778 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1779 specialized parameter to a method to something that is not TYPEP
1780 the specializer is now possible.
1781 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1782 face of package deletion.
1783 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1784 pathname. (reported by David Tolpin on c.l.l)
1785 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1786 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1787 STRING-STREAM and FILE-STREAM, respectively.
1788 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1789 than previously; in addition, its test suite is now run on build.
1790 (thanks to Luke Gorrie)
1791 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1792 * optimization: added a immediate representation for single-floats
1794 * fixed some bugs revealed by Paul Dietz' test suite:
1795 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1796 ** attempting to create a package with a colliding nickname causes
1797 correctable errors to be signalled.
1798 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1799 ** several x86-64 backend bugs related to sign-extension of immediate
1802 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1803 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1804 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1806 * fixed loading of multiply forward-referenced layouts.
1807 (thanks to Cheuksan Wang)
1808 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1809 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1810 (see COMPILE-FILE documentation for details of :PRINT option)
1811 * fixed bugs 19 and 317: fixed-format floating point printing is
1812 more accurate. This also fixes a bug reported by Adam Warner
1813 related to the ~@F format directive.
1814 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1816 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1817 dispatch table if the from-char is a dispatch macro character.
1818 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1819 by the wordsize no longer ignore the last word. (reported by Lutz
1821 * fixed bug in type checking of dynamic-extent variables. (reported
1823 * optimization: sequence traversal functions use their freedom to
1824 coerce function designators to functions.
1825 * optimization: code with many calls to CLOS methods specialized on
1826 CLOS classes has had redundant type checks removed.
1827 * optimization: added declarations to speed up operations that access
1828 the internal character database (for example STRING-UPCASE)
1829 * optimization: comparison operations between floats and sufficiently small
1830 fixnums no longer create extra rationals
1831 * fixed some bugs related to Unicode integration:
1832 ** portions of multibyte characters at the end of buffers for
1833 character-based file input are correctly transferred to the
1834 start of the buffer at the next read.
1835 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1836 passing it through to OPEN.
1837 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1838 argument given to internal calls to OPEN.
1839 ** the built-in comment readers (introduced by character sequences
1840 ";" and "#|") are more forgiving to encoding errors; they will
1841 STYLE-WARN and then attempt to resync the stream at a character
1842 boundary. (thanks to Teemu Kalvas)
1843 * fixed some bugs in the x86-64 port:
1844 ** Negative short int return values from c-calls are sign-extended
1846 ** The stack is aligned to 16-bytes for c-calls, as required by
1847 the ABI. (thanks to Cheuksan Wang)
1848 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1849 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1850 for enums. (thanks to Vincent Arkesteijn)
1851 ** Multiple small optimizations and bugfixes for floating point
1853 * fixed some bugs revealed by Paul Dietz' test suite:
1854 ** Space, Tab, Linefeed, Return and Page have the invalid
1855 secondary constituent character trait.
1856 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1858 ** WITH-INPUT-FROM-STRING should only update the index place on
1860 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1861 works more reliably.
1862 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1863 with one-dimensional arrays with fill pointers.
1864 ** TYPE-OF failed on a complex with an integer realpart and a
1866 ** compiler failure during type inference for the code of form
1867 (IF (EQL X (THE ...)) ...) (MISC.535).
1869 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1870 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1871 platform. The port supports 61-bit fixnums, large memory spaces
1872 and reloading shared object files.
1873 * enhancement: saving cores with foreign code loaded is now
1874 supported on ppc/Darwin in addition to the previously supported
1876 * enhancement: the statistical profiler now walks deeper into the
1877 call stack for more meaningful call-graphs and accrued time
1878 reports (x86/x86-64 only). It also now reports time spent in
1880 * enhancement: it is now possible to trace most individual methods
1881 of a generic function in addition to tracing the generic function
1883 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1885 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1886 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1887 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1888 produces an error. (thanks to Vincent Arkesteijn)
1889 * bug fix: NAMESTRING on pathnames with :WILD components in their
1890 directories works correctly. (thanks to Artem V. Andreev)
1891 * fixed bug 125: compiler preserves identity of closures. (reported
1893 * bug fixed: functions with &REST arguments sometimes failed with
1894 "Undefined function" when compiled with (DEBUG 3). (reported by
1896 * bug fix: overflow during compiling of setting element of a bit
1897 vector with constant index and value. (reported by Timmy Douglas)
1898 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1899 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1900 * on x86 compiler partially supports stack allocation of dynamic-extent
1902 * GO and RETURN-FROM do not check the extent of their exit points
1903 when compiled with SAFETY 0.
1904 * fixed some bugs related to Unicode integration:
1905 ** encoding and decoding errors are now much more robustly
1906 handled; it should now be possible to recover even from invalid
1907 input or output to the terminal. (thanks to Teemu Kalvas)
1908 ** provided a first cut at implementing STRING-TO-OCTETS and
1909 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1910 ** altered the SB-MD5 contributed module slightly, changing the
1911 interface just enough for it to be supportable for builds where
1912 lisp characters are not eight bits.
1913 * fixed some bugs revealed by Paul Dietz' test suite:
1914 ** the FORMATTER-generated functions for ~V[ conditionals require
1915 the correct number of arguments.
1916 ** READ-FROM-STRING returns the mandated second value when applied
1917 to displaced strings.
1918 ** the #\Rubout and #\Backspace characters are treated as invalid
1919 constituent characters by the tokenizer.
1921 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1922 * new feature: reloading changed shared object files with
1923 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1924 * new feature: references to foreign variables and functions
1925 can now be compiled and loaded before the corresponding shared
1926 object file is loaded, as long as the foreign definitions are
1927 available at runtime.
1928 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1929 supported. (thanks to Dan Debertin)
1930 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1931 just keywords. (thanks to Vincent Arkesteijn)
1932 * fixed bug #331: structure-class instances corresponding to
1933 DEFSTRUCT forms are now created eagerly.
1934 * fixed bug #345: backtraces from calls to undefined functions work
1935 on x86 as well. Related bug #61 is now also partially fixed on x86
1936 (backtraces from throws to unknown catch tags.)
1937 * bug fix: lambda-list parsing is now stricter vrt. order and number
1938 of lambda-list keywords.
1939 * bug fix: as specified by AMOP, an error is signalled if a
1940 class-option appears multiple times in a DEFCLASS form. (reported
1942 * bug fix: RANDOM can be compiled when the compiler derives the type
1943 of its numeric argument as a disjoint set of small integers.
1944 (reported by Paul Dietz)
1945 * bug fix: starting a core saved with shared objects loaded when
1946 those objects are not available no longer causes threaded SBCL to
1947 hang. (reported by Sean Ross)
1948 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1950 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1951 argument, which must always be NIL. (reported by Kalle Niemitalo)
1952 * bug fix: printing 1.0d+23 no longer results in an error.
1953 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1954 * bug fix: structure slot setters preserve evaluation order. (thanks
1956 * fixed some bugs related to Unicode integration:
1957 ** RUN-PROGRAM can allow its child to take input from a Lisp
1958 stream. (reported by Stefan Scholl)
1959 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1960 locales. (reported by Ken Causey)
1961 * fixed some bugs revealed by Paul Dietz' test suite:
1962 ** INCF, DECF and REMF evaluate their place form as specified in
1964 ** FORMATTER expands ~{ ~} iteration directives with V or #
1965 parameters correctly.
1966 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1967 consequent uses no arguments correctly.
1968 ** the system has a partial understanding of the (COMPLEX RATIO)
1970 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1973 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1974 * new feature: a build-time option (controlled by the :SB-UNICODE
1975 keyword feature, enabled by default) for building the system with
1976 support for the entire 21-bit character space defined by the
1978 * new feature: the system now has rudimentary external-format
1979 support; the primary user-visible change at this time is that
1980 characters with the high bit set (such as lower-case-e-acute) will
1981 print correctly to a terminal in a UTF-8 environment.
1982 * minor incompatible change: BASE-CHAR no longer names a class;
1983 however, CHARACTER continues to do so, as required by ANSI.
1984 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1985 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1986 has been moved to the SB-EXT package (temporarily re-exported from
1988 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1989 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1991 * fixed bug #351: better error-handlind and reporting for malformed
1993 * fixed bug #350: bignum-printing is now more memory-efficient,
1994 allowing printing of very large bignums, eg. (expt 2 10000000).
1995 (reported by Bruno Haible)
1996 * fixed bug #302: better primitive-type selection for intersection
1998 * fixed bug #308: non-graphic characters now all have names, as
1999 required. (reported by Bruno Haible)
2000 * bug fix: redefining a class with different superclasses now correctly
2001 removes it from the direct-subclasses of its previous superclasses.
2002 (reported by David Morse)
2003 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2004 new value now works. (reported by Bruno Haible)
2005 * bug fix: correct canonicalization of multiple non-standard slot
2006 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2007 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2008 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2009 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2010 now exists, an signals an error.
2011 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2012 during macroexpansion for non-iteration variables as well. (reported
2013 by Bruno Haible for CMUCL)
2014 * bug fix: Cyclic structures and unprintable objects in compiler
2015 messages no longer cause errors. (reported by Bruno Haible)
2016 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2017 and READ-FROM-STRING all now return a primary value of NIL if
2018 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2019 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2020 (reported by Bruno Haible for CMUCL)
2021 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2022 do not designate a condition. (reported by Bruno Haible for
2024 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2025 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2026 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2027 permitting resolution in favour of any of the conflicting symbols.
2028 (reported by Bruno Haible for CMUCL)
2029 * bug fix: EQUAL compiler optimizations is less aggressive on
2030 strings which can potentially compare true despite having distinct
2031 specialized array element types.
2032 * bug fix: unit enumerations can be defined without dividing by
2033 zero. (thanks to Vincent Arkesteijn)
2034 * FORMAT compile-time argument count checking has been enhanced.
2035 (report from Bruno Haible for CMUCL)
2036 * a partial workaround for the bug 262: the compiler does not try to
2037 inline-expand a local function doing RETURN-FROM from a deleted
2038 BLOCK. (thanks to Peter Denno for the bug report and to David
2039 Wragg for the simple test case)
2040 * fixed some bugs revealed by Paul Dietz' test suite:
2041 ** RENAME-PACKAGE allows all package designators as new package
2043 ** constraint propagation and lambda variable substitution are
2044 more cautious in dealing with partially deleted code.
2045 ** compiler sometimes lost reoptimization passes.
2046 ** CERROR, given a condition as condition designator, passes its
2047 remaining arguments to the continue format control without
2049 ** Case-altering FORMAT directives work correctly on non-ASCII
2051 ** The REAL type specifier handles bounds outside the floating
2052 point ranges without signalling FLOATING-POINT-OVERFLOW.
2053 ** Functions with IR1-transformations can create intercomponent
2054 references to global functions.
2055 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2057 ** FORMAT ~R treats a nil value for its first parameter correctly.
2059 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2060 * enhancement: saving cores with foreign code loaded is now
2061 supported on x86/NetBSD and sparc/Linux in addition to the previously
2062 supported platforms.
2063 * bug fix: on some platforms repeated installations caused multiple
2064 copies of HTML documentation to be installed -- should not happen
2065 any more. (reported by Stefan Scholl)
2066 * bug fix: parsing self-recursive alien record types multiple times
2067 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2068 original patch by Helmut Eller for CMUCL)
2069 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2070 (thanks to Richard Kreuter)
2071 * bug fix: defining classes whose accessors are methods on existing
2072 generic functions in other (locked) packages no longer signals
2073 bogus package lock violations. (reported by François-René Rideau)
2074 * bug fix: special variables as DEFMETHOD parameters no longer have
2075 associated bogus type declarations. (reported by David Wragg and
2077 * bug fix: read-write consistency on streams of element-type
2078 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2079 * bug fix: redefiniton of the only method of a generic function with
2080 no DEFGENERIC no longer emits a full WARNING. In addition,
2081 redefinition of generic functions with no DEFGENERIC to an
2082 incompatible lambda list now signals an error. (thanks to Zach
2084 * bug fix: DEFGENERIC now works even when there's a function of the
2085 same name in an enclosing lexical environment. (thanks to Zach
2087 * fixed compiler failure, caused by instrumenting code during
2088 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2089 * optimization: added loop analysis and improved register allocation
2090 * fixed some bugs revealed by Paul Dietz' test suite:
2091 ** POSITION on displaced vectors with non-zero displacement
2092 returns the right answer.
2093 ** (SIMPLE-STRING) is a valid type specifier for sequence
2095 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2097 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2098 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2100 ** The justification version of the FORMAT ~< directive treats
2101 non-zero minpad parameter correctly.
2103 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2104 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2105 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2106 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2107 the supported interface.
2108 * new feature: Single-stepping of code compiled with DEBUG 2 or
2109 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2110 * new feature: saving cores with foreign code loaded is now
2111 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2112 Timothy Moore's work for CMUCL)
2113 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2114 parameters to * instead of NIL if no initform is supplied.
2115 (reported by Johan Bockgaard)
2116 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2117 correctly when FUNCALL appears as the car of the form. Note:
2118 despite this FUNCALL forms are not currently subject to
2119 compiler-macro expansion. (port of Raymond Toy's fix for the
2120 same from CMUCL, reported by Johan Bockgaard)
2121 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2122 (thanks for Teemu Kalvas)
2123 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2124 case instead of terminating the iteration (thanks for Julian
2125 Squires, Sean Champ and Raymond Toy)
2126 * bug fix: incorrect expansion of defgeneric that caused a style
2127 warning. (thanks for Zach Beane)
2128 * bug fix: slot accessor effective method computation works properly
2129 for classes with multiple non-standard applicable methods on
2130 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2131 * on x86 compiler supports stack allocation of results of LIST and
2132 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2133 CMUCL implementation by Gerd Moellmann)
2134 * fixed some bugs revealed by Paul Dietz' test suite:
2135 ** FORMAT strings with both the ~<~:;~> form of the justification
2136 directive and pretty-printing directives cause an error.
2138 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2139 * incompatible change: the internal functions
2140 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2141 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2142 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2143 instead of the old functions.
2144 * new feature: on platforms where "dladdr" is available foreign
2145 function names now appear in backtraces. (based on Helmut Eller's
2147 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2148 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2150 * documentation: ASDF manual is now distributed with SBCL.
2151 * bug fix: SBCL can now load its contributed modules with REQUIRE
2152 even if the system-provided entries have been removed from
2153 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2155 * bug fix: backtraces involving undefined functions or assembly
2156 routines are more informative. (thanks to Brian Downing)
2157 * bug fix: mutually referent alien structures now work correctly.
2158 (reported by Rick Taube)
2159 * bug fix: structures defined by WITH-ALIEN can be referred to
2160 within other definitions in the same WITH-ALIEN.
2161 * bug fix: division operators (MOD, TRUNCATE and the like) with
2162 constant zero divisors and integer dividends no longer generate
2164 * bug fix: provide default methods for INPUT-STREAM-P and
2165 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2166 * bug fix: improve the detection and resolution of MOP metacycles
2167 (where effective method computation depends on the generic function
2168 having its effective method computed). (reported by Bruno Haible)
2169 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2170 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2171 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2172 * bug fix: pathnames with relative directory components can be
2173 represented relative to default pathnames.
2174 * optimization: in taking the GCD of bignums, reduce the two bignums
2175 to approximately the same size (using Euclid's algorithm) before
2176 applying the more sophisticated binary GCD. (thanks to Juho
2178 * optimization: COUNT on bitvectors now operates word-at-a-time.
2179 * optimization: ASH with a positive, but not necessarily constant,
2180 (leftwards) shift, when in a modular context, is compiled to a
2182 * fixed some bugs revealed by Paul Dietz' test suite:
2183 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2184 if the corresponding argument is NIL.
2185 ** FORMAT directives accepting a minpad parameter treat negative
2187 ** literal commas as character parameters to FORMAT directives are
2189 ** literal spaces directly after ~<Newline> directives within a
2190 format-logical-block (~:< ~@:>) do not induce :FILL-style
2191 conditional newlines.
2192 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2193 as specified: it no longer includes conditional newlines).
2194 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2196 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2197 :PER-LINE-PREFIX argument does not evaluate to a string.
2199 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2200 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2201 the manual for details; for now, package locks can be disabled by
2202 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2203 if no major problems are found then it is likely that they will be
2204 compiled in unconditionally.
2205 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2206 unsupported operators on all platforms. To load a shared library
2207 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2208 object file, link it into a shared library outside of SBCL and
2209 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2211 * bug workaround: ROOM T and the GC were not working together
2212 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2213 reporting facility aren't preserved. That reporting has been
2214 disabled, so now until and unless someone figures out how to make it
2215 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2216 * minor incompatible change: as threatened around sbcl-0.8.0, the
2217 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2218 The symbols are also exported from SB-PCL for backwards
2219 compatibility, but more so than before SB-PCL should be treated as
2220 an implementation-internal package.
2221 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2223 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2224 It is known as of this release not to work on the Alpha, however.
2225 * fixed bug #167: errors signalled due to illegal syntax in method
2226 bodies are now more legible.
2227 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2228 designators and can hence be used with TYPEP.
2229 * fixed bug #333: CHECK-TYPE now ensures that the type error
2230 signalled, if any, has the right object to be accessed by
2231 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2232 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2233 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2235 * fixed bug #334: programmatic addition of slots using specialized
2236 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2237 and :ALLOCATION :CLASS slots.
2238 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2239 float exponents. (rereported by Peter Seibel)
2240 * fixed a bug: #\Space (and other whitespace characters) are no
2241 longer considered to be macro characters in standard syntax by
2242 GET-MACRO-CHARACTER.
2243 * fixed bug: initialization of condition class metaobjects no longer
2244 causes an instance of the condition to be created. (reported by
2246 * fixed bug: it is now possible to have slots such that
2247 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2248 description is neither :INSTANCE nor :CLASS.
2249 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2250 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2251 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2252 non-standard SLOT-DEFINITION-ALLOCATION values.
2253 * fixed bugs: various race conditions handling exiting threads.
2254 CL-PPCRE's thread torture test now passes.
2255 * fixed arguably-a-bug: GC time proportional to number of threads in
2256 system even when most of them are idle
2257 * optimization: improved performance of BIT and SBIT on bit-vectors.
2258 * .fasl file incompatibility: The fasl file version number has
2259 been incremented because of changes associated with package locks.
2261 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2262 * minor incompatible change: the system no longer provides
2263 optimization hints (conditions which are TYPEP
2264 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2265 mode; these hints are still emitted when higher SPEED optimization
2267 * new contrib module: a sampling profiler (profiling by statistical
2268 sampling, rather than by instrumenting functions) is available as
2269 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2270 * the behaviour of the standard function ED is now customizeable by
2271 third parties through a hook variable: see ED's documentation
2272 string for information on the protocol.
2273 * the compiler no longer emits efficiency notes for (FUNCALL X)
2274 when the type of X is uncertain under default optimization
2276 * fixed bug 276: mutating a binding of a specialized parameter to a
2277 method to something that is not TYPEP the specializer is now
2279 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2280 exhibit the required behaviour on the x86 platform. (thanks to
2281 Peter van Eynde, Eric Marsden and Bruno Haible)
2282 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2283 even for difficult arguments. (reported by Peter Graves)
2284 * fixed bug 141a: the backquote printer now descends quoted
2286 * fixed another bug in backquote printing: no more destructive
2287 modification of the form's list structure. (reported by Brian
2289 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2290 move between its address being taken and the call to
2291 interrupt_thread, fixing a crashing race condition.
2292 * the SB-POSIX contrib implementation has been adjusted so that it
2293 no longer exhibits ridiculously poor performance when constructing
2294 instances corresponding to C structs.
2296 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2297 * minor incompatible change: the sb-grovel contrib now treats C
2298 structures as alien (in the sense of SB-ALIEN) objects rather than
2299 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2300 has implications for memory management of client code
2301 (sb-grovel-returned objects must now be manually managed) and for
2302 type safety (alien objects now have full types).
2303 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2304 used to control emission of compiler diagnostics, rather than the
2305 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2306 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2307 quality should be considered deprecated.
2308 * install.sh now installs the user manual as well
2309 * (not quite a new documentable feature, but worth considering in
2310 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2311 beginnings of a semantically meaningful condition hierarchy is
2312 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2314 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2315 designator as the defaults argument.
2316 * fixed bug: Displaced arrays whose displaced-to array has become
2317 too small now cause ARRAY-DIMENSION to signal an error, providing
2318 for safer bounds-checking. (reported by Bruno Haible)
2319 * fixed bug: DEFCLASS slot definitions with identical :READER and
2320 :WRITER names now signal a reasonable error. (reported by Thomas
2322 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2324 * fixed bug: Invalid slot specification errors now print correctly.
2325 (thanks to Zach Beane)
2326 * fixed bug 320: Shared to local slot value transfers in class
2327 redefinitions now happen corrently from superclasses as
2328 well. (reported by Bruno Haible)
2329 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2331 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2332 improper lists correctly. (reported by Bruno Haible)
2333 * fixed bug 313: source-transform for <fun-name> was erroneously
2334 applied to a call of a value of a variable with name <fun-name>.
2335 (reported by Antonio Menezes Leitao)
2336 * fixed bug 307: The obsolete instance protocol ensures that
2337 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2338 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2339 remove dead unknown-values globs from the middle of the stack.
2340 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2341 easier to resume long computations after using *BREAK-ON-SIGNALS*
2342 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2343 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2344 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2345 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2346 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2347 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2349 * fixed a (fairly theoretical) bug in string printing: if
2350 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2351 does not have array-element-type equal to the most general string
2353 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2354 function name. (thanks to Nikodemus Siivola)
2355 * fixed bug: compile-time format string checker failed on
2356 non-closed ~{. (reported by Thomas F Burdick)
2357 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2358 don't warn on legal loop code involving more than one aggregate
2359 boolean. (thanks to Nikodemus Siivola)
2360 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2361 in keyword position to MAKE-PACKAGE was not being reported as
2363 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2364 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2366 * fixed bugs 280 and 312: the checking for multiple definitions in a
2367 file is less likely to become confused by uses of inline
2369 * fixed bug: the #S reader macro performs the keyword coercion
2370 specified for slot names. (reported by Kalle Niemitalo)
2371 * fixed bug: lambda lists may contain symbols whose names start with
2372 & but are not lambda-list-keywords; their occurrence triggers a
2374 * fixed bug 321: define-method-combination argument lambda lists do
2375 not cause invalid code to be generated when &optional or &aux
2376 variables have default values. (reported by Bruno Haible)
2377 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2378 have CLOS classes; this appears also to have fixed reported
2379 instability in stack exhaustion detection.
2380 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2381 MUFFLE-WARNING when no associated restart is present is now
2383 * optimization: rearranged the expansion of various defining macros
2384 so that each expands into only one top-level form in a
2385 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2387 * optimization: used a previously little-used slot in symbols to
2388 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2390 * fixed some bugs revealed by Paul Dietz' test suite:
2391 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2393 ** Fixed an optimization bug related to inheritance of initforms
2394 from local to shared slots.
2395 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2396 ** CAST optimizer forgot to flush argument derived type.
2397 ** print/read consistency on floats is now orders of magnitude
2398 more likely. (thanks also to Bruno Haible for a similar report
2400 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2401 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2402 for objects of type REAL. Make it so.
2403 ** SXHASH is specified (infelicitously) to respect similarity,
2404 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2405 it so. (thanks to Markus Ziegler)
2406 ** on the Alpha, the compiler succeeds in compiling functions
2407 returning a known number of arguments greater than 63.
2408 ** fixed handling of invalid NIL arguments in keyword position
2410 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2412 ** provided a readably-printable representation for RANDOM-STATE
2414 ** ensured that pathnames, where they have a namestring, always
2415 print using #P"..." syntax.
2417 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2418 * Support for the forthcoming 2.0 version of the NetBSD kernel
2419 running on x86 hardware has been added. (thanks to Perry
2420 E. Metzger most immediately, and others for their past work)
2421 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2422 current release version; SBCL's OpenBSD support had been broken
2423 since about the time of OpenBSD's switch to ELF binary format.)
2424 * [placeholder for DX summary]
2425 ** user code with &REST lists declared dynamic-extent, under high
2426 speed or space and low safety and debug optimization policy.
2427 * The manual has been converted to Texinfo format and the debugger
2428 chapter from the cmucl manual has been added.
2429 * A facility has been added to extract documentation strings from
2430 sbcl and store them as Texinfo-formatted snippets for inclusion in
2431 the manual (via Texinfo's @include directive)
2432 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2433 the test case to Dave Roberts)
2434 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2435 pointers. (reported by Sean Ross)
2436 * bug fix: PROFILE output is printed nicely even for large numerical
2437 values. (thanks to Zach Beane)
2438 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2439 greater than 32 handle EOF correctly.
2440 * bug fix: on X86 an immediate argument of the IMUL instruction is
2441 correctly printed in disassembly. (thanks to Lutz Euler)
2442 * bug fix: class slots in redefined classes preserve their old
2443 values. (thanks to Bruno Haible and Nikodemus Siivola)
2444 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2445 succeeds. (reported by Marco Baringer)
2446 * fixed some bugs revealed by Paul Dietz' test suite:
2447 ** READ-SEQUENCE now works on ECHO-STREAMs.
2448 ** RATIONALIZE works more according to its specification. (thanks
2451 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2452 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2453 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2454 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2455 should matter to you only if you rebind the printer control
2456 variables and then find you want different bindings in the
2457 debugger than in the ordinary execution of your program.)
2458 * The runtime build system has been tweaked to support building
2459 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2460 assembler and linker. (thanks to Nikodemus Siivola)
2461 * Unbound, undefined, undeclared variables now trigger full
2462 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2463 more often programmer error than deliberate exploitation of undefined
2465 * optimization: the hash algorithm for strings has changed to one
2466 that is less vulnerable to spurious collisions. (thanks to Juho
2468 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2469 needless bounds checking. (thanks to Juho Snellman)
2470 * optimization: implemented multiplication as a modular
2471 (UNSIGNED-BYTE 32) operation on the PPC backend.
2472 * fixed some bugs revealed by Paul Dietz' test suite:
2473 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2475 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2476 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2477 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2479 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2481 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2483 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2484 their output stream on EOF from read.
2485 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2486 have been read to end-of-file.
2487 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2489 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2490 description of determination of which consecutive characters
2492 ** Printing the "Space" character with escaping on now yields "#\\ ",
2493 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2494 ** Reading floating-point numbers with *READ-BASE* set to a number
2495 less than 10 works correctly.
2496 ** Reading floating-point numbers with *READ-BASE* set to a number
2497 more than 10 works correctly.
2498 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2499 the readtable currently in effect.
2501 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2502 * minor incompatible change: parsing of namestrings on a physical
2503 (Unix) host has changed; numbers after the final #\. in a
2504 namestring are no longer interpreted as a version field. This is
2505 intented to be largely invisible to the user, except that the
2506 meaning of the namestring "*.*.*" has changed: it now refers to a
2507 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2508 should usually be replaced by
2509 (make-pathname :name :wild :type :wild :version :wild)
2510 with the added benefit that this is more likely to be portable.
2511 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2512 option to OPEN now signals an error if the file being opened
2513 exists; this may have an impact on existing code.
2514 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2515 signals received, on the PowerPC platforms (both Linux and
2516 Darwin). (thanks to Pierre Mai for pointing out the location of
2518 * several fixes on OS X: The system now builds and runs cleanly on
2519 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2520 (thanks to Brian Mastenbrook)
2521 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2522 second-resolution: integer multiples of 1/3600 between -24 and 24.
2523 (thanks to Vincent Arkesteijn)
2524 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2525 type when called with 1 argument; PEEK-CHAR checked type of
2526 PEEK-TYPE only after having read first character from a
2527 stream. (reported by Peter Graves)
2528 * bug fix: the garbage collector now has much better locality
2529 behaviour, and in particular no longer treats all memory as being
2530 exhausted when many small objects point to each other in a deeply
2532 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2533 recognized as being TYPEP their class.
2534 * bug fix: the PUSHNEW documentation string has been corrected.
2535 (thanks to Vincent Arkesteijn)
2536 * bug fix: defaulting of the value for the last of an atypically
2537 large number of multiple values being bound was not being
2538 performed correctly on the Alpha or PPC platforms
2539 * optimization: implemented multiplication as a modular
2540 (UNSIGNED-BYTE 32) operation on the x86 backend.
2541 * optimization: SEARCH on simple-base-strings can now be open-coded.
2542 (see also contrib/compiler-extras.lisp for inspiration for
2543 teaching the compiler about the Boyer-Moore algorithm).
2544 * value, returned by MAX (and MIN) called with several EQUALP, but
2545 not EQL, arguments now does not depend on compiler settings.
2546 * fixed some bugs revealed by Paul Dietz' test suite:
2547 ** in stack analysis liveness information is propagated from
2548 non-local entry points.
2549 ** pathwise CAST removing failed when the CAST node did not start
2551 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2552 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2554 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2555 host is already defined.
2556 ** RENAME-FILE works on streams instead of signalling an internal
2558 ** PEEK-CHAR uses the current readtable when determining whether
2559 or not a character is whitespace.
2560 ** MERGE-PATHNAMES handles the case when the pathname does not
2561 specify a name while the default-pathname specifies a version.
2562 ** Pathnames now stand a better chance of respecting print/read
2564 ** Attempting to use standardized file system operators with a
2565 pathname with invalid :DIRECTORY components signals a
2567 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2569 ** DIRECTORY on logical pathnames is more correct.
2570 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2571 signal a TYPE-ERROR if their argument is not a stream.
2572 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2573 designator argument does not designate a stream.
2574 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2575 examining the synonym.
2576 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2578 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2579 element-type larger than ([UN]SIGNED-BYTE 32).
2581 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2582 * When built with the :SB-FUTEX feature, threaded builds now take
2583 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2584 for faster/more reliable mutex and condition variable support.
2585 * Incompatible change (but one you probably shouldn't have been using
2586 anyway): the interface and code for arbitrating between multiple
2587 threads in the same user session has been redesigned.
2588 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2589 error when called without an explicit environment argument.
2590 (thanks to Nikodemus Siivola)
2591 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2592 confused about their position. (thanks to Adam Warner and Gerd
2594 * bug fix: Pretty printing backquoted forms with unquotations in the
2595 argument list position of various code constructs such as LAMBDA
2596 now works correctly. (reported by Paul Dietz)
2597 * bug fix: Pretty printing unquotations no longer loses all
2598 stream position information.
2599 * optimization: performance of string output streams is now less
2600 poor for multiple small sequence writes.
2601 * optimization: performance of CSUBTYPEP in the presence of complex
2602 expressions involving CONS and NOT many times has been improved.
2603 (reported by Paul Dietz)
2604 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2606 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2607 streams for the REPL, for the debugger, and so on. (thanks to
2609 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2610 * fixed some bugs revealed by Paul Dietz' test suite:
2611 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2612 is ignored if it is too small, rather than propagating through
2613 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2614 ** extremely complex negations of CONS types were not being
2615 sufficiently canonicalized, leading to inconsistencies in
2617 ** VALUES tranformer lost derived type.
2619 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2620 * fixed a bootstrapping bug: the build process no longer assumes
2621 that the various BOOLE-related constants have the same value in
2622 host and target lisps. (noted by Paul Dietz' test suite on an
2623 SBCL binary built from CLISP)
2624 * The system can now be dynamically linked on the MIPS platform,
2625 which enables dynamic loading of foreign code from Lisp. (thanks
2626 to Ralf Baechle for discussions on the MIPS ABI)
2627 * The system now records debugging information for its own source
2628 files in a filesystem-position-independent manner, relative to
2629 the "SYS" logical host.
2630 * fixed a compiler bug: MV-LET convertion did not check references
2631 to the "max args" entry point. (reported by Brian Downing)
2632 * tweaked disassembly notes to be less confident about proclaiming
2633 some instruction as an LRA. (thanks to Brian Downing)
2634 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2635 now each have their own history, command character, and other
2636 characteristics. (thanks to David Lichteblau)
2637 * fixed some bugs revealed by Paul Dietz' test suite:
2638 ** compiler failure in compiling LOGAND expressions including a
2640 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2641 shift greater than 32.
2642 ** FLUSH-DEST did not mark blocks for type check regeneration.
2643 ** HANDLER-CASE failed to accept declarations in handler clauses
2644 in some circumstances.
2646 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2647 * New code in contrib/sb-introspect (still probably not entirely
2648 stable yet) provides some support for smart Lisp development
2649 environments like SLIME.
2650 * The conditions signalled for errors occurring when loading .fasl
2651 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2652 in a way which should help ASDF recover gracefully.
2653 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2654 stuff has been cleaned up. If you code contrib/ stuff, this might
2655 affect you, and you can look at contrib/README, contrib/STANDARDS,
2656 and/or the 0.8.4.27 diff to check.
2657 * In full calls the compiler now does not generate checks for declared
2658 argument types for all arguments.
2659 * various threading fixes
2660 ** and some experimental patches which didn't make it into
2661 the main tree for this release, but which are shipped in
2662 contrib/experimental-thread.patch as a possible fix for some
2663 failures (deadlock, spinning...) in GC-intensive multithreaded
2665 * fixed PPC build problem (source code incompatibility of different
2666 library versions): added offsetof() hackery which attempts to divine
2667 where glibc maintainers put uc_mcontext today
2668 * fixed bug 282: compiler does not trust type assertions while passing
2669 arguments to a full call.
2670 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2671 &OPTIONAL VALUES type parameter.
2672 * fix bug 214: algorithm for noting rejected templates is now more
2673 similar to that of template seletion. (also reported by rydis on
2675 * fixed bug 141b: printing backquoted information readably and prettily
2676 inserts a space where necessary.
2677 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2678 (CHARACTER 10) are now reported as errors, rather than propagated
2679 as unknown types. (reported by piso on #lisp)
2680 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2681 with values NIL and :ERROR. (thanks to Milan Zamazal)
2682 * fixed bug 191c: CLOS now does proper keyword argument checking as
2683 described in CLHS 7.6.5 and 7.6.5.1.
2684 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2685 longer raise an error; further, using a list as a for-as-arithmetic
2686 counter now raises a meaningful error.
2687 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2688 now understood by sequence creation functions such as MAKE-SEQUENCE
2690 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2691 a STRING-INPUT-STREAM.
2692 * compiler enhancement: SIGNUM is now better able to derive the type
2694 * type declarations inside WITH-SLOTS are checked. (reported by
2696 * fixed some bugs revealed by Paul Dietz' test suite:
2697 ** incorrect optimization of TRUNCATE for a positive first
2698 argument and negative second.
2699 ** compiler failure in let-convertion during flushing dead code.
2700 ** compiler failure while deriving type of TRUNCATE on an
2701 interval, containing 0.
2702 ** ASH of a negative bignum by a negative bignum count now returns
2704 ** intersection of CONS types now canonicalizes properly, fixing
2705 inconsistencies in SUBTYPEP.
2707 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2708 * incompatible change: The --disable-debugger command line
2709 option now clobbers the debugger at a more fundamental
2710 level, by redefining #'INVOKE-DEBUGGER instead of by
2711 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2712 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2713 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2714 enter the debugger REPL and then suffer recursive errors
2715 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2716 while under the new system, BREAK in --disable-debugger mode
2717 terminates the system just as an unhandled error would.
2718 * fixed compiler performance when processing loops with a step >1;
2719 * bug fix: DOCUMENTATION now retrieves generic function
2720 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2721 support has been systematized, and now supports the methods
2722 specified by ANSI, along with a default method and a method for
2723 slot documentation. (reported by Nathan Froyd)
2724 * bug fix: effective methods associated with a generic function are
2725 no longer cached over a change of that generic function's method
2726 combination. (reported by Andreas Fuchs)
2727 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2728 names a non-existent directory. (thanks to Andreas Fuchs)
2729 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2730 the PPC platform, lead to stack corruption; this has been fixed.
2731 (reported by Rainer Joswig)
2732 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2733 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2734 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2735 arithmetic in general, is now much more reliable on the PPC
2737 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2738 platform now returns the right answer.
2739 * optimization: restored some effective method precomputation in
2740 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2741 precomputation is now tunable.
2742 * optimization: compiler-internal data structure use has been
2743 reviewed, and changes have been made that should improve the
2744 performance of the compiler by about 20%.
2745 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2746 simple-base-strings and simple-bit-vectors is improved.
2747 * optimization: the optimization of 32-bit logical and arithmetic
2748 functions introduced in version 0.8.3 on the x86 has been
2749 implemented on the mips, ppc and sparc platforms; an
2750 implementation of the same facility, but for 64-bit arithmetic,
2751 has been added for the alpha.
2752 * microoptimization: the compiler is better able to make use of the
2753 x86 LEA instruction for multiplication by constants.
2754 * bug fix: in some situations compiler did not report usage of
2755 generic arithmetic in (SPEED 3) policy.
2756 * bug 145b fix: compiler used wrong type specifier while converting
2757 MEMBER-types to numeric.
2758 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2760 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2761 index argument now works properly on the Alpha platform.
2762 * bug fix: floating point exception treatment on the Alpha platform
2764 * bug fix: FILE-POSITION works much better on string input and
2765 output streams. (thanks to Nikodemus Siivola)
2766 * bug fix: many threading/garbage collection symptoms sorted.
2767 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2768 might be pseudo-atomic.
2769 * internal change: Stopping for GC is now done with signals not
2770 ptrace. GC is now done in whichever thread wanted it, instead of
2772 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2774 * bug fix: result form in DO is not contained in the implicit
2776 * incompatible change: ICR structure is changed; the value part of
2777 CONTINUATION is now called LVAR; corresponding functions are
2778 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2779 * added type deriver for ISQRT (thanks to Robert E. Brown).
2780 * bug fix: better support for loading from the command line when an
2781 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2783 * fixed some bugs revealed by Paul Dietz' test suite:
2784 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2785 ** ROUND and FROUND now give the right answer when given very
2786 small float arguments.
2787 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2789 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2790 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2791 ** LCM should return a non-negative integer.
2792 ** PARSE-INTEGER returned the index of a terminator instead of the
2793 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2794 ** PARSE-INTEGER returned an incorrect index being applied to a
2796 ** LCM with two arguments of 0 returns 0 rather than signalling
2798 ** unsigned addition of a 32-bit constant with the high bit set no
2799 longer causes an internal compiler error on the x86.
2800 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2801 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2802 with negative last argument.
2803 ** byte specifiers with zero size and position no longer cause
2804 an error during type derivation.
2805 ** bignum multiplication on the Alpha platform now returns the
2807 * porting: The system now builds on SuSE AMD64, although it still
2808 generates a 32-bit binary.
2809 * .fasl file incompatibility: The fasl file version number has
2810 been incremented (because of the changes to internal compiler
2811 data structures referred to above).
2813 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2814 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2815 more accurately, on the Darwin kernel running on PowerPC hardware.
2816 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2817 * Compiler code deletion notes now signal a condition of type
2818 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2819 an associated MUFFLE-WARNING restart.
2820 * The compiler now performs limited argument count validation of
2821 constant format strings in FORMAT, and where appropriate in ERROR,
2822 CERROR and WARN. (thanks to Gerd Moellmann)
2823 * New ASDF-INSTALL contrib can be used for automatic download and
2824 installation of third-party Lisp code from CCLAN or other sites
2826 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2827 which forces another thread to execute a function supplied by the
2829 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2830 now accept and act upon their :ELEMENT-TYPE keyword argument.
2831 (reported by Martin Atzmueller, Edi Weitz)
2832 * bug fix: FILE-POSITION now accepts position designators up to
2833 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2834 is the greater. (thanks to Patrik Nordebo)
2835 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2836 Kalle Olavi Niemitalo)
2837 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2838 printable. (reported by Eric Marsden)
2839 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2840 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2841 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2842 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2843 (thanks to Stig Sandoe)
2844 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2845 UNREAD-CHAR now correctly includes the unread character in the
2846 target sequence. (thanks to Gerd Moellmann)
2847 * bug fix in threaded builds: the system can now be suspended and
2848 resumed by shell job control with minimal disruption.
2849 * bug fixes in times and timezones >2038AD
2850 * better handling of "where is GNU make?" problem in build scripts
2851 (thanks to Nikodemus Siivola)
2852 * new optimization: inside a named function any reference to a
2853 function with the same name is considered to be a self-reference;
2854 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2855 optimization quality.
2856 * new optimization on x86: logical functions and + now have
2857 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2858 used when the result is truncated to 32 bits.
2859 * VALUES declaration is partially enabled.
2860 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2861 * fixed some bugs revealed by Paul Dietz' test suite:
2862 ** The system now obeys the constraint imposed by
2863 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2864 types form a lattice under type intersection.
2865 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2866 ** ASSOC now ignores NIL elements in an alist.
2867 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2868 and (1+ MOST-POSITIVE-FIXNUM) answers.
2869 ** The addition of a method with invalid qualifiers to a generic
2870 function does not cause an error to be signalled immediately;
2871 a warning is signalled, and the error is generated only on
2872 calling the generic function.
2873 * changed .fasl file version number, in part to add type codes for
2874 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2875 obscure ANSI requirements
2877 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2878 * fixed bug 148: failure to inline-expand a local function left
2879 garbage, confusing the compiler.
2880 * fixed bugs 3cd: structure slot readers perform type check if the
2881 slot can have an invalid value (i.e. it is either not initialized
2882 or can be written with a less specific slot writer).
2883 * bug fix: the compiler now traps array references to elements off
2884 the end of an array; previously, the bounds checking in some
2885 circumstances could go off-by-one.
2886 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2888 * type declarations for array element types now obey the description
2889 on the CLHS page "Declaration TYPE", as per discussions on
2890 sbcl-help around 2003-05-08. This means that a declaration
2891 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2892 declaration, all references to BAR will be asserted or assumed
2893 (with THE, so dependent on compiler policy) to involve objects of
2894 type FOO. Note that no such declaration is implied in
2895 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2896 * declared types of functions from the "Conditions"
2897 chapter. (reported by Paul Dietz)
2898 * bug fix: CERROR accepts a function as its first argument.
2899 * bug fix: NTH an NTHCDR accept a bignum as index
2900 arguments. (reported by Adam Warner)
2901 * optimization: character compare routines now optimize comparing
2902 against a constant character. (reported by Gilbert Baumann)
2903 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2904 argument now works properly on the MIPS platform.
2905 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2906 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2907 declared to have a functional type, does not cause an error
2909 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2910 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2911 * fixed bug: if last continuation of a deleted block has a
2912 destination, this destination should be deleted too. (reported by
2914 * fixed a bug in the bootstrap process: the host compiler's values
2915 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2916 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2917 test case from Patrik Nordebo)
2918 * improved the ability of the disassembler on the PPC platform to
2919 provide helpful disassembly notes.
2920 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2921 the class in more cases than previously.
2922 * bug fix: FILE-POSITION now understands :START and :END for
2923 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2924 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2925 8) data. (thanks to David Lichteblau)
2926 * bug fix: it is possible to add a method to a generic function
2927 without lambda list.
2928 * bug fix: reader failed to signal END-OF-FILE inside an
2929 object representation. (reported by Nikodemus Siivola)
2930 * fixed some bugs revealed by Paul Dietz' test suite:
2931 ** LAST and [N]BUTLAST should accept a bignum.
2932 ** condition slot accessors are methods.
2933 ** (VECTOR NIL) is a subtype of STRING.
2935 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2936 * minor incompatible change: some nonsensical specialized lambda
2937 lists (used in DEFMETHOD) which were previously ignored now signal
2939 * minor incompatible change: the system is now aware of the types of
2940 variables in the COMMON-LISP package, and will signal errors for
2941 most violations of these type constraints (where previously they
2942 were silently accepted).
2943 * minor incompatible change: COMPILE-FILE now uses the freedom
2944 afforded (ANSI 3.2.2.3) to use derived function types for
2945 functions defined in the same file. This also permits the system
2946 to warn on static type mismatches and function
2947 redefinition. (Currently it does not work with high DEBUG level.)
2948 * minor incompatible change: VALUES declaration is disabled.
2949 * When issuing notes, the compiler now signals a condition of type
2950 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2951 restart for use in user handlers. It is expected that the
2952 COMPILER-NOTE condition will eventually become a condition
2953 supertype to a hierarchy of note types, which will then be
2954 handleable in a similar fashion. However, at the moment, no such
2955 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2956 but it's an implementation detail, not a classification for the
2958 * Changes in type checking closed the following bugs:
2959 ** type checking of unused values (192b, 194d, 203);
2960 ** template selection based on unsafe type assertions (192c, 236);
2961 ** type checking in branches (194bc).
2962 * A short form of VALUES type specifier has ANSI meaning (it has
2963 increased the number of situations when SBCL cannot perform type
2965 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2966 keywords or constants is permissible.
2967 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2968 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2969 classes. (thanks to Antonio Martinez)
2970 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2971 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2972 argument to a pretty-print stream. Instead, it leaves any such
2973 operations to DESCRIBE-OBJECT methods.
2974 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2976 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2977 lambda lists. (thanks to David Lichteblau)
2978 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2979 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2980 (thanks to Gerd Moellmann)
2981 * fixed bug 246: increased compilation speed of long
2982 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2984 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2986 * DEFCONSTANT now throws a condition of type
2987 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2988 constant to a non-EQL value; CONTINUE and ABORT restarts
2989 respectively change and preserve the value.
2990 * fixed bug 63: The code walker, part of the implementation of CLOS,
2991 is now better at handling symbol macros.
2992 * bug fix: There is no longer an internal implementation type named
2993 CL:LENGTH. (reported by Raymond Toy)
2994 * bug fix: In macro-like defining macros/special operators the
2995 implicit block does not enclose the lambda list.
2996 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2997 enclosed in parentheses) are not suitable as type specifiers, and
2998 their use properly signals an error now.
2999 * bug fix: An argument count mismatch for a type specifier in code
3000 being compiled no longer causes an unhandled error at compile
3001 time, but signals a compile-time warning.
3002 * fixed simple vector readable printing
3003 * bug fix: DESCRIBE takes more care over whether the class
3004 precedence list slot of a class is bound before accessing it.
3005 (reported by Markus Krummenacker)
3006 * bug fix: FORMATTER can successfully compile pretty-printer format
3007 strings which use variants of the ~* directive inside.
3008 * bug fix: SEARCH now applies its TEST predicate to the elements of
3009 the arguments in the correct order. (thanks to Wolfhard Buss)
3010 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3011 in type inference. (thanks to Robert E. Brown)
3012 * bug fix: Reading in symbols with an explicit package name of ""
3013 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3014 * fixed some bugs revealed by Paul Dietz' test suite:
3015 ** NIL is now allowed as a structure slot name.
3016 ** Arbitrary numbers, not just REALs, are allowed in certain
3017 circumstances in LOOP for-as-arithmetic clauses.
3018 ** Multiple class redefinitions before slot access no longer
3019 causes a type error.
3020 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3021 association between the name and a class.
3022 ** Generic functions with non-standard method-combination and over
3023 five methods all of which return constants no longer return NIL
3024 after the first few invocations. (thanks to Gerd Moellmann)
3025 ** CALL-NEXT-METHOD with no arguments now passes the original
3026 values of the arguments, even in the presence of assignment.
3027 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3029 ** DISASSEMBLE works with closures and funcallable instances.
3030 ** ADD-METHOD now returns the generic function, not the new method.
3031 ** FIND-METHOD signals an error if the lengths of the specializers
3032 is incompatible with the generic function, even if the ERRORP
3034 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3035 which its argument is a member.
3036 ** DEFCLASS only redefines the class named by its class-name
3037 argument if that name is the proper name of the class;
3038 otherwise, it creates a new class.
3039 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3040 of the UNBOUND-SLOT condition to the name of the slot.
3041 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3042 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3043 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3044 SLOT-MAKUNBOUND in the specified fashion.
3046 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3047 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3048 cross-compilation host. As a consequence, we can now bootstrap our
3049 way up to SBCL starting with a bare gcc toolchain and human-readable
3050 source code (first the source to CLISP, then the source to SBCL).
3051 * A contributed module containing a partial implementation of the
3052 simple-streams interface has been included. (thanks to Rudi
3054 * A contributed module implementing the RFC1321 Message Digest
3055 Algorithm, known as MD5, has been included.
3056 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3057 no longer has any effect, as the code controlled by this feature
3058 has been deleted. (As far as we know, no-one has ever built using
3059 this feature, and its semantics were confused in any case).
3060 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3061 work on conditions (as required by the ANSI specification),
3062 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3063 expected behaviour on conditions. Users should note, however,
3064 that such behaviour is not required by the ANSI specification,
3065 and so use of this behaviour may render their code unportable.
3066 * fixed some bugs revealed by Paul Dietz' test suite:
3067 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3069 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3070 ** SXHASH on condition objects no longer returns NIL.
3071 ** :ALLOCATION :CLASS slots are better treated; their values are
3072 updated on class redefinition, and initforms inherited from
3073 superclasses are applied.
3074 ** REMOVE-METHOD returns its generic function argument even when
3075 no method was removed.
3076 ** SHARED-INITIALIZE now initializes the values of the requested
3077 slots, including those with :ALLOCATION :CLASS.
3078 ** ALLOCATE-INSTANCE now works on structure classes defined via
3079 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3081 ** SLOT-EXISTS-P now works on conditions, as well as structures
3083 ** MAKE-LOAD-FORM now has the required methods on
3084 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3085 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3086 keyword, and now implements the SLOT-NAMES argument.
3087 ** methods with &OPTIONAL arguments no longer allow too many
3088 arguments to be passed in the call without error.
3089 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3090 option is consistent with the required arguments of the generic
3091 function lambda list.
3092 * bug fix: REQUIRE accepts a string designator. (Thanks to
3094 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3095 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3096 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3097 * bug fix: The compiler now checks for duplicated variables in macro
3099 * bug fix: SETQ on globals returns the correct value.
3100 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3101 not a COMPILER-ERROR (followed by some other strange error on
3102 choosing the CONTINUE restart).
3103 * bug fix: make.sh and friends are now more consistent in the way that
3104 they look for GNU "make".
3106 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3107 * experimental native threads support (on x86 Linux >=2.4 only).
3108 This is not compiled in by default: you need to add :SB-THREAD to
3109 the target features. See the "Beyond ANSI" chapter of the manual
3111 * fix for longstanding nonANSIism: The old distinction between
3112 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3113 The return value from CL:FIND-CLASS is now a CLOS class, and
3114 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3115 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3116 * An interface to the MetaObject Protocol, as described in Kiczales,
3117 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3118 MIT Press, 1991, is available from the SB-MOP package.
3119 * incompatible change: the SB-PCL package should now be considered
3120 a private implementation detail, and no longer a semi-private MOP
3122 * minor incompatible change: due to rearrangement for threads, the
3123 control stack and binding stack are now allocated at arbitrary
3124 addresses instead of being hardcoded per-port. Users affected by
3125 this probably have to be doing advanced things with shared
3126 libraries, and will know who they are.
3127 * minor incompatible change: Previously, all --eval forms used were
3128 processed with READ before any of them were processed with EVAL.
3129 Now each --eval form is processed with both READ and EVAL before
3130 the next --eval form is processed. (Thus package operations like
3131 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3132 work as the user might reasonably expect.)
3133 * minor incompatible change: *STANDARD-INPUT* is now only an
3134 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3136 * minor incompatible change: Y-OR-N-P is now character-oriented,
3137 not line oriented. Also, YES-OR-NO-P now works without errors.
3138 (thanks to Antonio Martinez)
3139 * sb-aclrepl module improvements: an integrated inspector, added
3140 repl features, and a bug fix to :trace command.
3141 * Known functions, which cannot be open coded by the backend, are
3142 considered to be able to check types of their arguments. (fixing
3143 a bug report by Nathan J. Froyd)
3144 * fixed a bug in computing method discriminating functions: It is
3145 now possible to define methods specialized on classes which have
3146 forward-referenced superclasses. (thanks to Gerd Moellmann)
3147 * fixed evaluation order in optional entries (reported by Gilbert
3149 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3150 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3151 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3152 required-by-AMOP NAME argument, as well as CLASS and
3153 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3154 * fixed bug 20: DEFMETHOD can define methods using names that are
3155 not the proper names of classes to designate class specializers.
3156 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3157 Unix file descriptors, instead of blowing up. (thanks to Antonio
3159 * Garbage collection refactoring: user-visible change is that a
3160 call to the GC function during WITHOUT-GCING will not do garbage
3161 collection until the end of the WITHOUT-GCING. If you were doing
3162 this you were probably losing anyway.
3163 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3164 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3165 TYPEP the latter but not the former.
3166 * The compiler issues a full WARNING for calls to undefined functions
3167 with names from the CL package.
3168 * MAP-INTO for a vector destination is open coded. (reported by
3169 Brian Downing on c.l.l)
3170 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3171 documentation string.
3172 * fixed some bugs revealed by Paul Dietz' test suite:
3173 ** COPY-ALIST now signals an error if its argument is a dotted
3175 ** Condition slots are now accessed more correctly in the presence
3176 of multiple initargs for a given slot.
3177 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3178 exclude restarts of the same name associated with a different
3180 ** DEFCLASS of forward-referenced classes with another
3181 forward-referenced class in the superclasses list no longer
3183 ** Condition slots are now initialized once each, not multiple
3184 times. (thanks to Gerd Moellmann)
3185 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3187 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3188 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3189 arguments contain duplicated elements.
3190 ** RESTART-CASE understands local macros.
3191 ** RESTART-CASE associates exactly its own restarts with a condition.
3192 ** ENDP in safe mode checks its argument to be of type LIST.
3193 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3194 in question is unbound.
3195 ** Optimized MAKE-INSTANCE functions no longer cause internal
3196 assertion failures in the presence of duplicate initargs.
3197 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3198 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3199 propagated even in the ctor optimized implementation of
3201 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3203 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3204 * a better implementation of SXHASH on (simple) bit vectors,
3205 measured both in execution speed and in distribution of results
3206 over the positive fixnums, has been installed. Likewise, a better
3207 implementation of EQUAL for simple bit vectors is now available.
3208 * fixed CEILING optimization for a divisor of form 2^k.
3209 * fixed bug 240 (emitting extra style warnings "using the lexical
3210 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3211 by Antonio Martinez)
3212 * fixed SXHASH, giving different results for NIL depending on type
3213 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3214 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3215 variable. (found by Rolf Wester)
3216 * SBCL does not ignore type declarations for special
3217 variables. (reported by rif on c.l.l 2003-03-05)
3218 * some bug fixes in contrib/sb-aclrepl/
3219 * fixed some bugs revealed by Paul Dietz' test suite:
3220 ** a bug in the CONS type specifier, whereby the CAR and CDR
3221 types got intertwined, has been fixed;
3222 ** the type system is now able to reason about the interaction
3223 between INTEGER and RATIO types more completely;
3224 ** APPEND, [N]REVERSE and NRECONC check that those their
3225 arguments, which must be proper lists, are really so;
3226 ** An array specialized to be unable to hold elements has been
3227 implemented, as required -- yes, really -- by ANSI;
3228 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3229 on malformed property lists;
3231 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3232 * incompatible packaging change: in line with Unix convention,
3233 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3234 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3235 * REQUIRE and PROVIDE are now optionally capable of doing something
3236 useful. See the documentation string for REQUIRE.
3237 * infrastructure for a managed SBCL contrib system: contributed
3238 modules in this release include:
3239 ** the ASDF system definition facility;
3240 ** an interface to the BSD Sockets API;
3241 ** an ACL-like convenience interface to the repl;
3242 (thanks to Kevin Rosenberg)
3243 ** an implementation of ROTATE-BYTE, with efficient implementation
3245 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3246 gives a closer approximation to the right answer than previously.
3247 (thanks to Raymond Toy)
3248 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3249 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3250 optional environment arguments, as required by ANSI.
3251 * fixed bugs in other functions taking environment objects, allowing
3252 calls with an explicit NIL environment argument to be compiled
3254 * fixed bug 228: primary return values from
3255 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3256 COMPILE or FUNCTION.
3257 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3258 :TYPE will no longer signal errors on innocuous objects.
3259 * fixed bug 231b: SETQ is better at respecting type declarations in
3260 the lexical environment.
3261 * fixed a bug in DEFCLASS: classes named by symbols with no or
3262 unprintable packages can now be defined.
3263 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3264 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3265 * fixed compiler failure related to checking types of functions.
3266 (reported by Robert E. Brown)
3267 * the compiler is now much more consistent in its error-checking
3268 treatment of bounding index arguments to sequence functions: in
3269 (SAFETY 3) code, errors will be signalled in almost all cases if
3270 invalid sequence bounding indices are passed to functions defined
3271 by ANSI to operate on sequences.
3272 * fixed a bug in the build procedure: documentation of SBCL-specific
3273 packages is now preserved and available in the final Lisp image.
3274 * lifted FDEFINITION lookup out of loops in the implementation of
3275 many list operations. (thanks to Robert E. Brown)
3276 * fixed a bug in the reader: the #n# reader macro now works for
3277 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3278 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3279 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3280 E. Brown and Tony Martinez respectively)
3281 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3282 SLOT-BOUNDP: in particular, it is now able to optimize them much
3283 better, and is now not vulnerable to having packages renamed.
3284 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3285 always called when a slot is not present in an instance. (thanks
3287 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3288 longer signal an error on unreading a character following EOF on
3289 the previous constituent stream. (thanks to Tony Martinez)
3290 * fixed some bugs revealed by Paul Dietz' test suite:
3291 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3292 not just nonnegative fixnums;
3293 ** the logical bit-array operators such as BIT-AND now accept an
3294 explicit NIL for their "opt-arg" argument (to indicate a
3295 freshly-consed result bit-array);
3296 ** ELT now signals an error on an invalid sequence index in safe
3298 ** the type system is now cleverer about negations of numeric
3299 types, and consequently understands the BIGNUM and RATIO types
3301 ** the type system is now cleverer about the interaction between
3302 INTEGER and RATIO types: while bugs still remain, many more
3303 cases are accurately computed;
3304 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3305 if it is in the last clause;
3306 ** CONSTANTLY now correctly returns a side-effect-free function in
3308 ** DECLARE is no longer treated as a special-operator; in
3309 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3310 * incremented fasl file version number due to the change in the
3311 DEFSTRUCT-SLOT-DESCRIPTION structure.
3313 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3314 * minor incompatible change: code processed by the "interpreter" or
3315 EVAL now has a compilation optimization policy of (DEBUG 2)
3316 (changed from (DEBUG 1)) to improve debuggability of interactive
3317 development, and to allow the use of the debug RETURN command in
3319 * an experimental implementation of the RETURN command for the
3320 debugger has been included. (thanks to Frederik Kuivinen)
3321 * fixed bug 62: constraints were not propagated into a loop.
3322 * fixed bug in embedded calls of SORT (reported and investigated by
3324 * fixed some bugs revealed by Paul Dietz' test suite:
3325 ** printing and reading of arrays with some dimensions having
3326 length 0 (thanks to Gerd Moellmann);
3327 ** BOA constructor with &AUX argument without a default value does
3328 not cause a type error;
3329 ** CONSTANTP now returns true for all self-evaluating objects.
3331 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3332 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3333 accessors that are related by inheritance, as specified in the
3334 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3335 Valtteri Vuorikoski)
3336 * The compiler is now able to inline functions that were defined in
3337 a complex lexical environment (e.g. inside a MACROLET).
3338 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3340 * Support for the upcoming FreeBSD-5.0 release has been included.
3341 (thanks to Dag-Erling Smorgrav)
3342 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3343 effect when it is not in a toplevel context.
3344 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3345 stand a better chance of being correct. (thanks to Gerd
3347 * fixed bug in COERCE, which now signals an error on coercing a
3348 rational to a bounded real type which excludes the expected
3350 * The compiler is now able to derive types more accurately from the
3351 COERCE and COMPILE functions.
3352 * fixed bug 223: functional binding is considered to be constant
3353 only for symbols in the CL package.
3354 * fixed bug 231: SETQ did not check the type of a variable being set
3355 (reported by Robert E. Brown)
3356 * A new optimization for MAKE-INSTANCE has been included, fixing
3357 various bugs (including relating to :ALLOCATION :CLASS slots and
3358 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3360 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3361 ** As required by ANSI, LOOP now disallows anonymous collection
3362 clauses such as COLLECT I in conjunction with aggregate boolean
3363 clauses such as THEREIS (= I 1);
3364 ** LOOP now signals an error when any variable is reused in the
3365 same loop (including the potentially useful construct analogous
3366 to WITH A = 1 WITH A = (1+ A);
3367 ** IT is only a special loop symbol within the first clause of a
3368 conditional loop clause;
3369 ** LOOP with a typed iteration variable over a hashtable now
3370 signals a type error iff it should.
3371 * fixed some other bugs revealed by Paul Dietz' test suite:
3372 ** FILE-STREAM now names the class previously known as FD-STREAM;
3373 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3374 argument) no longer signals an error;
3375 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3376 of the slot symbol, rather than using the current package
3377 ((:CONC-NAME "") continues to intern the slot's name in the
3379 * incremented fasl file version number, because of the incompatible
3380 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3381 of the new implementation of DEFINE-COMPILER-MACRO.
3383 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3384 * Support for building SBCL for MIPS platforms running in
3385 little-endian mode has now been checked in, and basic
3386 functionality on said platforms verified.
3387 * minor incompatible change: PCL now records the pathname of a file
3388 in which methods and the like are defined, rather than its
3390 * minor incompatible change: TRUENAME now considers the truename of
3391 a file naming a directory to be the pathname with :DIRECTORY
3392 component indicating that directory.
3393 * minor incompatible change: a NAMED clause in the extended form of
3394 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3395 reason for the previous behaviour is unclear.
3396 * more systematization and improvement of CLOS and MOP conformance
3397 in PCL (thanks to Gerd Moellman and Pierre Mai):
3398 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3400 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3401 primary methods with no specializers;
3402 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3404 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3405 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3406 FORWARD-REFERENCED-CLASSes; error reporting on
3407 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3409 ** SXHASH on CLOS instances now uses a slot internal to the
3410 instance to return different numbers on distinct instances,
3411 while preserving the same return value through invocations of
3413 ** DEFMETHOD signals errors when methods with longer incongruent
3414 lambda lists are added to generic functions;
3415 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3416 CLASS, as specified in AMOP;
3417 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3418 on the order returned by the primary method for classes of
3419 class STANDARD-CLASS;
3420 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3421 * fixed some bugs shown by Paul Dietz' test suite:
3422 ** DOLIST puts its body in TAGBODY;
3423 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3425 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3426 value producing form;
3427 ** if more variables are given to PROGV than values, extra
3428 variables are bound and made to have no value;
3429 ** NSUBSTITUTE on list arguments gets the right answer with
3431 ** ELT signals an error of type TYPE-ERROR when the index argument
3432 is not a valid sequence index;
3433 ** LOOP signals (at macroexpansion time) an error of type
3434 PROGRAM-ERROR when duplicate variable names are found;
3435 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3436 ** FUNCALL of special-operators now cause an error of type
3438 ** PSETQ now works as required in the presence of side-effecting
3439 symbol-macro places;
3440 ** NCONC accepts any object as its last argument;
3441 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3443 ** loop-for-as-package does not require a package to be explicitely
3445 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3447 * fixed bug 166: compiler preserves "there is a way to go"
3448 invariant when deleting code.
3449 * fixed bug 172: macro lambda lists with required arguments after
3450 &REST arguments now cause an error to be signalled. (thanks to
3452 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3453 bug. (thanks to Matthew Danish)
3454 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3456 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3457 arguments are lexically rebound. (thanks to Gerd Moellmann and
3459 * fixed bug 194: error messages are now more informative when there
3460 is no primary method applicable in a call to a generic
3461 function. (thanks to Gerd Moellmann)
3462 * fixed bug in command line argument checking (thanks to Julian
3464 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3465 (thanks to Matthew Danish)
3466 * incremented fasl file version number, because of the
3467 SXHASH-related changes in the layout of CLOS data structures
3469 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3470 * minor incompatible change: The runtime (the Unix executable named
3471 "sbcl") is now much pickier about the .core files it will load.
3472 Essentially it now requires .core files to descend from the same
3473 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3474 as the runtime does. (The intent is to prevent the crashes which
3475 can occur, and which can even be reported as mysterious failures,
3476 when people patch the sources or change the build parameters
3477 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3478 sbcl and .core files.)
3479 * fixed bug: VALUES-LIST is no longer optimized away.
3480 * fixed bug 142: The FFI conversion of C string values to Lisp
3481 string values no longer conses excessively. (thanks to Nathan
3482 Froyd porting Raymond Toy's fix to CMU CL)
3483 * began to systematize and improve MOP conformance in PCL (thanks to
3484 Nathan Froyd, Gerd Moellman and Pierre Mai):
3485 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3487 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3488 ** FINALIZE-INHERITANCE is now called on class finalization;
3489 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3490 argument precedence order.
3491 * fixed bug 202: The compiler no longer fails on functions whose
3492 derived types contradict their declared type.
3493 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3494 so it can be non-toplevel.
3495 * The fasl file version number has changed (because of the new
3496 implementation of DEFMACRO).
3497 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3498 safe code, that any length requirement by their type-specifier
3499 argument is valid. The exceptions to this are described in bug
3501 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3502 functions now check the types of their inputs as required by ANSI.
3503 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3504 introduced symbol is DECLAREd to be SPECIAL.
3505 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3506 * fixed bug: :COUNT argument to sequence functions may be negative
3507 * fixed bug: body of DO-SYMBOLS may contain declarations
3508 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3509 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3510 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3511 symbol macro only once
3512 * fixed printing of call frame when argument list is unavailable
3513 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3514 * compiler no longer signals WARNING on unknown keyword
3517 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3518 * A beta-quality port to the mips architecture running Linux,
3519 based on the old CMUCL backend, has been made. It has been tested
3520 on a big-endian kernel, and works sufficiently well to be able to
3521 rebuild itself; it has not been tested in little-endian mode.
3522 * fixed an inconsistency between gencgc.c and purify.c which made
3523 dumping/loading .core files unreliable
3524 * fixed bug 120a: The compiler now deals correctly with IFs where
3525 the consequent is the same as the alternative, instead of
3526 misderiving the return type. (thanks to Alexey Dejneka)
3527 * fixed bug 113: Logical pathnames are now dumpable (the logical
3528 host is resolved at load-time, throwing an error if it is not
3530 * fixed bug 174: FORMAT's error message is slightly clearer when a
3531 non-printing character is used in a format directive.
3532 * fixed several bugs in compiler checking of type declarations, i.e.
3533 violations of the Python "declarations are assertions" principle
3534 (thanks to Alexey Dejneka)
3535 * fixed several bugs in PCL's error checking (thanks to Gerd
3537 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3539 * fixed bug in compilation of functions as first class values
3540 (thanks to Antonio Martinez-Shotton)
3541 * The compiler's handling TYPE-ERRORs which it can prove will
3542 inevitably happen at runtime has been cleaned up and corrected
3543 in several ways. (thanks to Alexey Dejneka)
3544 * improved argument type checking for various basic arithmetic
3545 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3546 lavished on them in the past that they can be compiled in many
3547 ways in different special cases
3548 * fixed bug 181: compiler checks validity of user supplied type
3550 * cleaned up code flushing in optimization: Function calls which
3551 should signal errors for safety purposes (e.g. which ANSI says
3552 should signal errors when their arguments are of incorrect type)
3553 are no longer optimized away.
3554 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3555 * incremented fasl file version number, because changes in the
3556 implementation of sequence functions like COERCE caused
3557 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3558 in old inline expansions) to become undefined. (Actually these
3559 changes were later undone, so we might very well be binary
3560 compatible with 0.7.7 after all, but leaving the version number
3561 incremented seemed like the simplest and most conservative
3564 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3565 * An alpha-quality port to the parisc architecture running Linux,
3566 based on the old CMUCL backend, has been made. This, even more so
3567 than the other backends, should be considered still a work in
3568 progress; known problems include that the Linux kernel in 64-bit
3569 mode does not propagate the correct sigcontext structure to
3570 userspace, and consequently SBCL on a parisc64 kernel will not
3572 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3573 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3574 suppressed.) Also now that the compiler is looking at declarations
3575 in the environment, it checks optimization declarations as well,
3576 and suppresses inlining when (> DEBUG SPEED).
3577 * More fixes have been made to treatment of floating point exception
3578 treatment and other Unix signals. In particular, floating point
3579 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3580 * The detection and handling of control stack exhaustion (infinite
3581 or very deeply nested recursion) has changed. Stack exhaustion
3582 detection is now done by write-protecting pages at the OS level
3583 and applies at all optimization settings; when found, a
3584 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3585 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3586 be caught using IGNORE-ERRORS.
3587 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3588 correctly when transforming recursive LABELS functions to LETs.
3589 (thanks to Alexey Dejneka)
3590 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3591 that are names of constants or global variables.
3592 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3593 alien routines with docstrings.
3594 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3595 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3597 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3598 (thanks to Eric Marsden)
3599 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3600 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3601 * Bug fix: The SPARC backend can now compile functions involving
3602 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3603 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3604 to a functions expecting a PATHNAME-DESIGNATOR.
3605 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3606 lambda lists. (thanks to Alexey Dejneka)
3607 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3608 functions so that EXPORTing the name of the class doesn't cause
3609 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3610 to Antonio Martinez for reporting this)
3611 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3612 constant-folded without failing an assertion. (thanks to Einar
3613 Floystad Dorum for reporting this)
3614 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3615 types are now dealt with more correctly by the compiler.
3616 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3617 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3618 behaviour analogous to RENAME-FILE. This puts its behaviour more
3619 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3621 * The fasl file version number has changed again. (because of the
3622 bug fix involving the names of PCL MAKE-INSTANCE functions)
3624 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3625 * bug fix: Floating point exceptions are treated much more
3626 consistently on the x86/Linux and PPC/Linux platforms.
3627 * Array initialization with :INITIAL-ELEMENT is now much faster for
3628 cases when the compiler cannot open code the array creation, but
3629 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3630 array accesses have also seen a speed increase.
3631 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3632 specified. (This was at the root of some bad interactions between
3633 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3635 * bug fix: Internal error arguments for undefined functions are now
3636 computed correctly on the PPC/Linux platform.
3637 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3638 Raymond Toy's patch for CMU CL)
3639 * Support for the Solaris 9 operating environment has been included
3640 (thanks to Daniel Merritt)
3641 * A very ugly but hopefully complete draft of the missing FFI chapter
3642 of the manual has been created by reformatting the corresponding
3643 CMU CL manual chapter into (currently very ugly and incoherent)
3644 DocBook and bringing it up to date for SBCL behavior. Thus, the
3645 manual is now essentially complete, at least by my extreme
3646 once-and-only-once standards, whereby it's acceptable to refer to
3647 the doc strings of SB-EXT functions as the primary documentation.
3648 * The fasl file version number has changed again, due to cleanup of
3649 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3650 is no longer a static symbol.)
3652 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3653 * SBCL now builds with OpenMCL (version 0.12) as the
3654 cross-compilation host; also, more progress has been made toward
3655 bootstrapping under CLISP.
3656 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3658 * bug 158 fixed: The compiler can now deal with integer loop
3659 increments different from 1; fixing this turned out also to fix
3661 * bug 169 fixed: no more bogus warnings about using lexical bindings
3662 despite the presence of perfectly good SPECIAL declarations (thanks
3663 to David Lichteblau)
3664 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3665 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3666 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3668 * bug 180 fixed: Method combination specifications no longer ignore
3669 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3670 * bug fix: Structure type predicate functions now check their argument
3671 count as they should.
3672 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3673 correctly. (thanks to Pierre Mai)
3674 * minor incompatible change: The --noprogrammer option is deprecated
3675 in favor of the new --disable-debugger option, which is very similar.
3676 (The major difference is that it takes effect at a slightly different
3677 time at startup, causing handling of errors in --sysinit and
3678 --userinit files will be affected differently.) The
3679 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3680 been added to allow this functionality to be controlled from ordinary
3681 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3682 maintainers, who might want to run non-interactive scripts to
3683 build SBCL cores which will later be used interactively.)
3684 * minor incompatible change: The LOAD function no longer, when given
3685 a wild pathname to load, loads all files matching that pathname.
3686 Instead, an error of type FILE-ERROR is signalled.
3688 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3689 * bug 147 fixed: The compiler preserves its block link/count
3690 invariants more correctly now so that it doesn't crash. (thanks
3692 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3694 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3695 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3696 no-value-for-this-character as NIL (as specified by ANSI).
3697 * HOST-NAMESTRING on physical pathnames now returns a string that is
3698 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3699 (thanks to Christophe Rhodes)
3700 * The Alpha port handles icache flushing more correctly. (thanks to
3702 * More progress has been made toward bootstrapping under CLISP. (thanks
3703 to Christophe Rhodes)
3704 * The fasl file format has changed again, because dynamic loading
3705 on OpenBSD (which has non-ELF object files) motivated some cleanups
3706 in the way that foreign symbols are transformed and passed around.
3707 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3708 is no longer treated as whitespace by the reader, but instead as
3709 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3710 |A
\7fB|, instead of A as it used to.
3712 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3713 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3714 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3715 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3716 this as of 0.7.1.45, but was left out of the previous news section
3717 (thanks to Dan Barlow)
3718 * SBCL now runs on the Solaris operating system on SPARC architectures
3719 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3720 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3721 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3722 * SPARC backend cleanups, allowing builds of cores optimized for V8
3723 and V9 SPARCS, and also emission of code targeted to a particular
3724 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3726 * SBCL is closer to bootstrapping under CLISP, thanks to various
3727 fixes by Christophe Rhodes.
3728 * The fasl file format has changed again, to allow the compiler's
3729 INFO database to support symbol macros.
3730 * The user manual (in doc/) is formatted into HTML more nicely.
3731 (thanks to coreythomas)
3732 * The system is smarter about SUBTYPEP relationships, especially
3733 those involving NOT types (including types such as ATOM which are
3734 represented internally using NOT types). Thus SUBTYPEP is less
3735 likely to return (VALUES NIL NIL) in general, and in particular
3736 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3738 * The fasl file format has changed again, because the internal
3739 representation of types now includes a new slot to support the new
3740 SUBTYPEP-of-NOT-types logic.
3741 * (not a change in the main branch of SBCL, but a related prototype
3742 which can hopefully be merged into the main branch of SBCL in the
3743 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3744 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3745 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3746 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3747 dispatch macro character combinations. (thanks to Alexey Dejneka)
3748 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3749 that superficially look like logical namestrings correctly.
3750 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3752 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3753 default supplied-p) for &optional and &key arguments. (thanks to
3756 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3757 * incompatible change: The compiler is now less aggressive about
3758 tail call optimization, doing it only when (> SPACE DEBUG) or
3759 (> SPEED DEBUG). (This is an incompatible change because there are
3760 programs which relied on the old CMU-CL-style behavior to optimize
3761 away their unbounded recursion which will now die of stack overflow.)
3762 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3763 for non-GENCGC systems has been increased to 20M (since that
3764 seems much closer to the likely performance optimum for modern
3765 systems than the old 4M value was)
3766 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3767 and *DEBUG-PRINT-LEVEL*
3768 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3769 of CMU CL's support for SPARC, and various endianness and other
3770 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3771 * new syntactic sugar for the Unix command line: --load foo.bar is now
3772 an alternate notation for --eval '(load "foo.bar")'.
3774 ** The system now detects stack overflow and handles it gracefully,
3775 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3776 optimization settings. (This is a good thing in general, and
3777 its introduction in this version should be particularly timely
3778 for anyone whose code fails because of suppression of tail
3780 ** The system now hunts for the C variable "environ" in a more
3781 devious way, to avoid segfaults when the C library version differs
3782 between compile time and run time. (thanks to Christophe Rhodes)
3783 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3784 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3785 ** The compiler no longer issues bogus style warnings for undefined
3786 classes in the same source file as the DEFCLASSes which defined
3787 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3789 ** fixes in CONDITION class precedence list for undefined function
3790 errors (thanks to Alexei Dejneka)
3791 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3792 correctly. (thanks to Dan Barlow)
3793 ** portability fixes aiming at bootstrapping under CLISP (thanks
3794 to Dave McDonald and Christophe Rhodes)
3795 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3796 ** fixes in type translation and and type inference (thanks to
3798 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3799 ** various fixes in the new ports (thanks to Dan Barlow)
3800 * several changes related to debugging:
3801 ** suppression of tail recursion, as noted above
3802 ** stack overflow detection, as noted above
3803 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3804 is now the default. (For some time encapsulation has been more
3805 reliable than the breakpoint-based :ENCAPSULATE NIL
3806 implementation, at least on X86 systems; and I just noticed that
3807 encapsulation also seems closer to the spirit of the ANSI
3810 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3812 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3813 up properly again. (There was a packaging bug in 0.7.0 which
3814 left their definitions in SB-SYS::LOAD-FOREIGN and
3815 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3816 vital for most things which interface to C-level interfaces,
3817 like extensions working with sockets or databases or
3818 Perl-compatible regexes or whatever, and the need to fix
3819 this bug is the main reason that 0.7.1 was released so
3821 ** DEFGENERIC is now choosier about the methods it redefines, so that
3822 reLOADing a previously-LOADed file containing DEFGENERICs does
3823 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3824 cycle now works as it should. (thanks to Alexey Dejneka)
3825 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3826 fixed by Christophe Rhodes. (He actually submitted this patch
3827 months ago, and I delayed until after 0.7.0.)
3828 ** Bug 111 (internal compiler confusion about runtime checks on
3829 FUNCTION types) was fixed by Alexey Dejneka.
3830 * Some internal cleanups (getting rid of variables which aren't
3831 needed now that the byte interpreter is gone) caused the fasl
3832 file format number to change again.
3834 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3835 * major incompatible change: The default fasl file extension, i.e. the
3836 default extension for files produced by COMPILE-FILE, has changed
3837 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3839 ** There are many changes in the implementation of the compiler.
3840 SBCL is now essentially a compiler-only implementation of ANSI
3841 Common Lisp. EVAL still "interprets" a few special cases, but
3842 almost all the interesting cases are handled by creating
3843 a LAMBDA expression, calling COMPILE on it, then calling
3844 FUNCALL on the result.
3845 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3846 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3847 Since the code is newer, there might still be some new bugs
3848 (though not as many as before Martin Atzmueller's fixes:-). But
3849 the new code is substantially simpler and clearer, and hopefully
3850 any remaining bugs will be simpler, less fundamental, and more
3851 fixable then the bugs in the old code.
3852 ** The revised compiler is still a little unsteady on its feet.
3854 *** The debugging information it produces (particularly the names
3855 of FUNCTION objects) is sometimes much less useful than what
3856 the old compiler produced.
3857 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3858 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3859 within a MACROLET) has been temporarily weakened.
3860 ** There are new compiler optimizations for various functions:
3861 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3862 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3863 *** the math functions TRUNCATE, FLOOR, and CEILING
3864 *** the function-of-all-trades COERCE
3865 Mostly these should be transparent, but there's one
3866 potentially-annoying problem (bug 117): when the compiler
3867 inline-expands a function and does type analysis on the result,
3868 it can create control paths which have type mismatches, and
3869 when it can't prove that those control paths aren't taken,
3870 it will issue WARNINGs about the type mismatches. This is
3871 a particular problem in practice for the new sequence functions.
3872 It's not clear how this should be fixed, and for now, a
3873 workaround is given in the entry for 117 in the BUGS file.
3874 ** (Because of the interaction between the two previous items --
3875 occasional inlining problems and new inline expansions -- some
3876 of the new sequence function optimizations won't really kick in
3877 completely until debugging information, and then inlining, are
3878 straightened out in some future version.)
3879 * minor incompatible changes:
3880 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3881 DIRECTORY no longer implicitly promotes NIL slots of its
3882 pathname argument to :WILD. In particular, when you ask for the
3883 contents of a directory (which you used to be able to do without
3884 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3885 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3886 ** changes in behavior that ANSI explicitly defines to be
3887 implementation dependent:
3888 *** The new compiler-only implementation still conforms with ANSI,
3889 but acts a little different than before. Besides the obvious
3890 changes in performance tradeoffs (that the cost per form passed
3891 to EVAL has gone up, and the cost per form executed by EVAL
3892 has gone down), the behavior of the system changes a little
3893 because there are no longer any interpreted function objects.
3894 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3895 e.g. doing COMPILE on the output of interactive DEFUN is
3897 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3899 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3900 has changed from (:CL) to NIL.
3901 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3902 e.g. #\Newline and #\Tab, don't have names specified in the
3903 ANSI Common Lisp standard, is now based on their ASCII symbolic
3904 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3905 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3906 are no longer used for output.
3907 ** changes in internal implementation constants:
3908 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3909 4 million. (If your application spends a lot of time GCing and
3910 you have a lot of RAM, you might want to experiment with
3911 increasing it even more.)
3912 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3913 However, almost all old code should still continue to work without
3914 immediate update, as SB-C-CALL is now a (deprecated) nickname
3916 ** Old operator names in the style DEF-FOO are now deprecated in
3917 favor of new corresponding names DEFINE-FOO, for consistency with
3918 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3919 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3920 internal symbols, but a few supported extensions like
3921 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3922 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3923 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3924 etc. as you get deeper into recursive calls to the debugger
3925 command loop, instead of the old "5]", "5]]", "5]]]"
3926 sequence. (I was motivated to do this when squabbles between
3927 ILISP and SBCL left me very deeply nested in the debugger. In the
3928 short term, this change will probably provoke more ILISP/SBCL
3929 squabbles, but hopefully it will be an improvement in the long run.)
3930 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3931 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3932 function, since declaiming return types involving aliens is
3933 (1) annoyingly messy to do by hand and (2) vital to efficient
3934 compilation of code which calls such functions.
3935 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3936 longer reexported by the SB-EXT package. They're solely useful
3937 for alien code, so it seems more logical that you should get
3938 them from the SB-ALIEN package, not in SB-EXT.
3939 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3940 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3941 features. Instead, the code that they used to control is always
3942 built into the system.
3943 * many other bug fixes
3944 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3945 advantage of the new EVAL-WHEN stuff and to clean them up in
3946 general, and they are now more ANSI-compliant in a number of
3947 ways. Martin Atzmueller is responsible for a lot of this.
3948 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3950 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3951 *** correct ERROR type for various file operations
3952 *** some fixes for Lisp streams
3953 *** DEFMETHOD syntax checking
3954 *** changing old weird representation of debug information as
3955 strings (which, among their other deficiencies, don't transform
3956 correctly when you rename packages, and don't change their
3957 print representation when you change things like *PACKAGE*
3958 and *PRINT-LENGTH*) to symbols and lists of symbols
3959 He also made several improvements and fixed several bugs in DESCRIBE.
3960 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3962 *** misbehavior of WRITE-STRING/WRITE-LINE
3963 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3964 and several other LOOP problems as well
3965 *** DIRECTORY when similar filenames are present
3966 *** DEFGENERIC with :METHOD options
3967 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3968 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3969 *** argument ordering in FIND with :TEST option
3970 *** mishandled package designator argument in APROPOS-LIST
3971 *** various problems in the backquote readmacro
3972 *** a bug in APROPOS
3973 *** probably some others that I'm not describing very well here,
3974 since the CVS log documents them by reference to sbcl-devel
3975 messages, and the SourceForge archives aren't working well.:-(
3976 ** Dan Barlow improved the Alpha port (and is making progress on the
3977 PPC port, for those of you who think different).
3978 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3979 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3980 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3981 continues to work on the SPARC port (for those of you in a position
3982 to look down upon our little PC-compatible boxes from a great height).
3983 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3984 and uses that copy, rather than the current dynamic value, when
3985 it's trying to decide whether to truncate output. Thus e.g.
3986 (let ((*print-lines* 50))
3987 (pprint-logical-block (stream nil)
3989 (let ((*print-lines* 8))
3990 (print (aref possiblybigthings i) stream)))))
3991 should now truncate the logical block only at 50 lines, instead of
3992 often truncating it at 8 lines, as it did before.
3993 * The doc/cmucl/ directory, containing old CMU CL documentation
3994 from the time of the fork, is no longer part of the base system.
3995 SourceForge has shut down its anonymous FTP service, and with it
3996 my original plan for distributing the old CMU CL documentation
3997 there. For now, if you need these files you can download an old
3998 SBCL source release and extract them from it.
3999 * The fasl file version number changed again, for dozens of reasons,
4000 some of which are apparent above.
4002 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4003 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4004 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4005 patch, so that DISASSEMBLE output is much nicer.
4006 * The code in the SB-PROFILE package now seems reasonably stable.
4007 I still haven't decided what the final interface should look like
4008 (I'd like PROFILE to interact cleanly with TRACE, since both
4009 facilities use function encapsulation) but if you have a need
4010 for profiling now, you can probably use it successfully with
4011 the current CMU-CL-style interface.
4012 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4013 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4014 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4015 link now returns the dangling link itself, and for similar
4016 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4017 link itself. (In these cases the old code signalled an error and
4018 looped endlessly, respectively.) Thus, DIRECTORY now works even
4019 in the presence of dangling and cyclic symbolic links.
4020 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4021 is now a supported extension again, since the consensus on
4022 sbcl-devel was that it can be useful for ordinary development
4023 work, not just for debugging SBCL itself.
4024 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4025 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4026 that currently-defined functions might be redefined later with
4027 different return types.
4028 * Hash tables can be printed readably, as inspired by CMU CL code
4029 of Eric Marsden and SBCL code of Martin Atzmueller.
4030 * better error handling in CLOS method combination, thanks to
4031 Martin Atzmueller porting Pierre Mai's CMU CL patches
4032 * more overflow fixes for >16Mbyte I/O buffers
4033 * A bug in READ has been fixed, so that now a single Ctrl-D
4034 character suffices to cause end-of-file on character streams.
4035 In particular, now you only need one Ctrl-D at the command
4036 line (not two) to exit SBCL.
4037 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4039 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4040 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4041 does the right thing.
4042 * The implementation of some type tests, especially for CONDITION
4043 types, is now tidier and maybe faster, due to CMU CL code
4044 originally by Douglas Crosher, ported by Martin Atzmueller.
4045 * Some math functions have been fixed, and there are new
4046 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4047 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4048 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4049 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4050 likely to be built into sbcl-0.7.0.)
4051 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4052 CL, ported to SBCL by Martin Atzmueller)
4053 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4054 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4055 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4056 * There's a new slam.sh hack to shorten the edit/compile/debug
4057 cycle for low-level changes to SBCL itself, and a new
4058 :SB-AFTER-XC-CORE target feature to control the generation of
4059 the after-xc.core file needed by slam.sh.
4060 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4061 COMPILE-FILE is no longer supported, so that now every function
4062 gets an entry point, so that block compilation looks a little
4063 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4064 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4065 returns the number of bytes consed since the system started,
4066 rather than the number consed since the first time the function
4067 was called. (The new definition parallels ANSI functions like
4068 CL:GET-INTERNAL-RUN-TIME.)
4069 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4070 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4071 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4072 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4073 * Fasl file version numbers are now independent of the target CPU,
4074 since historically most system changes which required version
4075 number changes have affected all CPUs equally. Similarly,
4076 the byte fasl file version is now equal to the ordinary
4079 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4080 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4081 is no longer recognized. I apologize for this, because it was
4082 listed in SB-EXT as a supported extension, but I found that
4083 its existing behavior was poorly specified, as well as incorrectly
4084 specified, and it looked like too much of a mess to straighten it
4085 out. I have enough on my hands trying to get ANSI stuff to work..
4086 * many patches ported from CMU CL by Martin Atzmueller, with
4087 half a dozen bug fixes in pretty-printing and the debugger, and
4088 half a dozen others elsewhere
4089 * fixed bug 13: Floating point infinities are now supported again.
4090 They might still be a little bit flaky, but thanks to bug reports
4091 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4092 as flaky as they were.
4093 * The --noprogrammer command line option is now supported. (Its
4094 behavior is slightly different in detail from what the old man
4095 page claimed it would do, but it's still appropriate under the
4096 same circumstances that the man page talks about.)
4097 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4098 are now supported, and enabled by default. Thus, the compiler can
4099 handle many floating point and complex operations much less
4100 inefficiently. (Thus e.g. you can implement a complex FFT
4102 * The compiler now detects type mismatches between DECLAIM FTYPE
4103 and DEFUN better, and implements CHECK-TYPE more correctly, and
4104 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4105 thanks to patches from Martin Atzmueller.
4106 * various fixes to make the cross-compiler more portable to
4107 ANSI-conforming-but-different cross-compilation hosts (notably
4108 Lispworks for Windows, following bug reports from Arthur Lemmens)
4109 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4110 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4111 fixed thanks to Pierre Mai's CMU CL patches.
4112 * fixes to keep the system from overflowing internal counters when
4113 it tries to use i/o buffers larger than 16M bytes
4114 * fixed bug 45a: Various internal functions required to support
4115 complex special functions have been merged from CMU CL sources.
4116 (When I was first setting up SBCL, I misunderstood a compile-time
4117 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4118 * improved support for type intersection and union, fixing bug 12
4119 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4120 more obscure bugs as well
4121 * some steps toward byte-compiling non-performance-critical
4122 parts of the system, courtesy of patches from Martin Atzmueller
4123 * Christophe Rhodes has made some debian packages of sbcl at
4124 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4125 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4126 stable, but are nonetheless usable. When he's ready, I'd be happy
4127 to add them to the SourceForge "File Releases" section. (And if
4128 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4129 * new fasl file format version number (because of changes in
4130 internal representation of (OR ..) types to accommodate the new
4131 support for (AND ..) types, among other things)
4133 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4134 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4136 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4138 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4139 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4140 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4141 * DESCRIBE now works on structure objects again.
4142 * Most function call argument type mismatches are now handled as
4143 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4144 know whether the function will be redefined before the call is
4145 executed. (The compiler could flag local calls with full WARNINGs,
4146 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4147 it doesn't keep track of enough information to know whether calls
4148 are local in this sense.)
4149 * Compiler output is now more verbose, with messages truncated
4150 later than before. (There should be some supported way for users
4151 to override the default verbosity, but I haven't decided how to
4152 provide it yet, so this behavior is still controlled by the internal
4153 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4154 src/compiler/ir1util.lisp.)
4155 * Fasl file format version numbers have increased again, because
4156 support for the Gray streams extension changes the layout of the
4157 system's STREAM objects.
4158 * The Gray subclassable streams extension now works, thanks to a
4159 patch from Martin Atzmueller.
4160 * The full LOAD-FOREIGN extension (not just the primitive
4161 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4162 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4163 but like most other programs, it defaults to copying the Unix
4164 environment from the original process instead of starting the
4165 new process in an empty environment.
4166 * Extensions which manipulate the Unix environment now support
4167 an :ENVIRONMENT keyword option which doesn't smash case or
4168 do other bad things. The CMU-CL-style :ENV option is retained
4169 for porting convenience.
4170 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4171 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4173 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4175 * A patch from Martin Atzmueller seems to have solved the SIGINT
4176 problem, and as far as we know, signal-handling now works cleanly.
4177 (If you find any new bugs, please report them!)
4178 * The system no longer defaults Lisp source file names to types
4179 ".l", ".cl", or ".lsp", but only to ".lisp".
4180 * The compiler no longer uses special default file extensions for
4181 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4182 seems to expect a single default extension for all compiled code,
4183 and there's no compelling reason to try to stretch the standard
4184 to allow two different extensions.) Instead, byte-compiled files
4185 default to the same extension as native-compiled files.
4186 * Fasl file format version numbers have increased again, because
4187 a rearrangement of internal implementation packages made some
4188 dumped symbols in old fasl files unreadable in new cores.
4189 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4190 many fewer weird special cases.
4191 * Bug #17 (differing COMPILE-FILE behavior between logical and
4192 physical pathnames) has been fixed, and some related misbehavior too,
4193 thanks to a patch from Martin Atzmueller.
4194 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4195 by Tim Moore, ported to SBCL by Martin Atzmueller.
4196 * Martin Atzmueller fixed several filesystem-related problems,
4197 including bug #36, in part by porting CMU CL patches, which were
4198 written in part by Paul Werkowski.
4199 * More compiler warnings in src/runtime/ are gone, thanks to
4200 more patches from Martin Atzmueller.
4201 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4204 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4206 * DESCRIBE now works on CONDITION objects.
4207 * The debugger now handles errors which arise when trying to print
4208 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4210 * The build system now uses an additional file, customize-target-features.lisp,
4211 to allow local modifications to the target *FEATURES* list. (The point of
4212 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4213 debugging features enabled, without having to worry about propagating it
4214 into everyone's system when I do a "cvs update".) When no
4215 customize-target-features.lisp file exists, the target *FEATURES* list
4216 should be constructed the same way as before.
4217 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4218 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4219 warn about behavior which is undefined under the ANSI spec. Note: This
4220 is specified by ANSI, but it's not very popular with programmers.
4221 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4222 macro in the SBCL sources for an example of a workaround which you
4223 might use to make portable ANSI-standard code which does what you want.
4224 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4225 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4226 when it's used as a non-toplevel form. (This is required in order
4227 to implement the DEFCONSTANT-EQX macro.)
4228 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4229 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4230 funny things, due to bugs in the implementation of EVAL-WHEN
4231 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4232 non-toplevel DEFCONSTANTs are uncommon.)
4233 * The core file version number and fasl file version number have been
4234 incremented, because the old noncompliant DEFCONSTANT behavior involved
4235 calling functions which no longer exist, and because I also took the
4236 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4237 * fixed bug 1 (error handling before read-eval-print loop starts), and
4238 redid debugger restarts and related debugger commands somewhat while
4240 ** The QUIT debugger command is gone, since it did something
4241 rather different than the SB-EXT:QUIT command, and since it never
4242 worked properly outside the main toplevel read/eval/print loop.
4243 Invoking the new TOPLEVEL restart provides the same functionality.
4244 ** The GO debugger command is also gone, since you can just invoke
4245 the CONTINUE restart directly instead.
4246 ** The TOP debugger command is also gone, since it's redundant with the
4247 FRAME 0 command, and since it interfered with abbreviations for the
4249 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4251 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4252 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4253 some porting work by Martin Atzmueller)
4254 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4255 with a boatload of patches which he ported to SBCL. Now that those
4257 ** The system tries to make sure that its low-priority messages
4258 are prefixed by semicolons, to help people who like to use
4259 syntax highlighting in their ILISP buffer. (This patch
4260 was originally due to Raymond Toy.)
4261 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4262 patches originally written by Raymond Toy.
4263 ** The compiler understands coercion between single-value and
4264 multiple-VALUES type expressions better, getting rid of some very
4265 weird behavior, thanks to patches originally by Robert MacLachlan
4266 and Douglas Crosher.
4267 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4268 lambda lists, thanks to a patch originally by Pierre Mai.
4269 ** The system no longer bogusly warns about "abbreviated type
4271 ** The compiler gets less confused by inlining and RETURN-FROM,
4272 thanks to some patches originally by Tim Moore.
4273 ** The system no longer hangs when dumping circular lists to fasl
4274 files, thanks to a patch originally from Douglas Crosher.
4275 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4276 undefined function error.
4277 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4278 in structure slot initforms) for now, documented workaround instead:-|
4279 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4280 * fixed bug 5: added stubs for various Gray stream functions called
4281 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4282 installed, at least appropriate type errors are generated
4283 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4284 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4285 consistently in DEFMETHOD forms.
4286 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4287 it doesn't seem to affect SBCL after all
4288 * The C runtime system now builds with better optimization and many
4289 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4291 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4293 * The system is now under CVS at SourceForge (instead of the
4294 CVS repository on my home machine).
4295 * The new signal handling code has been tweaked to treat register
4296 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4297 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4298 downstream, e.g. in debug-int.lisp, has implicit dependencies
4299 on the unsignedness of integer representation of machine words,
4300 and that caused the system to bomb out with infinite regress
4301 when trying to recover from type errors involving signed values,
4302 e.g. (BUTLAST '(1 2 3) -1).)
4303 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4304 (This was one of the bugs Peter Van Eynde reported back in July.)
4305 * The system now uses code inspired by Colin Walters' O(N)
4306 implementation of MAP (from the cmucl-imp@cons.org mailing
4307 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4308 inline the MAP operation, and there is more than one
4309 sequence argument to the MAP call (so that it can't just
4310 do ETYPECASE once and for all based on the type of the
4311 single sequence argument). (The old non-inline implementation
4312 of the general M-argument sequence-of-length-N case required
4313 O(M*N*N) time when any of the sequence arguments were LISTs.)
4314 * The QUIT :UNIX-CODE keyword argument has been renamed to
4315 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4316 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4318 * Raymond Wiker's patches to port dynamic loading from Linux to
4319 FreeBSD have been added.
4320 * The BUGS file is now more nearly up to date, thanks in large part
4321 to Martin Atzmueller's review of it.
4322 * The debugger now flushes standard output streams before it begins
4323 its output ("debugger invoked" and so forth).
4324 * The core version number and fasl file version number have both
4325 been incremented, because of incompatible changes in the layout
4327 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4328 used to not be called for a saved Lisp image.)
4329 * Martin Atzmueller's version of a patch to fix a compiler crash,
4330 as posted on sbcl-devel 13 September 2000, has been installed.
4331 * Instead of installing Martin Atzmueller's patch for the
4332 compiler transform for SUBSEQ, I deleted the compiler transform,
4333 and transforms for some similar consing operations.
4334 * A bug in signal handling which kept TRACE from working on OpenBSD
4336 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4337 away by constant folding
4338 * The system now defines its address space constants in one place
4339 (in the Lisp sources), and propagates them automatically elsewhere
4340 (through GENESIS and the sbcl.h file). Therefore, patching the
4341 address map is less unnecessarily tedious and error-prone. The
4342 Lisp names of address space constants have also been systematized.
4343 * CVS tags like dollar-Header-dollar have been removed from
4344 the sources, because they have never saved me trouble and
4345 they've been source of trouble working with patches and other
4346 diff-related operations.
4347 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4348 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4350 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4352 * The system has been ported to OpenBSD.
4353 * The system now compiles with a simple "sh make.sh" on the systems
4354 that it's supported on. I.e., now you no longer need to tweak
4355 text in the target-features.lisp-expr and symlinks in src/runtime/
4356 by hand, the make.sh takes care of it for you.
4357 * The system is no longer so grossly inefficient when compiling code
4358 involving vectors implemented as general (not simple) vectors (VECTOR T),
4359 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4360 which dares to use the various sequence functions on non-simple
4361 vectors, takes less of a performance hit.
4362 * There is now a primitive type predicate VECTOR-T-P
4363 to test for the (VECTOR T) type, so that e.g.
4364 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4365 can now be compiled with some semblance of efficiency. (The old code
4366 turned the type declaration into a full call to %TYPEP at runtime!)
4367 * AREF on (VECTOR T) is still not fast, since it's still compiled
4368 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4369 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4370 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4371 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4372 without making full calls to SUBTYPEP at runtime.
4373 (Some analogous efficiency issues for non-simple vectors specialized to
4374 element types other than T, or for non-simple multidimensional arrays,
4375 have not been addressed. They could almost certainly be handled the
4376 same way if anyone is motivated to do so.)
4377 * The changes in array handling break binary compatibility, so
4378 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4379 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4380 instead of (VALUES T T).
4381 * By following the instructions that Dan Barlow posted to sbcl-devel
4382 on 2 July 2000, I was able to enable primitive dynamic object
4383 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4384 functionality is not implemented (since it calls ld to resolve
4385 library references automatically, requiring RUN-PROGRAM for its
4386 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4387 try to resolve library references) is now supported.
4388 * The system now flushes the standard output streams when it terminates,
4389 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4390 them at several other probably-convenient times, e.g. in each pass of
4391 the toplevel read-eval-print loop, and after evaluating a form given
4392 as an "--eval" command-line option. (These changes were motivated by a
4393 discussion of stream flushing issues on cmucl-imp in August 2000.)
4394 * The source transform for TYPEP of array types no longer assumes
4395 that an array whose element type is a not-yet-defined type
4396 is implemented as an array of T, but instead punts, so that the
4397 type will be interpreted at runtime.
4398 * There is now some support for cross-compiling in make.sh: each of
4399 the phases of make.sh has its own script. (This should be transparent
4400 to people doing ordinary, non-cross-compile builds.)
4401 * Since my laptop doesn't have hundreds of megabytes of memory like
4402 my desktop machine, I became more motivated to do some items on
4403 my to-do list in order to reduce the size of the system a little:
4404 ** Arrange for various needed-only-at-cold-init things to be
4405 uninterned after cold init. To support this, those things have
4406 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4407 symbols with such names are now uninterned after cold init).
4408 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4409 fasl files for cold load.
4410 ** Remove the old compiler structure pooling code (which used to
4411 be conditional on the target feature :SB-ALLOC) completely.
4412 ** Redo the representation of some data in cold init to be more compact.
4413 (I also looked into supporting byte compiled code at bootstrap time,
4414 which would probably reduce the size of the system a lot, but that
4415 looked too complicated, so I punted for now.)
4416 * The maximum signal nesting depth in the src/runtime/ support code has
4417 been reduced from 4096 to 256. (I don't know any reason for the very
4418 large old value. If the new smaller value turns out to break something,
4419 I'll probably just bump it back up.)
4420 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4422 * Many, many bugs reported by Peter Van Eynde have been added to
4423 the BUGS list; some have even been fixed.
4424 * While enabling dynamic object file loading, I tried to make the
4425 code easier to understand, renaming various functions and variables
4426 with less ambiguous names, and changing some function calling
4427 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4428 * While trying to figure out how to do the OpenBSD port, I tried to
4429 clean up some of the code in src/runtime/. In particular, I dropped
4430 support for non-POSIX signal handling, added various comments,
4431 tweaked the code to reduce the number of compilation warnings, and
4432 renamed some files to increase consistency.
4433 * To support the new automatic configuration functionality in make.sh,
4434 the source file target-features.lisp-expr has been replaced with the
4435 source file base-target-features.lisp-expr and the machine-generated
4436 file local-target-features.lisp-expr.
4437 * fixed a stupid quoting error in make.sh so that using CMU CL
4438 "lisp -batch" as cross-compilation host works again
4440 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4442 * DESCRIBE no longer tries to call itself recursively to describe
4443 bound/fbound values, so that it no longer fails on symbols which are
4444 bound to themselves (like keywords, T, and NIL).
4445 * DESCRIBE now works on generic functions.
4446 * The printer now prints less-screwed-up representations of closures
4447 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4448 * A private symbol is used instead of the :EMPTY keyword previously
4449 used to mark empty slots in hash tables. Thus
4450 (DEFVAR *HT* (MAKE-HASH-TABLE))
4451 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4452 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4453 now does what ANSI says that it should. (You can still get
4454 similar noncompliant behavior if bang on the hash table
4455 implementation with all the symbols you get back from
4456 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4457 This breaks binary compatibility, since tests for equality to
4458 :EMPTY are wired into things like the macroexpansion of
4459 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4461 * There's now a minimal placeholder implementation for CL:STEP,
4462 as required by ANSI.
4463 * An obscure bug in the interaction of the normal compiler, the byte
4464 compiler, inlining, and structure predicates has been patched
4465 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4466 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4468 * Missing ordinary arguments in a macro call are now detected even
4469 when the macro lambda list contains &KEY or &REST.
4470 * The debugger no longer complains about encountering the top of the
4471 stack when you type "FRAME 0" to explicitly instruct it to go to
4472 the top of the stack. And it now prints the frame you request even
4473 if it's the current frame (instead of saying "You are here.").
4474 * As specified by ANSI, the system now always prints keywords
4475 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4477 * The default initial SIZE of HASH-TABLEs is now smaller.
4478 * Type information from CLOS class dispatch is now propagated
4479 into DEFMETHOD bodies, so that e.g.
4480 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4482 is now basically equivalent to
4483 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4484 (DECLARE (TYPE SINGLE-FLOAT X))
4486 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4487 operation, without having to do run-time type dispatch.
4488 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4489 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4490 * Since I don't seem to be making big file reorganizations very often
4491 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4492 is overflowing my ability to conveniently back them up), I've finally
4493 checked the system into CVS. (The CVS repository is on my home system,
4494 not at SourceForge -- putting it on SourceForge might come later.)
4495 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4496 high-level GC-NOTIFY-FOO functions send their output. (There's
4497 still very little control of where low-level verbose GC functions
4498 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4499 less than it used to -- the GC-NOTIFY-FOO functions are now under
4500 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4501 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4502 in only one place in the source code, and propagates it automatically
4503 everywhere that it's needed. Thus e.g. when I bump the version from
4504 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4505 * The C source files now include boilerplate legalese and documentation
4506 at the head of each file (just as the Lisp source files already did).
4507 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4508 to his page will be replaced with a link to his new CLiki service.
4510 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4512 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4513 * The build process now looks for GNU make under the default name "gmake",
4514 instead of "make" as it used to. If GNU make is not available as "gmake"
4515 on your system, you can change this default behavior by setting the
4516 GNUMAKE environment variable.
4517 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4518 can build without error under CMU CL.
4520 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4522 * There is now a partial SBCL user manual (with some new text and some
4523 text cribbed from the CMU CL manual).
4524 * The beginnings of a profiler have been added (starting with the
4525 CMU CL profiler and simplifying and cleaning up). Eventually the
4526 main interface should be through the TRACE macro, but for now,
4527 it's still accessed through vaguely CMU-CL-style functions and macros
4528 exported from the package SB-PROFILE.
4529 * Some problems left over from porting CMU CL to the new
4530 cross-compilation bootstrap process have been cleaned up:
4531 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4532 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4533 part when getting rid of the compiler *BACKEND* stuff.)
4534 ** The value of *NULL-TYPE* was screwed up, because it was
4535 being initialized before the type system knew the final
4536 definition of the 'NULL type. This screwed up several key
4537 optimizations in the compiler, causing inefficiency in all sorts
4538 of places. (I found it because I wanted to understand why
4539 GET-INTERNAL-RUN-TIME was consing.)
4540 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4541 PROCLAIM FTYPE information. Unfortunately this broke binary
4542 compatibility again, since now the forms output by DEFGENERIC
4543 to refer to functions which didn't exist in 0.6.3.
4544 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4545 can use the new (as of 0.6.3) transform for SOME into MAP into
4547 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4548 values to (INTEGER 0 1000000), so that the time code will no longer
4549 occasionally get blown up by Linux returning 1000000 microseconds
4550 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4551 its output conform to the ANSI spec. (Alas, this makes its output
4552 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4553 * A full call to MAP NIL with a single sequence argument no longer conses.
4554 * fixes to problems pointed out by Martin Atzmueller:
4555 * The manual page no longer talks about multiprocessing as though
4556 it were currently supported.
4557 * The ILISP support patches have been removed from the distribution,
4558 because as of version 5.10.1, ILISP now supports SBCL without us
4559 having to maintain patches.
4560 * added a modified version of Raymond Toy's recent CMU CL patch for
4561 EQUALP comparison of HASH-TABLE
4563 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4565 * The system still can't cross-compile itself with
4566 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4567 entails), but at least it can compile more of itself that way
4568 than it used to be able to, and various buglets which were uncovered
4569 by trying to cross-compile itself that way have now been fixed.
4570 * This release breaks binary compatibility again. This time
4571 at least I've incremented the FASL file format version to 2, so that the
4572 problem can be detected reliably instead of just causing weird errors.
4573 * various new style warnings:
4574 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4575 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4576 ** using the lexical binding of a variable named in the *FOO* style
4577 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4578 as specified by ANSI.
4579 * *RANDOM-STATE* is no longer automatically initialized from
4580 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4581 default behavior of the system is to repeat its behavior every time
4582 it's run. If you'd like to change this behavior, you can always
4583 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4584 old convention there was no comparably easy way to get the system to
4585 repeat its behavior every time it was run.
4586 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4587 FTYPE declarations has been removed, in favor of their ANSI
4588 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4589 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4590 the types of their sequence arguments can be determined at compile time.
4591 This is done through a new open code expansion for MAP which eliminates
4592 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4593 argument types can be determined at compile time.
4594 * The optimizer now transforms COERCE into an identity operation when it
4595 can prove that the coerced object is already of the correct type. (This
4596 can be a win for machine generated code, including the output of other
4597 optimization transforms, such as the MAP transform above.)
4598 * Credit information has been moved from source file headers into CREDITS.
4599 * Source file headers have been made more standard.
4600 * The CASE macro now compiles without complaining even when it has
4603 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4605 * (Note that the way that the PCL macroexpansions were rewritten
4606 to accommodate the change in DEFGENERIC below breaks binary
4607 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4608 not run under sbcl-0.6.2. Once we get out of alpha releases,
4609 i.e. hit release 1.0.0, we'll probably try to maintain binary
4610 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4611 and sbcl-1.4.4. Until then, however, it might be fairly common
4612 for maintenance releases to break binary compatibility.)
4613 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4615 * The reporting of unhandled signals has been changed to print some
4616 explanatory text as well as the report form. (Previously only
4617 the report form was printed.)
4618 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4619 it defines, so that the compiler no longer issues undefined function
4620 warnings for compiled-but-not-yet-loaded generic functions.
4621 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4622 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4623 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4624 section of the ANSI spec are supported.
4625 * The "" nickname for the "KEYWORD" package has been removed.
4626 The reader still handles symbol tokens which begin with a package marker
4627 as keywords, but it doesn't expose its mechanism for doing so in the
4628 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4629 * The system now issues STYLE-WARNINGs for contradictory TYPE
4630 proclamations. (Warnings for contradictory FTYPE proclamations would
4631 be nice too, but those can't be done usefully unless the type system
4632 is made smarter about FUNCTION types.)
4633 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4634 systematized, so that "*target-*.lisp is supposed to exist only on the
4635 target and imply that there's a related file which exists on the
4636 host, and *host-*.lisp is supposed to exist only on the host and imply
4637 that there's a related file which exists on the target. This involves a
4638 lot of renaming. Hopefully the acute confusion caused by the renaming
4639 will be justified by the reduction in chronic confusion..
4640 ** runtime-type.lisp -> early-target-type.lisp
4641 ** target-type.lisp -> late-target-type.lisp
4642 ** early-host-format.lisp -> early-format.lisp
4643 ** late-host-format.lisp -> late-format.lisp
4644 ** host-error.lisp -> misc-error.lisp
4645 ** early-error.lisp -> early-target-error.lisp
4646 ** late-error.lisp -> late-target-error.lisp
4647 ** host-defboot.lisp -> early-defboot.lisp
4648 ** code/misc.lisp -> code/target-misc.lisp
4649 ** code/host-misc.lisp -> code/misc.lisp
4650 ** code/numbers.lisp -> code/target-numbers.lisp
4651 ** code/early-numbers.lisp -> numbers.lisp
4652 ** early-host-type.lisp -> early-type.lisp
4653 ** late-host-type.lisp -> late-type.lisp
4654 ** host-typep.lisp -> typep.lisp
4655 ** load.lisp -> target-load.lisp
4656 ** host-load.lisp -> load.lisp
4657 ** host-disassem.lisp -> disassem.lisp
4658 ** host-insts.lisp -> insts.lisp
4659 ** byte-comp.lisp -> target-byte-comp.lisp
4660 ** host-byte-comp.lisp -> byte-comp.lisp
4661 ** host-signal.lisp -> signal.lisp
4662 ** host-defstruct.lisp -> defstruct.lisp
4663 ** late-target-type.lisp -> deftypes-for-target.lisp
4664 Furthermore, several other previously target-only files foo.lisp (e.g.
4665 hash-table.lisp and random.lisp) have been split into a target-and-host
4666 foo.lisp file and a target-only target-foo.lisp file, with their key type
4667 definitions in the target-and-host part, so that the cross-compiler will
4668 know more about target types.
4669 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4670 gone away. In their place are various *BACKEND-FOO* variables
4671 corresponding to the slots of the old structure.
4672 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4673 into the target SBCL has been fixed.
4674 * The chill.lisp system for loading cold code into a running SBCL
4676 * Support for the CMU CL "scavenger hook" extension has been removed.
4677 (It was undocumented and unused in the CMU CL sources that SBCL was
4678 derived from, and stale in sbcl-0.6.1.)
4679 * Various errors in the cross-compiler type system were detected
4680 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4681 (enabling various consistency checks). Many of them were fixed,
4682 but some hard problems remain, so the compiler is back to
4683 running without *TYPE-SYSTEM-INITIALIZED* for now.
4684 * As part of the cross-compiler type system cleanup, I implemented
4685 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4686 * I have started adding UNCROSS calls throughout the type system
4687 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4688 on cross-compiler input files will be able to go away, and various
4690 * CONSTANTP now returns true for quoted forms (as explicitly required
4693 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4695 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4696 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4697 (MAKE-LIST -1). In the longer term, it ought to become true
4698 that declarations are assertions even at SAFETY 1. For now, it's not
4699 quite true even at SAFETY 3, but it's at least more nearly true..
4700 (Note that this change seems to increases the size of the system by
4701 O(5%) and to decrease the speed of the compiler by 20% or more.)
4702 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4703 for the problem of printing dozens of lines of distracting information
4704 about low-level system machinery as part of the top stack frame
4705 on entry to the debugger when an undefined function was called.
4706 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4707 is not reset to COMMON-LISP-USER.
4708 * Compilation of stuff related to dyncount.lisp has been made conditional
4709 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4710 smaller. The various dyncount-related symbols have been moved into
4711 a new "SB-DYNCOUNT" package.
4712 * tty-inspect.lisp has been renamed to inspect.lisp.
4713 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4714 feature has gone away. (When we eventually port to other flavors of
4715 libc and/or Unix, we'll try to make the differences between flavors
4716 invisible at the user level.)
4717 * Various other *FEATURES* tags, and/or their associated conditionals,
4718 have been removed if obsolescent, or given better documentation, or
4719 sometimes given more-mnemonic names.
4721 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4723 * tidied up "make.sh" script
4724 * tidied up system directory structure
4725 * better "clean.sh" behavior
4726 * added doc/FOR-CMUCL-DEVELOPERS
4727 * many many small tweaks to output format, e.g. removing possibly-confusing
4728 trailing #\. character in DESCRIBE-INSTANCE
4729 * (EQUALP #\A 'A) no longer signals an error.
4730 * new hashing code, including EQUALP hashing
4731 * tidied up Lisp initialization and toplevel
4732 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4733 * command line argument processing
4734 * added POSIX-GETENV function to deal with Unix-ish environment variables
4735 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4736 terminating SBCL on EOF
4737 * non-verbose GC by default
4738 * There is no more "sbcl" shell script; the sbcl file is now the C
4739 runtime executable (just like CMU CL).
4740 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4742 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4743 symbol and package information as Lisp native symbol and package objects
4744 instead of strings naming symbols and strings naming packages. This way,
4745 whenever packages are renamed (as in warm init), debug information is
4746 transformed along with everything else.
4747 * tweaked the optimization policy declarations which control the building
4748 of SBCL itself. Now, among other things, the system no longer saves
4749 source location debugging information. (This helps two problems at once
4750 by reducing SBCL size and by keeping SBCL from trying to look for its
4751 sources -- which may not exist -- when reporting errors.)
4752 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4753 debugging and testing purposes
4754 * cleaned up printing, making the printer call PRINT-OBJECT for
4755 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4756 methods, giving nearly-ANSI behavior
4757 * converted almost all special variables to use *FOO* naming convention
4758 * deleted PARSE-TIME functionality, since it can be done portably
4759 * moved some files out of cold init into warm init
4760 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4762 * regularized formatting of source files
4763 * added an install.sh script
4764 * fixed ridiculous memory usage of cross-compiler by making
4765 compiler/alloc.lisp not try to do pooling unless it can hook
4766 itself into the GC of the cross-compilation host. Now the system
4767 builds nicely on my old laptop.
4768 * added :SB-ALLOC in target-features.lisp-expr
4769 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4770 was not implemented)
4771 * re-did condition handling and note reporting in the compiler. Notes
4772 are no longer handled by signalling conditions. Style warnings
4773 and warnings are handled more correctly and reported in such a way
4774 that it's easy to find one or the other in your output (so that you
4775 can e.g. figure out which of many problems caused COMPILE-FILE to
4777 * changed the severity of several compiler warnings from full WARNING
4778 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4779 compiler note reporting so that it doesn't use the condition system
4780 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4781 COMPILE-FILE command)
4782 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4783 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4784 no longer supported, since their implementation was deeply intertwingled
4785 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4786 * removed bogus "support" for compiler macros named (SETF FOO), and
4787 removed the compiler macro for SETF INFO (but only after making a fool
4788 of myself on the cmucl-imp mailing list by posting a bogus patch for
4789 DEFINE-COMPILER-MACRO..)
4790 * Compiled files containing forms which have side effects on the Lisp
4791 reader (such as DEFPACKAGE forms) are now handled more correctly.
4792 (Compiler queuing of top level lambdas has been suppressed by setting
4793 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4794 * deleted various currently-unused source files, e.g. gengc.lisp. They
4795 may be added back at some point e.g. when porting to other architectures,
4796 but until they are it's distracting to distribute them and to try to
4798 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4799 there were so many of them they're just distractions, and UNCROSS is
4800 known to be able to handle the current sources
4801 * moved PROFILE functionality into TRACE, so that it will be clear
4802 how the wrapping and unwrapping of functions when you profile them
4803 interacts with the wrapping and unwrapping of functions when you
4804 trace them. (Actually, the functionality isn't there yet, but at least
4805 the interface specification is there. Hopefully, the functionality will
4806 arrive with some future maintenance release.)
4807 * removed host-oops.lisp
4808 * changed signature of QUIT function to allow UNIX-CODE argument
4809 * fixed READ-SEQUENCE bug
4810 * tweaked verbose GC output so that it looks more like the progress
4811 output that ANSI specifies for functions like LOAD
4812 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4813 * added <http://sbcl.sourceforge.com> to the banner information printed by