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 * optimization: REPLACE deftransforms don't punt when :START1 or
26 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
27 functions with an odd number of &OPTIONAL arguments, a &REST
28 argument, and one or more &KEY arguments at the call site.
29 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
30 arguments appeared at call sites as well.
31 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
32 READ-LINE &co can no longer read from them afterwards. (reported
34 * bug fix: fixed #427: unused local aliens no longer cause compiler
35 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
37 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
38 alien-stack leakage. (reported by Andy Hefner)
39 * bug fix: PROGV signals an error when an attempt to violate declared
40 type of a variable or bind a constant is made.
41 * bug fix: SET signals an error when an attempt to violate declared
42 type of a variable is made.
43 * bug fix: restart computation during the execution of a restart
44 test function no longer causes infinite recursion. (reported by
46 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
47 signals an error. (thanks to Pierre Mai)
48 * bug fix: EXPORT left symbol unexported in conflict situations.
49 (thanks to Michael Weber)
50 * bug fix: correctly handle name conflicts involving CL:NIL.
51 (thanks to Michael Weber)
52 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
53 conflicts arising from USEing package with conflicting symbols
54 correctly. (thanks to Michael Weber)
56 changes in sbcl-1.0.19 relative to 1.0.18:
57 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
58 warnings that go otherwise unhandled will be muffled if they are
59 of the type that's the value of this variable.
60 * optimization: stack allocation is slightly more efficient on x86
62 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
63 where a raw slot always is initialized using the initform whose
64 type is not know sufficiently well a compile-time are now compiled
65 correctly. (reported by John Morrison)
66 * bug fix: compiler no longer makes erronous assumptions in the
67 presense of non-foldable SATISFIES types.
68 * bug fix: stack analysis missed cleanups of dynamic-extent
69 arguments in non-let-converted calls to local functions.
70 * improvements to the Windows port:
71 ** adjusted address spaces for building on both Win32 and
72 Win64. (thanks for John Connors)
73 * fixed some bugs revealed by Paul Dietz' test suite:
74 ** interval arithmetic during type derivation used inexact integer
75 to single-float coercions.
76 ** arithmetic operations involving large integers and single
77 floats give the same results in compiled and interpreted code.
78 ** deriving the result type of COERCE no longer signals an error
79 if the derived type of the second argument is a MEMBER type
80 containing invalid type specifiers.
81 ** ADJOIN with constant NIL as second argument works correctly.
82 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
84 changes in sbcl-1.0.18 relative to 1.0.17:
85 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
86 profiles only the current thread.
87 * minor incompatible change: changes to SYMBOL-VALUE of constants
88 defined with DEFCONSTANT now signal an error.
89 * enhancement: SB-SPROF now has support for wallclock profiling,
90 and is also able to profile specific threads. REPORT output
91 has also additional sorting options.
92 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
94 * optimization: structure allocation has been improved
95 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
96 ** out of line constructors are ~10% faster.
97 ** inline constructors are ~15% faster.
98 ** inline constructors are capable of dynamic extent allocation
99 (generally on x86 and x86-64, in some cases on other platforms
101 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
103 * optimization: file compiler is now able to coalesce non-circular
104 lists, non-base strings, and bit-vectors. Additionally, constants
105 are never referenced through SYMBOL-VALUE at runtime.
106 * optimization: code defining methods on PRINT-OBJECT (and other
107 generic functions in the COMMON-LISP package) now loads faster.
108 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
110 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
111 fasl is now deleted. (reported by Attila Lendvai)
112 * bug fix: READ-LINE always returned NIL for the last line in files.
113 (reported by Yoshinori Tahara)
114 * bug fix: more accurate disassembly annotations of foreign function
115 calls. (thanks to Andy Hefner)
116 * bug fix: trimming non-simple strings and non-string string
117 designators when the there is nothing to trim works properly.
118 (thanks to James Knight)
119 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
120 (thanks to Travis Cross)
121 * fixed some bugs revealed by Paul Dietz' test suite:
122 ** NIL is a valid function name (regression at 1.0.13.38)
123 ** FILL on lists was missing its return value (regression at 1.0.12.27)
124 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
125 fill pointers properly (regression at 1.0.12.23)
126 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
127 displacement indices properly (regression at 1.0.12.23)
129 changes in sbcl-1.0.17 relative to 1.0.16:
130 * temporary regression: user code can no longer allocate closure
131 variable storage on stack, due to bug 419 without explicitly
132 requesting it. Please consult sbcl-devel for advice if you need to
133 use this feature in the meanwhile.
134 * new feature: runtime argument --control-stack-size can be used to
135 adjust thread default control stack size.
136 * enhancement: improved TIME output
137 ** all times are reported using the measured accuracy (milliseconds
138 for real and GC times, microseconds for everything else.)
139 ** processor cycle counts on x86 and x86-64.
140 ** interpreted forms are counted for both evaluator modes.
141 ** number of lambdas converted by the compiler is reported.
142 ** CPU percentage report (computed from real and total run time.)
143 ** more comprehensive run time reporting, using a condenced format
144 ** interperted form, lambda, and page fault counts are omitted
146 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
148 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
149 * optimization: two argument forms of LAST are upto ~10% faster
150 in normal SPEED policies.
151 * optimization: NCONC no longer needs to heap cons its &REST list
152 in normal SPEED policies.
153 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
155 * bug fix: UNION and NUNION work with :TEST-NOT once more,
156 regression since 1.0.9.1. (thanks to Eric Marsden)
157 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
158 since 1.0.15.36. (thanks to Paul Khuong)
159 * bug fix: LAST when always returned the whole list when given a bignum
160 as the second argument.
161 * bug fix: dynamic extent allocation of nested lists and vectors
162 could leak to otherwise accessible parts.
163 * bug fix: invalid optimization of heap-allocated alien variable
165 * bug fix: fasl header checking is less vulnerable to different
166 platform word lengths.
167 * bug fix: more correct assembler syntax for GNU binutils
168 2.18.50.0.4 support. (thanks to Marijn Schouten)
169 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
171 * bug fix: the fopcompiler can handle LOCALLY forms (with no
172 declarations) successfully. (reported by Attila Lendvai)
174 changes in sbcl-1.0.16 relative to 1.0.15:
175 * minor incompatible change: revert the changes to sb-posix's error
176 signaling added in 1.0.14.
177 * minor incompatible change: change PROBE-FILE back to returning
178 NIL whenever we can't get a truename, as was the case before 1.0.14.
179 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
180 inherits from ERROR instead of just CONDITION.
181 * new feature: SB-INTROSPECT can provide source locations for instances
182 as well. (thanks to Tobian Ritterweiler)
183 * optimization: binding special variables now generates smaller code
184 on threaded platforms.
185 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
186 and cases where no :TEST is given but the compiler can infer that
187 the element to search is of type (OR FIXNUM (NOT NUMBER)).
188 * optimization: better LOGNOT on fixnums.
189 * optimization: modular arithmetic for a particular requested width
190 is implemented using a tagged representation unless a better
191 representation is available.
192 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
193 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
194 for wellformedness and returns a specialized vector. (reported by
195 Francois-Rene Rideau)
196 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
197 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
198 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
199 unscheduling timer at the same time as another timer fires.
200 * bug fix: don't reschedule timers for dead threads.
201 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
202 * bug fix: copying output from RUN-PROGRAM to a stream signalled
203 bogus errors if select() was interrupted.
204 * enhancement: add support for fcntl's struct flock to SB-POSIX.
206 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
207 * enhancement: cleaner backtraces for interactive interrupts, as
208 well as other cases where the interesting frames used to be
209 obscured by interrupt handling frames.
210 * enhancement: untracing a whole package using (UNTRACE "FOO") is
211 now supported, and tracing a whole package using (TRACE "FOO") now
212 traces SETF-functions as well.
213 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
214 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
215 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
216 * unreadably printed representation of hash-tables now includes
218 * bug fix: partially fixed #188: type propagation from assignments
219 is now more efficient.
220 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
221 'DOUBLE-FLOAT) are not flushable.
222 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
223 full address of the object, and none of the tag bits.
224 * bug fix: readably printing hash-tables now respects other printer
225 control variables. (reported by Cedric St-Jean)
226 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
227 keyword with MAKE-HASH-TABLE.
228 * bug fix: export SB-POSIX:MKSTEMP.
229 * bug fix: SORT was not interrupt safe.
230 * bug fix: XREF accounts for the last node of each basic-block as
232 * bug fix: MAKE-INSTANCE optimizations interacted badly with
233 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
234 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
235 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
236 create obsolete instances in the case of redefinition or
237 obsoletion of a superclass. (thanks to Andy Hefner)
238 * bug fix: Support for the Alpha architecture has been revived; it had
239 suffered somewhat from lack of maintenance since sbcl-1.0.
240 * improvements to the Windows port:
241 ** The system detects the codepage to use at startup, instead of
242 using the value from the saved core. (thanks to Kei Suzuki)
244 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
245 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
246 (see documentation for details.)
247 * revived support for OpenBSD (contributed by Josh Elsasser)
248 * partially fixed bug #108: ROOM no longer suffers from occasional
249 (AVER (SAP= CURRENT END)) failures.
250 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
251 bodies no longer cause a WARNING to be signalled. (reported by
253 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
254 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
255 the bug in TRUNCATE.) (reported by Michael Weber)
256 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
257 no samples. (reported by Andy Hefner)
258 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
259 no longer appear as (NIL ...) frames in backtraces.
260 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
261 is now actually exported from SB-EXT as documented. (reported by
263 * bug fix: sb-aclrepl now correctly understands how to inspect
264 single-floats on 64-bit platforms where single-floats are not boxed.
265 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
266 yet been finalized. (reported by Levente Meszaros)
267 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
268 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
269 SBCL-specific optimize qualities.
271 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
272 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
273 an executable in the search path, and does so in the child
274 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
275 been removed; it can be found in the CVS history, for somebody who
276 needs that search behavior (see the manual).
277 * minor incompatible change: compiler policy re. weakening type
278 checks has changed: now type checks are weakened only if SAFETY < 2
280 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
281 unparsing of directory pathnames as files. Analogously,
282 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
283 filename to parse into a directory pathname.
284 * enhancement: implicit generic function creation now signals a
285 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
286 users can bind handlers for and muffle around calls to LOAD.
287 * enhancement: RUN-PROGRAM allows unicode arguments and environments
288 to be used (using the default stream external format), and allows
289 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
290 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
291 strings and vectors whose element-type or simplicity is not fully
292 known at compile-time.
293 * optimization: STRING-TRIM and related functions no longer allocate
294 a new string when no trimming needs to be performed. These functions
295 are also faster than before when the input string has been declared
297 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
298 * optimization: READ-LINE is significantly faster for files containing
300 * optimization: non-open coded uses of character comparison operators
301 (e.g. char=) no longer cons when called with more than one parameter
302 on platforms supporting dynamic-extent allocation.
303 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
304 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
305 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
307 * bug fix: some sequence functions elided bounds checking when
309 * bug fix: too liberal weakening of union-type checks when SPEED >
311 * bug fix: more bogus fixnum declarations in ROOM implementation
314 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
315 * new feature: MAKE-HASH-TABLE now experimentally accepts a
316 :SYNCHRONIZED argument, which makes the hash-table safe for
317 concurrent accesses (but not iteration.) See also:
318 SB-EXT:WITH-LOCKED-HASH-TABLE, and
319 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
320 * optimization: CONCATENATE on strings is an order of magnitude faster
321 in code compiled with (> SPEED SPACE).
322 * optimization: SUBSEQ is ~50% faster on lists.
323 * bug fix: bug 417 fixed -- source location reporting is now more robust.
324 * bug fix: SUBSEQ on a list will now correctly signal an error if if
325 END is smaller then START.
326 * bug fix: SB-PROFILE will no longer report extra consing for nested
327 calls to profiled functions.
328 * bug fix: ROOM implementation had bogus fixnum declarations which
329 could cause type-errors when calling ROOM on large images.
330 * bug fix: if file compilation is aborted, the partial fasl is now
331 deleted, and COMPILE-FILE returns NIL as the primary value.
332 * bug fix: number of thread safety issues relating to SBCL's internal
333 hash-table usage have been fixed.
334 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
335 be returned from its body when the values were being returned
336 using unknown-values return convection and the W-P-O was wrapped
337 inside an UNWIND-PROTECT.
338 * bug fix: sb-posix should now compile again under Windows, enabling
341 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
342 * incompatible change: hash-table accessor functions are no longer
343 automatically protected by locks. Concurrent accesses on the same hash-table
344 from multiple threads can give inconsistent results or even corrupt the
345 hash-table completely. Multi-threaded applications should do their own
346 locking at the correct granularity. In the current implementation it is
347 still safe to have multiple readers access the same table, but it's not
348 guaranteed that this property will be maintained in future releases.
349 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
350 and will signal an error at runtime.
351 * enhancement: SB-THREAD package now exports a semaphore interface.
352 * enhancement: CONS can now stack-allocate on x86 and
353 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
355 * enhancement: nested lists can now be stack allocated on
356 platforms providing stack allocation support.
357 * enhancement: dynamic-extent support has been extended to support
358 cases where there are multiple possible sources for the stack
360 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
361 if the mutex is uncontested on Linux.
362 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
363 as specified by the CLHS. (thanks to Tobias Rittweiler)
364 * bug fix: NaN comparison now works on x86-64.
365 * bug fix: CLOSE :ABORT T on a stream with pending output now
367 * bug fix: instances of non-standard metaclasses using standard
368 instance structure protocol sometimes missed the slot type checks
370 * bug fix: known functions can be inlined in high-debug code.
372 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
373 * minor incompatible change: the MSI installer on Windows no longer
374 associates .lisp and .fasl files with the installed SBCL.
375 * minor incompatible change: :UNIX is no longer present in *FEATURES*
376 on Windows. (thanks to Luis Oliviera)
377 * new feature: SB-CLTL2 contrib module now implements
378 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
379 * optimization: scavenging weak pointers is now more efficient,
380 requiring O(1) instead of O(N) per weak pointer to identify
381 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
382 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
383 method bodies using constant slot-names when the first argument is
384 a specializer parameter for the method.
385 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
386 names, when the class of the instance is a direct instance
387 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
388 as slow as the constant slot-name case.)
389 * optimization: member type construction is now O(N) instead
391 * optimization: UNION and NUNION are now O(N+M) for large
392 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
394 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
395 now more readable in environments like Slime which display it.
396 (thanks to Tobias C. Rittweiler)
397 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
398 was able to infer the correct argument type for the object on which
399 the CAS operation was being performed.
400 * bug fix: copy propagation interfered with parallel assignment
401 semantics in local calls. (reported by Paul Khuong)
402 * bug fix: the signed modular fixnum shift compiled to wrong code on
403 x86 and x86-64. (spotted by a slight modification to some of
406 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
407 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
408 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
409 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
411 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
412 and generic function now print the object identity for anonymous
413 instances. (thanks to Attila Lendvai)
414 * enhancement: as an extension to MOP, SBCL now supports SETF of
415 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
416 (thanks to Attila Lendvai)
417 * enhancement: sb-cover has an alternate annotation mode which puts
418 the form annotations at the car of the form, instead of annotating
419 it completely. The benefit of this mode is that it shows better
420 which forms were instrumented by the compiler.
421 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
422 by printer control variables. (reported by Dan Corkill)
423 * bug fix: system leaked memory when delayed output was performed by
424 the OS in smaller chunks then expected. (thanks to David Smith)
425 * bug fix: system leaked memory when file streams were not closed
427 * bug fix: large objects written to slow streams that were modified
428 after the write could end up with the modified state written to
429 the underlying file descriptor.
430 * bug fix: multiple threads operating in parallel on the same stream
431 could cause buffer-overflows.
432 * bug fix: source location information is stored correctly
433 (broken since 1.0.6). This bug would generally show up as the
434 Slime debugger highlighting the wrong form when the sldb-show-source
436 * bug fix: Branch forms are again annotated as branches in the sb-cover
438 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
439 on x86-64. (reported by Gregory Vanuxem)
440 * bug fix: EQUALP could return wrong results for structures with raw
441 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
442 word). (reported by Vjacheslav Fyodorov)
443 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
446 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
447 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
448 atomic compare-and-swap operations on threaded platforms.
449 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
450 allows assigning a global minimum value to optimization qualities
451 (overriding proclamations and declarations).
452 * enhancement: closed over variables can be stack-allocated on x86
454 * performance bug fix: GETHASH and (SETF GETHASH) are once again
456 * optimization: slot definition lookup is now O(1). This speeds up
457 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
458 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
459 * optimization: ASSOC and MEMBER can now be open-coded for all
460 combinations of keyword arguments when second argument is constant
461 and SPEED >= SPACE. In other cases a specialized version is
463 * bug fix: using obsoleted structure instances with TYPEP and
464 generic functions now signals a sensible error.
465 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
466 (reported by Kristoffer Kvello)
467 * bug fix: backtrace construction is now more careful when making
468 lisp-objects from pointers on the stack, to avoid creating bogus
469 objects that can be seen by the GC.
470 * bug fix: defaulting of values in contexts expecting more than 7
471 variables now works on x86-64. (reported by Christopher Laux)
472 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
474 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
475 as the property-list of a symbol.
476 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
477 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
478 situations requested, are once again file-compileable. (reported
481 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
482 * MOP improvement: support for user-defined subclasses of
483 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
484 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
485 the expansion of DEFMETHOD, is responsible for generating a form
486 which creates a list of specializers when evaluated. Additional
487 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
488 debugging and introspective support.
489 * minor incompatible change: the (unsupported) spinlock interface
490 has changed: free spinlock now has the value NIL, and a held spinlock
491 has the owning thread as its value.
492 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
493 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
495 * enhancement: name of a socket-stream is now "a socket" instead of
497 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
498 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
499 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
500 for generic functions. (Thanks to Tobias C. Rittweiler)
501 * optimization: bignum printing speed has been improved by 20-40%
502 (depending on the bignum size.)
503 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
505 * bug fix: the cache used by the CLOS to store precomputed effective
506 methods, slot offsets, and constant return values is now thread and
508 * bug fix: generic function dispatch function updating is now thread
509 and interrupt safe (in the sense that the known issues have been
511 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
512 * bug fix: interning EQL-specializers is now thread and interrupt safe.
513 * bug fix: asdf systems with dependencies to the SB-POSIX or
514 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
515 * bug fix: interrupt safety of applicable method computation has been
518 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
519 * new contrib: sb-cover, an experimental code coverage tool, is included
521 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
522 significantly faster.
523 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
524 produce more efficient code on x86-64 if the compiler can determine
525 that (+ INDEX <constant>) does not require a bounds check and FOO
526 has an element type at least 8 bits wide.
527 * enhancement: a new, experimental synchronous timeout facility is
528 provided. Refer to SB-SYS:WITH-DEADLINE for details.
529 * enhancement: when a symbol name conflict error arises, the
530 conflicting symbols are always printed with a package prefix.
531 (thanks to Kevin Reid)
532 * enhancement: stepping is now once again supported on the SPARC and
533 MIPS platforms. (It is also now more likely to work on CheneyGC
535 * enhancement: sb-sprof can now also track and report accurate call
537 * bug fixes: the treatment of non-standard subclasses of
538 SB-MOP:SPECIALIZER is more correct.
539 * incompatible change: PURIFY no longer copies the data from the
540 dynamic space into the static and read-only spaces on platforms
541 that use the generational garbage collector
542 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
544 * bug fix: GC race condition occasionally resulting in crashes with
545 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
547 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
549 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
550 a STYLE-WARNING to be signalled (regression from 1.0.4.)
551 * bug fix: an asynchronous interrupt could previously leave the
552 system running with GC inhibited.
553 * bug fix: a DECLARE form evaluated at top-level now causes an error
554 rather than silently (or verbosely) returning NIL.
555 * bug fix: trying to dissassemble functions compiled at high DEBUG could
556 result in a error being signalled due to source form lookup errors.
557 (reported by Peter Graves)
559 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
560 * incompatible change: removed writer methods for host-ent-name,
561 host-ent-addresses -- changing the values did not update the DNS
563 * minor incompatible change: changed experimental JOIN-THREAD interface
564 * documentation: the manual now lists reader and writer methods
565 in class slot documentation sections. (thanks to Richard M Kreuter)
566 * documentation: unwinding from asyncronous events has been
567 documented as unsafe.
568 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
569 in multithreaded application code.
570 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
571 platforms. (thanks to James Anderson for the optimization hint)
572 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
574 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
575 variants no longer cons.
576 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
577 their NOT- variants no longer cons.
578 * optimization: EQUAL hash tables no longer use SXHASH for objects
579 of all data types, but instead use an EQL hash for types for which
580 EQUAL is the same as EQL.
581 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
582 are significantly faster.
583 * optimization: new STRING-OUTPUT-STREAM implementation is much
584 faster for large outputs and conses 30% less on average.
585 * enhancement: XREF information is now collected for references made
586 to global variables using SYMBOL-VALUE and a constant argument.
587 * enhancement: SIGINT now causes a specific condition
588 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
589 * bug fix: FILE-POSITION used to signal an error in some cases where
590 ANSI requires it to return NIL.
591 * bug fix: ADJUST-ARRAY is now interrupt-safe.
592 * bug fix: adding and removing fd-handlers is now interrupt-safe.
593 * bug fix: inlined calls to C now ensure 16byte stack alignment on
595 * bug fix: bad type declaration in the CLOS implementation has
596 been fixed. (reported by James Anderson)
597 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
598 (thanks to Lutz Euler)
599 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
600 required by the CLHS. (reported by Samium Gromoff)
601 * bug fix: dead unbound variable references now signal an error.
602 (reported by Marco Monteiro)
603 * bug fix: / with an unused value was being deleted in safe code.
604 (thanks to Marco Monteiro and Kevin Reid)
605 * bug fix: number of characters that can be written onto a single
606 line in a file is unlimited.
607 * bug fix: some GC deadlocks caused by asynchronous interrupts have
608 been fixed by inhibiting interrupts for when GC is disabled.
609 * bug fix: some interrupt safety issues with GETHASH, (SETF
610 GETHASH), CLRHASH and REMHASH have been fixed.
611 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
612 specifier no longer causes infinite recursion.
613 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
614 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
615 * bug fix: an &environment argument with the correct variable information
616 is passed to macros that are expanded during byte compilation.
617 (reported by Samium Gromoff)
618 * bug fix: a logic error was causing the hash function for bignums to
619 have excessive amounts of collisions. (reported by Faré Rideau)
620 * bug fix: modifying the contents of an array could change the return
621 value of SXHASH on that array, which is only allowed for strings
622 and bit vectors. (bug introduced in 0.9.16)
623 * improvement: the x86-64/darwin port now passes all tests (except
624 for the debugger tests) but should still be considered
625 experimental until this is fixed.
626 * improvement: a style-warning is signaled for CASE (etc) clauses with
627 duplicate keys. (thanks to Kevin Reid)
628 * improvement: macos/x86 and macos/x86-64 now use mach exception
629 handlers for dealing with illegal instructions (for trapping and
630 error handling) and memory protection violations (for GC).
632 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
633 * new platform: experimental support for x86-64/darwin (MacOS).
634 * incompatible change: the thread-safe (on most platforms) getaddrinfo
635 and getnameinfo sockets functions are used instead of gethostbyaddr
636 and gethostbyname, on platforms where the newer functions are available.
637 As a result, the ALIASES field of HOST-ENT will always be NIL on these
639 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
640 don't print the startup banner, but behave as if --noinform was passed
641 as a command line argument. (thanks to Kevin Reid)
642 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
643 * optimization: code using alien values with undeclared types is much faster.
644 * optimization: the compiler is now able to open code SEARCH in more cases.
645 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
646 * bug fix: using standardized COMMON-LISP special variables as loop
647 variables no longer signals bogus package lock violations. (reported
649 * bug fix: declaring local loop variables to be of a range-limited type
650 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
651 (reported by Andras Simon)
652 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
653 bugs remain on x86-64.)
654 * bug fix: the #= and ## reader macros now interact reasonably with
655 funcallable instances.
656 * bug fix: type-checks for function arguments were compiled using the
657 compiler policy settings of the wrong lexical environment (in 1.0.2
659 * bug fix: SHADOW accepts characters as string designators, as required
660 by the spec (thanks to Eric Marsden)
661 * bug fix: fixed GC safety issues when foreign functions are called with
662 non-base strings as arguments
663 * bug fix: more consistent error output for fatal-compiler-warnings, like
665 * bug fix: functions with proper names were showing up as NIL in
666 backtraces (reported by Edi Weitz, regression in 1.0.2)
668 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
669 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
670 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
671 produce more efficient code on the x86 if the compiler can determine
672 that (+ INDEX <constant>) does not require a bounds check and FOO
673 has an element type at least 8 bits wide.
674 * bug fix: references to undefined variables are handled the same way
675 in toplevel forms as in the normal compiler
676 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
678 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
679 (thanks to Jon Buller)
680 * improvement: faster compilation times for complex functions
681 * improvement: added readlink support to SB-POSIX (thanks to Richard
684 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
685 * improvement: experimental support for mach exception handling on
686 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
688 * improvement: support for GBK external format.
689 (thanks to Chun Tian (binghe))
690 * improvement: the debugger now displays variables that have been closed
691 over, in code compiled with (DEBUG 2) or higher
692 * improvement: support for executable cores on NetBSD (thanks to
694 * new feature: added a RESTART-FRAME debugger command
695 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
696 be used to provide an implementation for FILE-POSITION on Gray streams
697 (thanks to Eric Marsden)
698 * new feature: add syslog, openlog and closelog support to SB-POSIX
699 (thanks to Richard Kreuter)
700 * optimization: the function call overhead in code compiled with
701 a high DEBUG optimization setting is significantly
702 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
703 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
704 (thanks to Tony Martinez)
705 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
706 evaluated once (reported by Kevin Reid)
707 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
708 works again (thanks to Kilian Sprotte)
709 * bug fix: an error is signaled for tagbodies with duplicate tags
710 (thanks to Stephen Wilson)
711 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
712 * bug fix: Win32 port can now handle foreign code unwinding Lisp
713 stack frames from alien callbacks.
714 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
715 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
716 65k pages (thanks to David Woodhouse)
717 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
719 changes in sbcl-1.0.1 relative to sbcl-1.0:
720 * new platform: FreeBSD/x86-64, including support for threading.
721 * new feature: the compiler stores cross-referencing information
722 about function calls (who-calls), macroexpansion (who-macroexpands)
723 and special variables (who-binds, who-sets, who-references) for code
724 compiled with (< SPACE 3). This information is available through the
725 sb-introspect contrib.
726 * new feature: users may subclass SEQUENCE, and have instances of
727 these classes interoperate with standard Common Lisp functions if
728 a number of methods are defined. (This feature is experimental
729 and the interface subject to change based on feedback from SBCL
730 users and the general community)
731 * improvement: sb-sprof traces call stacks to an arbitrary depth on
732 x86/x86-64, rather than the previous fixed depth of 8
733 * improvement: another pthread back-end of mutex "pthread-futex".
734 * bug fix: non-ascii command-line arguments are processed correctly
735 (thanks to Yaroslav Kavenchuk)
736 * bug fix: non-required arguments were not passed correctly when a method
737 defined using DEFMETHOD was called from a mop-generated method using
738 CALL-NEXT-METHOD (reported by Pascal Costanza)
739 * bug fix: recursion is now permitted in accessors through
740 SLOT-UNBOUND. (reported by Pascal Costanza)
741 * bug fix: an error was signaled at startup if the HOME environment
742 variable was defined, but had an empty value (reported by Peter Van Eynde)
743 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
744 signal a PROGRAM-ERROR, not a BUG.
745 * bug fix: SB-POSIX:READDIR works when built with large file support.
746 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
748 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
749 for better automatic indentation support. (thanks to Matt Pillsbury)
750 * optimization: loading generic functions no longer takes O(n^2) time,
751 proportional to the amount of methods in the generic function
752 (reported by Todd Sabin and Jeremy Brown)
753 * optimization: the FIND and POSITION family of sequence functions
754 are significantly faster on arrays whose element types have been
756 * improvements to the Windows port:
757 ** Intermittent heap corruption problems have been fixed. (thanks
758 to Alastair Bridgewater)
759 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
761 ** Lisp is able to unwind foreign exception frames from alien
762 callbacks. (thanks to Alastair Bridgewater)
764 changes in sbcl-1.0 relative to sbcl-0.9.18:
765 * improvement: experimental support for threading on FreeBSD/x86.
766 (thanks to NIIMI Satoshi)
767 * improvement: runtime option --dynamic-space-size can be used
768 to set the size of the dynamic space reserved on startup.
769 * improvement: floating point modes in effect are now saved in
770 core, and restored on startup.
771 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
772 startup, not time since first call to GET-INTERNAL-REAL-TIME.
773 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
774 threads are not running after *SAVE-HOOKS* have run.
775 * improvement: writes to CLOS instance slots are type-checked in code
776 compiled with (SAFETY 3)
777 * improvement: floating-point exception handling on FreeBSD (thanks to
779 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
780 (thanks to Zach Beane)
781 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
783 * improvement: added support for the Shift-JIS external format.
784 (contributed by NIIMI Satoshi)
785 * improvement: callbacks are supported on Linux/PPC. (thanks to
787 * bug fix: compiler bug triggered by a (non-standard) VALUES
788 declaration in a LET* was fixed. (reported by Kaersten Poeck)
789 * bug fix: file compiler no longer confuses validated and already
790 dumped structurres. (reported by Kaersten Poeck)
791 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
792 fill-pointer signals a type-error as required. (thanks to
794 * bug fix: disassemly of funcallable instances works.
795 * bug fix: single stepping on PPC.
796 * bug fix: fix thread-safety problems in the type system (generally
797 manifesting as nonsensical errors like "STRING is a bad type specifier
798 for sequences" or "The value 1 is not of type FIXNUM")
799 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
800 (reported by Josip Gracin)
801 * bug fix: an error is signaled for attempts to displace arrays with
802 incompatible element types (thanks to Mario Mommer)
803 * bug fix: more correct handling of wide characters in debug info
804 (bug reported by Attila Lendvai and fixed by Juho Snellman)
805 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
806 and don't cause extra consing
807 * optimization: MAP and MAP-INTO are significantly faster on vectors
808 whose elements types have been declared.
809 * Improvements to SB-SPROF:
810 ** Support for allocation profiling
811 ** Reduced profiling overhead, especially for long profiling runs
812 * Improvements to the Windows port:
813 ** floating point exceptions are now reported correctly.
814 ** stack exhaustion detection works partially.
815 ** more accurate GET-INTERNAL-REAL-TIME.
816 ** better RUN-PROGRAM behaviour in case of errors in the child
818 ** PROBE-FILE now simplifies pathnames correctly.
819 ** DIRECTORY now works correctly with :WILD-INFERIORS.
820 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
821 1st 1970 (although time timezone data will be unreliable).
822 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
824 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
825 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
826 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
828 * bug fix: two potential GC deadlocks affecting threaded builds.
829 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
830 returns T (reported by Anton Kazennikov)
831 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
832 with non-variable places
833 * bug fix: remove a race condition in the setting of
834 funcallable-instance functions, this should make threaded CLOS
835 code more stable against memory faults.
836 * bug fix: corruption of specials when unbinding is interrupted by an
837 asynchronous unwind (reported by Hannu Koivisto)
838 * improvement: the debugger will now also display local variables that
839 are only used once, for code compiled with a DEBUG optimization quality
842 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
843 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
844 * incompatible change: External-format support for FFI calls. The
845 SB-ALIEN:C-STRING no longer implies an ASCII
846 external-format. Instead the string is subject to external-format
847 conversion. Additionally return values of type SB-ALIEN:C-STRING
848 are by default of type (SIMPLE-ARRAY CHARACTER), not
849 SIMPLE-BASE-STRING. For an alien type that behaves like the old
850 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
851 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
852 * incompatible change: SB-EXT package no longer contains the
853 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
854 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
856 * incompatible change: the single-stepper is no longer available
857 on Alpha, Mips and Sparc platforms.
858 * minor incompatible change: the direct superclasses of
859 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
860 not (STANDARD-OBJECT FUNCTION). This makes the
861 class-precedence-lists of GENERIC-FUNCTION and
862 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
864 * minor incompatible change: the default stream external format on
865 non-unicode SBCL is no longer detected from the locale, but is always
867 * new feature: Add a version of evaluator that uses an interpreter instead
868 of the compiler. EVAL still uses the compiler by default, to switch it
869 to use the interpreter, set the value of the variable
870 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
871 * minor incompatible change: the single-stepper REPL has been merged
872 with the normal debugger (see the "Stepping" heading of the debugger help
873 for more details). The debugger command STEP will no longer switch
874 to the single-stepper REPL.
875 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
876 for a type now works.
877 * bug fix: loading of default sysinit file works. (thanks to Leonid
879 * bug fix: better detection of circularities in the file-compiler.
880 (reported by Marco Monteiro)
881 * bug fix: the CL pathname functions now work with files that have
882 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
883 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
884 systems with Unix98 pty semantics.
885 * bug fix: ASDF-INSTALL will now work with bsd tar.
886 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
888 * bug fix: timers expiring in dead threads no longer cause a
889 type-error (reported by Paul "Nonny Mouse").
890 * bug fix: thanks to more lightweight single-stepper instrumentation,
891 code compiled with (DEBUG 3) will compile and execute significantly faster,
892 and will have more accurate type-inferencing than before
893 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
894 whose bindings are modified
895 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
896 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
897 ** replace dummy implementations of CL:MACHINE-INSTANCE and
898 CL:SOFTWARE-VERSION with proper version
900 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
901 * feature: implemented the READER-METHOD-CLASS and
902 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
903 as specified by AMOP.
904 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
906 * optimization: faster LOGCOUNT implementation on x86 and x86-64
907 (thanks to Lutz Euler)
908 * optimization: hashing of general arrays and vectors has been
909 improved. (reported by Any Fingerhut)
910 * enhancement: SB-INTROSPECT is now able to find definitions of
911 profiled functions. (thanks to Troels Henriksen)
912 * enhancement: compiler-macro expansion applies now to FUNCALL forms
913 as well, allowing compiler-macros for SETF-functions to expand.
914 * enhancement: step-instrumentation no longer wraps known
915 single-value functions with multiple-value context, allowing
916 better type inference.
917 * fixed bug #337: use of MAKE-METHOD in method combination now works
918 even in the presence of user-defined method classes. (reported by
919 Bruno Haible and Pascal Costanza)
920 * fixed bug #339(c): if there are applicable methods not part of any
921 long-form method-combination group, call INVALID-METHOD-ERROR.
922 (reported by Bruno Haible)
923 * fixed bug #361: the :FUNCTION initarg in the protocol for
924 initialization of methods can now be used to override
925 internally-produced optimized functions. (reported by Bruno
927 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
928 system-provided lambda expression no longer cause warnings about
929 unbound #:|pv-table| symbols.
930 * bug fix: improved the handling of type declarations and the
931 detection of violations for keyword arguments with non-constant
933 * bug fix: potentially erronous calls to PATHNAME and
934 MERGE-PATHNAMES were being flushed in some cases.
935 (reported by Richard Kreuter)
936 * bug fix: compiled calls to TYPEP were mishandling obsolete
937 instances. (reported by James Bielman and Attila Lendvai)
938 * bug fix: format strings with ~> without matching ~< no longer
939 trigger an AVER, but signal an understandable error instead.
940 (reported by Antonio Martinez)
941 * bug fix: specifying an output-file without a pathname-type for
942 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
943 of input-file instead of "fasl". (reported by Robert Dodier)
944 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
945 (reported by James Y Knight).
946 * bug fix: compiler-macros-function did not consider the environment
947 argument for shadowing by local functions.
948 * bug fix: compiler-macros expansion was inhibited by local INLINE
950 * bug fix: inline expansions of known functions were subject to
951 step-instrumentation in high debug policies, leading to problems
953 * bug fix: compiler failed to differentiate between different CONS
955 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
956 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
957 * bug fix: #368: incorrect use of expressed vs. upgraded array
959 * bug fix: #306a: more precise unions of array types.
960 * thread-safety improvements:
961 ** CONDITION-WAIT could return early on Linux, if the thread was
962 interrupted and subsequently continued with SIGCONT.
963 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
965 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
966 * added support for the ucs-2 external format. (contributed by Ivan
968 * minor incompatible change: pretty printing of objects of type
969 (cons symbol) is, in the default pprint-dispatch-table, now
970 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
972 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
973 called later in a class's lifetime, possibly as late as when the
974 first instance of the class is created. Previously,
975 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
976 class became finalizeable.
977 * fixed bug: FILE-POSITION sometimes returned inconsistent results
978 for multibyte external-format streams. (thanks to "vbzoli")
979 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
980 with :ALLOCATION :CLASS inherited from superclasses of the
982 * fixed bug: anonymous classes can now be created using the :NAME
983 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
984 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
985 * fixed bug: core-files saved with :EXECUTABLE T can again be
986 executed when SBCL_HOME isn't set. (reported by James Knight)
987 * fixed bug: toplevel LOCALLY forms with declarations could
988 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
989 * fixed bug: printing from several different threads using different
990 values of *print-case* could cause invalid output, due to
991 some internal special variables of the printer not being bound
992 thread-locally (reported by Max Mikhanosha)
993 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
994 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
995 (reported by Pascal Costanza)
996 * fixed bug: FINALIZE-INHERITANCE is called from
997 REINITIALIZE-INSTANCE on classes when the class has previously
998 been finalized, as required by AMOP.
999 * minor code generation optimizations:
1000 ** better register allocation in CLOS dispatching functions
1001 ** overflow detection when coercing signed bytes to fixnums on x86-64
1002 ** is now implemented with one IMUL instruction instead of three shifts
1003 ** more efficient bit-vector access on x86 and x86-64
1004 ** more efficient access to raw structure slots on x86 and x86-64
1005 * fixed some bugs revealed by Paul Dietz' test suite:
1006 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1007 return its argument.
1009 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1010 * feature: thread support on Solaris/x86, and experimental thread support
1012 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1014 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1015 default method for SLOT-UNBOUND.
1016 * minor incompatible change: prevent the user from specializing the
1017 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1018 somewhat counter to the intent of the protocol, I (CSR) think, and
1019 additionally it just doesn't work in SBCL as currently
1020 implemented, thanks to optimizations (that are always valid for
1021 the other three SLOT-VALUEish functions, but not for the setter).
1022 * minor incompatible change: the :SB-LDB feature is now enabled by
1023 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1024 the low-level debugger.
1025 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1026 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1027 down to an SB-SYS:FD-STREAM, allowing direct communication in
1028 more cases. Temporary files and pipes are still used as a fallback
1030 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1031 on the non-reentrant C functions localtime(3) and gmtime(3).
1032 * bug fix: global optimization policy was not visible in LOCALLY and
1034 * bug fix: class objects can be used as specializers in methods.
1035 (reported by Pascal Costanza)
1036 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1037 with a type error. (reported by blitz_ on #lisp)
1038 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1039 failed with a type error. (reported by Pascal Bourguignon)
1040 * bug fix: merging pathnames against defaults with :DIRECTORY
1041 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1042 (reported by James Y Knight)
1043 * bug fix: saving large (>2GB) cores on x86-64 now works
1044 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1045 constant index and a (simple-array (signed-byte 32)) array
1046 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1047 error (patch by Robert J. Macomber)
1048 * feature: TIME now displays the amount of run-time spent in GC
1049 * bug fix: The debugger now does a better job of respecting
1050 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1051 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1052 workaround for bug 403.)
1053 * bug fix: floating point exception handling now works on FreeBSD
1054 (thanks to NIIMI Satoshi)
1055 * fixed some bugs revealed by Paul Dietz' test suite:
1056 ** MISC.641: LET-conversion were not supposed to work in late
1058 * improvements to the Win32/x86 port:
1059 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1060 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1062 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1063 * new feature: source path information is generated for macro-expansion
1064 errors for use in IDE's like Slime (thanks to Helmut Eller)
1065 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1066 * bug fix: compiler does not loop forever on an invalid type in
1068 * improvement: compilation of most CLOS applications is significantly
1070 * optimization: added a limited bytecode compiler for simple toplevel
1071 forms, speeding up compilation and FASL loading
1072 * bug fix: the statistical profiler now properly distinguishes anonymous
1075 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1076 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1077 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1078 into subclasses of the finalized class. Now user calls to
1079 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1080 system return before any subclasses are finalized.
1081 * minor incompatible change: The reader no longer ignores errors
1082 regarding non-existent packages in #+ and #- feature tests.
1083 * new feature: command line options --no-sysinit, --no-userinit to
1084 inhibit loading the corresponding init files
1085 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1086 for sending data through UDP sockets (thanks to François-René Rideau)
1087 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1088 error is signalled instead (thanks to François-René Rideau)
1089 * minor incompatible change: Error signalling behaviour of lexical
1090 operations violating package locks has changed slightly. Refer to
1091 documentation on package locks for details.
1092 * enhancement: EVAL can process IF-expressions without resorting to the
1094 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1095 constant-folded properly due to missing out-of-line functions.
1096 (reported by Mika Pihlajamäki)
1097 * bug fix: LISTEN sometimes returned T even in cases where no data was
1098 immediately available from the stream
1099 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1100 were reversed. (reported by Levente Mészáros)
1101 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1102 when given a vector displaced to an adjustable array. (reported
1104 * bug fix: derivation of float boundaries from numbers outside the
1105 appropriate float range (reported by John Wiseman)
1106 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1108 * bug fix: precomputing cache entries for generic functions with
1109 some subclasses of specializers as yet invalid does not attempt to
1110 fill a cache line with a negative offset. (reported by Levente
1112 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1113 allow condition class objects as arguments to DOCUMENTATION and
1114 (SETF DOCUMENTATION); only find and set documentation for
1115 structure names for the STRUCTURE doc-type. (suggested by Gary
1117 * improvements to the Win32/x86 port:
1118 ** (user-homedir-pathname) and default initialization file
1119 locations now know about the user's "Documents and Settings"
1120 directory (thanks to Yaroslav Kavenchuk)
1121 ** run-program is implemented (thanks to Mike Thomas)
1122 ** sockets support (thanks to Timothy Ritchey)
1123 ** better backtrace support (thanks to Alastair Bridgewater)
1124 ** sb-grovel supported
1125 ** asdf-install and sb-posix work somewhat
1126 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1127 * improvements to the Solaris/x86 port:
1128 ** works on Solaris 11/Solaris Express
1129 ** floating-point exception handling support
1130 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1131 * fixed some bugs revealed by Paul Dietz' test suite:
1132 ** REMOVE-METHOD returns its generic function argument even when
1133 the method is not one of the generic functions' methods.
1134 ** objects declared as MEMBER types can be admissible arguments to
1135 structure accessors.
1136 ** printing characters should simply be printed by the FORMAT ~:C
1138 ** compiler failure when compiling functions with hairy constant
1139 defaults for optional parameters.
1140 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1141 function, which is already optimized.
1143 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1144 * new platform: experimental support for SBCL x86/Darwin, including
1145 MacOS X 10.4.5 on Intel.
1146 * new feature: Unicode character names are now known to the system
1147 (through CHAR-NAME and NAME-CHAR).
1148 * new feature: the filesystem location of SBCL's core file is
1149 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1150 * minor incompatible change: the contrib modules SB-POSIX and
1151 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1152 this change is to make it easier to distribute
1153 location-independent binaries.
1154 * bug fix: as implied by AMOP, standardized classes no longer have
1155 slots named by external symbols of public packages. (reported by
1157 * optimization: calling functions via constant symbols -- as in
1158 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1159 via the function object as in (FUNCALL #'FOO).
1160 * optimization: on x86 and x86-64, the calling convention for the
1161 case of dealing with an unknown number of values has been altered
1162 to be friendlier to the prediction heuristics implemented,
1163 particularly with reference to CALL and RET pairing. (thanks to
1164 Alastair Bridgewater)
1165 * optimization: on x86, the code for access to thread-local symbol
1166 values and binding/unbinding of thread-local symbols is smaller.
1167 (thanks to Alastair Bridgewater)
1168 * enhancement: CONSTANTP is now able to determine constantness of
1169 more complex forms, including calls to constant-foldable standardized
1170 functions and some special forms beyond QUOTE.
1171 * fixed bug: occasional GC crashes on Solaris/x86
1172 * optimization: x86-64 supports stack allocation of results of simple
1173 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1174 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1175 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1176 (thanks to James Knight)
1177 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1178 by the standard (thanks to Stephen Compall)
1180 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1181 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1182 be used for bundling the runtime and the core file into one
1183 executable binary. This feature is not currently supported on all SBCL
1184 platforms. (thanks to James Bielman and NIIMI Satoshi)
1185 * new feature: a generational or ephemeral garbage collector is now
1186 the default on the PowerPC platform (both Linux and Darwin). The
1187 old Cheney (stop and copy) collector is a build-time option.
1188 (thanks to Cyrus Harmon, after Raymond Toy)
1189 * minor incompatible change: the method by which SBCL finds its
1190 contributed modules has changed; it no longer relies on symbolic
1191 links from an $SBCL_HOME/systems directory, but searches directly
1192 in the subdirectories of $SBCL_HOME.
1193 * enhancement: the dynamic heap size on the Linux/PPC platform is
1194 markedly larger, even using the older Cheney garbage collector.
1195 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1196 certain complicated string types. (reported by Gary King)
1197 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1198 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1199 * fixed bug: DOCUMENTATION on structures no longer signals an
1200 error if no structure type of the right name exists. (reported by
1202 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1203 applicable non-standard primary or :AROUND method on
1204 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1205 no longer cause unbound variable errors. (reported by Kevin Reid)
1206 * fixed bug: metacircle resolution in cases where methods have slots
1207 added before the slots from STANDARD-METHOD. (reported by Jean
1209 * fixed bug: the Power PC assembly code for calling into lisp
1210 sometimes computed its return address wrongly (depending on the
1211 15th bit of the address, fixed at link time). This used to
1212 manifest itself in a segmentation violation while building PCL.
1213 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1214 many others over the years)
1215 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1216 the stack pointer (treated as a signed integer) is negative.
1217 (thanks to Peter van Eynde)
1219 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1220 * new platform: experimental support for the Windows operating
1221 system has been added. (thanks to Alastair Bridgewater)
1222 * fixed several bugs in and robustified the PPC FFI (including
1223 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1224 * bug fix: allow non-simple string symbol names (reported by Paul
1226 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1228 * bug fix: saving a core corrupted callbacks on x86/x86-64
1229 * bug fix: closed a loophole in metacircularity detection and
1230 grounding in the PCL implementation of CLOS.
1231 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1232 (reported by tomppa on #lisp)
1233 * bug fix: clear the direction flag on Lisp -> C transitions, as
1234 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1235 (reported by Andrej Grozin and Hendrik Maryns)
1236 * optimization: major improvements to GC efficiency on GENCGC platforms
1237 * optimization: faster implementation of EQUAL
1238 * optimization: emit more efficient opcodes for some common
1239 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1240 * optimization: several other minor code-generation improvements on x86-64
1241 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1243 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1244 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1245 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1246 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1247 by ANSI as a generic function, and for consistency (SETF
1248 GENERIC-FUNCTION-NAME) is treated likewise.
1249 * fixed bug #233.b: make constraint propagation notice when a variable
1250 value is changed after it is referenced but before it is used
1251 * fixed bug #296: no more arbitrary behaviour from filenames with
1252 odd characters as --load arguments. (reported by Adam Warner)
1253 * fixed bug #390: :CHARACTER-SET pathname components now work as
1254 expected. (reported by Tim Daly Jr)
1255 * fixed bug #391: complicated :TYPE intersections in slot
1256 definitions no longer cause an error in PCL internals.
1257 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1258 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1259 returning the number of octets which would be written to the
1260 file-stream. (thanks to Robert J. Macomber)
1261 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1262 arguments to the internal stream functions. (thanks to David
1264 * bug fix: stack exhaustion is now handled on the x86/SunOS
1265 platform. (thanks to Daisuke Homma)
1266 * bug fix: more accurate ROOM results on GENCGC platforms
1267 * optimization: improved type inference for arithmetic-for
1268 index variables in LOOP
1269 * optimization: faster floating-point SQRT on x86-64
1270 * fixed some bugs revealed by Paul Dietz' test suite:
1271 ** DOCUMENTATION returns NIL instead of "" for method combinations
1272 that don't have a docstring
1274 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1275 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1276 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1277 therefore are not customizeable by user code (as seems to be at
1278 least permitted and maybe required by AMOP). As a consolation,
1279 however, the SBCL implementation of these functions now calls
1280 REINITIALIZE-INSTANCE as specified by AMOP.
1281 * bug fix: slot-definition documentation is propagated as per ANSI
1282 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1283 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1284 argument, as do eslotd initialization methods. (from Pascal
1285 Costanza's "Closer" project)
1286 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1287 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1289 * bug fix: it is now possible to have more than one subclass of
1290 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1291 (reported by Bruno Haible, Pascal Costanza and others)
1292 * bug fix: the dependent update protocol now works for generic
1293 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1294 and Pascal Costanza)
1295 * bug fix: condition-class instances corresponding to
1296 DEFINE-CONDITION forms are now created eagerly. (reported by
1297 Kalle Olavi Niemitalo on comp.lang.lisp)
1298 * bug fix: floating point printing is more accurate in some
1299 circumstances. (thanks to Simon Alexander)
1300 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1301 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1302 * bug fix: callbacks on OS X now preserve stack-alignment.
1303 (thanks to Cyrus Harmon)
1304 * enhancement: the x86-64 disassembler is much better at
1305 disassembling SSE instructions. (thanks to Lutz Euler)
1306 * enhancement: CHECK-TYPE on a variable now allows type-inference
1307 (thanks to Vincent Arkesteijn)
1308 * optimization: improved performance of EUC-JP external format.
1309 (thanks to NIIMI Satoshi)
1310 * optimization: performance improvements to IO on file streams of
1311 :ELEMENT-TYPE CHARACTER
1312 * optimization: much faster memory allocation on x86-64
1313 * optimization: faster garbage collections (latency and throughput)
1315 * optimization: faster bignum subtraction and division on x86 and x86-64
1316 (thanks to Lutz Euler)
1317 * optimization: more accurate type-inference for loops that use a
1318 floating point index variable or a negative step.
1320 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1321 * bug fix: add a workaround to SBCL looping infinitely at startup on
1322 Linux kernels with apparently buggy implementations of personality().
1323 (thanks to Svein Ove Aas)
1324 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1325 (thanks to David Lichteblau)
1326 * bug fix: local bindings shadow global symbol macros for
1327 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1328 * optimization: non-open coded uses of numeric comparison operators
1329 (e.g. >) no longer cons when called with more than one parameter
1330 on platforms supporting dynamic-extent allocation.
1331 * enhancement: saving cores with foreign code loaded is now supported
1332 on MIPS/Linux in addition to the previously supported platforms.
1333 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1334 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1335 * bug fix: division by zero in sb-sprof when no samples were collected
1336 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1337 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1338 (thanks to Alexander Kjeldaas)
1339 * new feature: ignoring the return values of destructive functions
1340 like NREVERSE signals a compile-time style-warning.
1341 (thanks to Kevin Reid)
1342 * new feature: the alignment of alien structure fields can be
1343 explicitly specified. (thanks to Cyrus Harmon)
1344 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1345 and dump core on SIGQUIT
1347 ** incompatible change: threads do not inherit values of specials
1348 from their parents (see manual)
1349 ** bug fix: threads stacks belonging to dead threads are freed by the
1350 next exiting thread, no need to gc to collect thread stacks anymore
1351 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1352 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1353 ** bug fix: dynamic variable and thread start related gc lossage
1354 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1356 * fixed some bugs revealed by Paul Dietz' test suite:
1357 ** SUBTYPEP is slightly more accurate on heinously complicated
1358 CONS types where some of the members have uncertain (in the
1359 NIL, NIL sense) type relationships to each other.
1361 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1362 no longer purified unless :PURIFY T is explicitly specified.
1363 ** Non-purified cores are significantly smaller than before
1365 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1366 * new feature: timers based on Zach Beane's excellent timer package
1367 * added support for the following external formats: koi8-u,
1368 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1369 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1370 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1371 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1372 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1373 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1374 * incompatible change: a threaded SBCL will no longer revert to
1375 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1376 * bug fix: interrupts are disabled until startup is complete; no
1377 more sigsegvs when receiving a signal to soon
1378 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1380 * bug fix: add a workaround for the memory randomization features in
1381 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1382 workaround will only be in effect on systems with the proc filesystem
1384 * bug fix: printing objects of type HASH-TABLE signals a
1385 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1387 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1388 have been running for over 50 days. (reported by Gilbert Baumann)
1389 * bug fix: the logic for getting names of functions gets less
1390 confused when confronded with alternate-metaclass
1391 funcallable-instances. (reported by Cyrus Harmon)
1392 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1393 consistent, even on internal alternate-metaclass objects.
1394 * bug fix: flush closure information collected by physical
1395 environment analysis prepass before the main pass. (bug reported
1397 * bug fix: compiler pack phase does not modify a hashtable, which is
1398 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1399 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1400 may be expanded. (reported by Alexander <malishev> on c.l.l)
1401 * bug fix: no more style-warnings when compiling code using LAST
1402 under high-SPEED low-SPACE settings. (reported by David Wallin
1404 * fixed some incompatibilities between SBCL's MOP and the MOP
1406 ** the METAOBJECT class is now implemented;
1407 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1408 STANDARD-OBJECT, as required;
1409 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1410 now compatible (as required by VALIDATE-SUPERCLASS); there
1411 remains a consistency requirement over the presence or absence
1412 of the FUNCTION class in the superclass list of finalized
1413 classes; see the manual for more details;
1414 ** the :AROUND method for COMPUTE-SLOTS on
1415 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1416 requested slot ordering.
1418 ** bug fix: parent thread now can be gc'ed even with a live
1420 ** bug fix: binding a special with PROGV to no value is not
1421 the same as that symbol not having been bound (thanks to
1423 ** bug fix: binding specials is thread safe (thanks to
1425 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1426 and SB-SPROF do not die with 'no handler for signal XX in
1427 interrupt_handle_now(..)' anymore
1428 ** bug fix: WITH-TIMEOUT works with multiple running threads
1429 * fixed some bugs revealed by Paul Dietz' test suite:
1430 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1431 the :method-class keyword argument.
1433 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1434 * new port: the Solaris operating system on x86 processors is now
1435 mostly supported, though some rough edges in the environment
1436 remain. (thanks to Daisuke Homma)
1437 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1438 space for its heap. (thanks to Thiemo Seufer)
1439 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1440 closures. (thanks to Thiemo Seufer)
1441 * minor incompatible change: eof selects abort in the debugger.
1442 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1443 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1445 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1446 deprecated, and will go away in a future revision of SBCL.
1447 * minor incompatible change: GC-ON and GC-OFF are no longer
1448 implemented with a counter, it does not matter how many times gc
1449 is switched on or off
1450 * bug fix: discriminating functions for generic function classes
1451 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1452 make invalid assumptions about method precedence order. (reported
1454 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1455 platforms. (thanks to Thiemo Seufer)
1456 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1457 * bug fix: correct pathname printing with printer escaping is on.
1458 (thanks to Kevin Reid)
1459 * bug fix: complex VOP definitions in "user-space" no longer trigger
1460 package locks. (reported by Zach Beane)
1461 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1462 for funcallable-instance functions, and is no different from
1464 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1465 characters in the current readtable. (reported by Nicholas Neuss)
1466 * bug fix: SUBTYPEP on various CONS types returns more a more
1467 accurate acknowledgment of its certainty.
1468 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1469 lists in some cases. This partially fixes bug 384.
1470 * flush all standard streams before prompting in the REPL and the
1472 * bug fix: signal handling and triggering gc do not conflict
1473 directly anymore, in particular a high frequency sb-sprof does
1474 not prevent gc from running
1475 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1476 approximation for timezone and DST information between the
1477 universal time epoch and the smallest negative 32-bit time_t.
1478 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1479 year 1899 so long as the encoded time is non-negative.
1480 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1481 the correct answer for non-constant rotate amounts when performing
1482 an inline 32-bit rotation.
1484 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1485 there is only one thread in the session
1486 ** bug fix: memory leak for streams created in one thread and
1487 written to in another
1488 ** bug fix: lockup when compiled with gcc4
1489 ** bug fix: race that allows the gc to be triggered when gc is
1491 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1492 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1494 ** bug fix: finished threads are gc'ed properly
1495 * fixed some bugs revealed by Paul Dietz' test suite:
1496 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1497 has arguments with a different set of applicable methods from
1498 the orignal arguments.
1499 ** The type error thrown by MAP now has a correct expected-type
1501 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1502 name a compiled function.
1503 ** Three MISC tests where a large bignum was improperly coerced to
1504 a float that couldn't represent that bignum during type
1505 derivation were fixed.
1506 ** SUBTYPEP can now handle types involving the negation of a
1507 list-form FUNCTION type.
1508 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1509 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1510 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1512 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1513 * New feature: Experimental support for bivalent streams: streams
1514 opened with :element-type :default now allow character and binary
1515 (unsigned-byte 8) I/O
1516 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1517 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1519 * bug fix: no more highly sporadic "couldn't check whether ~S is
1520 readable" when reading a stream and an interrupt hits in the middle
1521 of a select system call
1522 * compiler better recognizes complex arrays (reported by Eduardo
1524 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1525 for variables on SBCL built with linkage-tables. (reported by Luis
1527 * various error reporting improvements.
1528 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1529 (Thanks to Thiemo Seufer)
1530 * bug fix: nested reader invokations work correctly
1531 * bug fix: it is possible to have simultaneous references to foreign
1532 code and foreign data with the same name.
1534 ** added x86-64 support
1535 ** incompatible change: the threading api now works with thread
1536 objects instead of thread ids
1537 ** bug fix: threads are protected from signals and interruption when
1538 starting up or going down
1539 ** bug fix: a race where an exiting thread could lose its stack to gc
1540 ** bug fix: don't halt on infinite error in threads if possible
1541 ** fixed numerous gc deadlocks introduced in the pthread merge
1542 ** bug fix: fixed thread safety issues in read and print
1543 ** bug fix: debugger doesn't hang on session lock if interrupted at
1544 an inappropriate moment
1545 ** bug fix: run-program is now thread safe(r)
1546 ** bug fix: inner with-recursive-lock no longer releases the mutex
1547 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1548 * fixed some bugs revealed by Paul Dietz' test suite:
1549 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1550 EXPECTED-TYPE slots filled.
1551 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1553 ** MAKE-CONDITION accepts classes as type-designators.
1554 ** COMPILE may never return NIL.
1555 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1556 range before calling Unix time functions
1558 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1559 * numerous signal handling fixes to increase stability
1560 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1561 * minor incompatible change: we now correctly canonize default
1562 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1563 as per the MOP, rather than the historical (INITARG INITFUNCTION
1564 INITFORM). (reported by Bruno Haible)
1565 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1566 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1567 for more information.
1568 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1569 pathname is a directory pathname.
1570 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1571 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1573 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1574 :METACLASS argument in addition to a class name. (reported by
1575 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1576 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1577 (reported by Bruno Haible; patch from Gerd Moellmann)
1578 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1580 * bug fix: more cleanups to the floating point exception handling on
1581 x86-64 (thanks to James Knight)
1582 * bug fix: the compiler does not try to constant fold calls of
1583 COERCE and BIT-* functions when they return freshly constructed
1584 objects. (reported by Nikodemus Siivola and Paul Dietz)
1585 * optimization: improved the MIPS versions of generic (in the
1586 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1587 * optimization: direct conversion of (unsigned-byte 32) to floats on
1588 the PowerPC platform.
1589 * optimization: structure instances with raw slots now use less
1590 memory, and probably show better memory locality. (thanks to
1592 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1593 are treated as requests for stack allocation on the x86-64,
1594 Alpha-32, PPC and SPARC platforms.
1595 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1596 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1597 (thanks to Tony Martinez)
1599 ** gcing a dead thread can no longer lead to lockups
1600 ** threads block signals until they are set up properly
1601 ** errno is no longer shared by threads
1602 ** interrupt-thread restores the eflags register on x86
1603 ** fixed some lockups due to gc/thread interaction
1604 * fixed some bugs revealed by Paul Dietz' test suite:
1605 ** invalid dotted lists no longer raise a read error when
1606 *READ-SUPPRESS* is T
1607 ** signal an error if a symbol that names a declaration is used
1608 as the name of a type, or vice versa
1609 ** allow using the (declare (typespec var*)) abbreviation for
1610 (declare (type typespec var*)) with all type specifiers
1611 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1612 ** always return NIL from PPRINT-POP when OBJECT is NIL
1613 ** don't signal errors when pretty-printing malformed LABELS,
1614 FLET or MACROLET forms
1615 ** declarations in a DOLIST body are also in scope for the
1617 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1619 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1622 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1623 * fixed cross-compiler leakages that prevented building a 32-bit
1624 target with a 64-bit host compiler.
1625 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1626 opened with :IF-EXISTS :OVERWRITE.
1627 * fixed bug 281: error for an invalid qualifier in a short-form method
1628 combination method is not signalled until the faulty method is called.
1629 * bug fix: iteration variable type inferrer failed to deal with open
1630 intervals. (reported by Alan Shields)
1631 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1632 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1633 or more runtime options were provided to the sbcl binary.
1634 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1635 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1636 * optimization: there's now a fast-path for fixnum arguments in the
1637 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1638 * optimization: the code generated on x86-64 is more compact thanks
1639 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1640 * bug fix: floating-point exception handling now partly works on
1641 x86-64. (Thanks to James Knight)
1642 * improvement to the MIPS backend from Thiemo Seufer: C-style
1643 64-bit long long arguments and return values to alien functions
1645 * fixed some bugs revealed by Paul Dietz' test suite:
1646 ** the type-error signalled from WARN has a filled-in DATUM slot.
1647 ** the type-error required when a stream is not associated with
1648 a file has the stream as its datum.
1649 ** type-errors on single-floats on x86-64 no longer have
1650 :INVALID-OBJECT as the datum
1651 ** the type-errors signalled for invalid function names now have
1652 a correct expected type
1653 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1654 for structures defined with a :TYPE. Documentation strings for
1655 typed structures are no longer immediately discarded
1656 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1657 broadcast streams and synonym streams. FILE-LENGTH now also works
1658 on broadcast streams.
1660 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1661 * incompatible change: the --noprogrammer option, deprecated since
1662 version 0.7.5, has been removed. Please use the equivalent
1663 --disable-debugger option instead.
1664 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1666 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1667 inoperational for a while now) has been completely removed.
1668 * null lexical environments are now printed as #<NULL-LEXENV>,
1669 significantly reducing the amount of clutter in typical backtraces.
1670 * documentation on weak pointers, finalization, and after GC hooks
1671 has been added to the manual.
1672 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1673 as other specialized array types, is much faster. SUBSEQ and
1674 COPY-SEQ on such arrays have also been sped up.
1675 * optimization: EQL is now more efficient when at least other argument
1676 is known to be of type (OR FIXNUM (NOT NUMBER)).
1677 * fixed bug: the runtime is now less vulnerable to changes in the
1678 size of the SBCL object on OS X, and virtual memory is reserved for
1679 all spaces that need to be at a fixed address.
1680 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1681 * fixed bug: finalizers and after GC hooks that cause consing are now
1683 * fixed bug: compiler error messages and summaries are now printed to
1684 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1685 * fixed inference of the upper bound of an iteration variable.
1686 (reported by Rajat Datta).
1687 * fixed bug 211e: calling local functions with duplicated constant
1688 keyword argument no longer causes a bogus style warning about an
1690 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1691 ftype declaration to be disregarded. (reported by Dave Roberts)
1692 * fixed bug 373: caused by erronous compilation of references to alien
1693 variables in the runtime on ppc/darwin.
1694 * fixed bug 376: CONJUGATE type deriver.
1695 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1696 (reported by Baughn on #lisp)
1697 * a cross-compiler bug on non-x86 platforms has been identified and
1698 fixed. (thanks to Bruno Haible)
1699 * improvements to the MIPS runtime code for increased stability.
1700 (thanks to Thiemo Seufer)
1701 * increased the maximimum compact environment size to allow
1702 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1703 * partial workaround for bug 135: don't name the function we're
1704 calling for hairy cases of EVAL, so as not to accumulate one environment
1705 entry per call to EVAL. (reported by Kevin Reid)
1706 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1707 * optimization: type testing for non-vector arrays should be faster.
1708 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1710 * bug fix: setting 31st element of a bit vector to zero did not work
1712 * bug fix: redefining a class definition which failed due to a
1713 previous accessor / function clash now works (but see BUGS entry
1714 #380 for more problems in this area). (thanks to Zach Beane)
1715 * the long form of DEFINE-METHOD-COMBINATION disables method group
1716 checking when given a single method group with pattern *. (thanks
1718 * on x86 compiler supports stack allocation of results of simple
1719 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1720 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1721 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1722 * fixed some bugs related to Unicode integration:
1723 ** the restarts for recovering from input and output encoding
1724 errors only appear when there is in fact such an error to
1726 * fixed some bugs revealed by Paul Dietz' test suite:
1727 ** MISC.549 and similar: late transformation of unsafe type
1728 assertions into derived types caused unexpected code
1730 ** SCALE-FLOAT type deriver is less wrong.
1731 ** type derivers for EXP, LOG and similar functions compute result
1732 types for complex arguments better.
1733 ** (MISC.563) CONJUGATE type deriver works for very restricted
1735 ** out-of-line type testers for character strings are available.
1736 ** EQUAL compiler transform understands specialness of objects
1738 ** accessing double-floats stored on the stack now works on x86-64.
1739 ** debugger internals could sometimes create invalid lispobjs,
1740 resulting in GC crashes.
1741 ** MISC.548: type check weakening can convert required type into
1743 ** initialization forms for bindings are not in scope of free special
1746 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1747 * incompatible change: thread support for non-NPTL systems has
1748 been removed - locking is buggy and unreliable. A threaded
1749 SBCL build will now warn at startup and refuse to create
1750 new threads, unless futex support is detected in the kernel
1751 * incompatible change: the top level REPL now has only an ABORT
1752 restart associated with it, not TOPLEVEL and ABORT as it used to.
1753 TOP and TOPLEVEL are now available as debugger commands for
1754 returning to the top level.
1755 * incompatible change: forms evaluated in the REPL now use the
1756 global optimization policy.
1757 * incompatible change: user- and system-initialization files are
1758 no longer processed with LOAD, but by READ and EVAL; hence the
1759 global optimization policy, startup package, readtable, etc,
1761 * The .fasl file format number has been incremented because of
1762 various incompatible changes.
1763 * internal entry point details and argument counts no longer appear
1764 in backtraces unless explicitly requested by setting
1765 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1766 * built-in and standard functions no longer have names like "top
1767 level local call to FOO".
1768 * fixed bug 32: functions defined in non-null lexical environments
1769 now have more legible printed representation
1770 * fixed bug 33: functions defined in non-null lexical environemnts
1771 are now more amenable to inspection by INSPECT.
1772 * workaround for bug 354: XEPs no longer appear in backtraces unless
1773 explicitly requested.
1774 * fixed bug: receiving the signal which results from attempting to
1775 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1776 SIGBUS on other *BSDs) on architectures where the C stack is also
1777 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1778 to debugger confusion or wild instability. (reported by Cheuksan
1780 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1781 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1782 (reported by Lutz Euler)
1783 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1784 compiled code. (reported by Paul Dietz)
1785 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1786 specialized parameter to a method to something that is not TYPEP
1787 the specializer is now possible.
1788 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1789 face of package deletion.
1790 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1791 pathname. (reported by David Tolpin on c.l.l)
1792 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1793 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1794 STRING-STREAM and FILE-STREAM, respectively.
1795 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1796 than previously; in addition, its test suite is now run on build.
1797 (thanks to Luke Gorrie)
1798 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1799 * optimization: added a immediate representation for single-floats
1801 * fixed some bugs revealed by Paul Dietz' test suite:
1802 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1803 ** attempting to create a package with a colliding nickname causes
1804 correctable errors to be signalled.
1805 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1806 ** several x86-64 backend bugs related to sign-extension of immediate
1809 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1810 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1811 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1813 * fixed loading of multiply forward-referenced layouts.
1814 (thanks to Cheuksan Wang)
1815 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1816 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1817 (see COMPILE-FILE documentation for details of :PRINT option)
1818 * fixed bugs 19 and 317: fixed-format floating point printing is
1819 more accurate. This also fixes a bug reported by Adam Warner
1820 related to the ~@F format directive.
1821 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1823 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1824 dispatch table if the from-char is a dispatch macro character.
1825 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1826 by the wordsize no longer ignore the last word. (reported by Lutz
1828 * fixed bug in type checking of dynamic-extent variables. (reported
1830 * optimization: sequence traversal functions use their freedom to
1831 coerce function designators to functions.
1832 * optimization: code with many calls to CLOS methods specialized on
1833 CLOS classes has had redundant type checks removed.
1834 * optimization: added declarations to speed up operations that access
1835 the internal character database (for example STRING-UPCASE)
1836 * optimization: comparison operations between floats and sufficiently small
1837 fixnums no longer create extra rationals
1838 * fixed some bugs related to Unicode integration:
1839 ** portions of multibyte characters at the end of buffers for
1840 character-based file input are correctly transferred to the
1841 start of the buffer at the next read.
1842 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1843 passing it through to OPEN.
1844 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1845 argument given to internal calls to OPEN.
1846 ** the built-in comment readers (introduced by character sequences
1847 ";" and "#|") are more forgiving to encoding errors; they will
1848 STYLE-WARN and then attempt to resync the stream at a character
1849 boundary. (thanks to Teemu Kalvas)
1850 * fixed some bugs in the x86-64 port:
1851 ** Negative short int return values from c-calls are sign-extended
1853 ** The stack is aligned to 16-bytes for c-calls, as required by
1854 the ABI. (thanks to Cheuksan Wang)
1855 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1856 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1857 for enums. (thanks to Vincent Arkesteijn)
1858 ** Multiple small optimizations and bugfixes for floating point
1860 * fixed some bugs revealed by Paul Dietz' test suite:
1861 ** Space, Tab, Linefeed, Return and Page have the invalid
1862 secondary constituent character trait.
1863 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1865 ** WITH-INPUT-FROM-STRING should only update the index place on
1867 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1868 works more reliably.
1869 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1870 with one-dimensional arrays with fill pointers.
1871 ** TYPE-OF failed on a complex with an integer realpart and a
1873 ** compiler failure during type inference for the code of form
1874 (IF (EQL X (THE ...)) ...) (MISC.535).
1876 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1877 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1878 platform. The port supports 61-bit fixnums, large memory spaces
1879 and reloading shared object files.
1880 * enhancement: saving cores with foreign code loaded is now
1881 supported on ppc/Darwin in addition to the previously supported
1883 * enhancement: the statistical profiler now walks deeper into the
1884 call stack for more meaningful call-graphs and accrued time
1885 reports (x86/x86-64 only). It also now reports time spent in
1887 * enhancement: it is now possible to trace most individual methods
1888 of a generic function in addition to tracing the generic function
1890 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1892 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1893 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1894 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1895 produces an error. (thanks to Vincent Arkesteijn)
1896 * bug fix: NAMESTRING on pathnames with :WILD components in their
1897 directories works correctly. (thanks to Artem V. Andreev)
1898 * fixed bug 125: compiler preserves identity of closures. (reported
1900 * bug fixed: functions with &REST arguments sometimes failed with
1901 "Undefined function" when compiled with (DEBUG 3). (reported by
1903 * bug fix: overflow during compiling of setting element of a bit
1904 vector with constant index and value. (reported by Timmy Douglas)
1905 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1906 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1907 * on x86 compiler partially supports stack allocation of dynamic-extent
1909 * GO and RETURN-FROM do not check the extent of their exit points
1910 when compiled with SAFETY 0.
1911 * fixed some bugs related to Unicode integration:
1912 ** encoding and decoding errors are now much more robustly
1913 handled; it should now be possible to recover even from invalid
1914 input or output to the terminal. (thanks to Teemu Kalvas)
1915 ** provided a first cut at implementing STRING-TO-OCTETS and
1916 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1917 ** altered the SB-MD5 contributed module slightly, changing the
1918 interface just enough for it to be supportable for builds where
1919 lisp characters are not eight bits.
1920 * fixed some bugs revealed by Paul Dietz' test suite:
1921 ** the FORMATTER-generated functions for ~V[ conditionals require
1922 the correct number of arguments.
1923 ** READ-FROM-STRING returns the mandated second value when applied
1924 to displaced strings.
1925 ** the #\Rubout and #\Backspace characters are treated as invalid
1926 constituent characters by the tokenizer.
1928 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1929 * new feature: reloading changed shared object files with
1930 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1931 * new feature: references to foreign variables and functions
1932 can now be compiled and loaded before the corresponding shared
1933 object file is loaded, as long as the foreign definitions are
1934 available at runtime.
1935 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1936 supported. (thanks to Dan Debertin)
1937 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1938 just keywords. (thanks to Vincent Arkesteijn)
1939 * fixed bug #331: structure-class instances corresponding to
1940 DEFSTRUCT forms are now created eagerly.
1941 * fixed bug #345: backtraces from calls to undefined functions work
1942 on x86 as well. Related bug #61 is now also partially fixed on x86
1943 (backtraces from throws to unknown catch tags.)
1944 * bug fix: lambda-list parsing is now stricter vrt. order and number
1945 of lambda-list keywords.
1946 * bug fix: as specified by AMOP, an error is signalled if a
1947 class-option appears multiple times in a DEFCLASS form. (reported
1949 * bug fix: RANDOM can be compiled when the compiler derives the type
1950 of its numeric argument as a disjoint set of small integers.
1951 (reported by Paul Dietz)
1952 * bug fix: starting a core saved with shared objects loaded when
1953 those objects are not available no longer causes threaded SBCL to
1954 hang. (reported by Sean Ross)
1955 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1957 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1958 argument, which must always be NIL. (reported by Kalle Niemitalo)
1959 * bug fix: printing 1.0d+23 no longer results in an error.
1960 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1961 * bug fix: structure slot setters preserve evaluation order. (thanks
1963 * fixed some bugs related to Unicode integration:
1964 ** RUN-PROGRAM can allow its child to take input from a Lisp
1965 stream. (reported by Stefan Scholl)
1966 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1967 locales. (reported by Ken Causey)
1968 * fixed some bugs revealed by Paul Dietz' test suite:
1969 ** INCF, DECF and REMF evaluate their place form as specified in
1971 ** FORMATTER expands ~{ ~} iteration directives with V or #
1972 parameters correctly.
1973 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1974 consequent uses no arguments correctly.
1975 ** the system has a partial understanding of the (COMPLEX RATIO)
1977 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1980 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1981 * new feature: a build-time option (controlled by the :SB-UNICODE
1982 keyword feature, enabled by default) for building the system with
1983 support for the entire 21-bit character space defined by the
1985 * new feature: the system now has rudimentary external-format
1986 support; the primary user-visible change at this time is that
1987 characters with the high bit set (such as lower-case-e-acute) will
1988 print correctly to a terminal in a UTF-8 environment.
1989 * minor incompatible change: BASE-CHAR no longer names a class;
1990 however, CHARACTER continues to do so, as required by ANSI.
1991 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1992 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1993 has been moved to the SB-EXT package (temporarily re-exported from
1995 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1996 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1998 * fixed bug #351: better error-handlind and reporting for malformed
2000 * fixed bug #350: bignum-printing is now more memory-efficient,
2001 allowing printing of very large bignums, eg. (expt 2 10000000).
2002 (reported by Bruno Haible)
2003 * fixed bug #302: better primitive-type selection for intersection
2005 * fixed bug #308: non-graphic characters now all have names, as
2006 required. (reported by Bruno Haible)
2007 * bug fix: redefining a class with different superclasses now correctly
2008 removes it from the direct-subclasses of its previous superclasses.
2009 (reported by David Morse)
2010 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2011 new value now works. (reported by Bruno Haible)
2012 * bug fix: correct canonicalization of multiple non-standard slot
2013 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2014 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2015 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2016 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2017 now exists, an signals an error.
2018 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2019 during macroexpansion for non-iteration variables as well. (reported
2020 by Bruno Haible for CMUCL)
2021 * bug fix: Cyclic structures and unprintable objects in compiler
2022 messages no longer cause errors. (reported by Bruno Haible)
2023 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2024 and READ-FROM-STRING all now return a primary value of NIL if
2025 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2026 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2027 (reported by Bruno Haible for CMUCL)
2028 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2029 do not designate a condition. (reported by Bruno Haible for
2031 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2032 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2033 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2034 permitting resolution in favour of any of the conflicting symbols.
2035 (reported by Bruno Haible for CMUCL)
2036 * bug fix: EQUAL compiler optimizations is less aggressive on
2037 strings which can potentially compare true despite having distinct
2038 specialized array element types.
2039 * bug fix: unit enumerations can be defined without dividing by
2040 zero. (thanks to Vincent Arkesteijn)
2041 * FORMAT compile-time argument count checking has been enhanced.
2042 (report from Bruno Haible for CMUCL)
2043 * a partial workaround for the bug 262: the compiler does not try to
2044 inline-expand a local function doing RETURN-FROM from a deleted
2045 BLOCK. (thanks to Peter Denno for the bug report and to David
2046 Wragg for the simple test case)
2047 * fixed some bugs revealed by Paul Dietz' test suite:
2048 ** RENAME-PACKAGE allows all package designators as new package
2050 ** constraint propagation and lambda variable substitution are
2051 more cautious in dealing with partially deleted code.
2052 ** compiler sometimes lost reoptimization passes.
2053 ** CERROR, given a condition as condition designator, passes its
2054 remaining arguments to the continue format control without
2056 ** Case-altering FORMAT directives work correctly on non-ASCII
2058 ** The REAL type specifier handles bounds outside the floating
2059 point ranges without signalling FLOATING-POINT-OVERFLOW.
2060 ** Functions with IR1-transformations can create intercomponent
2061 references to global functions.
2062 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2064 ** FORMAT ~R treats a nil value for its first parameter correctly.
2066 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2067 * enhancement: saving cores with foreign code loaded is now
2068 supported on x86/NetBSD and sparc/Linux in addition to the previously
2069 supported platforms.
2070 * bug fix: on some platforms repeated installations caused multiple
2071 copies of HTML documentation to be installed -- should not happen
2072 any more. (reported by Stefan Scholl)
2073 * bug fix: parsing self-recursive alien record types multiple times
2074 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2075 original patch by Helmut Eller for CMUCL)
2076 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2077 (thanks to Richard Kreuter)
2078 * bug fix: defining classes whose accessors are methods on existing
2079 generic functions in other (locked) packages no longer signals
2080 bogus package lock violations. (reported by François-René Rideau)
2081 * bug fix: special variables as DEFMETHOD parameters no longer have
2082 associated bogus type declarations. (reported by David Wragg and
2084 * bug fix: read-write consistency on streams of element-type
2085 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2086 * bug fix: redefiniton of the only method of a generic function with
2087 no DEFGENERIC no longer emits a full WARNING. In addition,
2088 redefinition of generic functions with no DEFGENERIC to an
2089 incompatible lambda list now signals an error. (thanks to Zach
2091 * bug fix: DEFGENERIC now works even when there's a function of the
2092 same name in an enclosing lexical environment. (thanks to Zach
2094 * fixed compiler failure, caused by instrumenting code during
2095 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2096 * optimization: added loop analysis and improved register allocation
2097 * fixed some bugs revealed by Paul Dietz' test suite:
2098 ** POSITION on displaced vectors with non-zero displacement
2099 returns the right answer.
2100 ** (SIMPLE-STRING) is a valid type specifier for sequence
2102 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2104 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2105 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2107 ** The justification version of the FORMAT ~< directive treats
2108 non-zero minpad parameter correctly.
2110 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2111 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2112 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2113 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2114 the supported interface.
2115 * new feature: Single-stepping of code compiled with DEBUG 2 or
2116 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2117 * new feature: saving cores with foreign code loaded is now
2118 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2119 Timothy Moore's work for CMUCL)
2120 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2121 parameters to * instead of NIL if no initform is supplied.
2122 (reported by Johan Bockgaard)
2123 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2124 correctly when FUNCALL appears as the car of the form. Note:
2125 despite this FUNCALL forms are not currently subject to
2126 compiler-macro expansion. (port of Raymond Toy's fix for the
2127 same from CMUCL, reported by Johan Bockgaard)
2128 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2129 (thanks for Teemu Kalvas)
2130 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2131 case instead of terminating the iteration (thanks for Julian
2132 Squires, Sean Champ and Raymond Toy)
2133 * bug fix: incorrect expansion of defgeneric that caused a style
2134 warning. (thanks for Zach Beane)
2135 * bug fix: slot accessor effective method computation works properly
2136 for classes with multiple non-standard applicable methods on
2137 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2138 * on x86 compiler supports stack allocation of results of LIST and
2139 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2140 CMUCL implementation by Gerd Moellmann)
2141 * fixed some bugs revealed by Paul Dietz' test suite:
2142 ** FORMAT strings with both the ~<~:;~> form of the justification
2143 directive and pretty-printing directives cause an error.
2145 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2146 * incompatible change: the internal functions
2147 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2148 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2149 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2150 instead of the old functions.
2151 * new feature: on platforms where "dladdr" is available foreign
2152 function names now appear in backtraces. (based on Helmut Eller's
2154 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2155 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2157 * documentation: ASDF manual is now distributed with SBCL.
2158 * bug fix: SBCL can now load its contributed modules with REQUIRE
2159 even if the system-provided entries have been removed from
2160 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2162 * bug fix: backtraces involving undefined functions or assembly
2163 routines are more informative. (thanks to Brian Downing)
2164 * bug fix: mutually referent alien structures now work correctly.
2165 (reported by Rick Taube)
2166 * bug fix: structures defined by WITH-ALIEN can be referred to
2167 within other definitions in the same WITH-ALIEN.
2168 * bug fix: division operators (MOD, TRUNCATE and the like) with
2169 constant zero divisors and integer dividends no longer generate
2171 * bug fix: provide default methods for INPUT-STREAM-P and
2172 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2173 * bug fix: improve the detection and resolution of MOP metacycles
2174 (where effective method computation depends on the generic function
2175 having its effective method computed). (reported by Bruno Haible)
2176 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2177 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2178 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2179 * bug fix: pathnames with relative directory components can be
2180 represented relative to default pathnames.
2181 * optimization: in taking the GCD of bignums, reduce the two bignums
2182 to approximately the same size (using Euclid's algorithm) before
2183 applying the more sophisticated binary GCD. (thanks to Juho
2185 * optimization: COUNT on bitvectors now operates word-at-a-time.
2186 * optimization: ASH with a positive, but not necessarily constant,
2187 (leftwards) shift, when in a modular context, is compiled to a
2189 * fixed some bugs revealed by Paul Dietz' test suite:
2190 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2191 if the corresponding argument is NIL.
2192 ** FORMAT directives accepting a minpad parameter treat negative
2194 ** literal commas as character parameters to FORMAT directives are
2196 ** literal spaces directly after ~<Newline> directives within a
2197 format-logical-block (~:< ~@:>) do not induce :FILL-style
2198 conditional newlines.
2199 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2200 as specified: it no longer includes conditional newlines).
2201 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2203 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2204 :PER-LINE-PREFIX argument does not evaluate to a string.
2206 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2207 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2208 the manual for details; for now, package locks can be disabled by
2209 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2210 if no major problems are found then it is likely that they will be
2211 compiled in unconditionally.
2212 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2213 unsupported operators on all platforms. To load a shared library
2214 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2215 object file, link it into a shared library outside of SBCL and
2216 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2218 * bug workaround: ROOM T and the GC were not working together
2219 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2220 reporting facility aren't preserved. That reporting has been
2221 disabled, so now until and unless someone figures out how to make it
2222 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2223 * minor incompatible change: as threatened around sbcl-0.8.0, the
2224 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2225 The symbols are also exported from SB-PCL for backwards
2226 compatibility, but more so than before SB-PCL should be treated as
2227 an implementation-internal package.
2228 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2230 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2231 It is known as of this release not to work on the Alpha, however.
2232 * fixed bug #167: errors signalled due to illegal syntax in method
2233 bodies are now more legible.
2234 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2235 designators and can hence be used with TYPEP.
2236 * fixed bug #333: CHECK-TYPE now ensures that the type error
2237 signalled, if any, has the right object to be accessed by
2238 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2239 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2240 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2242 * fixed bug #334: programmatic addition of slots using specialized
2243 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2244 and :ALLOCATION :CLASS slots.
2245 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2246 float exponents. (rereported by Peter Seibel)
2247 * fixed a bug: #\Space (and other whitespace characters) are no
2248 longer considered to be macro characters in standard syntax by
2249 GET-MACRO-CHARACTER.
2250 * fixed bug: initialization of condition class metaobjects no longer
2251 causes an instance of the condition to be created. (reported by
2253 * fixed bug: it is now possible to have slots such that
2254 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2255 description is neither :INSTANCE nor :CLASS.
2256 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2257 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2258 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2259 non-standard SLOT-DEFINITION-ALLOCATION values.
2260 * fixed bugs: various race conditions handling exiting threads.
2261 CL-PPCRE's thread torture test now passes.
2262 * fixed arguably-a-bug: GC time proportional to number of threads in
2263 system even when most of them are idle
2264 * optimization: improved performance of BIT and SBIT on bit-vectors.
2265 * .fasl file incompatibility: The fasl file version number has
2266 been incremented because of changes associated with package locks.
2268 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2269 * minor incompatible change: the system no longer provides
2270 optimization hints (conditions which are TYPEP
2271 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2272 mode; these hints are still emitted when higher SPEED optimization
2274 * new contrib module: a sampling profiler (profiling by statistical
2275 sampling, rather than by instrumenting functions) is available as
2276 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2277 * the behaviour of the standard function ED is now customizeable by
2278 third parties through a hook variable: see ED's documentation
2279 string for information on the protocol.
2280 * the compiler no longer emits efficiency notes for (FUNCALL X)
2281 when the type of X is uncertain under default optimization
2283 * fixed bug 276: mutating a binding of a specialized parameter to a
2284 method to something that is not TYPEP the specializer is now
2286 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2287 exhibit the required behaviour on the x86 platform. (thanks to
2288 Peter van Eynde, Eric Marsden and Bruno Haible)
2289 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2290 even for difficult arguments. (reported by Peter Graves)
2291 * fixed bug 141a: the backquote printer now descends quoted
2293 * fixed another bug in backquote printing: no more destructive
2294 modification of the form's list structure. (reported by Brian
2296 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2297 move between its address being taken and the call to
2298 interrupt_thread, fixing a crashing race condition.
2299 * the SB-POSIX contrib implementation has been adjusted so that it
2300 no longer exhibits ridiculously poor performance when constructing
2301 instances corresponding to C structs.
2303 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2304 * minor incompatible change: the sb-grovel contrib now treats C
2305 structures as alien (in the sense of SB-ALIEN) objects rather than
2306 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2307 has implications for memory management of client code
2308 (sb-grovel-returned objects must now be manually managed) and for
2309 type safety (alien objects now have full types).
2310 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2311 used to control emission of compiler diagnostics, rather than the
2312 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2313 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2314 quality should be considered deprecated.
2315 * install.sh now installs the user manual as well
2316 * (not quite a new documentable feature, but worth considering in
2317 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2318 beginnings of a semantically meaningful condition hierarchy is
2319 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2321 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2322 designator as the defaults argument.
2323 * fixed bug: Displaced arrays whose displaced-to array has become
2324 too small now cause ARRAY-DIMENSION to signal an error, providing
2325 for safer bounds-checking. (reported by Bruno Haible)
2326 * fixed bug: DEFCLASS slot definitions with identical :READER and
2327 :WRITER names now signal a reasonable error. (reported by Thomas
2329 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2331 * fixed bug: Invalid slot specification errors now print correctly.
2332 (thanks to Zach Beane)
2333 * fixed bug 320: Shared to local slot value transfers in class
2334 redefinitions now happen corrently from superclasses as
2335 well. (reported by Bruno Haible)
2336 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2338 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2339 improper lists correctly. (reported by Bruno Haible)
2340 * fixed bug 313: source-transform for <fun-name> was erroneously
2341 applied to a call of a value of a variable with name <fun-name>.
2342 (reported by Antonio Menezes Leitao)
2343 * fixed bug 307: The obsolete instance protocol ensures that
2344 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2345 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2346 remove dead unknown-values globs from the middle of the stack.
2347 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2348 easier to resume long computations after using *BREAK-ON-SIGNALS*
2349 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2350 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2351 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2352 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2353 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2354 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2356 * fixed a (fairly theoretical) bug in string printing: if
2357 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2358 does not have array-element-type equal to the most general string
2360 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2361 function name. (thanks to Nikodemus Siivola)
2362 * fixed bug: compile-time format string checker failed on
2363 non-closed ~{. (reported by Thomas F Burdick)
2364 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2365 don't warn on legal loop code involving more than one aggregate
2366 boolean. (thanks to Nikodemus Siivola)
2367 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2368 in keyword position to MAKE-PACKAGE was not being reported as
2370 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2371 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2373 * fixed bugs 280 and 312: the checking for multiple definitions in a
2374 file is less likely to become confused by uses of inline
2376 * fixed bug: the #S reader macro performs the keyword coercion
2377 specified for slot names. (reported by Kalle Niemitalo)
2378 * fixed bug: lambda lists may contain symbols whose names start with
2379 & but are not lambda-list-keywords; their occurrence triggers a
2381 * fixed bug 321: define-method-combination argument lambda lists do
2382 not cause invalid code to be generated when &optional or &aux
2383 variables have default values. (reported by Bruno Haible)
2384 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2385 have CLOS classes; this appears also to have fixed reported
2386 instability in stack exhaustion detection.
2387 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2388 MUFFLE-WARNING when no associated restart is present is now
2390 * optimization: rearranged the expansion of various defining macros
2391 so that each expands into only one top-level form in a
2392 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2394 * optimization: used a previously little-used slot in symbols to
2395 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2397 * fixed some bugs revealed by Paul Dietz' test suite:
2398 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2400 ** Fixed an optimization bug related to inheritance of initforms
2401 from local to shared slots.
2402 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2403 ** CAST optimizer forgot to flush argument derived type.
2404 ** print/read consistency on floats is now orders of magnitude
2405 more likely. (thanks also to Bruno Haible for a similar report
2407 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2408 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2409 for objects of type REAL. Make it so.
2410 ** SXHASH is specified (infelicitously) to respect similarity,
2411 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2412 it so. (thanks to Markus Ziegler)
2413 ** on the Alpha, the compiler succeeds in compiling functions
2414 returning a known number of arguments greater than 63.
2415 ** fixed handling of invalid NIL arguments in keyword position
2417 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2419 ** provided a readably-printable representation for RANDOM-STATE
2421 ** ensured that pathnames, where they have a namestring, always
2422 print using #P"..." syntax.
2424 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2425 * Support for the forthcoming 2.0 version of the NetBSD kernel
2426 running on x86 hardware has been added. (thanks to Perry
2427 E. Metzger most immediately, and others for their past work)
2428 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2429 current release version; SBCL's OpenBSD support had been broken
2430 since about the time of OpenBSD's switch to ELF binary format.)
2431 * [placeholder for DX summary]
2432 ** user code with &REST lists declared dynamic-extent, under high
2433 speed or space and low safety and debug optimization policy.
2434 * The manual has been converted to Texinfo format and the debugger
2435 chapter from the cmucl manual has been added.
2436 * A facility has been added to extract documentation strings from
2437 sbcl and store them as Texinfo-formatted snippets for inclusion in
2438 the manual (via Texinfo's @include directive)
2439 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2440 the test case to Dave Roberts)
2441 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2442 pointers. (reported by Sean Ross)
2443 * bug fix: PROFILE output is printed nicely even for large numerical
2444 values. (thanks to Zach Beane)
2445 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2446 greater than 32 handle EOF correctly.
2447 * bug fix: on X86 an immediate argument of the IMUL instruction is
2448 correctly printed in disassembly. (thanks to Lutz Euler)
2449 * bug fix: class slots in redefined classes preserve their old
2450 values. (thanks to Bruno Haible and Nikodemus Siivola)
2451 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2452 succeeds. (reported by Marco Baringer)
2453 * fixed some bugs revealed by Paul Dietz' test suite:
2454 ** READ-SEQUENCE now works on ECHO-STREAMs.
2455 ** RATIONALIZE works more according to its specification. (thanks
2458 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2459 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2460 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2461 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2462 should matter to you only if you rebind the printer control
2463 variables and then find you want different bindings in the
2464 debugger than in the ordinary execution of your program.)
2465 * The runtime build system has been tweaked to support building
2466 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2467 assembler and linker. (thanks to Nikodemus Siivola)
2468 * Unbound, undefined, undeclared variables now trigger full
2469 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2470 more often programmer error than deliberate exploitation of undefined
2472 * optimization: the hash algorithm for strings has changed to one
2473 that is less vulnerable to spurious collisions. (thanks to Juho
2475 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2476 needless bounds checking. (thanks to Juho Snellman)
2477 * optimization: implemented multiplication as a modular
2478 (UNSIGNED-BYTE 32) operation on the PPC backend.
2479 * fixed some bugs revealed by Paul Dietz' test suite:
2480 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2482 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2483 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2484 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2486 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2488 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2490 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2491 their output stream on EOF from read.
2492 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2493 have been read to end-of-file.
2494 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2496 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2497 description of determination of which consecutive characters
2499 ** Printing the "Space" character with escaping on now yields "#\\ ",
2500 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2501 ** Reading floating-point numbers with *READ-BASE* set to a number
2502 less than 10 works correctly.
2503 ** Reading floating-point numbers with *READ-BASE* set to a number
2504 more than 10 works correctly.
2505 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2506 the readtable currently in effect.
2508 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2509 * minor incompatible change: parsing of namestrings on a physical
2510 (Unix) host has changed; numbers after the final #\. in a
2511 namestring are no longer interpreted as a version field. This is
2512 intented to be largely invisible to the user, except that the
2513 meaning of the namestring "*.*.*" has changed: it now refers to a
2514 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2515 should usually be replaced by
2516 (make-pathname :name :wild :type :wild :version :wild)
2517 with the added benefit that this is more likely to be portable.
2518 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2519 option to OPEN now signals an error if the file being opened
2520 exists; this may have an impact on existing code.
2521 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2522 signals received, on the PowerPC platforms (both Linux and
2523 Darwin). (thanks to Pierre Mai for pointing out the location of
2525 * several fixes on OS X: The system now builds and runs cleanly on
2526 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2527 (thanks to Brian Mastenbrook)
2528 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2529 second-resolution: integer multiples of 1/3600 between -24 and 24.
2530 (thanks to Vincent Arkesteijn)
2531 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2532 type when called with 1 argument; PEEK-CHAR checked type of
2533 PEEK-TYPE only after having read first character from a
2534 stream. (reported by Peter Graves)
2535 * bug fix: the garbage collector now has much better locality
2536 behaviour, and in particular no longer treats all memory as being
2537 exhausted when many small objects point to each other in a deeply
2539 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2540 recognized as being TYPEP their class.
2541 * bug fix: the PUSHNEW documentation string has been corrected.
2542 (thanks to Vincent Arkesteijn)
2543 * bug fix: defaulting of the value for the last of an atypically
2544 large number of multiple values being bound was not being
2545 performed correctly on the Alpha or PPC platforms
2546 * optimization: implemented multiplication as a modular
2547 (UNSIGNED-BYTE 32) operation on the x86 backend.
2548 * optimization: SEARCH on simple-base-strings can now be open-coded.
2549 (see also contrib/compiler-extras.lisp for inspiration for
2550 teaching the compiler about the Boyer-Moore algorithm).
2551 * value, returned by MAX (and MIN) called with several EQUALP, but
2552 not EQL, arguments now does not depend on compiler settings.
2553 * fixed some bugs revealed by Paul Dietz' test suite:
2554 ** in stack analysis liveness information is propagated from
2555 non-local entry points.
2556 ** pathwise CAST removing failed when the CAST node did not start
2558 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2559 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2561 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2562 host is already defined.
2563 ** RENAME-FILE works on streams instead of signalling an internal
2565 ** PEEK-CHAR uses the current readtable when determining whether
2566 or not a character is whitespace.
2567 ** MERGE-PATHNAMES handles the case when the pathname does not
2568 specify a name while the default-pathname specifies a version.
2569 ** Pathnames now stand a better chance of respecting print/read
2571 ** Attempting to use standardized file system operators with a
2572 pathname with invalid :DIRECTORY components signals a
2574 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2576 ** DIRECTORY on logical pathnames is more correct.
2577 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2578 signal a TYPE-ERROR if their argument is not a stream.
2579 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2580 designator argument does not designate a stream.
2581 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2582 examining the synonym.
2583 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2585 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2586 element-type larger than ([UN]SIGNED-BYTE 32).
2588 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2589 * When built with the :SB-FUTEX feature, threaded builds now take
2590 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2591 for faster/more reliable mutex and condition variable support.
2592 * Incompatible change (but one you probably shouldn't have been using
2593 anyway): the interface and code for arbitrating between multiple
2594 threads in the same user session has been redesigned.
2595 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2596 error when called without an explicit environment argument.
2597 (thanks to Nikodemus Siivola)
2598 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2599 confused about their position. (thanks to Adam Warner and Gerd
2601 * bug fix: Pretty printing backquoted forms with unquotations in the
2602 argument list position of various code constructs such as LAMBDA
2603 now works correctly. (reported by Paul Dietz)
2604 * bug fix: Pretty printing unquotations no longer loses all
2605 stream position information.
2606 * optimization: performance of string output streams is now less
2607 poor for multiple small sequence writes.
2608 * optimization: performance of CSUBTYPEP in the presence of complex
2609 expressions involving CONS and NOT many times has been improved.
2610 (reported by Paul Dietz)
2611 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2613 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2614 streams for the REPL, for the debugger, and so on. (thanks to
2616 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2617 * fixed some bugs revealed by Paul Dietz' test suite:
2618 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2619 is ignored if it is too small, rather than propagating through
2620 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2621 ** extremely complex negations of CONS types were not being
2622 sufficiently canonicalized, leading to inconsistencies in
2624 ** VALUES tranformer lost derived type.
2626 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2627 * fixed a bootstrapping bug: the build process no longer assumes
2628 that the various BOOLE-related constants have the same value in
2629 host and target lisps. (noted by Paul Dietz' test suite on an
2630 SBCL binary built from CLISP)
2631 * The system can now be dynamically linked on the MIPS platform,
2632 which enables dynamic loading of foreign code from Lisp. (thanks
2633 to Ralf Baechle for discussions on the MIPS ABI)
2634 * The system now records debugging information for its own source
2635 files in a filesystem-position-independent manner, relative to
2636 the "SYS" logical host.
2637 * fixed a compiler bug: MV-LET convertion did not check references
2638 to the "max args" entry point. (reported by Brian Downing)
2639 * tweaked disassembly notes to be less confident about proclaiming
2640 some instruction as an LRA. (thanks to Brian Downing)
2641 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2642 now each have their own history, command character, and other
2643 characteristics. (thanks to David Lichteblau)
2644 * fixed some bugs revealed by Paul Dietz' test suite:
2645 ** compiler failure in compiling LOGAND expressions including a
2647 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2648 shift greater than 32.
2649 ** FLUSH-DEST did not mark blocks for type check regeneration.
2650 ** HANDLER-CASE failed to accept declarations in handler clauses
2651 in some circumstances.
2653 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2654 * New code in contrib/sb-introspect (still probably not entirely
2655 stable yet) provides some support for smart Lisp development
2656 environments like SLIME.
2657 * The conditions signalled for errors occurring when loading .fasl
2658 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2659 in a way which should help ASDF recover gracefully.
2660 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2661 stuff has been cleaned up. If you code contrib/ stuff, this might
2662 affect you, and you can look at contrib/README, contrib/STANDARDS,
2663 and/or the 0.8.4.27 diff to check.
2664 * In full calls the compiler now does not generate checks for declared
2665 argument types for all arguments.
2666 * various threading fixes
2667 ** and some experimental patches which didn't make it into
2668 the main tree for this release, but which are shipped in
2669 contrib/experimental-thread.patch as a possible fix for some
2670 failures (deadlock, spinning...) in GC-intensive multithreaded
2672 * fixed PPC build problem (source code incompatibility of different
2673 library versions): added offsetof() hackery which attempts to divine
2674 where glibc maintainers put uc_mcontext today
2675 * fixed bug 282: compiler does not trust type assertions while passing
2676 arguments to a full call.
2677 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2678 &OPTIONAL VALUES type parameter.
2679 * fix bug 214: algorithm for noting rejected templates is now more
2680 similar to that of template seletion. (also reported by rydis on
2682 * fixed bug 141b: printing backquoted information readably and prettily
2683 inserts a space where necessary.
2684 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2685 (CHARACTER 10) are now reported as errors, rather than propagated
2686 as unknown types. (reported by piso on #lisp)
2687 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2688 with values NIL and :ERROR. (thanks to Milan Zamazal)
2689 * fixed bug 191c: CLOS now does proper keyword argument checking as
2690 described in CLHS 7.6.5 and 7.6.5.1.
2691 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2692 longer raise an error; further, using a list as a for-as-arithmetic
2693 counter now raises a meaningful error.
2694 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2695 now understood by sequence creation functions such as MAKE-SEQUENCE
2697 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2698 a STRING-INPUT-STREAM.
2699 * compiler enhancement: SIGNUM is now better able to derive the type
2701 * type declarations inside WITH-SLOTS are checked. (reported by
2703 * fixed some bugs revealed by Paul Dietz' test suite:
2704 ** incorrect optimization of TRUNCATE for a positive first
2705 argument and negative second.
2706 ** compiler failure in let-convertion during flushing dead code.
2707 ** compiler failure while deriving type of TRUNCATE on an
2708 interval, containing 0.
2709 ** ASH of a negative bignum by a negative bignum count now returns
2711 ** intersection of CONS types now canonicalizes properly, fixing
2712 inconsistencies in SUBTYPEP.
2714 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2715 * incompatible change: The --disable-debugger command line
2716 option now clobbers the debugger at a more fundamental
2717 level, by redefining #'INVOKE-DEBUGGER instead of by
2718 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2719 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2720 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2721 enter the debugger REPL and then suffer recursive errors
2722 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2723 while under the new system, BREAK in --disable-debugger mode
2724 terminates the system just as an unhandled error would.
2725 * fixed compiler performance when processing loops with a step >1;
2726 * bug fix: DOCUMENTATION now retrieves generic function
2727 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2728 support has been systematized, and now supports the methods
2729 specified by ANSI, along with a default method and a method for
2730 slot documentation. (reported by Nathan Froyd)
2731 * bug fix: effective methods associated with a generic function are
2732 no longer cached over a change of that generic function's method
2733 combination. (reported by Andreas Fuchs)
2734 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2735 names a non-existent directory. (thanks to Andreas Fuchs)
2736 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2737 the PPC platform, lead to stack corruption; this has been fixed.
2738 (reported by Rainer Joswig)
2739 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2740 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2741 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2742 arithmetic in general, is now much more reliable on the PPC
2744 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2745 platform now returns the right answer.
2746 * optimization: restored some effective method precomputation in
2747 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2748 precomputation is now tunable.
2749 * optimization: compiler-internal data structure use has been
2750 reviewed, and changes have been made that should improve the
2751 performance of the compiler by about 20%.
2752 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2753 simple-base-strings and simple-bit-vectors is improved.
2754 * optimization: the optimization of 32-bit logical and arithmetic
2755 functions introduced in version 0.8.3 on the x86 has been
2756 implemented on the mips, ppc and sparc platforms; an
2757 implementation of the same facility, but for 64-bit arithmetic,
2758 has been added for the alpha.
2759 * microoptimization: the compiler is better able to make use of the
2760 x86 LEA instruction for multiplication by constants.
2761 * bug fix: in some situations compiler did not report usage of
2762 generic arithmetic in (SPEED 3) policy.
2763 * bug 145b fix: compiler used wrong type specifier while converting
2764 MEMBER-types to numeric.
2765 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2767 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2768 index argument now works properly on the Alpha platform.
2769 * bug fix: floating point exception treatment on the Alpha platform
2771 * bug fix: FILE-POSITION works much better on string input and
2772 output streams. (thanks to Nikodemus Siivola)
2773 * bug fix: many threading/garbage collection symptoms sorted.
2774 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2775 might be pseudo-atomic.
2776 * internal change: Stopping for GC is now done with signals not
2777 ptrace. GC is now done in whichever thread wanted it, instead of
2779 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2781 * bug fix: result form in DO is not contained in the implicit
2783 * incompatible change: ICR structure is changed; the value part of
2784 CONTINUATION is now called LVAR; corresponding functions are
2785 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2786 * added type deriver for ISQRT (thanks to Robert E. Brown).
2787 * bug fix: better support for loading from the command line when an
2788 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2790 * fixed some bugs revealed by Paul Dietz' test suite:
2791 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2792 ** ROUND and FROUND now give the right answer when given very
2793 small float arguments.
2794 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2796 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2797 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2798 ** LCM should return a non-negative integer.
2799 ** PARSE-INTEGER returned the index of a terminator instead of the
2800 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2801 ** PARSE-INTEGER returned an incorrect index being applied to a
2803 ** LCM with two arguments of 0 returns 0 rather than signalling
2805 ** unsigned addition of a 32-bit constant with the high bit set no
2806 longer causes an internal compiler error on the x86.
2807 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2808 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2809 with negative last argument.
2810 ** byte specifiers with zero size and position no longer cause
2811 an error during type derivation.
2812 ** bignum multiplication on the Alpha platform now returns the
2814 * porting: The system now builds on SuSE AMD64, although it still
2815 generates a 32-bit binary.
2816 * .fasl file incompatibility: The fasl file version number has
2817 been incremented (because of the changes to internal compiler
2818 data structures referred to above).
2820 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2821 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2822 more accurately, on the Darwin kernel running on PowerPC hardware.
2823 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2824 * Compiler code deletion notes now signal a condition of type
2825 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2826 an associated MUFFLE-WARNING restart.
2827 * The compiler now performs limited argument count validation of
2828 constant format strings in FORMAT, and where appropriate in ERROR,
2829 CERROR and WARN. (thanks to Gerd Moellmann)
2830 * New ASDF-INSTALL contrib can be used for automatic download and
2831 installation of third-party Lisp code from CCLAN or other sites
2833 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2834 which forces another thread to execute a function supplied by the
2836 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2837 now accept and act upon their :ELEMENT-TYPE keyword argument.
2838 (reported by Martin Atzmueller, Edi Weitz)
2839 * bug fix: FILE-POSITION now accepts position designators up to
2840 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2841 is the greater. (thanks to Patrik Nordebo)
2842 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2843 Kalle Olavi Niemitalo)
2844 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2845 printable. (reported by Eric Marsden)
2846 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2847 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2848 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2849 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2850 (thanks to Stig Sandoe)
2851 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2852 UNREAD-CHAR now correctly includes the unread character in the
2853 target sequence. (thanks to Gerd Moellmann)
2854 * bug fix in threaded builds: the system can now be suspended and
2855 resumed by shell job control with minimal disruption.
2856 * bug fixes in times and timezones >2038AD
2857 * better handling of "where is GNU make?" problem in build scripts
2858 (thanks to Nikodemus Siivola)
2859 * new optimization: inside a named function any reference to a
2860 function with the same name is considered to be a self-reference;
2861 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2862 optimization quality.
2863 * new optimization on x86: logical functions and + now have
2864 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2865 used when the result is truncated to 32 bits.
2866 * VALUES declaration is partially enabled.
2867 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2868 * fixed some bugs revealed by Paul Dietz' test suite:
2869 ** The system now obeys the constraint imposed by
2870 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2871 types form a lattice under type intersection.
2872 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2873 ** ASSOC now ignores NIL elements in an alist.
2874 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2875 and (1+ MOST-POSITIVE-FIXNUM) answers.
2876 ** The addition of a method with invalid qualifiers to a generic
2877 function does not cause an error to be signalled immediately;
2878 a warning is signalled, and the error is generated only on
2879 calling the generic function.
2880 * changed .fasl file version number, in part to add type codes for
2881 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2882 obscure ANSI requirements
2884 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2885 * fixed bug 148: failure to inline-expand a local function left
2886 garbage, confusing the compiler.
2887 * fixed bugs 3cd: structure slot readers perform type check if the
2888 slot can have an invalid value (i.e. it is either not initialized
2889 or can be written with a less specific slot writer).
2890 * bug fix: the compiler now traps array references to elements off
2891 the end of an array; previously, the bounds checking in some
2892 circumstances could go off-by-one.
2893 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2895 * type declarations for array element types now obey the description
2896 on the CLHS page "Declaration TYPE", as per discussions on
2897 sbcl-help around 2003-05-08. This means that a declaration
2898 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2899 declaration, all references to BAR will be asserted or assumed
2900 (with THE, so dependent on compiler policy) to involve objects of
2901 type FOO. Note that no such declaration is implied in
2902 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2903 * declared types of functions from the "Conditions"
2904 chapter. (reported by Paul Dietz)
2905 * bug fix: CERROR accepts a function as its first argument.
2906 * bug fix: NTH an NTHCDR accept a bignum as index
2907 arguments. (reported by Adam Warner)
2908 * optimization: character compare routines now optimize comparing
2909 against a constant character. (reported by Gilbert Baumann)
2910 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2911 argument now works properly on the MIPS platform.
2912 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2913 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2914 declared to have a functional type, does not cause an error
2916 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2917 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2918 * fixed bug: if last continuation of a deleted block has a
2919 destination, this destination should be deleted too. (reported by
2921 * fixed a bug in the bootstrap process: the host compiler's values
2922 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2923 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2924 test case from Patrik Nordebo)
2925 * improved the ability of the disassembler on the PPC platform to
2926 provide helpful disassembly notes.
2927 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2928 the class in more cases than previously.
2929 * bug fix: FILE-POSITION now understands :START and :END for
2930 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2931 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2932 8) data. (thanks to David Lichteblau)
2933 * bug fix: it is possible to add a method to a generic function
2934 without lambda list.
2935 * bug fix: reader failed to signal END-OF-FILE inside an
2936 object representation. (reported by Nikodemus Siivola)
2937 * fixed some bugs revealed by Paul Dietz' test suite:
2938 ** LAST and [N]BUTLAST should accept a bignum.
2939 ** condition slot accessors are methods.
2940 ** (VECTOR NIL) is a subtype of STRING.
2942 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2943 * minor incompatible change: some nonsensical specialized lambda
2944 lists (used in DEFMETHOD) which were previously ignored now signal
2946 * minor incompatible change: the system is now aware of the types of
2947 variables in the COMMON-LISP package, and will signal errors for
2948 most violations of these type constraints (where previously they
2949 were silently accepted).
2950 * minor incompatible change: COMPILE-FILE now uses the freedom
2951 afforded (ANSI 3.2.2.3) to use derived function types for
2952 functions defined in the same file. This also permits the system
2953 to warn on static type mismatches and function
2954 redefinition. (Currently it does not work with high DEBUG level.)
2955 * minor incompatible change: VALUES declaration is disabled.
2956 * When issuing notes, the compiler now signals a condition of type
2957 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2958 restart for use in user handlers. It is expected that the
2959 COMPILER-NOTE condition will eventually become a condition
2960 supertype to a hierarchy of note types, which will then be
2961 handleable in a similar fashion. However, at the moment, no such
2962 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2963 but it's an implementation detail, not a classification for the
2965 * Changes in type checking closed the following bugs:
2966 ** type checking of unused values (192b, 194d, 203);
2967 ** template selection based on unsafe type assertions (192c, 236);
2968 ** type checking in branches (194bc).
2969 * A short form of VALUES type specifier has ANSI meaning (it has
2970 increased the number of situations when SBCL cannot perform type
2972 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2973 keywords or constants is permissible.
2974 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2975 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2976 classes. (thanks to Antonio Martinez)
2977 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2978 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2979 argument to a pretty-print stream. Instead, it leaves any such
2980 operations to DESCRIBE-OBJECT methods.
2981 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2983 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2984 lambda lists. (thanks to David Lichteblau)
2985 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2986 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2987 (thanks to Gerd Moellmann)
2988 * fixed bug 246: increased compilation speed of long
2989 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2991 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2993 * DEFCONSTANT now throws a condition of type
2994 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2995 constant to a non-EQL value; CONTINUE and ABORT restarts
2996 respectively change and preserve the value.
2997 * fixed bug 63: The code walker, part of the implementation of CLOS,
2998 is now better at handling symbol macros.
2999 * bug fix: There is no longer an internal implementation type named
3000 CL:LENGTH. (reported by Raymond Toy)
3001 * bug fix: In macro-like defining macros/special operators the
3002 implicit block does not enclose the lambda list.
3003 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3004 enclosed in parentheses) are not suitable as type specifiers, and
3005 their use properly signals an error now.
3006 * bug fix: An argument count mismatch for a type specifier in code
3007 being compiled no longer causes an unhandled error at compile
3008 time, but signals a compile-time warning.
3009 * fixed simple vector readable printing
3010 * bug fix: DESCRIBE takes more care over whether the class
3011 precedence list slot of a class is bound before accessing it.
3012 (reported by Markus Krummenacker)
3013 * bug fix: FORMATTER can successfully compile pretty-printer format
3014 strings which use variants of the ~* directive inside.
3015 * bug fix: SEARCH now applies its TEST predicate to the elements of
3016 the arguments in the correct order. (thanks to Wolfhard Buss)
3017 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3018 in type inference. (thanks to Robert E. Brown)
3019 * bug fix: Reading in symbols with an explicit package name of ""
3020 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3021 * fixed some bugs revealed by Paul Dietz' test suite:
3022 ** NIL is now allowed as a structure slot name.
3023 ** Arbitrary numbers, not just REALs, are allowed in certain
3024 circumstances in LOOP for-as-arithmetic clauses.
3025 ** Multiple class redefinitions before slot access no longer
3026 causes a type error.
3027 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3028 association between the name and a class.
3029 ** Generic functions with non-standard method-combination and over
3030 five methods all of which return constants no longer return NIL
3031 after the first few invocations. (thanks to Gerd Moellmann)
3032 ** CALL-NEXT-METHOD with no arguments now passes the original
3033 values of the arguments, even in the presence of assignment.
3034 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3036 ** DISASSEMBLE works with closures and funcallable instances.
3037 ** ADD-METHOD now returns the generic function, not the new method.
3038 ** FIND-METHOD signals an error if the lengths of the specializers
3039 is incompatible with the generic function, even if the ERRORP
3041 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3042 which its argument is a member.
3043 ** DEFCLASS only redefines the class named by its class-name
3044 argument if that name is the proper name of the class;
3045 otherwise, it creates a new class.
3046 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3047 of the UNBOUND-SLOT condition to the name of the slot.
3048 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3049 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3050 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3051 SLOT-MAKUNBOUND in the specified fashion.
3053 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3054 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3055 cross-compilation host. As a consequence, we can now bootstrap our
3056 way up to SBCL starting with a bare gcc toolchain and human-readable
3057 source code (first the source to CLISP, then the source to SBCL).
3058 * A contributed module containing a partial implementation of the
3059 simple-streams interface has been included. (thanks to Rudi
3061 * A contributed module implementing the RFC1321 Message Digest
3062 Algorithm, known as MD5, has been included.
3063 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3064 no longer has any effect, as the code controlled by this feature
3065 has been deleted. (As far as we know, no-one has ever built using
3066 this feature, and its semantics were confused in any case).
3067 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3068 work on conditions (as required by the ANSI specification),
3069 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3070 expected behaviour on conditions. Users should note, however,
3071 that such behaviour is not required by the ANSI specification,
3072 and so use of this behaviour may render their code unportable.
3073 * fixed some bugs revealed by Paul Dietz' test suite:
3074 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3076 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3077 ** SXHASH on condition objects no longer returns NIL.
3078 ** :ALLOCATION :CLASS slots are better treated; their values are
3079 updated on class redefinition, and initforms inherited from
3080 superclasses are applied.
3081 ** REMOVE-METHOD returns its generic function argument even when
3082 no method was removed.
3083 ** SHARED-INITIALIZE now initializes the values of the requested
3084 slots, including those with :ALLOCATION :CLASS.
3085 ** ALLOCATE-INSTANCE now works on structure classes defined via
3086 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3088 ** SLOT-EXISTS-P now works on conditions, as well as structures
3090 ** MAKE-LOAD-FORM now has the required methods on
3091 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3092 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3093 keyword, and now implements the SLOT-NAMES argument.
3094 ** methods with &OPTIONAL arguments no longer allow too many
3095 arguments to be passed in the call without error.
3096 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3097 option is consistent with the required arguments of the generic
3098 function lambda list.
3099 * bug fix: REQUIRE accepts a string designator. (Thanks to
3101 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3102 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3103 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3104 * bug fix: The compiler now checks for duplicated variables in macro
3106 * bug fix: SETQ on globals returns the correct value.
3107 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3108 not a COMPILER-ERROR (followed by some other strange error on
3109 choosing the CONTINUE restart).
3110 * bug fix: make.sh and friends are now more consistent in the way that
3111 they look for GNU "make".
3113 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3114 * experimental native threads support (on x86 Linux >=2.4 only).
3115 This is not compiled in by default: you need to add :SB-THREAD to
3116 the target features. See the "Beyond ANSI" chapter of the manual
3118 * fix for longstanding nonANSIism: The old distinction between
3119 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3120 The return value from CL:FIND-CLASS is now a CLOS class, and
3121 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3122 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3123 * An interface to the MetaObject Protocol, as described in Kiczales,
3124 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3125 MIT Press, 1991, is available from the SB-MOP package.
3126 * incompatible change: the SB-PCL package should now be considered
3127 a private implementation detail, and no longer a semi-private MOP
3129 * minor incompatible change: due to rearrangement for threads, the
3130 control stack and binding stack are now allocated at arbitrary
3131 addresses instead of being hardcoded per-port. Users affected by
3132 this probably have to be doing advanced things with shared
3133 libraries, and will know who they are.
3134 * minor incompatible change: Previously, all --eval forms used were
3135 processed with READ before any of them were processed with EVAL.
3136 Now each --eval form is processed with both READ and EVAL before
3137 the next --eval form is processed. (Thus package operations like
3138 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3139 work as the user might reasonably expect.)
3140 * minor incompatible change: *STANDARD-INPUT* is now only an
3141 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3143 * minor incompatible change: Y-OR-N-P is now character-oriented,
3144 not line oriented. Also, YES-OR-NO-P now works without errors.
3145 (thanks to Antonio Martinez)
3146 * sb-aclrepl module improvements: an integrated inspector, added
3147 repl features, and a bug fix to :trace command.
3148 * Known functions, which cannot be open coded by the backend, are
3149 considered to be able to check types of their arguments. (fixing
3150 a bug report by Nathan J. Froyd)
3151 * fixed a bug in computing method discriminating functions: It is
3152 now possible to define methods specialized on classes which have
3153 forward-referenced superclasses. (thanks to Gerd Moellmann)
3154 * fixed evaluation order in optional entries (reported by Gilbert
3156 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3157 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3158 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3159 required-by-AMOP NAME argument, as well as CLASS and
3160 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3161 * fixed bug 20: DEFMETHOD can define methods using names that are
3162 not the proper names of classes to designate class specializers.
3163 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3164 Unix file descriptors, instead of blowing up. (thanks to Antonio
3166 * Garbage collection refactoring: user-visible change is that a
3167 call to the GC function during WITHOUT-GCING will not do garbage
3168 collection until the end of the WITHOUT-GCING. If you were doing
3169 this you were probably losing anyway.
3170 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3171 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3172 TYPEP the latter but not the former.
3173 * The compiler issues a full WARNING for calls to undefined functions
3174 with names from the CL package.
3175 * MAP-INTO for a vector destination is open coded. (reported by
3176 Brian Downing on c.l.l)
3177 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3178 documentation string.
3179 * fixed some bugs revealed by Paul Dietz' test suite:
3180 ** COPY-ALIST now signals an error if its argument is a dotted
3182 ** Condition slots are now accessed more correctly in the presence
3183 of multiple initargs for a given slot.
3184 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3185 exclude restarts of the same name associated with a different
3187 ** DEFCLASS of forward-referenced classes with another
3188 forward-referenced class in the superclasses list no longer
3190 ** Condition slots are now initialized once each, not multiple
3191 times. (thanks to Gerd Moellmann)
3192 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3194 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3195 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3196 arguments contain duplicated elements.
3197 ** RESTART-CASE understands local macros.
3198 ** RESTART-CASE associates exactly its own restarts with a condition.
3199 ** ENDP in safe mode checks its argument to be of type LIST.
3200 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3201 in question is unbound.
3202 ** Optimized MAKE-INSTANCE functions no longer cause internal
3203 assertion failures in the presence of duplicate initargs.
3204 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3205 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3206 propagated even in the ctor optimized implementation of
3208 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3210 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3211 * a better implementation of SXHASH on (simple) bit vectors,
3212 measured both in execution speed and in distribution of results
3213 over the positive fixnums, has been installed. Likewise, a better
3214 implementation of EQUAL for simple bit vectors is now available.
3215 * fixed CEILING optimization for a divisor of form 2^k.
3216 * fixed bug 240 (emitting extra style warnings "using the lexical
3217 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3218 by Antonio Martinez)
3219 * fixed SXHASH, giving different results for NIL depending on type
3220 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3221 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3222 variable. (found by Rolf Wester)
3223 * SBCL does not ignore type declarations for special
3224 variables. (reported by rif on c.l.l 2003-03-05)
3225 * some bug fixes in contrib/sb-aclrepl/
3226 * fixed some bugs revealed by Paul Dietz' test suite:
3227 ** a bug in the CONS type specifier, whereby the CAR and CDR
3228 types got intertwined, has been fixed;
3229 ** the type system is now able to reason about the interaction
3230 between INTEGER and RATIO types more completely;
3231 ** APPEND, [N]REVERSE and NRECONC check that those their
3232 arguments, which must be proper lists, are really so;
3233 ** An array specialized to be unable to hold elements has been
3234 implemented, as required -- yes, really -- by ANSI;
3235 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3236 on malformed property lists;
3238 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3239 * incompatible packaging change: in line with Unix convention,
3240 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3241 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3242 * REQUIRE and PROVIDE are now optionally capable of doing something
3243 useful. See the documentation string for REQUIRE.
3244 * infrastructure for a managed SBCL contrib system: contributed
3245 modules in this release include:
3246 ** the ASDF system definition facility;
3247 ** an interface to the BSD Sockets API;
3248 ** an ACL-like convenience interface to the repl;
3249 (thanks to Kevin Rosenberg)
3250 ** an implementation of ROTATE-BYTE, with efficient implementation
3252 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3253 gives a closer approximation to the right answer than previously.
3254 (thanks to Raymond Toy)
3255 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3256 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3257 optional environment arguments, as required by ANSI.
3258 * fixed bugs in other functions taking environment objects, allowing
3259 calls with an explicit NIL environment argument to be compiled
3261 * fixed bug 228: primary return values from
3262 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3263 COMPILE or FUNCTION.
3264 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3265 :TYPE will no longer signal errors on innocuous objects.
3266 * fixed bug 231b: SETQ is better at respecting type declarations in
3267 the lexical environment.
3268 * fixed a bug in DEFCLASS: classes named by symbols with no or
3269 unprintable packages can now be defined.
3270 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3271 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3272 * fixed compiler failure related to checking types of functions.
3273 (reported by Robert E. Brown)
3274 * the compiler is now much more consistent in its error-checking
3275 treatment of bounding index arguments to sequence functions: in
3276 (SAFETY 3) code, errors will be signalled in almost all cases if
3277 invalid sequence bounding indices are passed to functions defined
3278 by ANSI to operate on sequences.
3279 * fixed a bug in the build procedure: documentation of SBCL-specific
3280 packages is now preserved and available in the final Lisp image.
3281 * lifted FDEFINITION lookup out of loops in the implementation of
3282 many list operations. (thanks to Robert E. Brown)
3283 * fixed a bug in the reader: the #n# reader macro now works for
3284 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3285 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3286 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3287 E. Brown and Tony Martinez respectively)
3288 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3289 SLOT-BOUNDP: in particular, it is now able to optimize them much
3290 better, and is now not vulnerable to having packages renamed.
3291 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3292 always called when a slot is not present in an instance. (thanks
3294 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3295 longer signal an error on unreading a character following EOF on
3296 the previous constituent stream. (thanks to Tony Martinez)
3297 * fixed some bugs revealed by Paul Dietz' test suite:
3298 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3299 not just nonnegative fixnums;
3300 ** the logical bit-array operators such as BIT-AND now accept an
3301 explicit NIL for their "opt-arg" argument (to indicate a
3302 freshly-consed result bit-array);
3303 ** ELT now signals an error on an invalid sequence index in safe
3305 ** the type system is now cleverer about negations of numeric
3306 types, and consequently understands the BIGNUM and RATIO types
3308 ** the type system is now cleverer about the interaction between
3309 INTEGER and RATIO types: while bugs still remain, many more
3310 cases are accurately computed;
3311 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3312 if it is in the last clause;
3313 ** CONSTANTLY now correctly returns a side-effect-free function in
3315 ** DECLARE is no longer treated as a special-operator; in
3316 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3317 * incremented fasl file version number due to the change in the
3318 DEFSTRUCT-SLOT-DESCRIPTION structure.
3320 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3321 * minor incompatible change: code processed by the "interpreter" or
3322 EVAL now has a compilation optimization policy of (DEBUG 2)
3323 (changed from (DEBUG 1)) to improve debuggability of interactive
3324 development, and to allow the use of the debug RETURN command in
3326 * an experimental implementation of the RETURN command for the
3327 debugger has been included. (thanks to Frederik Kuivinen)
3328 * fixed bug 62: constraints were not propagated into a loop.
3329 * fixed bug in embedded calls of SORT (reported and investigated by
3331 * fixed some bugs revealed by Paul Dietz' test suite:
3332 ** printing and reading of arrays with some dimensions having
3333 length 0 (thanks to Gerd Moellmann);
3334 ** BOA constructor with &AUX argument without a default value does
3335 not cause a type error;
3336 ** CONSTANTP now returns true for all self-evaluating objects.
3338 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3339 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3340 accessors that are related by inheritance, as specified in the
3341 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3342 Valtteri Vuorikoski)
3343 * The compiler is now able to inline functions that were defined in
3344 a complex lexical environment (e.g. inside a MACROLET).
3345 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3347 * Support for the upcoming FreeBSD-5.0 release has been included.
3348 (thanks to Dag-Erling Smorgrav)
3349 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3350 effect when it is not in a toplevel context.
3351 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3352 stand a better chance of being correct. (thanks to Gerd
3354 * fixed bug in COERCE, which now signals an error on coercing a
3355 rational to a bounded real type which excludes the expected
3357 * The compiler is now able to derive types more accurately from the
3358 COERCE and COMPILE functions.
3359 * fixed bug 223: functional binding is considered to be constant
3360 only for symbols in the CL package.
3361 * fixed bug 231: SETQ did not check the type of a variable being set
3362 (reported by Robert E. Brown)
3363 * A new optimization for MAKE-INSTANCE has been included, fixing
3364 various bugs (including relating to :ALLOCATION :CLASS slots and
3365 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3367 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3368 ** As required by ANSI, LOOP now disallows anonymous collection
3369 clauses such as COLLECT I in conjunction with aggregate boolean
3370 clauses such as THEREIS (= I 1);
3371 ** LOOP now signals an error when any variable is reused in the
3372 same loop (including the potentially useful construct analogous
3373 to WITH A = 1 WITH A = (1+ A);
3374 ** IT is only a special loop symbol within the first clause of a
3375 conditional loop clause;
3376 ** LOOP with a typed iteration variable over a hashtable now
3377 signals a type error iff it should.
3378 * fixed some other bugs revealed by Paul Dietz' test suite:
3379 ** FILE-STREAM now names the class previously known as FD-STREAM;
3380 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3381 argument) no longer signals an error;
3382 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3383 of the slot symbol, rather than using the current package
3384 ((:CONC-NAME "") continues to intern the slot's name in the
3386 * incremented fasl file version number, because of the incompatible
3387 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3388 of the new implementation of DEFINE-COMPILER-MACRO.
3390 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3391 * Support for building SBCL for MIPS platforms running in
3392 little-endian mode has now been checked in, and basic
3393 functionality on said platforms verified.
3394 * minor incompatible change: PCL now records the pathname of a file
3395 in which methods and the like are defined, rather than its
3397 * minor incompatible change: TRUENAME now considers the truename of
3398 a file naming a directory to be the pathname with :DIRECTORY
3399 component indicating that directory.
3400 * minor incompatible change: a NAMED clause in the extended form of
3401 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3402 reason for the previous behaviour is unclear.
3403 * more systematization and improvement of CLOS and MOP conformance
3404 in PCL (thanks to Gerd Moellman and Pierre Mai):
3405 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3407 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3408 primary methods with no specializers;
3409 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3411 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3412 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3413 FORWARD-REFERENCED-CLASSes; error reporting on
3414 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3416 ** SXHASH on CLOS instances now uses a slot internal to the
3417 instance to return different numbers on distinct instances,
3418 while preserving the same return value through invocations of
3420 ** DEFMETHOD signals errors when methods with longer incongruent
3421 lambda lists are added to generic functions;
3422 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3423 CLASS, as specified in AMOP;
3424 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3425 on the order returned by the primary method for classes of
3426 class STANDARD-CLASS;
3427 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3428 * fixed some bugs shown by Paul Dietz' test suite:
3429 ** DOLIST puts its body in TAGBODY;
3430 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3432 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3433 value producing form;
3434 ** if more variables are given to PROGV than values, extra
3435 variables are bound and made to have no value;
3436 ** NSUBSTITUTE on list arguments gets the right answer with
3438 ** ELT signals an error of type TYPE-ERROR when the index argument
3439 is not a valid sequence index;
3440 ** LOOP signals (at macroexpansion time) an error of type
3441 PROGRAM-ERROR when duplicate variable names are found;
3442 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3443 ** FUNCALL of special-operators now cause an error of type
3445 ** PSETQ now works as required in the presence of side-effecting
3446 symbol-macro places;
3447 ** NCONC accepts any object as its last argument;
3448 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3450 ** loop-for-as-package does not require a package to be explicitely
3452 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3454 * fixed bug 166: compiler preserves "there is a way to go"
3455 invariant when deleting code.
3456 * fixed bug 172: macro lambda lists with required arguments after
3457 &REST arguments now cause an error to be signalled. (thanks to
3459 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3460 bug. (thanks to Matthew Danish)
3461 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3463 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3464 arguments are lexically rebound. (thanks to Gerd Moellmann and
3466 * fixed bug 194: error messages are now more informative when there
3467 is no primary method applicable in a call to a generic
3468 function. (thanks to Gerd Moellmann)
3469 * fixed bug in command line argument checking (thanks to Julian
3471 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3472 (thanks to Matthew Danish)
3473 * incremented fasl file version number, because of the
3474 SXHASH-related changes in the layout of CLOS data structures
3476 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3477 * minor incompatible change: The runtime (the Unix executable named
3478 "sbcl") is now much pickier about the .core files it will load.
3479 Essentially it now requires .core files to descend from the same
3480 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3481 as the runtime does. (The intent is to prevent the crashes which
3482 can occur, and which can even be reported as mysterious failures,
3483 when people patch the sources or change the build parameters
3484 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3485 sbcl and .core files.)
3486 * fixed bug: VALUES-LIST is no longer optimized away.
3487 * fixed bug 142: The FFI conversion of C string values to Lisp
3488 string values no longer conses excessively. (thanks to Nathan
3489 Froyd porting Raymond Toy's fix to CMU CL)
3490 * began to systematize and improve MOP conformance in PCL (thanks to
3491 Nathan Froyd, Gerd Moellman and Pierre Mai):
3492 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3494 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3495 ** FINALIZE-INHERITANCE is now called on class finalization;
3496 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3497 argument precedence order.
3498 * fixed bug 202: The compiler no longer fails on functions whose
3499 derived types contradict their declared type.
3500 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3501 so it can be non-toplevel.
3502 * The fasl file version number has changed (because of the new
3503 implementation of DEFMACRO).
3504 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3505 safe code, that any length requirement by their type-specifier
3506 argument is valid. The exceptions to this are described in bug
3508 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3509 functions now check the types of their inputs as required by ANSI.
3510 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3511 introduced symbol is DECLAREd to be SPECIAL.
3512 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3513 * fixed bug: :COUNT argument to sequence functions may be negative
3514 * fixed bug: body of DO-SYMBOLS may contain declarations
3515 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3516 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3517 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3518 symbol macro only once
3519 * fixed printing of call frame when argument list is unavailable
3520 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3521 * compiler no longer signals WARNING on unknown keyword
3524 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3525 * A beta-quality port to the mips architecture running Linux,
3526 based on the old CMUCL backend, has been made. It has been tested
3527 on a big-endian kernel, and works sufficiently well to be able to
3528 rebuild itself; it has not been tested in little-endian mode.
3529 * fixed an inconsistency between gencgc.c and purify.c which made
3530 dumping/loading .core files unreliable
3531 * fixed bug 120a: The compiler now deals correctly with IFs where
3532 the consequent is the same as the alternative, instead of
3533 misderiving the return type. (thanks to Alexey Dejneka)
3534 * fixed bug 113: Logical pathnames are now dumpable (the logical
3535 host is resolved at load-time, throwing an error if it is not
3537 * fixed bug 174: FORMAT's error message is slightly clearer when a
3538 non-printing character is used in a format directive.
3539 * fixed several bugs in compiler checking of type declarations, i.e.
3540 violations of the Python "declarations are assertions" principle
3541 (thanks to Alexey Dejneka)
3542 * fixed several bugs in PCL's error checking (thanks to Gerd
3544 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3546 * fixed bug in compilation of functions as first class values
3547 (thanks to Antonio Martinez-Shotton)
3548 * The compiler's handling TYPE-ERRORs which it can prove will
3549 inevitably happen at runtime has been cleaned up and corrected
3550 in several ways. (thanks to Alexey Dejneka)
3551 * improved argument type checking for various basic arithmetic
3552 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3553 lavished on them in the past that they can be compiled in many
3554 ways in different special cases
3555 * fixed bug 181: compiler checks validity of user supplied type
3557 * cleaned up code flushing in optimization: Function calls which
3558 should signal errors for safety purposes (e.g. which ANSI says
3559 should signal errors when their arguments are of incorrect type)
3560 are no longer optimized away.
3561 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3562 * incremented fasl file version number, because changes in the
3563 implementation of sequence functions like COERCE caused
3564 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3565 in old inline expansions) to become undefined. (Actually these
3566 changes were later undone, so we might very well be binary
3567 compatible with 0.7.7 after all, but leaving the version number
3568 incremented seemed like the simplest and most conservative
3571 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3572 * An alpha-quality port to the parisc architecture running Linux,
3573 based on the old CMUCL backend, has been made. This, even more so
3574 than the other backends, should be considered still a work in
3575 progress; known problems include that the Linux kernel in 64-bit
3576 mode does not propagate the correct sigcontext structure to
3577 userspace, and consequently SBCL on a parisc64 kernel will not
3579 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3580 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3581 suppressed.) Also now that the compiler is looking at declarations
3582 in the environment, it checks optimization declarations as well,
3583 and suppresses inlining when (> DEBUG SPEED).
3584 * More fixes have been made to treatment of floating point exception
3585 treatment and other Unix signals. In particular, floating point
3586 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3587 * The detection and handling of control stack exhaustion (infinite
3588 or very deeply nested recursion) has changed. Stack exhaustion
3589 detection is now done by write-protecting pages at the OS level
3590 and applies at all optimization settings; when found, a
3591 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3592 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3593 be caught using IGNORE-ERRORS.
3594 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3595 correctly when transforming recursive LABELS functions to LETs.
3596 (thanks to Alexey Dejneka)
3597 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3598 that are names of constants or global variables.
3599 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3600 alien routines with docstrings.
3601 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3602 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3604 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3605 (thanks to Eric Marsden)
3606 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3607 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3608 * Bug fix: The SPARC backend can now compile functions involving
3609 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3610 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3611 to a functions expecting a PATHNAME-DESIGNATOR.
3612 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3613 lambda lists. (thanks to Alexey Dejneka)
3614 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3615 functions so that EXPORTing the name of the class doesn't cause
3616 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3617 to Antonio Martinez for reporting this)
3618 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3619 constant-folded without failing an assertion. (thanks to Einar
3620 Floystad Dorum for reporting this)
3621 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3622 types are now dealt with more correctly by the compiler.
3623 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3624 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3625 behaviour analogous to RENAME-FILE. This puts its behaviour more
3626 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3628 * The fasl file version number has changed again. (because of the
3629 bug fix involving the names of PCL MAKE-INSTANCE functions)
3631 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3632 * bug fix: Floating point exceptions are treated much more
3633 consistently on the x86/Linux and PPC/Linux platforms.
3634 * Array initialization with :INITIAL-ELEMENT is now much faster for
3635 cases when the compiler cannot open code the array creation, but
3636 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3637 array accesses have also seen a speed increase.
3638 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3639 specified. (This was at the root of some bad interactions between
3640 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3642 * bug fix: Internal error arguments for undefined functions are now
3643 computed correctly on the PPC/Linux platform.
3644 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3645 Raymond Toy's patch for CMU CL)
3646 * Support for the Solaris 9 operating environment has been included
3647 (thanks to Daniel Merritt)
3648 * A very ugly but hopefully complete draft of the missing FFI chapter
3649 of the manual has been created by reformatting the corresponding
3650 CMU CL manual chapter into (currently very ugly and incoherent)
3651 DocBook and bringing it up to date for SBCL behavior. Thus, the
3652 manual is now essentially complete, at least by my extreme
3653 once-and-only-once standards, whereby it's acceptable to refer to
3654 the doc strings of SB-EXT functions as the primary documentation.
3655 * The fasl file version number has changed again, due to cleanup of
3656 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3657 is no longer a static symbol.)
3659 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3660 * SBCL now builds with OpenMCL (version 0.12) as the
3661 cross-compilation host; also, more progress has been made toward
3662 bootstrapping under CLISP.
3663 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3665 * bug 158 fixed: The compiler can now deal with integer loop
3666 increments different from 1; fixing this turned out also to fix
3668 * bug 169 fixed: no more bogus warnings about using lexical bindings
3669 despite the presence of perfectly good SPECIAL declarations (thanks
3670 to David Lichteblau)
3671 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3672 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3673 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3675 * bug 180 fixed: Method combination specifications no longer ignore
3676 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3677 * bug fix: Structure type predicate functions now check their argument
3678 count as they should.
3679 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3680 correctly. (thanks to Pierre Mai)
3681 * minor incompatible change: The --noprogrammer option is deprecated
3682 in favor of the new --disable-debugger option, which is very similar.
3683 (The major difference is that it takes effect at a slightly different
3684 time at startup, causing handling of errors in --sysinit and
3685 --userinit files will be affected differently.) The
3686 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3687 been added to allow this functionality to be controlled from ordinary
3688 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3689 maintainers, who might want to run non-interactive scripts to
3690 build SBCL cores which will later be used interactively.)
3691 * minor incompatible change: The LOAD function no longer, when given
3692 a wild pathname to load, loads all files matching that pathname.
3693 Instead, an error of type FILE-ERROR is signalled.
3695 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3696 * bug 147 fixed: The compiler preserves its block link/count
3697 invariants more correctly now so that it doesn't crash. (thanks
3699 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3701 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3702 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3703 no-value-for-this-character as NIL (as specified by ANSI).
3704 * HOST-NAMESTRING on physical pathnames now returns a string that is
3705 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3706 (thanks to Christophe Rhodes)
3707 * The Alpha port handles icache flushing more correctly. (thanks to
3709 * More progress has been made toward bootstrapping under CLISP. (thanks
3710 to Christophe Rhodes)
3711 * The fasl file format has changed again, because dynamic loading
3712 on OpenBSD (which has non-ELF object files) motivated some cleanups
3713 in the way that foreign symbols are transformed and passed around.
3714 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3715 is no longer treated as whitespace by the reader, but instead as
3716 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3717 |A
\x7fB|, instead of A as it used to.
3719 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3720 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3721 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3722 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3723 this as of 0.7.1.45, but was left out of the previous news section
3724 (thanks to Dan Barlow)
3725 * SBCL now runs on the Solaris operating system on SPARC architectures
3726 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3727 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3728 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3729 * SPARC backend cleanups, allowing builds of cores optimized for V8
3730 and V9 SPARCS, and also emission of code targeted to a particular
3731 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3733 * SBCL is closer to bootstrapping under CLISP, thanks to various
3734 fixes by Christophe Rhodes.
3735 * The fasl file format has changed again, to allow the compiler's
3736 INFO database to support symbol macros.
3737 * The user manual (in doc/) is formatted into HTML more nicely.
3738 (thanks to coreythomas)
3739 * The system is smarter about SUBTYPEP relationships, especially
3740 those involving NOT types (including types such as ATOM which are
3741 represented internally using NOT types). Thus SUBTYPEP is less
3742 likely to return (VALUES NIL NIL) in general, and in particular
3743 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3745 * The fasl file format has changed again, because the internal
3746 representation of types now includes a new slot to support the new
3747 SUBTYPEP-of-NOT-types logic.
3748 * (not a change in the main branch of SBCL, but a related prototype
3749 which can hopefully be merged into the main branch of SBCL in the
3750 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3751 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3752 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3753 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3754 dispatch macro character combinations. (thanks to Alexey Dejneka)
3755 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3756 that superficially look like logical namestrings correctly.
3757 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3759 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3760 default supplied-p) for &optional and &key arguments. (thanks to
3763 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3764 * incompatible change: The compiler is now less aggressive about
3765 tail call optimization, doing it only when (> SPACE DEBUG) or
3766 (> SPEED DEBUG). (This is an incompatible change because there are
3767 programs which relied on the old CMU-CL-style behavior to optimize
3768 away their unbounded recursion which will now die of stack overflow.)
3769 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3770 for non-GENCGC systems has been increased to 20M (since that
3771 seems much closer to the likely performance optimum for modern
3772 systems than the old 4M value was)
3773 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3774 and *DEBUG-PRINT-LEVEL*
3775 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3776 of CMU CL's support for SPARC, and various endianness and other
3777 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3778 * new syntactic sugar for the Unix command line: --load foo.bar is now
3779 an alternate notation for --eval '(load "foo.bar")'.
3781 ** The system now detects stack overflow and handles it gracefully,
3782 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3783 optimization settings. (This is a good thing in general, and
3784 its introduction in this version should be particularly timely
3785 for anyone whose code fails because of suppression of tail
3787 ** The system now hunts for the C variable "environ" in a more
3788 devious way, to avoid segfaults when the C library version differs
3789 between compile time and run time. (thanks to Christophe Rhodes)
3790 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3791 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3792 ** The compiler no longer issues bogus style warnings for undefined
3793 classes in the same source file as the DEFCLASSes which defined
3794 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3796 ** fixes in CONDITION class precedence list for undefined function
3797 errors (thanks to Alexei Dejneka)
3798 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3799 correctly. (thanks to Dan Barlow)
3800 ** portability fixes aiming at bootstrapping under CLISP (thanks
3801 to Dave McDonald and Christophe Rhodes)
3802 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3803 ** fixes in type translation and and type inference (thanks to
3805 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3806 ** various fixes in the new ports (thanks to Dan Barlow)
3807 * several changes related to debugging:
3808 ** suppression of tail recursion, as noted above
3809 ** stack overflow detection, as noted above
3810 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3811 is now the default. (For some time encapsulation has been more
3812 reliable than the breakpoint-based :ENCAPSULATE NIL
3813 implementation, at least on X86 systems; and I just noticed that
3814 encapsulation also seems closer to the spirit of the ANSI
3817 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3819 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3820 up properly again. (There was a packaging bug in 0.7.0 which
3821 left their definitions in SB-SYS::LOAD-FOREIGN and
3822 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3823 vital for most things which interface to C-level interfaces,
3824 like extensions working with sockets or databases or
3825 Perl-compatible regexes or whatever, and the need to fix
3826 this bug is the main reason that 0.7.1 was released so
3828 ** DEFGENERIC is now choosier about the methods it redefines, so that
3829 reLOADing a previously-LOADed file containing DEFGENERICs does
3830 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3831 cycle now works as it should. (thanks to Alexey Dejneka)
3832 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3833 fixed by Christophe Rhodes. (He actually submitted this patch
3834 months ago, and I delayed until after 0.7.0.)
3835 ** Bug 111 (internal compiler confusion about runtime checks on
3836 FUNCTION types) was fixed by Alexey Dejneka.
3837 * Some internal cleanups (getting rid of variables which aren't
3838 needed now that the byte interpreter is gone) caused the fasl
3839 file format number to change again.
3841 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3842 * major incompatible change: The default fasl file extension, i.e. the
3843 default extension for files produced by COMPILE-FILE, has changed
3844 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3846 ** There are many changes in the implementation of the compiler.
3847 SBCL is now essentially a compiler-only implementation of ANSI
3848 Common Lisp. EVAL still "interprets" a few special cases, but
3849 almost all the interesting cases are handled by creating
3850 a LAMBDA expression, calling COMPILE on it, then calling
3851 FUNCALL on the result.
3852 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3853 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3854 Since the code is newer, there might still be some new bugs
3855 (though not as many as before Martin Atzmueller's fixes:-). But
3856 the new code is substantially simpler and clearer, and hopefully
3857 any remaining bugs will be simpler, less fundamental, and more
3858 fixable then the bugs in the old code.
3859 ** The revised compiler is still a little unsteady on its feet.
3861 *** The debugging information it produces (particularly the names
3862 of FUNCTION objects) is sometimes much less useful than what
3863 the old compiler produced.
3864 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3865 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3866 within a MACROLET) has been temporarily weakened.
3867 ** There are new compiler optimizations for various functions:
3868 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3869 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3870 *** the math functions TRUNCATE, FLOOR, and CEILING
3871 *** the function-of-all-trades COERCE
3872 Mostly these should be transparent, but there's one
3873 potentially-annoying problem (bug 117): when the compiler
3874 inline-expands a function and does type analysis on the result,
3875 it can create control paths which have type mismatches, and
3876 when it can't prove that those control paths aren't taken,
3877 it will issue WARNINGs about the type mismatches. This is
3878 a particular problem in practice for the new sequence functions.
3879 It's not clear how this should be fixed, and for now, a
3880 workaround is given in the entry for 117 in the BUGS file.
3881 ** (Because of the interaction between the two previous items --
3882 occasional inlining problems and new inline expansions -- some
3883 of the new sequence function optimizations won't really kick in
3884 completely until debugging information, and then inlining, are
3885 straightened out in some future version.)
3886 * minor incompatible changes:
3887 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3888 DIRECTORY no longer implicitly promotes NIL slots of its
3889 pathname argument to :WILD. In particular, when you ask for the
3890 contents of a directory (which you used to be able to do without
3891 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3892 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3893 ** changes in behavior that ANSI explicitly defines to be
3894 implementation dependent:
3895 *** The new compiler-only implementation still conforms with ANSI,
3896 but acts a little different than before. Besides the obvious
3897 changes in performance tradeoffs (that the cost per form passed
3898 to EVAL has gone up, and the cost per form executed by EVAL
3899 has gone down), the behavior of the system changes a little
3900 because there are no longer any interpreted function objects.
3901 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3902 e.g. doing COMPILE on the output of interactive DEFUN is
3904 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3906 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3907 has changed from (:CL) to NIL.
3908 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3909 e.g. #\Newline and #\Tab, don't have names specified in the
3910 ANSI Common Lisp standard, is now based on their ASCII symbolic
3911 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3912 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3913 are no longer used for output.
3914 ** changes in internal implementation constants:
3915 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3916 4 million. (If your application spends a lot of time GCing and
3917 you have a lot of RAM, you might want to experiment with
3918 increasing it even more.)
3919 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3920 However, almost all old code should still continue to work without
3921 immediate update, as SB-C-CALL is now a (deprecated) nickname
3923 ** Old operator names in the style DEF-FOO are now deprecated in
3924 favor of new corresponding names DEFINE-FOO, for consistency with
3925 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3926 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3927 internal symbols, but a few supported extensions like
3928 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3929 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3930 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3931 etc. as you get deeper into recursive calls to the debugger
3932 command loop, instead of the old "5]", "5]]", "5]]]"
3933 sequence. (I was motivated to do this when squabbles between
3934 ILISP and SBCL left me very deeply nested in the debugger. In the
3935 short term, this change will probably provoke more ILISP/SBCL
3936 squabbles, but hopefully it will be an improvement in the long run.)
3937 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3938 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3939 function, since declaiming return types involving aliens is
3940 (1) annoyingly messy to do by hand and (2) vital to efficient
3941 compilation of code which calls such functions.
3942 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3943 longer reexported by the SB-EXT package. They're solely useful
3944 for alien code, so it seems more logical that you should get
3945 them from the SB-ALIEN package, not in SB-EXT.
3946 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3947 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3948 features. Instead, the code that they used to control is always
3949 built into the system.
3950 * many other bug fixes
3951 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3952 advantage of the new EVAL-WHEN stuff and to clean them up in
3953 general, and they are now more ANSI-compliant in a number of
3954 ways. Martin Atzmueller is responsible for a lot of this.
3955 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3957 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3958 *** correct ERROR type for various file operations
3959 *** some fixes for Lisp streams
3960 *** DEFMETHOD syntax checking
3961 *** changing old weird representation of debug information as
3962 strings (which, among their other deficiencies, don't transform
3963 correctly when you rename packages, and don't change their
3964 print representation when you change things like *PACKAGE*
3965 and *PRINT-LENGTH*) to symbols and lists of symbols
3966 He also made several improvements and fixed several bugs in DESCRIBE.
3967 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3969 *** misbehavior of WRITE-STRING/WRITE-LINE
3970 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3971 and several other LOOP problems as well
3972 *** DIRECTORY when similar filenames are present
3973 *** DEFGENERIC with :METHOD options
3974 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3975 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3976 *** argument ordering in FIND with :TEST option
3977 *** mishandled package designator argument in APROPOS-LIST
3978 *** various problems in the backquote readmacro
3979 *** a bug in APROPOS
3980 *** probably some others that I'm not describing very well here,
3981 since the CVS log documents them by reference to sbcl-devel
3982 messages, and the SourceForge archives aren't working well.:-(
3983 ** Dan Barlow improved the Alpha port (and is making progress on the
3984 PPC port, for those of you who think different).
3985 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3986 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3987 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3988 continues to work on the SPARC port (for those of you in a position
3989 to look down upon our little PC-compatible boxes from a great height).
3990 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3991 and uses that copy, rather than the current dynamic value, when
3992 it's trying to decide whether to truncate output. Thus e.g.
3993 (let ((*print-lines* 50))
3994 (pprint-logical-block (stream nil)
3996 (let ((*print-lines* 8))
3997 (print (aref possiblybigthings i) stream)))))
3998 should now truncate the logical block only at 50 lines, instead of
3999 often truncating it at 8 lines, as it did before.
4000 * The doc/cmucl/ directory, containing old CMU CL documentation
4001 from the time of the fork, is no longer part of the base system.
4002 SourceForge has shut down its anonymous FTP service, and with it
4003 my original plan for distributing the old CMU CL documentation
4004 there. For now, if you need these files you can download an old
4005 SBCL source release and extract them from it.
4006 * The fasl file version number changed again, for dozens of reasons,
4007 some of which are apparent above.
4009 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4010 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4011 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4012 patch, so that DISASSEMBLE output is much nicer.
4013 * The code in the SB-PROFILE package now seems reasonably stable.
4014 I still haven't decided what the final interface should look like
4015 (I'd like PROFILE to interact cleanly with TRACE, since both
4016 facilities use function encapsulation) but if you have a need
4017 for profiling now, you can probably use it successfully with
4018 the current CMU-CL-style interface.
4019 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4020 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4021 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4022 link now returns the dangling link itself, and for similar
4023 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4024 link itself. (In these cases the old code signalled an error and
4025 looped endlessly, respectively.) Thus, DIRECTORY now works even
4026 in the presence of dangling and cyclic symbolic links.
4027 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4028 is now a supported extension again, since the consensus on
4029 sbcl-devel was that it can be useful for ordinary development
4030 work, not just for debugging SBCL itself.
4031 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4032 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4033 that currently-defined functions might be redefined later with
4034 different return types.
4035 * Hash tables can be printed readably, as inspired by CMU CL code
4036 of Eric Marsden and SBCL code of Martin Atzmueller.
4037 * better error handling in CLOS method combination, thanks to
4038 Martin Atzmueller porting Pierre Mai's CMU CL patches
4039 * more overflow fixes for >16Mbyte I/O buffers
4040 * A bug in READ has been fixed, so that now a single Ctrl-D
4041 character suffices to cause end-of-file on character streams.
4042 In particular, now you only need one Ctrl-D at the command
4043 line (not two) to exit SBCL.
4044 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4046 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4047 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4048 does the right thing.
4049 * The implementation of some type tests, especially for CONDITION
4050 types, is now tidier and maybe faster, due to CMU CL code
4051 originally by Douglas Crosher, ported by Martin Atzmueller.
4052 * Some math functions have been fixed, and there are new
4053 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4054 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4055 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4056 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4057 likely to be built into sbcl-0.7.0.)
4058 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4059 CL, ported to SBCL by Martin Atzmueller)
4060 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4061 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4062 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4063 * There's a new slam.sh hack to shorten the edit/compile/debug
4064 cycle for low-level changes to SBCL itself, and a new
4065 :SB-AFTER-XC-CORE target feature to control the generation of
4066 the after-xc.core file needed by slam.sh.
4067 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4068 COMPILE-FILE is no longer supported, so that now every function
4069 gets an entry point, so that block compilation looks a little
4070 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4071 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4072 returns the number of bytes consed since the system started,
4073 rather than the number consed since the first time the function
4074 was called. (The new definition parallels ANSI functions like
4075 CL:GET-INTERNAL-RUN-TIME.)
4076 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4077 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4078 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4079 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4080 * Fasl file version numbers are now independent of the target CPU,
4081 since historically most system changes which required version
4082 number changes have affected all CPUs equally. Similarly,
4083 the byte fasl file version is now equal to the ordinary
4086 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4087 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4088 is no longer recognized. I apologize for this, because it was
4089 listed in SB-EXT as a supported extension, but I found that
4090 its existing behavior was poorly specified, as well as incorrectly
4091 specified, and it looked like too much of a mess to straighten it
4092 out. I have enough on my hands trying to get ANSI stuff to work..
4093 * many patches ported from CMU CL by Martin Atzmueller, with
4094 half a dozen bug fixes in pretty-printing and the debugger, and
4095 half a dozen others elsewhere
4096 * fixed bug 13: Floating point infinities are now supported again.
4097 They might still be a little bit flaky, but thanks to bug reports
4098 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4099 as flaky as they were.
4100 * The --noprogrammer command line option is now supported. (Its
4101 behavior is slightly different in detail from what the old man
4102 page claimed it would do, but it's still appropriate under the
4103 same circumstances that the man page talks about.)
4104 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4105 are now supported, and enabled by default. Thus, the compiler can
4106 handle many floating point and complex operations much less
4107 inefficiently. (Thus e.g. you can implement a complex FFT
4109 * The compiler now detects type mismatches between DECLAIM FTYPE
4110 and DEFUN better, and implements CHECK-TYPE more correctly, and
4111 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4112 thanks to patches from Martin Atzmueller.
4113 * various fixes to make the cross-compiler more portable to
4114 ANSI-conforming-but-different cross-compilation hosts (notably
4115 Lispworks for Windows, following bug reports from Arthur Lemmens)
4116 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4117 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4118 fixed thanks to Pierre Mai's CMU CL patches.
4119 * fixes to keep the system from overflowing internal counters when
4120 it tries to use i/o buffers larger than 16M bytes
4121 * fixed bug 45a: Various internal functions required to support
4122 complex special functions have been merged from CMU CL sources.
4123 (When I was first setting up SBCL, I misunderstood a compile-time
4124 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4125 * improved support for type intersection and union, fixing bug 12
4126 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4127 more obscure bugs as well
4128 * some steps toward byte-compiling non-performance-critical
4129 parts of the system, courtesy of patches from Martin Atzmueller
4130 * Christophe Rhodes has made some debian packages of sbcl at
4131 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4132 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4133 stable, but are nonetheless usable. When he's ready, I'd be happy
4134 to add them to the SourceForge "File Releases" section. (And if
4135 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4136 * new fasl file format version number (because of changes in
4137 internal representation of (OR ..) types to accommodate the new
4138 support for (AND ..) types, among other things)
4140 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4141 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4143 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4145 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4146 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4147 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4148 * DESCRIBE now works on structure objects again.
4149 * Most function call argument type mismatches are now handled as
4150 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4151 know whether the function will be redefined before the call is
4152 executed. (The compiler could flag local calls with full WARNINGs,
4153 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4154 it doesn't keep track of enough information to know whether calls
4155 are local in this sense.)
4156 * Compiler output is now more verbose, with messages truncated
4157 later than before. (There should be some supported way for users
4158 to override the default verbosity, but I haven't decided how to
4159 provide it yet, so this behavior is still controlled by the internal
4160 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4161 src/compiler/ir1util.lisp.)
4162 * Fasl file format version numbers have increased again, because
4163 support for the Gray streams extension changes the layout of the
4164 system's STREAM objects.
4165 * The Gray subclassable streams extension now works, thanks to a
4166 patch from Martin Atzmueller.
4167 * The full LOAD-FOREIGN extension (not just the primitive
4168 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4169 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4170 but like most other programs, it defaults to copying the Unix
4171 environment from the original process instead of starting the
4172 new process in an empty environment.
4173 * Extensions which manipulate the Unix environment now support
4174 an :ENVIRONMENT keyword option which doesn't smash case or
4175 do other bad things. The CMU-CL-style :ENV option is retained
4176 for porting convenience.
4177 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4178 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4180 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4182 * A patch from Martin Atzmueller seems to have solved the SIGINT
4183 problem, and as far as we know, signal-handling now works cleanly.
4184 (If you find any new bugs, please report them!)
4185 * The system no longer defaults Lisp source file names to types
4186 ".l", ".cl", or ".lsp", but only to ".lisp".
4187 * The compiler no longer uses special default file extensions for
4188 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4189 seems to expect a single default extension for all compiled code,
4190 and there's no compelling reason to try to stretch the standard
4191 to allow two different extensions.) Instead, byte-compiled files
4192 default to the same extension as native-compiled files.
4193 * Fasl file format version numbers have increased again, because
4194 a rearrangement of internal implementation packages made some
4195 dumped symbols in old fasl files unreadable in new cores.
4196 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4197 many fewer weird special cases.
4198 * Bug #17 (differing COMPILE-FILE behavior between logical and
4199 physical pathnames) has been fixed, and some related misbehavior too,
4200 thanks to a patch from Martin Atzmueller.
4201 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4202 by Tim Moore, ported to SBCL by Martin Atzmueller.
4203 * Martin Atzmueller fixed several filesystem-related problems,
4204 including bug #36, in part by porting CMU CL patches, which were
4205 written in part by Paul Werkowski.
4206 * More compiler warnings in src/runtime/ are gone, thanks to
4207 more patches from Martin Atzmueller.
4208 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4211 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4213 * DESCRIBE now works on CONDITION objects.
4214 * The debugger now handles errors which arise when trying to print
4215 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4217 * The build system now uses an additional file, customize-target-features.lisp,
4218 to allow local modifications to the target *FEATURES* list. (The point of
4219 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4220 debugging features enabled, without having to worry about propagating it
4221 into everyone's system when I do a "cvs update".) When no
4222 customize-target-features.lisp file exists, the target *FEATURES* list
4223 should be constructed the same way as before.
4224 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4225 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4226 warn about behavior which is undefined under the ANSI spec. Note: This
4227 is specified by ANSI, but it's not very popular with programmers.
4228 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4229 macro in the SBCL sources for an example of a workaround which you
4230 might use to make portable ANSI-standard code which does what you want.
4231 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4232 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4233 when it's used as a non-toplevel form. (This is required in order
4234 to implement the DEFCONSTANT-EQX macro.)
4235 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4236 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4237 funny things, due to bugs in the implementation of EVAL-WHEN
4238 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4239 non-toplevel DEFCONSTANTs are uncommon.)
4240 * The core file version number and fasl file version number have been
4241 incremented, because the old noncompliant DEFCONSTANT behavior involved
4242 calling functions which no longer exist, and because I also took the
4243 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4244 * fixed bug 1 (error handling before read-eval-print loop starts), and
4245 redid debugger restarts and related debugger commands somewhat while
4247 ** The QUIT debugger command is gone, since it did something
4248 rather different than the SB-EXT:QUIT command, and since it never
4249 worked properly outside the main toplevel read/eval/print loop.
4250 Invoking the new TOPLEVEL restart provides the same functionality.
4251 ** The GO debugger command is also gone, since you can just invoke
4252 the CONTINUE restart directly instead.
4253 ** The TOP debugger command is also gone, since it's redundant with the
4254 FRAME 0 command, and since it interfered with abbreviations for the
4256 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4258 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4259 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4260 some porting work by Martin Atzmueller)
4261 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4262 with a boatload of patches which he ported to SBCL. Now that those
4264 ** The system tries to make sure that its low-priority messages
4265 are prefixed by semicolons, to help people who like to use
4266 syntax highlighting in their ILISP buffer. (This patch
4267 was originally due to Raymond Toy.)
4268 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4269 patches originally written by Raymond Toy.
4270 ** The compiler understands coercion between single-value and
4271 multiple-VALUES type expressions better, getting rid of some very
4272 weird behavior, thanks to patches originally by Robert MacLachlan
4273 and Douglas Crosher.
4274 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4275 lambda lists, thanks to a patch originally by Pierre Mai.
4276 ** The system no longer bogusly warns about "abbreviated type
4278 ** The compiler gets less confused by inlining and RETURN-FROM,
4279 thanks to some patches originally by Tim Moore.
4280 ** The system no longer hangs when dumping circular lists to fasl
4281 files, thanks to a patch originally from Douglas Crosher.
4282 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4283 undefined function error.
4284 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4285 in structure slot initforms) for now, documented workaround instead:-|
4286 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4287 * fixed bug 5: added stubs for various Gray stream functions called
4288 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4289 installed, at least appropriate type errors are generated
4290 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4291 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4292 consistently in DEFMETHOD forms.
4293 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4294 it doesn't seem to affect SBCL after all
4295 * The C runtime system now builds with better optimization and many
4296 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4298 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4300 * The system is now under CVS at SourceForge (instead of the
4301 CVS repository on my home machine).
4302 * The new signal handling code has been tweaked to treat register
4303 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4304 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4305 downstream, e.g. in debug-int.lisp, has implicit dependencies
4306 on the unsignedness of integer representation of machine words,
4307 and that caused the system to bomb out with infinite regress
4308 when trying to recover from type errors involving signed values,
4309 e.g. (BUTLAST '(1 2 3) -1).)
4310 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4311 (This was one of the bugs Peter Van Eynde reported back in July.)
4312 * The system now uses code inspired by Colin Walters' O(N)
4313 implementation of MAP (from the cmucl-imp@cons.org mailing
4314 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4315 inline the MAP operation, and there is more than one
4316 sequence argument to the MAP call (so that it can't just
4317 do ETYPECASE once and for all based on the type of the
4318 single sequence argument). (The old non-inline implementation
4319 of the general M-argument sequence-of-length-N case required
4320 O(M*N*N) time when any of the sequence arguments were LISTs.)
4321 * The QUIT :UNIX-CODE keyword argument has been renamed to
4322 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4323 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4325 * Raymond Wiker's patches to port dynamic loading from Linux to
4326 FreeBSD have been added.
4327 * The BUGS file is now more nearly up to date, thanks in large part
4328 to Martin Atzmueller's review of it.
4329 * The debugger now flushes standard output streams before it begins
4330 its output ("debugger invoked" and so forth).
4331 * The core version number and fasl file version number have both
4332 been incremented, because of incompatible changes in the layout
4334 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4335 used to not be called for a saved Lisp image.)
4336 * Martin Atzmueller's version of a patch to fix a compiler crash,
4337 as posted on sbcl-devel 13 September 2000, has been installed.
4338 * Instead of installing Martin Atzmueller's patch for the
4339 compiler transform for SUBSEQ, I deleted the compiler transform,
4340 and transforms for some similar consing operations.
4341 * A bug in signal handling which kept TRACE from working on OpenBSD
4343 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4344 away by constant folding
4345 * The system now defines its address space constants in one place
4346 (in the Lisp sources), and propagates them automatically elsewhere
4347 (through GENESIS and the sbcl.h file). Therefore, patching the
4348 address map is less unnecessarily tedious and error-prone. The
4349 Lisp names of address space constants have also been systematized.
4350 * CVS tags like dollar-Header-dollar have been removed from
4351 the sources, because they have never saved me trouble and
4352 they've been source of trouble working with patches and other
4353 diff-related operations.
4354 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4355 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4357 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4359 * The system has been ported to OpenBSD.
4360 * The system now compiles with a simple "sh make.sh" on the systems
4361 that it's supported on. I.e., now you no longer need to tweak
4362 text in the target-features.lisp-expr and symlinks in src/runtime/
4363 by hand, the make.sh takes care of it for you.
4364 * The system is no longer so grossly inefficient when compiling code
4365 involving vectors implemented as general (not simple) vectors (VECTOR T),
4366 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4367 which dares to use the various sequence functions on non-simple
4368 vectors, takes less of a performance hit.
4369 * There is now a primitive type predicate VECTOR-T-P
4370 to test for the (VECTOR T) type, so that e.g.
4371 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4372 can now be compiled with some semblance of efficiency. (The old code
4373 turned the type declaration into a full call to %TYPEP at runtime!)
4374 * AREF on (VECTOR T) is still not fast, since it's still compiled
4375 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4376 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4377 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4378 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4379 without making full calls to SUBTYPEP at runtime.
4380 (Some analogous efficiency issues for non-simple vectors specialized to
4381 element types other than T, or for non-simple multidimensional arrays,
4382 have not been addressed. They could almost certainly be handled the
4383 same way if anyone is motivated to do so.)
4384 * The changes in array handling break binary compatibility, so
4385 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4386 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4387 instead of (VALUES T T).
4388 * By following the instructions that Dan Barlow posted to sbcl-devel
4389 on 2 July 2000, I was able to enable primitive dynamic object
4390 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4391 functionality is not implemented (since it calls ld to resolve
4392 library references automatically, requiring RUN-PROGRAM for its
4393 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4394 try to resolve library references) is now supported.
4395 * The system now flushes the standard output streams when it terminates,
4396 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4397 them at several other probably-convenient times, e.g. in each pass of
4398 the toplevel read-eval-print loop, and after evaluating a form given
4399 as an "--eval" command-line option. (These changes were motivated by a
4400 discussion of stream flushing issues on cmucl-imp in August 2000.)
4401 * The source transform for TYPEP of array types no longer assumes
4402 that an array whose element type is a not-yet-defined type
4403 is implemented as an array of T, but instead punts, so that the
4404 type will be interpreted at runtime.
4405 * There is now some support for cross-compiling in make.sh: each of
4406 the phases of make.sh has its own script. (This should be transparent
4407 to people doing ordinary, non-cross-compile builds.)
4408 * Since my laptop doesn't have hundreds of megabytes of memory like
4409 my desktop machine, I became more motivated to do some items on
4410 my to-do list in order to reduce the size of the system a little:
4411 ** Arrange for various needed-only-at-cold-init things to be
4412 uninterned after cold init. To support this, those things have
4413 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4414 symbols with such names are now uninterned after cold init).
4415 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4416 fasl files for cold load.
4417 ** Remove the old compiler structure pooling code (which used to
4418 be conditional on the target feature :SB-ALLOC) completely.
4419 ** Redo the representation of some data in cold init to be more compact.
4420 (I also looked into supporting byte compiled code at bootstrap time,
4421 which would probably reduce the size of the system a lot, but that
4422 looked too complicated, so I punted for now.)
4423 * The maximum signal nesting depth in the src/runtime/ support code has
4424 been reduced from 4096 to 256. (I don't know any reason for the very
4425 large old value. If the new smaller value turns out to break something,
4426 I'll probably just bump it back up.)
4427 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4429 * Many, many bugs reported by Peter Van Eynde have been added to
4430 the BUGS list; some have even been fixed.
4431 * While enabling dynamic object file loading, I tried to make the
4432 code easier to understand, renaming various functions and variables
4433 with less ambiguous names, and changing some function calling
4434 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4435 * While trying to figure out how to do the OpenBSD port, I tried to
4436 clean up some of the code in src/runtime/. In particular, I dropped
4437 support for non-POSIX signal handling, added various comments,
4438 tweaked the code to reduce the number of compilation warnings, and
4439 renamed some files to increase consistency.
4440 * To support the new automatic configuration functionality in make.sh,
4441 the source file target-features.lisp-expr has been replaced with the
4442 source file base-target-features.lisp-expr and the machine-generated
4443 file local-target-features.lisp-expr.
4444 * fixed a stupid quoting error in make.sh so that using CMU CL
4445 "lisp -batch" as cross-compilation host works again
4447 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4449 * DESCRIBE no longer tries to call itself recursively to describe
4450 bound/fbound values, so that it no longer fails on symbols which are
4451 bound to themselves (like keywords, T, and NIL).
4452 * DESCRIBE now works on generic functions.
4453 * The printer now prints less-screwed-up representations of closures
4454 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4455 * A private symbol is used instead of the :EMPTY keyword previously
4456 used to mark empty slots in hash tables. Thus
4457 (DEFVAR *HT* (MAKE-HASH-TABLE))
4458 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4459 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4460 now does what ANSI says that it should. (You can still get
4461 similar noncompliant behavior if bang on the hash table
4462 implementation with all the symbols you get back from
4463 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4464 This breaks binary compatibility, since tests for equality to
4465 :EMPTY are wired into things like the macroexpansion of
4466 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4468 * There's now a minimal placeholder implementation for CL:STEP,
4469 as required by ANSI.
4470 * An obscure bug in the interaction of the normal compiler, the byte
4471 compiler, inlining, and structure predicates has been patched
4472 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4473 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4475 * Missing ordinary arguments in a macro call are now detected even
4476 when the macro lambda list contains &KEY or &REST.
4477 * The debugger no longer complains about encountering the top of the
4478 stack when you type "FRAME 0" to explicitly instruct it to go to
4479 the top of the stack. And it now prints the frame you request even
4480 if it's the current frame (instead of saying "You are here.").
4481 * As specified by ANSI, the system now always prints keywords
4482 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4484 * The default initial SIZE of HASH-TABLEs is now smaller.
4485 * Type information from CLOS class dispatch is now propagated
4486 into DEFMETHOD bodies, so that e.g.
4487 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4489 is now basically equivalent to
4490 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4491 (DECLARE (TYPE SINGLE-FLOAT X))
4493 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4494 operation, without having to do run-time type dispatch.
4495 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4496 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4497 * Since I don't seem to be making big file reorganizations very often
4498 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4499 is overflowing my ability to conveniently back them up), I've finally
4500 checked the system into CVS. (The CVS repository is on my home system,
4501 not at SourceForge -- putting it on SourceForge might come later.)
4502 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4503 high-level GC-NOTIFY-FOO functions send their output. (There's
4504 still very little control of where low-level verbose GC functions
4505 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4506 less than it used to -- the GC-NOTIFY-FOO functions are now under
4507 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4508 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4509 in only one place in the source code, and propagates it automatically
4510 everywhere that it's needed. Thus e.g. when I bump the version from
4511 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4512 * The C source files now include boilerplate legalese and documentation
4513 at the head of each file (just as the Lisp source files already did).
4514 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4515 to his page will be replaced with a link to his new CLiki service.
4517 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4519 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4520 * The build process now looks for GNU make under the default name "gmake",
4521 instead of "make" as it used to. If GNU make is not available as "gmake"
4522 on your system, you can change this default behavior by setting the
4523 GNUMAKE environment variable.
4524 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4525 can build without error under CMU CL.
4527 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4529 * There is now a partial SBCL user manual (with some new text and some
4530 text cribbed from the CMU CL manual).
4531 * The beginnings of a profiler have been added (starting with the
4532 CMU CL profiler and simplifying and cleaning up). Eventually the
4533 main interface should be through the TRACE macro, but for now,
4534 it's still accessed through vaguely CMU-CL-style functions and macros
4535 exported from the package SB-PROFILE.
4536 * Some problems left over from porting CMU CL to the new
4537 cross-compilation bootstrap process have been cleaned up:
4538 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4539 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4540 part when getting rid of the compiler *BACKEND* stuff.)
4541 ** The value of *NULL-TYPE* was screwed up, because it was
4542 being initialized before the type system knew the final
4543 definition of the 'NULL type. This screwed up several key
4544 optimizations in the compiler, causing inefficiency in all sorts
4545 of places. (I found it because I wanted to understand why
4546 GET-INTERNAL-RUN-TIME was consing.)
4547 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4548 PROCLAIM FTYPE information. Unfortunately this broke binary
4549 compatibility again, since now the forms output by DEFGENERIC
4550 to refer to functions which didn't exist in 0.6.3.
4551 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4552 can use the new (as of 0.6.3) transform for SOME into MAP into
4554 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4555 values to (INTEGER 0 1000000), so that the time code will no longer
4556 occasionally get blown up by Linux returning 1000000 microseconds
4557 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4558 its output conform to the ANSI spec. (Alas, this makes its output
4559 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4560 * A full call to MAP NIL with a single sequence argument no longer conses.
4561 * fixes to problems pointed out by Martin Atzmueller:
4562 * The manual page no longer talks about multiprocessing as though
4563 it were currently supported.
4564 * The ILISP support patches have been removed from the distribution,
4565 because as of version 5.10.1, ILISP now supports SBCL without us
4566 having to maintain patches.
4567 * added a modified version of Raymond Toy's recent CMU CL patch for
4568 EQUALP comparison of HASH-TABLE
4570 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4572 * The system still can't cross-compile itself with
4573 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4574 entails), but at least it can compile more of itself that way
4575 than it used to be able to, and various buglets which were uncovered
4576 by trying to cross-compile itself that way have now been fixed.
4577 * This release breaks binary compatibility again. This time
4578 at least I've incremented the FASL file format version to 2, so that the
4579 problem can be detected reliably instead of just causing weird errors.
4580 * various new style warnings:
4581 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4582 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4583 ** using the lexical binding of a variable named in the *FOO* style
4584 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4585 as specified by ANSI.
4586 * *RANDOM-STATE* is no longer automatically initialized from
4587 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4588 default behavior of the system is to repeat its behavior every time
4589 it's run. If you'd like to change this behavior, you can always
4590 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4591 old convention there was no comparably easy way to get the system to
4592 repeat its behavior every time it was run.
4593 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4594 FTYPE declarations has been removed, in favor of their ANSI
4595 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4596 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4597 the types of their sequence arguments can be determined at compile time.
4598 This is done through a new open code expansion for MAP which eliminates
4599 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4600 argument types can be determined at compile time.
4601 * The optimizer now transforms COERCE into an identity operation when it
4602 can prove that the coerced object is already of the correct type. (This
4603 can be a win for machine generated code, including the output of other
4604 optimization transforms, such as the MAP transform above.)
4605 * Credit information has been moved from source file headers into CREDITS.
4606 * Source file headers have been made more standard.
4607 * The CASE macro now compiles without complaining even when it has
4610 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4612 * (Note that the way that the PCL macroexpansions were rewritten
4613 to accommodate the change in DEFGENERIC below breaks binary
4614 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4615 not run under sbcl-0.6.2. Once we get out of alpha releases,
4616 i.e. hit release 1.0.0, we'll probably try to maintain binary
4617 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4618 and sbcl-1.4.4. Until then, however, it might be fairly common
4619 for maintenance releases to break binary compatibility.)
4620 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4622 * The reporting of unhandled signals has been changed to print some
4623 explanatory text as well as the report form. (Previously only
4624 the report form was printed.)
4625 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4626 it defines, so that the compiler no longer issues undefined function
4627 warnings for compiled-but-not-yet-loaded generic functions.
4628 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4629 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4630 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4631 section of the ANSI spec are supported.
4632 * The "" nickname for the "KEYWORD" package has been removed.
4633 The reader still handles symbol tokens which begin with a package marker
4634 as keywords, but it doesn't expose its mechanism for doing so in the
4635 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4636 * The system now issues STYLE-WARNINGs for contradictory TYPE
4637 proclamations. (Warnings for contradictory FTYPE proclamations would
4638 be nice too, but those can't be done usefully unless the type system
4639 is made smarter about FUNCTION types.)
4640 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4641 systematized, so that "*target-*.lisp is supposed to exist only on the
4642 target and imply that there's a related file which exists on the
4643 host, and *host-*.lisp is supposed to exist only on the host and imply
4644 that there's a related file which exists on the target. This involves a
4645 lot of renaming. Hopefully the acute confusion caused by the renaming
4646 will be justified by the reduction in chronic confusion..
4647 ** runtime-type.lisp -> early-target-type.lisp
4648 ** target-type.lisp -> late-target-type.lisp
4649 ** early-host-format.lisp -> early-format.lisp
4650 ** late-host-format.lisp -> late-format.lisp
4651 ** host-error.lisp -> misc-error.lisp
4652 ** early-error.lisp -> early-target-error.lisp
4653 ** late-error.lisp -> late-target-error.lisp
4654 ** host-defboot.lisp -> early-defboot.lisp
4655 ** code/misc.lisp -> code/target-misc.lisp
4656 ** code/host-misc.lisp -> code/misc.lisp
4657 ** code/numbers.lisp -> code/target-numbers.lisp
4658 ** code/early-numbers.lisp -> numbers.lisp
4659 ** early-host-type.lisp -> early-type.lisp
4660 ** late-host-type.lisp -> late-type.lisp
4661 ** host-typep.lisp -> typep.lisp
4662 ** load.lisp -> target-load.lisp
4663 ** host-load.lisp -> load.lisp
4664 ** host-disassem.lisp -> disassem.lisp
4665 ** host-insts.lisp -> insts.lisp
4666 ** byte-comp.lisp -> target-byte-comp.lisp
4667 ** host-byte-comp.lisp -> byte-comp.lisp
4668 ** host-signal.lisp -> signal.lisp
4669 ** host-defstruct.lisp -> defstruct.lisp
4670 ** late-target-type.lisp -> deftypes-for-target.lisp
4671 Furthermore, several other previously target-only files foo.lisp (e.g.
4672 hash-table.lisp and random.lisp) have been split into a target-and-host
4673 foo.lisp file and a target-only target-foo.lisp file, with their key type
4674 definitions in the target-and-host part, so that the cross-compiler will
4675 know more about target types.
4676 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4677 gone away. In their place are various *BACKEND-FOO* variables
4678 corresponding to the slots of the old structure.
4679 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4680 into the target SBCL has been fixed.
4681 * The chill.lisp system for loading cold code into a running SBCL
4683 * Support for the CMU CL "scavenger hook" extension has been removed.
4684 (It was undocumented and unused in the CMU CL sources that SBCL was
4685 derived from, and stale in sbcl-0.6.1.)
4686 * Various errors in the cross-compiler type system were detected
4687 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4688 (enabling various consistency checks). Many of them were fixed,
4689 but some hard problems remain, so the compiler is back to
4690 running without *TYPE-SYSTEM-INITIALIZED* for now.
4691 * As part of the cross-compiler type system cleanup, I implemented
4692 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4693 * I have started adding UNCROSS calls throughout the type system
4694 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4695 on cross-compiler input files will be able to go away, and various
4697 * CONSTANTP now returns true for quoted forms (as explicitly required
4700 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4702 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4703 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4704 (MAKE-LIST -1). In the longer term, it ought to become true
4705 that declarations are assertions even at SAFETY 1. For now, it's not
4706 quite true even at SAFETY 3, but it's at least more nearly true..
4707 (Note that this change seems to increases the size of the system by
4708 O(5%) and to decrease the speed of the compiler by 20% or more.)
4709 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4710 for the problem of printing dozens of lines of distracting information
4711 about low-level system machinery as part of the top stack frame
4712 on entry to the debugger when an undefined function was called.
4713 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4714 is not reset to COMMON-LISP-USER.
4715 * Compilation of stuff related to dyncount.lisp has been made conditional
4716 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4717 smaller. The various dyncount-related symbols have been moved into
4718 a new "SB-DYNCOUNT" package.
4719 * tty-inspect.lisp has been renamed to inspect.lisp.
4720 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4721 feature has gone away. (When we eventually port to other flavors of
4722 libc and/or Unix, we'll try to make the differences between flavors
4723 invisible at the user level.)
4724 * Various other *FEATURES* tags, and/or their associated conditionals,
4725 have been removed if obsolescent, or given better documentation, or
4726 sometimes given more-mnemonic names.
4728 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4730 * tidied up "make.sh" script
4731 * tidied up system directory structure
4732 * better "clean.sh" behavior
4733 * added doc/FOR-CMUCL-DEVELOPERS
4734 * many many small tweaks to output format, e.g. removing possibly-confusing
4735 trailing #\. character in DESCRIBE-INSTANCE
4736 * (EQUALP #\A 'A) no longer signals an error.
4737 * new hashing code, including EQUALP hashing
4738 * tidied up Lisp initialization and toplevel
4739 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4740 * command line argument processing
4741 * added POSIX-GETENV function to deal with Unix-ish environment variables
4742 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4743 terminating SBCL on EOF
4744 * non-verbose GC by default
4745 * There is no more "sbcl" shell script; the sbcl file is now the C
4746 runtime executable (just like CMU CL).
4747 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4749 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4750 symbol and package information as Lisp native symbol and package objects
4751 instead of strings naming symbols and strings naming packages. This way,
4752 whenever packages are renamed (as in warm init), debug information is
4753 transformed along with everything else.
4754 * tweaked the optimization policy declarations which control the building
4755 of SBCL itself. Now, among other things, the system no longer saves
4756 source location debugging information. (This helps two problems at once
4757 by reducing SBCL size and by keeping SBCL from trying to look for its
4758 sources -- which may not exist -- when reporting errors.)
4759 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4760 debugging and testing purposes
4761 * cleaned up printing, making the printer call PRINT-OBJECT for
4762 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4763 methods, giving nearly-ANSI behavior
4764 * converted almost all special variables to use *FOO* naming convention
4765 * deleted PARSE-TIME functionality, since it can be done portably
4766 * moved some files out of cold init into warm init
4767 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4769 * regularized formatting of source files
4770 * added an install.sh script
4771 * fixed ridiculous memory usage of cross-compiler by making
4772 compiler/alloc.lisp not try to do pooling unless it can hook
4773 itself into the GC of the cross-compilation host. Now the system
4774 builds nicely on my old laptop.
4775 * added :SB-ALLOC in target-features.lisp-expr
4776 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4777 was not implemented)
4778 * re-did condition handling and note reporting in the compiler. Notes
4779 are no longer handled by signalling conditions. Style warnings
4780 and warnings are handled more correctly and reported in such a way
4781 that it's easy to find one or the other in your output (so that you
4782 can e.g. figure out which of many problems caused COMPILE-FILE to
4784 * changed the severity of several compiler warnings from full WARNING
4785 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4786 compiler note reporting so that it doesn't use the condition system
4787 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4788 COMPILE-FILE command)
4789 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4790 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4791 no longer supported, since their implementation was deeply intertwingled
4792 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4793 * removed bogus "support" for compiler macros named (SETF FOO), and
4794 removed the compiler macro for SETF INFO (but only after making a fool
4795 of myself on the cmucl-imp mailing list by posting a bogus patch for
4796 DEFINE-COMPILER-MACRO..)
4797 * Compiled files containing forms which have side effects on the Lisp
4798 reader (such as DEFPACKAGE forms) are now handled more correctly.
4799 (Compiler queuing of top level lambdas has been suppressed by setting
4800 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4801 * deleted various currently-unused source files, e.g. gengc.lisp. They
4802 may be added back at some point e.g. when porting to other architectures,
4803 but until they are it's distracting to distribute them and to try to
4805 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4806 there were so many of them they're just distractions, and UNCROSS is
4807 known to be able to handle the current sources
4808 * moved PROFILE functionality into TRACE, so that it will be clear
4809 how the wrapping and unwrapping of functions when you profile them
4810 interacts with the wrapping and unwrapping of functions when you
4811 trace them. (Actually, the functionality isn't there yet, but at least
4812 the interface specification is there. Hopefully, the functionality will
4813 arrive with some future maintenance release.)
4814 * removed host-oops.lisp
4815 * changed signature of QUIT function to allow UNIX-CODE argument
4816 * fixed READ-SEQUENCE bug
4817 * tweaked verbose GC output so that it looks more like the progress
4818 output that ANSI specifies for functions like LOAD
4819 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4820 * added <http://sbcl.sourceforge.com> to the banner information printed by