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 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
8 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
10 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
11 to more efficient EQ-comparison versions more often.
12 * optimization: enhanced derivation of DOLIST iteration variable type
14 * optimization: constant folding of simple (LIST ...) forms as DOLIST
16 * optimization: runtime lookup of function definitions can be
17 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
18 * optimization: compiler is able to derive the return type of
19 (AREF (THE STRING X) Y) as being CHARACTER.
20 * bug fix: fixed #427: unused local aliens no longer cause compiler
21 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
23 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
24 alien-stack leakage. (reported by Andy Hefner)
25 * bug fix: PROGV signals an error when an attempt to violate declared
26 type of a variable or bind a constant is made.
27 * bug fix: SET signals an error when an attempt to violate declared
28 type of a variable is made.
29 * bug fix: restart computation during the execution of a restart
30 test function no longer causes infinite recursion. (reported by
32 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
33 signals an error. (thanks to Pierre Mai)
34 * bug fix: EXPORT left symbol unexported in conflict situations.
35 (thanks to Michael Weber)
36 * bug fix: correctly handle name conflicts involving CL:NIL.
37 (thanks to Michael Weber)
38 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
39 conflicts arising from USEing package with conflicting symbols
40 correctly. (thanks to Michael Weber)
42 changes in sbcl-1.0.19 relative to 1.0.18:
43 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
44 warnings that go otherwise unhandled will be muffled if they are
45 of the type that's the value of this variable.
46 * optimization: stack allocation is slightly more efficient on x86
48 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
49 where a raw slot always is initialized using the initform whose
50 type is not know sufficiently well a compile-time are now compiled
51 correctly. (reported by John Morrison)
52 * bug fix: compiler no longer makes erronous assumptions in the
53 presense of non-foldable SATISFIES types.
54 * bug fix: stack analysis missed cleanups of dynamic-extent
55 arguments in non-let-converted calls to local functions.
56 * improvements to the Windows port:
57 ** adjusted address spaces for building on both Win32 and
58 Win64. (thanks for John Connors)
59 * fixed some bugs revealed by Paul Dietz' test suite:
60 ** interval arithmetic during type derivation used inexact integer
61 to single-float coercions.
62 ** arithmetic operations involving large integers and single
63 floats give the same results in compiled and interpreted code.
64 ** deriving the result type of COERCE no longer signals an error
65 if the derived type of the second argument is a MEMBER type
66 containing invalid type specifiers.
67 ** ADJOIN with constant NIL as second argument works correctly.
68 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
70 changes in sbcl-1.0.18 relative to 1.0.17:
71 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
72 profiles only the current thread.
73 * minor incompatible change: changes to SYMBOL-VALUE of constants
74 defined with DEFCONSTANT now signal an error.
75 * enhancement: SB-SPROF now has support for wallclock profiling,
76 and is also able to profile specific threads. REPORT output
77 has also additional sorting options.
78 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
80 * optimization: structure allocation has been improved
81 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
82 ** out of line constructors are ~10% faster.
83 ** inline constructors are ~15% faster.
84 ** inline constructors are capable of dynamic extent allocation
85 (generally on x86 and x86-64, in some cases on other platforms
87 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
89 * optimization: file compiler is now able to coalesce non-circular
90 lists, non-base strings, and bit-vectors. Additionally, constants
91 are never referenced through SYMBOL-VALUE at runtime.
92 * optimization: code defining methods on PRINT-OBJECT (and other
93 generic functions in the COMMON-LISP package) now loads faster.
94 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
96 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
97 fasl is now deleted. (reported by Attila Lendvai)
98 * bug fix: READ-LINE always returned NIL for the last line in files.
99 (reported by Yoshinori Tahara)
100 * bug fix: more accurate disassembly annotations of foreign function
101 calls. (thanks to Andy Hefner)
102 * bug fix: trimming non-simple strings and non-string string
103 designators when the there is nothing to trim works properly.
104 (thanks to James Knight)
105 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
106 (thanks to Travis Cross)
107 * fixed some bugs revealed by Paul Dietz' test suite:
108 ** NIL is a valid function name (regression at 1.0.13.38)
109 ** FILL on lists was missing its return value (regression at 1.0.12.27)
110 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
111 fill pointers properly (regression at 1.0.12.23)
112 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
113 displacement indices properly (regression at 1.0.12.23)
115 changes in sbcl-1.0.17 relative to 1.0.16:
116 * temporary regression: user code can no longer allocate closure
117 variable storage on stack, due to bug 419 without explicitly
118 requesting it. Please consult sbcl-devel for advice if you need to
119 use this feature in the meanwhile.
120 * new feature: runtime argument --control-stack-size can be used to
121 adjust thread default control stack size.
122 * enhancement: improved TIME output
123 ** all times are reported using the measured accuracy (milliseconds
124 for real and GC times, microseconds for everything else.)
125 ** processor cycle counts on x86 and x86-64.
126 ** interpreted forms are counted for both evaluator modes.
127 ** number of lambdas converted by the compiler is reported.
128 ** CPU percentage report (computed from real and total run time.)
129 ** more comprehensive run time reporting, using a condenced format
130 ** interperted form, lambda, and page fault counts are omitted
132 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
134 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
135 * optimization: two argument forms of LAST are upto ~10% faster
136 in normal SPEED policies.
137 * optimization: NCONC no longer needs to heap cons its &REST list
138 in normal SPEED policies.
139 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
141 * bug fix: UNION and NUNION work with :TEST-NOT once more,
142 regression since 1.0.9.1. (thanks to Eric Marsden)
143 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
144 since 1.0.15.36. (thanks to Paul Khuong)
145 * bug fix: LAST when always returned the whole list when given a bignum
146 as the second argument.
147 * bug fix: dynamic extent allocation of nested lists and vectors
148 could leak to otherwise accessible parts.
149 * bug fix: invalid optimization of heap-allocated alien variable
151 * bug fix: fasl header checking is less vulnerable to different
152 platform word lengths.
153 * bug fix: more correct assembler syntax for GNU binutils
154 2.18.50.0.4 support. (thanks to Marijn Schouten)
155 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
157 * bug fix: the fopcompiler can handle LOCALLY forms (with no
158 declarations) successfully. (reported by Attila Lendvai)
160 changes in sbcl-1.0.16 relative to 1.0.15:
161 * minor incompatible change: revert the changes to sb-posix's error
162 signaling added in 1.0.14.
163 * minor incompatible change: change PROBE-FILE back to returning
164 NIL whenever we can't get a truename, as was the case before 1.0.14.
165 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
166 inherits from ERROR instead of just CONDITION.
167 * new feature: SB-INTROSPECT can provide source locations for instances
168 as well. (thanks to Tobian Ritterweiler)
169 * optimization: binding special variables now generates smaller code
170 on threaded platforms.
171 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
172 and cases where no :TEST is given but the compiler can infer that
173 the element to search is of type (OR FIXNUM (NOT NUMBER)).
174 * optimization: better LOGNOT on fixnums.
175 * optimization: modular arithmetic for a particular requested width
176 is implemented using a tagged representation unless a better
177 representation is available.
178 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
179 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
180 for wellformedness and returns a specialized vector. (reported by
181 Francois-Rene Rideau)
182 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
183 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
184 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
185 unscheduling timer at the same time as another timer fires.
186 * bug fix: don't reschedule timers for dead threads.
187 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
188 * bug fix: copying output from RUN-PROGRAM to a stream signalled
189 bogus errors if select() was interrupted.
190 * enhancement: add support for fcntl's struct flock to SB-POSIX.
192 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
193 * enhancement: cleaner backtraces for interactive interrupts, as
194 well as other cases where the interesting frames used to be
195 obscured by interrupt handling frames.
196 * enhancement: untracing a whole package using (UNTRACE "FOO") is
197 now supported, and tracing a whole package using (TRACE "FOO") now
198 traces SETF-functions as well.
199 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
200 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
201 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
202 * unreadably printed representation of hash-tables now includes
204 * bug fix: partially fixed #188: type propagation from assignments
205 is now more efficient.
206 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
207 'DOUBLE-FLOAT) are not flushable.
208 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
209 full address of the object, and none of the tag bits.
210 * bug fix: readably printing hash-tables now respects other printer
211 control variables. (reported by Cedric St-Jean)
212 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
213 keyword with MAKE-HASH-TABLE.
214 * bug fix: export SB-POSIX:MKSTEMP.
215 * bug fix: SORT was not interrupt safe.
216 * bug fix: XREF accounts for the last node of each basic-block as
218 * bug fix: MAKE-INSTANCE optimizations interacted badly with
219 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
220 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
221 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
222 create obsolete instances in the case of redefinition or
223 obsoletion of a superclass. (thanks to Andy Hefner)
224 * bug fix: Support for the Alpha architecture has been revived; it had
225 suffered somewhat from lack of maintenance since sbcl-1.0.
226 * improvements to the Windows port:
227 ** The system detects the codepage to use at startup, instead of
228 using the value from the saved core. (thanks to Kei Suzuki)
230 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
231 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
232 (see documentation for details.)
233 * revived support for OpenBSD (contributed by Josh Elsasser)
234 * partially fixed bug #108: ROOM no longer suffers from occasional
235 (AVER (SAP= CURRENT END)) failures.
236 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
237 bodies no longer cause a WARNING to be signalled. (reported by
239 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
240 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
241 the bug in TRUNCATE.) (reported by Michael Weber)
242 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
243 no samples. (reported by Andy Hefner)
244 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
245 no longer appear as (NIL ...) frames in backtraces.
246 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
247 is now actually exported from SB-EXT as documented. (reported by
249 * bug fix: sb-aclrepl now correctly understands how to inspect
250 single-floats on 64-bit platforms where single-floats are not boxed.
251 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
252 yet been finalized. (reported by Levente Meszaros)
253 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
254 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
255 SBCL-specific optimize qualities.
257 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
258 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
259 an executable in the search path, and does so in the child
260 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
261 been removed; it can be found in the CVS history, for somebody who
262 needs that search behavior (see the manual).
263 * minor incompatible change: compiler policy re. weakening type
264 checks has changed: now type checks are weakened only if SAFETY < 2
266 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
267 unparsing of directory pathnames as files. Analogously,
268 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
269 filename to parse into a directory pathname.
270 * enhancement: implicit generic function creation now signals a
271 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
272 users can bind handlers for and muffle around calls to LOAD.
273 * enhancement: RUN-PROGRAM allows unicode arguments and environments
274 to be used (using the default stream external format), and allows
275 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
276 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
277 strings and vectors whose element-type or simplicity is not fully
278 known at compile-time.
279 * optimization: STRING-TRIM and related functions no longer allocate
280 a new string when no trimming needs to be performed. These functions
281 are also faster than before when the input string has been declared
283 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
284 * optimization: READ-LINE is significantly faster for files containing
286 * optimization: non-open coded uses of character comparison operators
287 (e.g. char=) no longer cons when called with more than one parameter
288 on platforms supporting dynamic-extent allocation.
289 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
290 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
291 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
293 * bug fix: some sequence functions elided bounds checking when
295 * bug fix: too liberal weakening of union-type checks when SPEED >
297 * bug fix: more bogus fixnum declarations in ROOM implementation
300 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
301 * new feature: MAKE-HASH-TABLE now experimentally accepts a
302 :SYNCHRONIZED argument, which makes the hash-table safe for
303 concurrent accesses (but not iteration.) See also:
304 SB-EXT:WITH-LOCKED-HASH-TABLE, and
305 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
306 * optimization: CONCATENATE on strings is an order of magnitude faster
307 in code compiled with (> SPEED SPACE).
308 * optimization: SUBSEQ is ~50% faster on lists.
309 * bug fix: bug 417 fixed -- source location reporting is now more robust.
310 * bug fix: SUBSEQ on a list will now correctly signal an error if if
311 END is smaller then START.
312 * bug fix: SB-PROFILE will no longer report extra consing for nested
313 calls to profiled functions.
314 * bug fix: ROOM implementation had bogus fixnum declarations which
315 could cause type-errors when calling ROOM on large images.
316 * bug fix: if file compilation is aborted, the partial fasl is now
317 deleted, and COMPILE-FILE returns NIL as the primary value.
318 * bug fix: number of thread safety issues relating to SBCL's internal
319 hash-table usage have been fixed.
320 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
321 be returned from its body when the values were being returned
322 using unknown-values return convection and the W-P-O was wrapped
323 inside an UNWIND-PROTECT.
324 * bug fix: sb-posix should now compile again under Windows, enabling
327 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
328 * incompatible change: hash-table accessor functions are no longer
329 automatically protected by locks. Concurrent accesses on the same hash-table
330 from multiple threads can give inconsistent results or even corrupt the
331 hash-table completely. Multi-threaded applications should do their own
332 locking at the correct granularity. In the current implementation it is
333 still safe to have multiple readers access the same table, but it's not
334 guaranteed that this property will be maintained in future releases.
335 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
336 and will signal an error at runtime.
337 * enhancement: SB-THREAD package now exports a semaphore interface.
338 * enhancement: CONS can now stack-allocate on x86 and
339 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
341 * enhancement: nested lists can now be stack allocated on
342 platforms providing stack allocation support.
343 * enhancement: dynamic-extent support has been extended to support
344 cases where there are multiple possible sources for the stack
346 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
347 if the mutex is uncontested on Linux.
348 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
349 as specified by the CLHS. (thanks to Tobias Rittweiler)
350 * bug fix: NaN comparison now works on x86-64.
351 * bug fix: CLOSE :ABORT T on a stream with pending output now
353 * bug fix: instances of non-standard metaclasses using standard
354 instance structure protocol sometimes missed the slot type checks
356 * bug fix: known functions can be inlined in high-debug code.
358 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
359 * minor incompatible change: the MSI installer on Windows no longer
360 associates .lisp and .fasl files with the installed SBCL.
361 * minor incompatible change: :UNIX is no longer present in *FEATURES*
362 on Windows. (thanks to Luis Oliviera)
363 * new feature: SB-CLTL2 contrib module now implements
364 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
365 * optimization: scavenging weak pointers is now more efficient,
366 requiring O(1) instead of O(N) per weak pointer to identify
367 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
368 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
369 method bodies using constant slot-names when the first argument is
370 a specializer parameter for the method.
371 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
372 names, when the class of the instance is a direct instance
373 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
374 as slow as the constant slot-name case.)
375 * optimization: member type construction is now O(N) instead
377 * optimization: UNION and NUNION are now O(N+M) for large
378 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
380 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
381 now more readable in environments like Slime which display it.
382 (thanks to Tobias C. Rittweiler)
383 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
384 was able to infer the correct argument type for the object on which
385 the CAS operation was being performed.
386 * bug fix: copy propagation interfered with parallel assignment
387 semantics in local calls. (reported by Paul Khuong)
388 * bug fix: the signed modular fixnum shift compiled to wrong code on
389 x86 and x86-64. (spotted by a slight modification to some of
392 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
393 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
394 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
395 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
397 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
398 and generic function now print the object identity for anonymous
399 instances. (thanks to Attila Lendvai)
400 * enhancement: as an extension to MOP, SBCL now supports SETF of
401 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
402 (thanks to Attila Lendvai)
403 * enhancement: sb-cover has an alternate annotation mode which puts
404 the form annotations at the car of the form, instead of annotating
405 it completely. The benefit of this mode is that it shows better
406 which forms were instrumented by the compiler.
407 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
408 by printer control variables. (reported by Dan Corkill)
409 * bug fix: system leaked memory when delayed output was performed by
410 the OS in smaller chunks then expected. (thanks to David Smith)
411 * bug fix: system leaked memory when file streams were not closed
413 * bug fix: large objects written to slow streams that were modified
414 after the write could end up with the modified state written to
415 the underlying file descriptor.
416 * bug fix: multiple threads operating in parallel on the same stream
417 could cause buffer-overflows.
418 * bug fix: source location information is stored correctly
419 (broken since 1.0.6). This bug would generally show up as the
420 Slime debugger highlighting the wrong form when the sldb-show-source
422 * bug fix: Branch forms are again annotated as branches in the sb-cover
424 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
425 on x86-64. (reported by Gregory Vanuxem)
426 * bug fix: EQUALP could return wrong results for structures with raw
427 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
428 word). (reported by Vjacheslav Fyodorov)
429 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
432 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
433 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
434 atomic compare-and-swap operations on threaded platforms.
435 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
436 allows assigning a global minimum value to optimization qualities
437 (overriding proclamations and declarations).
438 * enhancement: closed over variables can be stack-allocated on x86
440 * performance bug fix: GETHASH and (SETF GETHASH) are once again
442 * optimization: slot definition lookup is now O(1). This speeds up
443 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
444 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
445 * optimization: ASSOC and MEMBER can now be open-coded for all
446 combinations of keyword arguments when second argument is constant
447 and SPEED >= SPACE. In other cases a specialized version is
449 * bug fix: using obsoleted structure instances with TYPEP and
450 generic functions now signals a sensible error.
451 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
452 (reported by Kristoffer Kvello)
453 * bug fix: backtrace construction is now more careful when making
454 lisp-objects from pointers on the stack, to avoid creating bogus
455 objects that can be seen by the GC.
456 * bug fix: defaulting of values in contexts expecting more than 7
457 variables now works on x86-64. (reported by Christopher Laux)
458 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
460 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
461 as the property-list of a symbol.
462 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
463 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
464 situations requested, are once again file-compileable. (reported
467 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
468 * MOP improvement: support for user-defined subclasses of
469 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
470 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
471 the expansion of DEFMETHOD, is responsible for generating a form
472 which creates a list of specializers when evaluated. Additional
473 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
474 debugging and introspective support.
475 * minor incompatible change: the (unsupported) spinlock interface
476 has changed: free spinlock now has the value NIL, and a held spinlock
477 has the owning thread as its value.
478 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
479 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
481 * enhancement: name of a socket-stream is now "a socket" instead of
483 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
484 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
485 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
486 for generic functions. (Thanks to Tobias C. Rittweiler)
487 * optimization: bignum printing speed has been improved by 20-40%
488 (depending on the bignum size.)
489 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
491 * bug fix: the cache used by the CLOS to store precomputed effective
492 methods, slot offsets, and constant return values is now thread and
494 * bug fix: generic function dispatch function updating is now thread
495 and interrupt safe (in the sense that the known issues have been
497 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
498 * bug fix: interning EQL-specializers is now thread and interrupt safe.
499 * bug fix: asdf systems with dependencies to the SB-POSIX or
500 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
501 * bug fix: interrupt safety of applicable method computation has been
504 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
505 * new contrib: sb-cover, an experimental code coverage tool, is included
507 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
508 significantly faster.
509 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
510 produce more efficient code on x86-64 if the compiler can determine
511 that (+ INDEX <constant>) does not require a bounds check and FOO
512 has an element type at least 8 bits wide.
513 * enhancement: a new, experimental synchronous timeout facility is
514 provided. Refer to SB-SYS:WITH-DEADLINE for details.
515 * enhancement: when a symbol name conflict error arises, the
516 conflicting symbols are always printed with a package prefix.
517 (thanks to Kevin Reid)
518 * enhancement: stepping is now once again supported on the SPARC and
519 MIPS platforms. (It is also now more likely to work on CheneyGC
521 * enhancement: sb-sprof can now also track and report accurate call
523 * bug fixes: the treatment of non-standard subclasses of
524 SB-MOP:SPECIALIZER is more correct.
525 * incompatible change: PURIFY no longer copies the data from the
526 dynamic space into the static and read-only spaces on platforms
527 that use the generational garbage collector
528 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
530 * bug fix: GC race condition occasionally resulting in crashes with
531 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
533 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
535 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
536 a STYLE-WARNING to be signalled (regression from 1.0.4.)
537 * bug fix: an asynchronous interrupt could previously leave the
538 system running with GC inhibited.
539 * bug fix: a DECLARE form evaluated at top-level now causes an error
540 rather than silently (or verbosely) returning NIL.
541 * bug fix: trying to dissassemble functions compiled at high DEBUG could
542 result in a error being signalled due to source form lookup errors.
543 (reported by Peter Graves)
545 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
546 * incompatible change: removed writer methods for host-ent-name,
547 host-ent-addresses -- changing the values did not update the DNS
549 * minor incompatible change: changed experimental JOIN-THREAD interface
550 * documentation: the manual now lists reader and writer methods
551 in class slot documentation sections. (thanks to Richard M Kreuter)
552 * documentation: unwinding from asyncronous events has been
553 documented as unsafe.
554 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
555 in multithreaded application code.
556 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
557 platforms. (thanks to James Anderson for the optimization hint)
558 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
560 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
561 variants no longer cons.
562 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
563 their NOT- variants no longer cons.
564 * optimization: EQUAL hash tables no longer use SXHASH for objects
565 of all data types, but instead use an EQL hash for types for which
566 EQUAL is the same as EQL.
567 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
568 are significantly faster.
569 * optimization: new STRING-OUTPUT-STREAM implementation is much
570 faster for large outputs and conses 30% less on average.
571 * enhancement: XREF information is now collected for references made
572 to global variables using SYMBOL-VALUE and a constant argument.
573 * enhancement: SIGINT now causes a specific condition
574 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
575 * bug fix: FILE-POSITION used to signal an error in some cases where
576 ANSI requires it to return NIL.
577 * bug fix: ADJUST-ARRAY is now interrupt-safe.
578 * bug fix: adding and removing fd-handlers is now interrupt-safe.
579 * bug fix: inlined calls to C now ensure 16byte stack alignment on
581 * bug fix: bad type declaration in the CLOS implementation has
582 been fixed. (reported by James Anderson)
583 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
584 (thanks to Lutz Euler)
585 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
586 required by the CLHS. (reported by Samium Gromoff)
587 * bug fix: dead unbound variable references now signal an error.
588 (reported by Marco Monteiro)
589 * bug fix: / with an unused value was being deleted in safe code.
590 (thanks to Marco Monteiro and Kevin Reid)
591 * bug fix: number of characters that can be written onto a single
592 line in a file is unlimited.
593 * bug fix: some GC deadlocks caused by asynchronous interrupts have
594 been fixed by inhibiting interrupts for when GC is disabled.
595 * bug fix: some interrupt safety issues with GETHASH, (SETF
596 GETHASH), CLRHASH and REMHASH have been fixed.
597 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
598 specifier no longer causes infinite recursion.
599 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
600 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
601 * bug fix: an &environment argument with the correct variable information
602 is passed to macros that are expanded during byte compilation.
603 (reported by Samium Gromoff)
604 * bug fix: a logic error was causing the hash function for bignums to
605 have excessive amounts of collisions. (reported by Faré Rideau)
606 * bug fix: modifying the contents of an array could change the return
607 value of SXHASH on that array, which is only allowed for strings
608 and bit vectors. (bug introduced in 0.9.16)
609 * improvement: the x86-64/darwin port now passes all tests (except
610 for the debugger tests) but should still be considered
611 experimental until this is fixed.
612 * improvement: a style-warning is signaled for CASE (etc) clauses with
613 duplicate keys. (thanks to Kevin Reid)
614 * improvement: macos/x86 and macos/x86-64 now use mach exception
615 handlers for dealing with illegal instructions (for trapping and
616 error handling) and memory protection violations (for GC).
618 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
619 * new platform: experimental support for x86-64/darwin (MacOS).
620 * incompatible change: the thread-safe (on most platforms) getaddrinfo
621 and getnameinfo sockets functions are used instead of gethostbyaddr
622 and gethostbyname, on platforms where the newer functions are available.
623 As a result, the ALIASES field of HOST-ENT will always be NIL on these
625 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
626 don't print the startup banner, but behave as if --noinform was passed
627 as a command line argument. (thanks to Kevin Reid)
628 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
629 * optimization: code using alien values with undeclared types is much faster.
630 * optimization: the compiler is now able to open code SEARCH in more cases.
631 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
632 * bug fix: using standardized COMMON-LISP special variables as loop
633 variables no longer signals bogus package lock violations. (reported
635 * bug fix: declaring local loop variables to be of a range-limited type
636 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
637 (reported by Andras Simon)
638 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
639 bugs remain on x86-64.)
640 * bug fix: the #= and ## reader macros now interact reasonably with
641 funcallable instances.
642 * bug fix: type-checks for function arguments were compiled using the
643 compiler policy settings of the wrong lexical environment (in 1.0.2
645 * bug fix: SHADOW accepts characters as string designators, as required
646 by the spec (thanks to Eric Marsden)
647 * bug fix: fixed GC safety issues when foreign functions are called with
648 non-base strings as arguments
649 * bug fix: more consistent error output for fatal-compiler-warnings, like
651 * bug fix: functions with proper names were showing up as NIL in
652 backtraces (reported by Edi Weitz, regression in 1.0.2)
654 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
655 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
656 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
657 produce more efficient code on the x86 if the compiler can determine
658 that (+ INDEX <constant>) does not require a bounds check and FOO
659 has an element type at least 8 bits wide.
660 * bug fix: references to undefined variables are handled the same way
661 in toplevel forms as in the normal compiler
662 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
664 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
665 (thanks to Jon Buller)
666 * improvement: faster compilation times for complex functions
667 * improvement: added readlink support to SB-POSIX (thanks to Richard
670 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
671 * improvement: experimental support for mach exception handling on
672 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
674 * improvement: support for GBK external format.
675 (thanks to Chun Tian (binghe))
676 * improvement: the debugger now displays variables that have been closed
677 over, in code compiled with (DEBUG 2) or higher
678 * improvement: support for executable cores on NetBSD (thanks to
680 * new feature: added a RESTART-FRAME debugger command
681 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
682 be used to provide an implementation for FILE-POSITION on Gray streams
683 (thanks to Eric Marsden)
684 * new feature: add syslog, openlog and closelog support to SB-POSIX
685 (thanks to Richard Kreuter)
686 * optimization: the function call overhead in code compiled with
687 a high DEBUG optimization setting is significantly
688 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
689 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
690 (thanks to Tony Martinez)
691 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
692 evaluated once (reported by Kevin Reid)
693 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
694 works again (thanks to Kilian Sprotte)
695 * bug fix: an error is signaled for tagbodies with duplicate tags
696 (thanks to Stephen Wilson)
697 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
698 * bug fix: Win32 port can now handle foreign code unwinding Lisp
699 stack frames from alien callbacks.
700 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
701 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
702 65k pages (thanks to David Woodhouse)
703 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
705 changes in sbcl-1.0.1 relative to sbcl-1.0:
706 * new platform: FreeBSD/x86-64, including support for threading.
707 * new feature: the compiler stores cross-referencing information
708 about function calls (who-calls), macroexpansion (who-macroexpands)
709 and special variables (who-binds, who-sets, who-references) for code
710 compiled with (< SPACE 3). This information is available through the
711 sb-introspect contrib.
712 * new feature: users may subclass SEQUENCE, and have instances of
713 these classes interoperate with standard Common Lisp functions if
714 a number of methods are defined. (This feature is experimental
715 and the interface subject to change based on feedback from SBCL
716 users and the general community)
717 * improvement: sb-sprof traces call stacks to an arbitrary depth on
718 x86/x86-64, rather than the previous fixed depth of 8
719 * improvement: another pthread back-end of mutex "pthread-futex".
720 * bug fix: non-ascii command-line arguments are processed correctly
721 (thanks to Yaroslav Kavenchuk)
722 * bug fix: non-required arguments were not passed correctly when a method
723 defined using DEFMETHOD was called from a mop-generated method using
724 CALL-NEXT-METHOD (reported by Pascal Costanza)
725 * bug fix: recursion is now permitted in accessors through
726 SLOT-UNBOUND. (reported by Pascal Costanza)
727 * bug fix: an error was signaled at startup if the HOME environment
728 variable was defined, but had an empty value (reported by Peter Van Eynde)
729 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
730 signal a PROGRAM-ERROR, not a BUG.
731 * bug fix: SB-POSIX:READDIR works when built with large file support.
732 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
734 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
735 for better automatic indentation support. (thanks to Matt Pillsbury)
736 * optimization: loading generic functions no longer takes O(n^2) time,
737 proportional to the amount of methods in the generic function
738 (reported by Todd Sabin and Jeremy Brown)
739 * optimization: the FIND and POSITION family of sequence functions
740 are significantly faster on arrays whose element types have been
742 * improvements to the Windows port:
743 ** Intermittent heap corruption problems have been fixed. (thanks
744 to Alastair Bridgewater)
745 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
747 ** Lisp is able to unwind foreign exception frames from alien
748 callbacks. (thanks to Alastair Bridgewater)
750 changes in sbcl-1.0 relative to sbcl-0.9.18:
751 * improvement: experimental support for threading on FreeBSD/x86.
752 (thanks to NIIMI Satoshi)
753 * improvement: runtime option --dynamic-space-size can be used
754 to set the size of the dynamic space reserved on startup.
755 * improvement: floating point modes in effect are now saved in
756 core, and restored on startup.
757 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
758 startup, not time since first call to GET-INTERNAL-REAL-TIME.
759 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
760 threads are not running after *SAVE-HOOKS* have run.
761 * improvement: writes to CLOS instance slots are type-checked in code
762 compiled with (SAFETY 3)
763 * improvement: floating-point exception handling on FreeBSD (thanks to
765 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
766 (thanks to Zach Beane)
767 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
769 * improvement: added support for the Shift-JIS external format.
770 (contributed by NIIMI Satoshi)
771 * improvement: callbacks are supported on Linux/PPC. (thanks to
773 * bug fix: compiler bug triggered by a (non-standard) VALUES
774 declaration in a LET* was fixed. (reported by Kaersten Poeck)
775 * bug fix: file compiler no longer confuses validated and already
776 dumped structurres. (reported by Kaersten Poeck)
777 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
778 fill-pointer signals a type-error as required. (thanks to
780 * bug fix: disassemly of funcallable instances works.
781 * bug fix: single stepping on PPC.
782 * bug fix: fix thread-safety problems in the type system (generally
783 manifesting as nonsensical errors like "STRING is a bad type specifier
784 for sequences" or "The value 1 is not of type FIXNUM")
785 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
786 (reported by Josip Gracin)
787 * bug fix: an error is signaled for attempts to displace arrays with
788 incompatible element types (thanks to Mario Mommer)
789 * bug fix: more correct handling of wide characters in debug info
790 (bug reported by Attila Lendvai and fixed by Juho Snellman)
791 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
792 and don't cause extra consing
793 * optimization: MAP and MAP-INTO are significantly faster on vectors
794 whose elements types have been declared.
795 * Improvements to SB-SPROF:
796 ** Support for allocation profiling
797 ** Reduced profiling overhead, especially for long profiling runs
798 * Improvements to the Windows port:
799 ** floating point exceptions are now reported correctly.
800 ** stack exhaustion detection works partially.
801 ** more accurate GET-INTERNAL-REAL-TIME.
802 ** better RUN-PROGRAM behaviour in case of errors in the child
804 ** PROBE-FILE now simplifies pathnames correctly.
805 ** DIRECTORY now works correctly with :WILD-INFERIORS.
806 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
807 1st 1970 (although time timezone data will be unreliable).
808 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
810 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
811 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
812 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
814 * bug fix: two potential GC deadlocks affecting threaded builds.
815 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
816 returns T (reported by Anton Kazennikov)
817 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
818 with non-variable places
819 * bug fix: remove a race condition in the setting of
820 funcallable-instance functions, this should make threaded CLOS
821 code more stable against memory faults.
822 * bug fix: corruption of specials when unbinding is interrupted by an
823 asynchronous unwind (reported by Hannu Koivisto)
824 * improvement: the debugger will now also display local variables that
825 are only used once, for code compiled with a DEBUG optimization quality
828 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
829 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
830 * incompatible change: External-format support for FFI calls. The
831 SB-ALIEN:C-STRING no longer implies an ASCII
832 external-format. Instead the string is subject to external-format
833 conversion. Additionally return values of type SB-ALIEN:C-STRING
834 are by default of type (SIMPLE-ARRAY CHARACTER), not
835 SIMPLE-BASE-STRING. For an alien type that behaves like the old
836 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
837 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
838 * incompatible change: SB-EXT package no longer contains the
839 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
840 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
842 * incompatible change: the single-stepper is no longer available
843 on Alpha, Mips and Sparc platforms.
844 * minor incompatible change: the direct superclasses of
845 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
846 not (STANDARD-OBJECT FUNCTION). This makes the
847 class-precedence-lists of GENERIC-FUNCTION and
848 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
850 * minor incompatible change: the default stream external format on
851 non-unicode SBCL is no longer detected from the locale, but is always
853 * new feature: Add a version of evaluator that uses an interpreter instead
854 of the compiler. EVAL still uses the compiler by default, to switch it
855 to use the interpreter, set the value of the variable
856 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
857 * minor incompatible change: the single-stepper REPL has been merged
858 with the normal debugger (see the "Stepping" heading of the debugger help
859 for more details). The debugger command STEP will no longer switch
860 to the single-stepper REPL.
861 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
862 for a type now works.
863 * bug fix: loading of default sysinit file works. (thanks to Leonid
865 * bug fix: better detection of circularities in the file-compiler.
866 (reported by Marco Monteiro)
867 * bug fix: the CL pathname functions now work with files that have
868 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
869 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
870 systems with Unix98 pty semantics.
871 * bug fix: ASDF-INSTALL will now work with bsd tar.
872 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
874 * bug fix: timers expiring in dead threads no longer cause a
875 type-error (reported by Paul "Nonny Mouse").
876 * bug fix: thanks to more lightweight single-stepper instrumentation,
877 code compiled with (DEBUG 3) will compile and execute significantly faster,
878 and will have more accurate type-inferencing than before
879 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
880 whose bindings are modified
881 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
882 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
883 ** replace dummy implementations of CL:MACHINE-INSTANCE and
884 CL:SOFTWARE-VERSION with proper version
886 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
887 * feature: implemented the READER-METHOD-CLASS and
888 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
889 as specified by AMOP.
890 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
892 * optimization: faster LOGCOUNT implementation on x86 and x86-64
893 (thanks to Lutz Euler)
894 * optimization: hashing of general arrays and vectors has been
895 improved. (reported by Any Fingerhut)
896 * enhancement: SB-INTROSPECT is now able to find definitions of
897 profiled functions. (thanks to Troels Henriksen)
898 * enhancement: compiler-macro expansion applies now to FUNCALL forms
899 as well, allowing compiler-macros for SETF-functions to expand.
900 * enhancement: step-instrumentation no longer wraps known
901 single-value functions with multiple-value context, allowing
902 better type inference.
903 * fixed bug #337: use of MAKE-METHOD in method combination now works
904 even in the presence of user-defined method classes. (reported by
905 Bruno Haible and Pascal Costanza)
906 * fixed bug #339(c): if there are applicable methods not part of any
907 long-form method-combination group, call INVALID-METHOD-ERROR.
908 (reported by Bruno Haible)
909 * fixed bug #361: the :FUNCTION initarg in the protocol for
910 initialization of methods can now be used to override
911 internally-produced optimized functions. (reported by Bruno
913 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
914 system-provided lambda expression no longer cause warnings about
915 unbound #:|pv-table| symbols.
916 * bug fix: improved the handling of type declarations and the
917 detection of violations for keyword arguments with non-constant
919 * bug fix: potentially erronous calls to PATHNAME and
920 MERGE-PATHNAMES were being flushed in some cases.
921 (reported by Richard Kreuter)
922 * bug fix: compiled calls to TYPEP were mishandling obsolete
923 instances. (reported by James Bielman and Attila Lendvai)
924 * bug fix: format strings with ~> without matching ~< no longer
925 trigger an AVER, but signal an understandable error instead.
926 (reported by Antonio Martinez)
927 * bug fix: specifying an output-file without a pathname-type for
928 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
929 of input-file instead of "fasl". (reported by Robert Dodier)
930 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
931 (reported by James Y Knight).
932 * bug fix: compiler-macros-function did not consider the environment
933 argument for shadowing by local functions.
934 * bug fix: compiler-macros expansion was inhibited by local INLINE
936 * bug fix: inline expansions of known functions were subject to
937 step-instrumentation in high debug policies, leading to problems
939 * bug fix: compiler failed to differentiate between different CONS
941 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
942 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
943 * bug fix: #368: incorrect use of expressed vs. upgraded array
945 * bug fix: #306a: more precise unions of array types.
946 * thread-safety improvements:
947 ** CONDITION-WAIT could return early on Linux, if the thread was
948 interrupted and subsequently continued with SIGCONT.
949 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
951 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
952 * added support for the ucs-2 external format. (contributed by Ivan
954 * minor incompatible change: pretty printing of objects of type
955 (cons symbol) is, in the default pprint-dispatch-table, now
956 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
958 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
959 called later in a class's lifetime, possibly as late as when the
960 first instance of the class is created. Previously,
961 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
962 class became finalizeable.
963 * fixed bug: FILE-POSITION sometimes returned inconsistent results
964 for multibyte external-format streams. (thanks to "vbzoli")
965 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
966 with :ALLOCATION :CLASS inherited from superclasses of the
968 * fixed bug: anonymous classes can now be created using the :NAME
969 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
970 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
971 * fixed bug: core-files saved with :EXECUTABLE T can again be
972 executed when SBCL_HOME isn't set. (reported by James Knight)
973 * fixed bug: toplevel LOCALLY forms with declarations could
974 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
975 * fixed bug: printing from several different threads using different
976 values of *print-case* could cause invalid output, due to
977 some internal special variables of the printer not being bound
978 thread-locally (reported by Max Mikhanosha)
979 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
980 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
981 (reported by Pascal Costanza)
982 * fixed bug: FINALIZE-INHERITANCE is called from
983 REINITIALIZE-INSTANCE on classes when the class has previously
984 been finalized, as required by AMOP.
985 * minor code generation optimizations:
986 ** better register allocation in CLOS dispatching functions
987 ** overflow detection when coercing signed bytes to fixnums on x86-64
988 ** is now implemented with one IMUL instruction instead of three shifts
989 ** more efficient bit-vector access on x86 and x86-64
990 ** more efficient access to raw structure slots on x86 and x86-64
991 * fixed some bugs revealed by Paul Dietz' test suite:
992 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
995 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
996 * feature: thread support on Solaris/x86, and experimental thread support
998 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1000 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1001 default method for SLOT-UNBOUND.
1002 * minor incompatible change: prevent the user from specializing the
1003 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1004 somewhat counter to the intent of the protocol, I (CSR) think, and
1005 additionally it just doesn't work in SBCL as currently
1006 implemented, thanks to optimizations (that are always valid for
1007 the other three SLOT-VALUEish functions, but not for the setter).
1008 * minor incompatible change: the :SB-LDB feature is now enabled by
1009 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1010 the low-level debugger.
1011 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1012 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1013 down to an SB-SYS:FD-STREAM, allowing direct communication in
1014 more cases. Temporary files and pipes are still used as a fallback
1016 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1017 on the non-reentrant C functions localtime(3) and gmtime(3).
1018 * bug fix: global optimization policy was not visible in LOCALLY and
1020 * bug fix: class objects can be used as specializers in methods.
1021 (reported by Pascal Costanza)
1022 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1023 with a type error. (reported by blitz_ on #lisp)
1024 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1025 failed with a type error. (reported by Pascal Bourguignon)
1026 * bug fix: merging pathnames against defaults with :DIRECTORY
1027 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1028 (reported by James Y Knight)
1029 * bug fix: saving large (>2GB) cores on x86-64 now works
1030 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1031 constant index and a (simple-array (signed-byte 32)) array
1032 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1033 error (patch by Robert J. Macomber)
1034 * feature: TIME now displays the amount of run-time spent in GC
1035 * bug fix: The debugger now does a better job of respecting
1036 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1037 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1038 workaround for bug 403.)
1039 * bug fix: floating point exception handling now works on FreeBSD
1040 (thanks to NIIMI Satoshi)
1041 * fixed some bugs revealed by Paul Dietz' test suite:
1042 ** MISC.641: LET-conversion were not supposed to work in late
1044 * improvements to the Win32/x86 port:
1045 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1046 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1048 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1049 * new feature: source path information is generated for macro-expansion
1050 errors for use in IDE's like Slime (thanks to Helmut Eller)
1051 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1052 * bug fix: compiler does not loop forever on an invalid type in
1054 * improvement: compilation of most CLOS applications is significantly
1056 * optimization: added a limited bytecode compiler for simple toplevel
1057 forms, speeding up compilation and FASL loading
1058 * bug fix: the statistical profiler now properly distinguishes anonymous
1061 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1062 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1063 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1064 into subclasses of the finalized class. Now user calls to
1065 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1066 system return before any subclasses are finalized.
1067 * minor incompatible change: The reader no longer ignores errors
1068 regarding non-existent packages in #+ and #- feature tests.
1069 * new feature: command line options --no-sysinit, --no-userinit to
1070 inhibit loading the corresponding init files
1071 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1072 for sending data through UDP sockets (thanks to François-René Rideau)
1073 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1074 error is signalled instead (thanks to François-René Rideau)
1075 * minor incompatible change: Error signalling behaviour of lexical
1076 operations violating package locks has changed slightly. Refer to
1077 documentation on package locks for details.
1078 * enhancement: EVAL can process IF-expressions without resorting to the
1080 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1081 constant-folded properly due to missing out-of-line functions.
1082 (reported by Mika Pihlajamäki)
1083 * bug fix: LISTEN sometimes returned T even in cases where no data was
1084 immediately available from the stream
1085 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1086 were reversed. (reported by Levente Mészáros)
1087 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1088 when given a vector displaced to an adjustable array. (reported
1090 * bug fix: derivation of float boundaries from numbers outside the
1091 appropriate float range (reported by John Wiseman)
1092 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1094 * bug fix: precomputing cache entries for generic functions with
1095 some subclasses of specializers as yet invalid does not attempt to
1096 fill a cache line with a negative offset. (reported by Levente
1098 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1099 allow condition class objects as arguments to DOCUMENTATION and
1100 (SETF DOCUMENTATION); only find and set documentation for
1101 structure names for the STRUCTURE doc-type. (suggested by Gary
1103 * improvements to the Win32/x86 port:
1104 ** (user-homedir-pathname) and default initialization file
1105 locations now know about the user's "Documents and Settings"
1106 directory (thanks to Yaroslav Kavenchuk)
1107 ** run-program is implemented (thanks to Mike Thomas)
1108 ** sockets support (thanks to Timothy Ritchey)
1109 ** better backtrace support (thanks to Alastair Bridgewater)
1110 ** sb-grovel supported
1111 ** asdf-install and sb-posix work somewhat
1112 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1113 * improvements to the Solaris/x86 port:
1114 ** works on Solaris 11/Solaris Express
1115 ** floating-point exception handling support
1116 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1117 * fixed some bugs revealed by Paul Dietz' test suite:
1118 ** REMOVE-METHOD returns its generic function argument even when
1119 the method is not one of the generic functions' methods.
1120 ** objects declared as MEMBER types can be admissible arguments to
1121 structure accessors.
1122 ** printing characters should simply be printed by the FORMAT ~:C
1124 ** compiler failure when compiling functions with hairy constant
1125 defaults for optional parameters.
1126 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1127 function, which is already optimized.
1129 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1130 * new platform: experimental support for SBCL x86/Darwin, including
1131 MacOS X 10.4.5 on Intel.
1132 * new feature: Unicode character names are now known to the system
1133 (through CHAR-NAME and NAME-CHAR).
1134 * new feature: the filesystem location of SBCL's core file is
1135 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1136 * minor incompatible change: the contrib modules SB-POSIX and
1137 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1138 this change is to make it easier to distribute
1139 location-independent binaries.
1140 * bug fix: as implied by AMOP, standardized classes no longer have
1141 slots named by external symbols of public packages. (reported by
1143 * optimization: calling functions via constant symbols -- as in
1144 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1145 via the function object as in (FUNCALL #'FOO).
1146 * optimization: on x86 and x86-64, the calling convention for the
1147 case of dealing with an unknown number of values has been altered
1148 to be friendlier to the prediction heuristics implemented,
1149 particularly with reference to CALL and RET pairing. (thanks to
1150 Alastair Bridgewater)
1151 * optimization: on x86, the code for access to thread-local symbol
1152 values and binding/unbinding of thread-local symbols is smaller.
1153 (thanks to Alastair Bridgewater)
1154 * enhancement: CONSTANTP is now able to determine constantness of
1155 more complex forms, including calls to constant-foldable standardized
1156 functions and some special forms beyond QUOTE.
1157 * fixed bug: occasional GC crashes on Solaris/x86
1158 * optimization: x86-64 supports stack allocation of results of simple
1159 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1160 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1161 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1162 (thanks to James Knight)
1163 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1164 by the standard (thanks to Stephen Compall)
1166 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1167 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1168 be used for bundling the runtime and the core file into one
1169 executable binary. This feature is not currently supported on all SBCL
1170 platforms. (thanks to James Bielman and NIIMI Satoshi)
1171 * new feature: a generational or ephemeral garbage collector is now
1172 the default on the PowerPC platform (both Linux and Darwin). The
1173 old Cheney (stop and copy) collector is a build-time option.
1174 (thanks to Cyrus Harmon, after Raymond Toy)
1175 * minor incompatible change: the method by which SBCL finds its
1176 contributed modules has changed; it no longer relies on symbolic
1177 links from an $SBCL_HOME/systems directory, but searches directly
1178 in the subdirectories of $SBCL_HOME.
1179 * enhancement: the dynamic heap size on the Linux/PPC platform is
1180 markedly larger, even using the older Cheney garbage collector.
1181 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1182 certain complicated string types. (reported by Gary King)
1183 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1184 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1185 * fixed bug: DOCUMENTATION on structures no longer signals an
1186 error if no structure type of the right name exists. (reported by
1188 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1189 applicable non-standard primary or :AROUND method on
1190 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1191 no longer cause unbound variable errors. (reported by Kevin Reid)
1192 * fixed bug: metacircle resolution in cases where methods have slots
1193 added before the slots from STANDARD-METHOD. (reported by Jean
1195 * fixed bug: the Power PC assembly code for calling into lisp
1196 sometimes computed its return address wrongly (depending on the
1197 15th bit of the address, fixed at link time). This used to
1198 manifest itself in a segmentation violation while building PCL.
1199 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1200 many others over the years)
1201 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1202 the stack pointer (treated as a signed integer) is negative.
1203 (thanks to Peter van Eynde)
1205 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1206 * new platform: experimental support for the Windows operating
1207 system has been added. (thanks to Alastair Bridgewater)
1208 * fixed several bugs in and robustified the PPC FFI (including
1209 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1210 * bug fix: allow non-simple string symbol names (reported by Paul
1212 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1214 * bug fix: saving a core corrupted callbacks on x86/x86-64
1215 * bug fix: closed a loophole in metacircularity detection and
1216 grounding in the PCL implementation of CLOS.
1217 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1218 (reported by tomppa on #lisp)
1219 * bug fix: clear the direction flag on Lisp -> C transitions, as
1220 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1221 (reported by Andrej Grozin and Hendrik Maryns)
1222 * optimization: major improvements to GC efficiency on GENCGC platforms
1223 * optimization: faster implementation of EQUAL
1224 * optimization: emit more efficient opcodes for some common
1225 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1226 * optimization: several other minor code-generation improvements on x86-64
1227 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1229 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1230 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1231 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1232 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1233 by ANSI as a generic function, and for consistency (SETF
1234 GENERIC-FUNCTION-NAME) is treated likewise.
1235 * fixed bug #233.b: make constraint propagation notice when a variable
1236 value is changed after it is referenced but before it is used
1237 * fixed bug #296: no more arbitrary behaviour from filenames with
1238 odd characters as --load arguments. (reported by Adam Warner)
1239 * fixed bug #390: :CHARACTER-SET pathname components now work as
1240 expected. (reported by Tim Daly Jr)
1241 * fixed bug #391: complicated :TYPE intersections in slot
1242 definitions no longer cause an error in PCL internals.
1243 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1244 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1245 returning the number of octets which would be written to the
1246 file-stream. (thanks to Robert J. Macomber)
1247 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1248 arguments to the internal stream functions. (thanks to David
1250 * bug fix: stack exhaustion is now handled on the x86/SunOS
1251 platform. (thanks to Daisuke Homma)
1252 * bug fix: more accurate ROOM results on GENCGC platforms
1253 * optimization: improved type inference for arithmetic-for
1254 index variables in LOOP
1255 * optimization: faster floating-point SQRT on x86-64
1256 * fixed some bugs revealed by Paul Dietz' test suite:
1257 ** DOCUMENTATION returns NIL instead of "" for method combinations
1258 that don't have a docstring
1260 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1261 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1262 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1263 therefore are not customizeable by user code (as seems to be at
1264 least permitted and maybe required by AMOP). As a consolation,
1265 however, the SBCL implementation of these functions now calls
1266 REINITIALIZE-INSTANCE as specified by AMOP.
1267 * bug fix: slot-definition documentation is propagated as per ANSI
1268 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1269 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1270 argument, as do eslotd initialization methods. (from Pascal
1271 Costanza's "Closer" project)
1272 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1273 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1275 * bug fix: it is now possible to have more than one subclass of
1276 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1277 (reported by Bruno Haible, Pascal Costanza and others)
1278 * bug fix: the dependent update protocol now works for generic
1279 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1280 and Pascal Costanza)
1281 * bug fix: condition-class instances corresponding to
1282 DEFINE-CONDITION forms are now created eagerly. (reported by
1283 Kalle Olavi Niemitalo on comp.lang.lisp)
1284 * bug fix: floating point printing is more accurate in some
1285 circumstances. (thanks to Simon Alexander)
1286 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1287 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1288 * bug fix: callbacks on OS X now preserve stack-alignment.
1289 (thanks to Cyrus Harmon)
1290 * enhancement: the x86-64 disassembler is much better at
1291 disassembling SSE instructions. (thanks to Lutz Euler)
1292 * enhancement: CHECK-TYPE on a variable now allows type-inference
1293 (thanks to Vincent Arkesteijn)
1294 * optimization: improved performance of EUC-JP external format.
1295 (thanks to NIIMI Satoshi)
1296 * optimization: performance improvements to IO on file streams of
1297 :ELEMENT-TYPE CHARACTER
1298 * optimization: much faster memory allocation on x86-64
1299 * optimization: faster garbage collections (latency and throughput)
1301 * optimization: faster bignum subtraction and division on x86 and x86-64
1302 (thanks to Lutz Euler)
1303 * optimization: more accurate type-inference for loops that use a
1304 floating point index variable or a negative step.
1306 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1307 * bug fix: add a workaround to SBCL looping infinitely at startup on
1308 Linux kernels with apparently buggy implementations of personality().
1309 (thanks to Svein Ove Aas)
1310 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1311 (thanks to David Lichteblau)
1312 * bug fix: local bindings shadow global symbol macros for
1313 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1314 * optimization: non-open coded uses of numeric comparison operators
1315 (e.g. >) no longer cons when called with more than one parameter
1316 on platforms supporting dynamic-extent allocation.
1317 * enhancement: saving cores with foreign code loaded is now supported
1318 on MIPS/Linux in addition to the previously supported platforms.
1319 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1320 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1321 * bug fix: division by zero in sb-sprof when no samples were collected
1322 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1323 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1324 (thanks to Alexander Kjeldaas)
1325 * new feature: ignoring the return values of destructive functions
1326 like NREVERSE signals a compile-time style-warning.
1327 (thanks to Kevin Reid)
1328 * new feature: the alignment of alien structure fields can be
1329 explicitly specified. (thanks to Cyrus Harmon)
1330 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1331 and dump core on SIGQUIT
1333 ** incompatible change: threads do not inherit values of specials
1334 from their parents (see manual)
1335 ** bug fix: threads stacks belonging to dead threads are freed by the
1336 next exiting thread, no need to gc to collect thread stacks anymore
1337 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1338 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1339 ** bug fix: dynamic variable and thread start related gc lossage
1340 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1342 * fixed some bugs revealed by Paul Dietz' test suite:
1343 ** SUBTYPEP is slightly more accurate on heinously complicated
1344 CONS types where some of the members have uncertain (in the
1345 NIL, NIL sense) type relationships to each other.
1347 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1348 no longer purified unless :PURIFY T is explicitly specified.
1349 ** Non-purified cores are significantly smaller than before
1351 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1352 * new feature: timers based on Zach Beane's excellent timer package
1353 * added support for the following external formats: koi8-u,
1354 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1355 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1356 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1357 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1358 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1359 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1360 * incompatible change: a threaded SBCL will no longer revert to
1361 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1362 * bug fix: interrupts are disabled until startup is complete; no
1363 more sigsegvs when receiving a signal to soon
1364 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1366 * bug fix: add a workaround for the memory randomization features in
1367 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1368 workaround will only be in effect on systems with the proc filesystem
1370 * bug fix: printing objects of type HASH-TABLE signals a
1371 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1373 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1374 have been running for over 50 days. (reported by Gilbert Baumann)
1375 * bug fix: the logic for getting names of functions gets less
1376 confused when confronded with alternate-metaclass
1377 funcallable-instances. (reported by Cyrus Harmon)
1378 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1379 consistent, even on internal alternate-metaclass objects.
1380 * bug fix: flush closure information collected by physical
1381 environment analysis prepass before the main pass. (bug reported
1383 * bug fix: compiler pack phase does not modify a hashtable, which is
1384 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1385 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1386 may be expanded. (reported by Alexander <malishev> on c.l.l)
1387 * bug fix: no more style-warnings when compiling code using LAST
1388 under high-SPEED low-SPACE settings. (reported by David Wallin
1390 * fixed some incompatibilities between SBCL's MOP and the MOP
1392 ** the METAOBJECT class is now implemented;
1393 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1394 STANDARD-OBJECT, as required;
1395 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1396 now compatible (as required by VALIDATE-SUPERCLASS); there
1397 remains a consistency requirement over the presence or absence
1398 of the FUNCTION class in the superclass list of finalized
1399 classes; see the manual for more details;
1400 ** the :AROUND method for COMPUTE-SLOTS on
1401 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1402 requested slot ordering.
1404 ** bug fix: parent thread now can be gc'ed even with a live
1406 ** bug fix: binding a special with PROGV to no value is not
1407 the same as that symbol not having been bound (thanks to
1409 ** bug fix: binding specials is thread safe (thanks to
1411 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1412 and SB-SPROF do not die with 'no handler for signal XX in
1413 interrupt_handle_now(..)' anymore
1414 ** bug fix: WITH-TIMEOUT works with multiple running threads
1415 * fixed some bugs revealed by Paul Dietz' test suite:
1416 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1417 the :method-class keyword argument.
1419 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1420 * new port: the Solaris operating system on x86 processors is now
1421 mostly supported, though some rough edges in the environment
1422 remain. (thanks to Daisuke Homma)
1423 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1424 space for its heap. (thanks to Thiemo Seufer)
1425 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1426 closures. (thanks to Thiemo Seufer)
1427 * minor incompatible change: eof selects abort in the debugger.
1428 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1429 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1431 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1432 deprecated, and will go away in a future revision of SBCL.
1433 * minor incompatible change: GC-ON and GC-OFF are no longer
1434 implemented with a counter, it does not matter how many times gc
1435 is switched on or off
1436 * bug fix: discriminating functions for generic function classes
1437 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1438 make invalid assumptions about method precedence order. (reported
1440 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1441 platforms. (thanks to Thiemo Seufer)
1442 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1443 * bug fix: correct pathname printing with printer escaping is on.
1444 (thanks to Kevin Reid)
1445 * bug fix: complex VOP definitions in "user-space" no longer trigger
1446 package locks. (reported by Zach Beane)
1447 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1448 for funcallable-instance functions, and is no different from
1450 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1451 characters in the current readtable. (reported by Nicholas Neuss)
1452 * bug fix: SUBTYPEP on various CONS types returns more a more
1453 accurate acknowledgment of its certainty.
1454 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1455 lists in some cases. This partially fixes bug 384.
1456 * flush all standard streams before prompting in the REPL and the
1458 * bug fix: signal handling and triggering gc do not conflict
1459 directly anymore, in particular a high frequency sb-sprof does
1460 not prevent gc from running
1461 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1462 approximation for timezone and DST information between the
1463 universal time epoch and the smallest negative 32-bit time_t.
1464 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1465 year 1899 so long as the encoded time is non-negative.
1466 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1467 the correct answer for non-constant rotate amounts when performing
1468 an inline 32-bit rotation.
1470 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1471 there is only one thread in the session
1472 ** bug fix: memory leak for streams created in one thread and
1473 written to in another
1474 ** bug fix: lockup when compiled with gcc4
1475 ** bug fix: race that allows the gc to be triggered when gc is
1477 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1478 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1480 ** bug fix: finished threads are gc'ed properly
1481 * fixed some bugs revealed by Paul Dietz' test suite:
1482 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1483 has arguments with a different set of applicable methods from
1484 the orignal arguments.
1485 ** The type error thrown by MAP now has a correct expected-type
1487 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1488 name a compiled function.
1489 ** Three MISC tests where a large bignum was improperly coerced to
1490 a float that couldn't represent that bignum during type
1491 derivation were fixed.
1492 ** SUBTYPEP can now handle types involving the negation of a
1493 list-form FUNCTION type.
1494 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1495 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1496 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1498 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1499 * New feature: Experimental support for bivalent streams: streams
1500 opened with :element-type :default now allow character and binary
1501 (unsigned-byte 8) I/O
1502 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1503 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1505 * bug fix: no more highly sporadic "couldn't check whether ~S is
1506 readable" when reading a stream and an interrupt hits in the middle
1507 of a select system call
1508 * compiler better recognizes complex arrays (reported by Eduardo
1510 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1511 for variables on SBCL built with linkage-tables. (reported by Luis
1513 * various error reporting improvements.
1514 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1515 (Thanks to Thiemo Seufer)
1516 * bug fix: nested reader invokations work correctly
1517 * bug fix: it is possible to have simultaneous references to foreign
1518 code and foreign data with the same name.
1520 ** added x86-64 support
1521 ** incompatible change: the threading api now works with thread
1522 objects instead of thread ids
1523 ** bug fix: threads are protected from signals and interruption when
1524 starting up or going down
1525 ** bug fix: a race where an exiting thread could lose its stack to gc
1526 ** bug fix: don't halt on infinite error in threads if possible
1527 ** fixed numerous gc deadlocks introduced in the pthread merge
1528 ** bug fix: fixed thread safety issues in read and print
1529 ** bug fix: debugger doesn't hang on session lock if interrupted at
1530 an inappropriate moment
1531 ** bug fix: run-program is now thread safe(r)
1532 ** bug fix: inner with-recursive-lock no longer releases the mutex
1533 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1534 * fixed some bugs revealed by Paul Dietz' test suite:
1535 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1536 EXPECTED-TYPE slots filled.
1537 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1539 ** MAKE-CONDITION accepts classes as type-designators.
1540 ** COMPILE may never return NIL.
1541 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1542 range before calling Unix time functions
1544 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1545 * numerous signal handling fixes to increase stability
1546 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1547 * minor incompatible change: we now correctly canonize default
1548 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1549 as per the MOP, rather than the historical (INITARG INITFUNCTION
1550 INITFORM). (reported by Bruno Haible)
1551 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1552 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1553 for more information.
1554 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1555 pathname is a directory pathname.
1556 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1557 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1559 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1560 :METACLASS argument in addition to a class name. (reported by
1561 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1562 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1563 (reported by Bruno Haible; patch from Gerd Moellmann)
1564 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1566 * bug fix: more cleanups to the floating point exception handling on
1567 x86-64 (thanks to James Knight)
1568 * bug fix: the compiler does not try to constant fold calls of
1569 COERCE and BIT-* functions when they return freshly constructed
1570 objects. (reported by Nikodemus Siivola and Paul Dietz)
1571 * optimization: improved the MIPS versions of generic (in the
1572 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1573 * optimization: direct conversion of (unsigned-byte 32) to floats on
1574 the PowerPC platform.
1575 * optimization: structure instances with raw slots now use less
1576 memory, and probably show better memory locality. (thanks to
1578 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1579 are treated as requests for stack allocation on the x86-64,
1580 Alpha-32, PPC and SPARC platforms.
1581 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1582 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1583 (thanks to Tony Martinez)
1585 ** gcing a dead thread can no longer lead to lockups
1586 ** threads block signals until they are set up properly
1587 ** errno is no longer shared by threads
1588 ** interrupt-thread restores the eflags register on x86
1589 ** fixed some lockups due to gc/thread interaction
1590 * fixed some bugs revealed by Paul Dietz' test suite:
1591 ** invalid dotted lists no longer raise a read error when
1592 *READ-SUPPRESS* is T
1593 ** signal an error if a symbol that names a declaration is used
1594 as the name of a type, or vice versa
1595 ** allow using the (declare (typespec var*)) abbreviation for
1596 (declare (type typespec var*)) with all type specifiers
1597 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1598 ** always return NIL from PPRINT-POP when OBJECT is NIL
1599 ** don't signal errors when pretty-printing malformed LABELS,
1600 FLET or MACROLET forms
1601 ** declarations in a DOLIST body are also in scope for the
1603 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1605 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1608 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1609 * fixed cross-compiler leakages that prevented building a 32-bit
1610 target with a 64-bit host compiler.
1611 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1612 opened with :IF-EXISTS :OVERWRITE.
1613 * fixed bug 281: error for an invalid qualifier in a short-form method
1614 combination method is not signalled until the faulty method is called.
1615 * bug fix: iteration variable type inferrer failed to deal with open
1616 intervals. (reported by Alan Shields)
1617 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1618 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1619 or more runtime options were provided to the sbcl binary.
1620 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1621 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1622 * optimization: there's now a fast-path for fixnum arguments in the
1623 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1624 * optimization: the code generated on x86-64 is more compact thanks
1625 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1626 * bug fix: floating-point exception handling now partly works on
1627 x86-64. (Thanks to James Knight)
1628 * improvement to the MIPS backend from Thiemo Seufer: C-style
1629 64-bit long long arguments and return values to alien functions
1631 * fixed some bugs revealed by Paul Dietz' test suite:
1632 ** the type-error signalled from WARN has a filled-in DATUM slot.
1633 ** the type-error required when a stream is not associated with
1634 a file has the stream as its datum.
1635 ** type-errors on single-floats on x86-64 no longer have
1636 :INVALID-OBJECT as the datum
1637 ** the type-errors signalled for invalid function names now have
1638 a correct expected type
1639 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1640 for structures defined with a :TYPE. Documentation strings for
1641 typed structures are no longer immediately discarded
1642 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1643 broadcast streams and synonym streams. FILE-LENGTH now also works
1644 on broadcast streams.
1646 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1647 * incompatible change: the --noprogrammer option, deprecated since
1648 version 0.7.5, has been removed. Please use the equivalent
1649 --disable-debugger option instead.
1650 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1652 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1653 inoperational for a while now) has been completely removed.
1654 * null lexical environments are now printed as #<NULL-LEXENV>,
1655 significantly reducing the amount of clutter in typical backtraces.
1656 * documentation on weak pointers, finalization, and after GC hooks
1657 has been added to the manual.
1658 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1659 as other specialized array types, is much faster. SUBSEQ and
1660 COPY-SEQ on such arrays have also been sped up.
1661 * optimization: EQL is now more efficient when at least other argument
1662 is known to be of type (OR FIXNUM (NOT NUMBER)).
1663 * fixed bug: the runtime is now less vulnerable to changes in the
1664 size of the SBCL object on OS X, and virtual memory is reserved for
1665 all spaces that need to be at a fixed address.
1666 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1667 * fixed bug: finalizers and after GC hooks that cause consing are now
1669 * fixed bug: compiler error messages and summaries are now printed to
1670 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1671 * fixed inference of the upper bound of an iteration variable.
1672 (reported by Rajat Datta).
1673 * fixed bug 211e: calling local functions with duplicated constant
1674 keyword argument no longer causes a bogus style warning about an
1676 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1677 ftype declaration to be disregarded. (reported by Dave Roberts)
1678 * fixed bug 373: caused by erronous compilation of references to alien
1679 variables in the runtime on ppc/darwin.
1680 * fixed bug 376: CONJUGATE type deriver.
1681 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1682 (reported by Baughn on #lisp)
1683 * a cross-compiler bug on non-x86 platforms has been identified and
1684 fixed. (thanks to Bruno Haible)
1685 * improvements to the MIPS runtime code for increased stability.
1686 (thanks to Thiemo Seufer)
1687 * increased the maximimum compact environment size to allow
1688 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1689 * partial workaround for bug 135: don't name the function we're
1690 calling for hairy cases of EVAL, so as not to accumulate one environment
1691 entry per call to EVAL. (reported by Kevin Reid)
1692 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1693 * optimization: type testing for non-vector arrays should be faster.
1694 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1696 * bug fix: setting 31st element of a bit vector to zero did not work
1698 * bug fix: redefining a class definition which failed due to a
1699 previous accessor / function clash now works (but see BUGS entry
1700 #380 for more problems in this area). (thanks to Zach Beane)
1701 * the long form of DEFINE-METHOD-COMBINATION disables method group
1702 checking when given a single method group with pattern *. (thanks
1704 * on x86 compiler supports stack allocation of results of simple
1705 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1706 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1707 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1708 * fixed some bugs related to Unicode integration:
1709 ** the restarts for recovering from input and output encoding
1710 errors only appear when there is in fact such an error to
1712 * fixed some bugs revealed by Paul Dietz' test suite:
1713 ** MISC.549 and similar: late transformation of unsafe type
1714 assertions into derived types caused unexpected code
1716 ** SCALE-FLOAT type deriver is less wrong.
1717 ** type derivers for EXP, LOG and similar functions compute result
1718 types for complex arguments better.
1719 ** (MISC.563) CONJUGATE type deriver works for very restricted
1721 ** out-of-line type testers for character strings are available.
1722 ** EQUAL compiler transform understands specialness of objects
1724 ** accessing double-floats stored on the stack now works on x86-64.
1725 ** debugger internals could sometimes create invalid lispobjs,
1726 resulting in GC crashes.
1727 ** MISC.548: type check weakening can convert required type into
1729 ** initialization forms for bindings are not in scope of free special
1732 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1733 * incompatible change: thread support for non-NPTL systems has
1734 been removed - locking is buggy and unreliable. A threaded
1735 SBCL build will now warn at startup and refuse to create
1736 new threads, unless futex support is detected in the kernel
1737 * incompatible change: the top level REPL now has only an ABORT
1738 restart associated with it, not TOPLEVEL and ABORT as it used to.
1739 TOP and TOPLEVEL are now available as debugger commands for
1740 returning to the top level.
1741 * incompatible change: forms evaluated in the REPL now use the
1742 global optimization policy.
1743 * incompatible change: user- and system-initialization files are
1744 no longer processed with LOAD, but by READ and EVAL; hence the
1745 global optimization policy, startup package, readtable, etc,
1747 * The .fasl file format number has been incremented because of
1748 various incompatible changes.
1749 * internal entry point details and argument counts no longer appear
1750 in backtraces unless explicitly requested by setting
1751 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1752 * built-in and standard functions no longer have names like "top
1753 level local call to FOO".
1754 * fixed bug 32: functions defined in non-null lexical environments
1755 now have more legible printed representation
1756 * fixed bug 33: functions defined in non-null lexical environemnts
1757 are now more amenable to inspection by INSPECT.
1758 * workaround for bug 354: XEPs no longer appear in backtraces unless
1759 explicitly requested.
1760 * fixed bug: receiving the signal which results from attempting to
1761 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1762 SIGBUS on other *BSDs) on architectures where the C stack is also
1763 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1764 to debugger confusion or wild instability. (reported by Cheuksan
1766 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1767 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1768 (reported by Lutz Euler)
1769 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1770 compiled code. (reported by Paul Dietz)
1771 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1772 specialized parameter to a method to something that is not TYPEP
1773 the specializer is now possible.
1774 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1775 face of package deletion.
1776 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1777 pathname. (reported by David Tolpin on c.l.l)
1778 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1779 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1780 STRING-STREAM and FILE-STREAM, respectively.
1781 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1782 than previously; in addition, its test suite is now run on build.
1783 (thanks to Luke Gorrie)
1784 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1785 * optimization: added a immediate representation for single-floats
1787 * fixed some bugs revealed by Paul Dietz' test suite:
1788 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1789 ** attempting to create a package with a colliding nickname causes
1790 correctable errors to be signalled.
1791 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1792 ** several x86-64 backend bugs related to sign-extension of immediate
1795 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1796 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1797 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1799 * fixed loading of multiply forward-referenced layouts.
1800 (thanks to Cheuksan Wang)
1801 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1802 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1803 (see COMPILE-FILE documentation for details of :PRINT option)
1804 * fixed bugs 19 and 317: fixed-format floating point printing is
1805 more accurate. This also fixes a bug reported by Adam Warner
1806 related to the ~@F format directive.
1807 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1809 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1810 dispatch table if the from-char is a dispatch macro character.
1811 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1812 by the wordsize no longer ignore the last word. (reported by Lutz
1814 * fixed bug in type checking of dynamic-extent variables. (reported
1816 * optimization: sequence traversal functions use their freedom to
1817 coerce function designators to functions.
1818 * optimization: code with many calls to CLOS methods specialized on
1819 CLOS classes has had redundant type checks removed.
1820 * optimization: added declarations to speed up operations that access
1821 the internal character database (for example STRING-UPCASE)
1822 * optimization: comparison operations between floats and sufficiently small
1823 fixnums no longer create extra rationals
1824 * fixed some bugs related to Unicode integration:
1825 ** portions of multibyte characters at the end of buffers for
1826 character-based file input are correctly transferred to the
1827 start of the buffer at the next read.
1828 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1829 passing it through to OPEN.
1830 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1831 argument given to internal calls to OPEN.
1832 ** the built-in comment readers (introduced by character sequences
1833 ";" and "#|") are more forgiving to encoding errors; they will
1834 STYLE-WARN and then attempt to resync the stream at a character
1835 boundary. (thanks to Teemu Kalvas)
1836 * fixed some bugs in the x86-64 port:
1837 ** Negative short int return values from c-calls are sign-extended
1839 ** The stack is aligned to 16-bytes for c-calls, as required by
1840 the ABI. (thanks to Cheuksan Wang)
1841 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1842 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1843 for enums. (thanks to Vincent Arkesteijn)
1844 ** Multiple small optimizations and bugfixes for floating point
1846 * fixed some bugs revealed by Paul Dietz' test suite:
1847 ** Space, Tab, Linefeed, Return and Page have the invalid
1848 secondary constituent character trait.
1849 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1851 ** WITH-INPUT-FROM-STRING should only update the index place on
1853 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1854 works more reliably.
1855 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1856 with one-dimensional arrays with fill pointers.
1857 ** TYPE-OF failed on a complex with an integer realpart and a
1859 ** compiler failure during type inference for the code of form
1860 (IF (EQL X (THE ...)) ...) (MISC.535).
1862 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1863 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1864 platform. The port supports 61-bit fixnums, large memory spaces
1865 and reloading shared object files.
1866 * enhancement: saving cores with foreign code loaded is now
1867 supported on ppc/Darwin in addition to the previously supported
1869 * enhancement: the statistical profiler now walks deeper into the
1870 call stack for more meaningful call-graphs and accrued time
1871 reports (x86/x86-64 only). It also now reports time spent in
1873 * enhancement: it is now possible to trace most individual methods
1874 of a generic function in addition to tracing the generic function
1876 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1878 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1879 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1880 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1881 produces an error. (thanks to Vincent Arkesteijn)
1882 * bug fix: NAMESTRING on pathnames with :WILD components in their
1883 directories works correctly. (thanks to Artem V. Andreev)
1884 * fixed bug 125: compiler preserves identity of closures. (reported
1886 * bug fixed: functions with &REST arguments sometimes failed with
1887 "Undefined function" when compiled with (DEBUG 3). (reported by
1889 * bug fix: overflow during compiling of setting element of a bit
1890 vector with constant index and value. (reported by Timmy Douglas)
1891 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1892 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1893 * on x86 compiler partially supports stack allocation of dynamic-extent
1895 * GO and RETURN-FROM do not check the extent of their exit points
1896 when compiled with SAFETY 0.
1897 * fixed some bugs related to Unicode integration:
1898 ** encoding and decoding errors are now much more robustly
1899 handled; it should now be possible to recover even from invalid
1900 input or output to the terminal. (thanks to Teemu Kalvas)
1901 ** provided a first cut at implementing STRING-TO-OCTETS and
1902 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1903 ** altered the SB-MD5 contributed module slightly, changing the
1904 interface just enough for it to be supportable for builds where
1905 lisp characters are not eight bits.
1906 * fixed some bugs revealed by Paul Dietz' test suite:
1907 ** the FORMATTER-generated functions for ~V[ conditionals require
1908 the correct number of arguments.
1909 ** READ-FROM-STRING returns the mandated second value when applied
1910 to displaced strings.
1911 ** the #\Rubout and #\Backspace characters are treated as invalid
1912 constituent characters by the tokenizer.
1914 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1915 * new feature: reloading changed shared object files with
1916 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1917 * new feature: references to foreign variables and functions
1918 can now be compiled and loaded before the corresponding shared
1919 object file is loaded, as long as the foreign definitions are
1920 available at runtime.
1921 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1922 supported. (thanks to Dan Debertin)
1923 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1924 just keywords. (thanks to Vincent Arkesteijn)
1925 * fixed bug #331: structure-class instances corresponding to
1926 DEFSTRUCT forms are now created eagerly.
1927 * fixed bug #345: backtraces from calls to undefined functions work
1928 on x86 as well. Related bug #61 is now also partially fixed on x86
1929 (backtraces from throws to unknown catch tags.)
1930 * bug fix: lambda-list parsing is now stricter vrt. order and number
1931 of lambda-list keywords.
1932 * bug fix: as specified by AMOP, an error is signalled if a
1933 class-option appears multiple times in a DEFCLASS form. (reported
1935 * bug fix: RANDOM can be compiled when the compiler derives the type
1936 of its numeric argument as a disjoint set of small integers.
1937 (reported by Paul Dietz)
1938 * bug fix: starting a core saved with shared objects loaded when
1939 those objects are not available no longer causes threaded SBCL to
1940 hang. (reported by Sean Ross)
1941 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1943 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1944 argument, which must always be NIL. (reported by Kalle Niemitalo)
1945 * bug fix: printing 1.0d+23 no longer results in an error.
1946 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1947 * bug fix: structure slot setters preserve evaluation order. (thanks
1949 * fixed some bugs related to Unicode integration:
1950 ** RUN-PROGRAM can allow its child to take input from a Lisp
1951 stream. (reported by Stefan Scholl)
1952 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1953 locales. (reported by Ken Causey)
1954 * fixed some bugs revealed by Paul Dietz' test suite:
1955 ** INCF, DECF and REMF evaluate their place form as specified in
1957 ** FORMATTER expands ~{ ~} iteration directives with V or #
1958 parameters correctly.
1959 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1960 consequent uses no arguments correctly.
1961 ** the system has a partial understanding of the (COMPLEX RATIO)
1963 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1966 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1967 * new feature: a build-time option (controlled by the :SB-UNICODE
1968 keyword feature, enabled by default) for building the system with
1969 support for the entire 21-bit character space defined by the
1971 * new feature: the system now has rudimentary external-format
1972 support; the primary user-visible change at this time is that
1973 characters with the high bit set (such as lower-case-e-acute) will
1974 print correctly to a terminal in a UTF-8 environment.
1975 * minor incompatible change: BASE-CHAR no longer names a class;
1976 however, CHARACTER continues to do so, as required by ANSI.
1977 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1978 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1979 has been moved to the SB-EXT package (temporarily re-exported from
1981 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1982 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1984 * fixed bug #351: better error-handlind and reporting for malformed
1986 * fixed bug #350: bignum-printing is now more memory-efficient,
1987 allowing printing of very large bignums, eg. (expt 2 10000000).
1988 (reported by Bruno Haible)
1989 * fixed bug #302: better primitive-type selection for intersection
1991 * fixed bug #308: non-graphic characters now all have names, as
1992 required. (reported by Bruno Haible)
1993 * bug fix: redefining a class with different superclasses now correctly
1994 removes it from the direct-subclasses of its previous superclasses.
1995 (reported by David Morse)
1996 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1997 new value now works. (reported by Bruno Haible)
1998 * bug fix: correct canonicalization of multiple non-standard slot
1999 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2000 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2001 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2002 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2003 now exists, an signals an error.
2004 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2005 during macroexpansion for non-iteration variables as well. (reported
2006 by Bruno Haible for CMUCL)
2007 * bug fix: Cyclic structures and unprintable objects in compiler
2008 messages no longer cause errors. (reported by Bruno Haible)
2009 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2010 and READ-FROM-STRING all now return a primary value of NIL if
2011 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2012 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2013 (reported by Bruno Haible for CMUCL)
2014 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2015 do not designate a condition. (reported by Bruno Haible for
2017 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2018 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2019 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2020 permitting resolution in favour of any of the conflicting symbols.
2021 (reported by Bruno Haible for CMUCL)
2022 * bug fix: EQUAL compiler optimizations is less aggressive on
2023 strings which can potentially compare true despite having distinct
2024 specialized array element types.
2025 * bug fix: unit enumerations can be defined without dividing by
2026 zero. (thanks to Vincent Arkesteijn)
2027 * FORMAT compile-time argument count checking has been enhanced.
2028 (report from Bruno Haible for CMUCL)
2029 * a partial workaround for the bug 262: the compiler does not try to
2030 inline-expand a local function doing RETURN-FROM from a deleted
2031 BLOCK. (thanks to Peter Denno for the bug report and to David
2032 Wragg for the simple test case)
2033 * fixed some bugs revealed by Paul Dietz' test suite:
2034 ** RENAME-PACKAGE allows all package designators as new package
2036 ** constraint propagation and lambda variable substitution are
2037 more cautious in dealing with partially deleted code.
2038 ** compiler sometimes lost reoptimization passes.
2039 ** CERROR, given a condition as condition designator, passes its
2040 remaining arguments to the continue format control without
2042 ** Case-altering FORMAT directives work correctly on non-ASCII
2044 ** The REAL type specifier handles bounds outside the floating
2045 point ranges without signalling FLOATING-POINT-OVERFLOW.
2046 ** Functions with IR1-transformations can create intercomponent
2047 references to global functions.
2048 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2050 ** FORMAT ~R treats a nil value for its first parameter correctly.
2052 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2053 * enhancement: saving cores with foreign code loaded is now
2054 supported on x86/NetBSD and sparc/Linux in addition to the previously
2055 supported platforms.
2056 * bug fix: on some platforms repeated installations caused multiple
2057 copies of HTML documentation to be installed -- should not happen
2058 any more. (reported by Stefan Scholl)
2059 * bug fix: parsing self-recursive alien record types multiple times
2060 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2061 original patch by Helmut Eller for CMUCL)
2062 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2063 (thanks to Richard Kreuter)
2064 * bug fix: defining classes whose accessors are methods on existing
2065 generic functions in other (locked) packages no longer signals
2066 bogus package lock violations. (reported by François-René Rideau)
2067 * bug fix: special variables as DEFMETHOD parameters no longer have
2068 associated bogus type declarations. (reported by David Wragg and
2070 * bug fix: read-write consistency on streams of element-type
2071 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2072 * bug fix: redefiniton of the only method of a generic function with
2073 no DEFGENERIC no longer emits a full WARNING. In addition,
2074 redefinition of generic functions with no DEFGENERIC to an
2075 incompatible lambda list now signals an error. (thanks to Zach
2077 * bug fix: DEFGENERIC now works even when there's a function of the
2078 same name in an enclosing lexical environment. (thanks to Zach
2080 * fixed compiler failure, caused by instrumenting code during
2081 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2082 * optimization: added loop analysis and improved register allocation
2083 * fixed some bugs revealed by Paul Dietz' test suite:
2084 ** POSITION on displaced vectors with non-zero displacement
2085 returns the right answer.
2086 ** (SIMPLE-STRING) is a valid type specifier for sequence
2088 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2090 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2091 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2093 ** The justification version of the FORMAT ~< directive treats
2094 non-zero minpad parameter correctly.
2096 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2097 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2098 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2099 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2100 the supported interface.
2101 * new feature: Single-stepping of code compiled with DEBUG 2 or
2102 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2103 * new feature: saving cores with foreign code loaded is now
2104 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2105 Timothy Moore's work for CMUCL)
2106 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2107 parameters to * instead of NIL if no initform is supplied.
2108 (reported by Johan Bockgaard)
2109 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2110 correctly when FUNCALL appears as the car of the form. Note:
2111 despite this FUNCALL forms are not currently subject to
2112 compiler-macro expansion. (port of Raymond Toy's fix for the
2113 same from CMUCL, reported by Johan Bockgaard)
2114 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2115 (thanks for Teemu Kalvas)
2116 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2117 case instead of terminating the iteration (thanks for Julian
2118 Squires, Sean Champ and Raymond Toy)
2119 * bug fix: incorrect expansion of defgeneric that caused a style
2120 warning. (thanks for Zach Beane)
2121 * bug fix: slot accessor effective method computation works properly
2122 for classes with multiple non-standard applicable methods on
2123 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2124 * on x86 compiler supports stack allocation of results of LIST and
2125 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2126 CMUCL implementation by Gerd Moellmann)
2127 * fixed some bugs revealed by Paul Dietz' test suite:
2128 ** FORMAT strings with both the ~<~:;~> form of the justification
2129 directive and pretty-printing directives cause an error.
2131 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2132 * incompatible change: the internal functions
2133 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2134 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2135 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2136 instead of the old functions.
2137 * new feature: on platforms where "dladdr" is available foreign
2138 function names now appear in backtraces. (based on Helmut Eller's
2140 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2141 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2143 * documentation: ASDF manual is now distributed with SBCL.
2144 * bug fix: SBCL can now load its contributed modules with REQUIRE
2145 even if the system-provided entries have been removed from
2146 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2148 * bug fix: backtraces involving undefined functions or assembly
2149 routines are more informative. (thanks to Brian Downing)
2150 * bug fix: mutually referent alien structures now work correctly.
2151 (reported by Rick Taube)
2152 * bug fix: structures defined by WITH-ALIEN can be referred to
2153 within other definitions in the same WITH-ALIEN.
2154 * bug fix: division operators (MOD, TRUNCATE and the like) with
2155 constant zero divisors and integer dividends no longer generate
2157 * bug fix: provide default methods for INPUT-STREAM-P and
2158 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2159 * bug fix: improve the detection and resolution of MOP metacycles
2160 (where effective method computation depends on the generic function
2161 having its effective method computed). (reported by Bruno Haible)
2162 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2163 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2164 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2165 * bug fix: pathnames with relative directory components can be
2166 represented relative to default pathnames.
2167 * optimization: in taking the GCD of bignums, reduce the two bignums
2168 to approximately the same size (using Euclid's algorithm) before
2169 applying the more sophisticated binary GCD. (thanks to Juho
2171 * optimization: COUNT on bitvectors now operates word-at-a-time.
2172 * optimization: ASH with a positive, but not necessarily constant,
2173 (leftwards) shift, when in a modular context, is compiled to a
2175 * fixed some bugs revealed by Paul Dietz' test suite:
2176 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2177 if the corresponding argument is NIL.
2178 ** FORMAT directives accepting a minpad parameter treat negative
2180 ** literal commas as character parameters to FORMAT directives are
2182 ** literal spaces directly after ~<Newline> directives within a
2183 format-logical-block (~:< ~@:>) do not induce :FILL-style
2184 conditional newlines.
2185 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2186 as specified: it no longer includes conditional newlines).
2187 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2189 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2190 :PER-LINE-PREFIX argument does not evaluate to a string.
2192 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2193 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2194 the manual for details; for now, package locks can be disabled by
2195 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2196 if no major problems are found then it is likely that they will be
2197 compiled in unconditionally.
2198 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2199 unsupported operators on all platforms. To load a shared library
2200 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2201 object file, link it into a shared library outside of SBCL and
2202 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2204 * bug workaround: ROOM T and the GC were not working together
2205 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2206 reporting facility aren't preserved. That reporting has been
2207 disabled, so now until and unless someone figures out how to make it
2208 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2209 * minor incompatible change: as threatened around sbcl-0.8.0, the
2210 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2211 The symbols are also exported from SB-PCL for backwards
2212 compatibility, but more so than before SB-PCL should be treated as
2213 an implementation-internal package.
2214 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2216 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2217 It is known as of this release not to work on the Alpha, however.
2218 * fixed bug #167: errors signalled due to illegal syntax in method
2219 bodies are now more legible.
2220 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2221 designators and can hence be used with TYPEP.
2222 * fixed bug #333: CHECK-TYPE now ensures that the type error
2223 signalled, if any, has the right object to be accessed by
2224 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2225 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2226 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2228 * fixed bug #334: programmatic addition of slots using specialized
2229 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2230 and :ALLOCATION :CLASS slots.
2231 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2232 float exponents. (rereported by Peter Seibel)
2233 * fixed a bug: #\Space (and other whitespace characters) are no
2234 longer considered to be macro characters in standard syntax by
2235 GET-MACRO-CHARACTER.
2236 * fixed bug: initialization of condition class metaobjects no longer
2237 causes an instance of the condition to be created. (reported by
2239 * fixed bug: it is now possible to have slots such that
2240 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2241 description is neither :INSTANCE nor :CLASS.
2242 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2243 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2244 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2245 non-standard SLOT-DEFINITION-ALLOCATION values.
2246 * fixed bugs: various race conditions handling exiting threads.
2247 CL-PPCRE's thread torture test now passes.
2248 * fixed arguably-a-bug: GC time proportional to number of threads in
2249 system even when most of them are idle
2250 * optimization: improved performance of BIT and SBIT on bit-vectors.
2251 * .fasl file incompatibility: The fasl file version number has
2252 been incremented because of changes associated with package locks.
2254 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2255 * minor incompatible change: the system no longer provides
2256 optimization hints (conditions which are TYPEP
2257 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2258 mode; these hints are still emitted when higher SPEED optimization
2260 * new contrib module: a sampling profiler (profiling by statistical
2261 sampling, rather than by instrumenting functions) is available as
2262 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2263 * the behaviour of the standard function ED is now customizeable by
2264 third parties through a hook variable: see ED's documentation
2265 string for information on the protocol.
2266 * the compiler no longer emits efficiency notes for (FUNCALL X)
2267 when the type of X is uncertain under default optimization
2269 * fixed bug 276: mutating a binding of a specialized parameter to a
2270 method to something that is not TYPEP the specializer is now
2272 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2273 exhibit the required behaviour on the x86 platform. (thanks to
2274 Peter van Eynde, Eric Marsden and Bruno Haible)
2275 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2276 even for difficult arguments. (reported by Peter Graves)
2277 * fixed bug 141a: the backquote printer now descends quoted
2279 * fixed another bug in backquote printing: no more destructive
2280 modification of the form's list structure. (reported by Brian
2282 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2283 move between its address being taken and the call to
2284 interrupt_thread, fixing a crashing race condition.
2285 * the SB-POSIX contrib implementation has been adjusted so that it
2286 no longer exhibits ridiculously poor performance when constructing
2287 instances corresponding to C structs.
2289 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2290 * minor incompatible change: the sb-grovel contrib now treats C
2291 structures as alien (in the sense of SB-ALIEN) objects rather than
2292 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2293 has implications for memory management of client code
2294 (sb-grovel-returned objects must now be manually managed) and for
2295 type safety (alien objects now have full types).
2296 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2297 used to control emission of compiler diagnostics, rather than the
2298 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2299 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2300 quality should be considered deprecated.
2301 * install.sh now installs the user manual as well
2302 * (not quite a new documentable feature, but worth considering in
2303 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2304 beginnings of a semantically meaningful condition hierarchy is
2305 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2307 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2308 designator as the defaults argument.
2309 * fixed bug: Displaced arrays whose displaced-to array has become
2310 too small now cause ARRAY-DIMENSION to signal an error, providing
2311 for safer bounds-checking. (reported by Bruno Haible)
2312 * fixed bug: DEFCLASS slot definitions with identical :READER and
2313 :WRITER names now signal a reasonable error. (reported by Thomas
2315 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2317 * fixed bug: Invalid slot specification errors now print correctly.
2318 (thanks to Zach Beane)
2319 * fixed bug 320: Shared to local slot value transfers in class
2320 redefinitions now happen corrently from superclasses as
2321 well. (reported by Bruno Haible)
2322 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2324 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2325 improper lists correctly. (reported by Bruno Haible)
2326 * fixed bug 313: source-transform for <fun-name> was erroneously
2327 applied to a call of a value of a variable with name <fun-name>.
2328 (reported by Antonio Menezes Leitao)
2329 * fixed bug 307: The obsolete instance protocol ensures that
2330 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2331 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2332 remove dead unknown-values globs from the middle of the stack.
2333 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2334 easier to resume long computations after using *BREAK-ON-SIGNALS*
2335 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2336 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2337 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2338 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2339 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2340 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2342 * fixed a (fairly theoretical) bug in string printing: if
2343 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2344 does not have array-element-type equal to the most general string
2346 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2347 function name. (thanks to Nikodemus Siivola)
2348 * fixed bug: compile-time format string checker failed on
2349 non-closed ~{. (reported by Thomas F Burdick)
2350 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2351 don't warn on legal loop code involving more than one aggregate
2352 boolean. (thanks to Nikodemus Siivola)
2353 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2354 in keyword position to MAKE-PACKAGE was not being reported as
2356 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2357 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2359 * fixed bugs 280 and 312: the checking for multiple definitions in a
2360 file is less likely to become confused by uses of inline
2362 * fixed bug: the #S reader macro performs the keyword coercion
2363 specified for slot names. (reported by Kalle Niemitalo)
2364 * fixed bug: lambda lists may contain symbols whose names start with
2365 & but are not lambda-list-keywords; their occurrence triggers a
2367 * fixed bug 321: define-method-combination argument lambda lists do
2368 not cause invalid code to be generated when &optional or &aux
2369 variables have default values. (reported by Bruno Haible)
2370 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2371 have CLOS classes; this appears also to have fixed reported
2372 instability in stack exhaustion detection.
2373 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2374 MUFFLE-WARNING when no associated restart is present is now
2376 * optimization: rearranged the expansion of various defining macros
2377 so that each expands into only one top-level form in a
2378 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2380 * optimization: used a previously little-used slot in symbols to
2381 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2383 * fixed some bugs revealed by Paul Dietz' test suite:
2384 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2386 ** Fixed an optimization bug related to inheritance of initforms
2387 from local to shared slots.
2388 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2389 ** CAST optimizer forgot to flush argument derived type.
2390 ** print/read consistency on floats is now orders of magnitude
2391 more likely. (thanks also to Bruno Haible for a similar report
2393 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2394 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2395 for objects of type REAL. Make it so.
2396 ** SXHASH is specified (infelicitously) to respect similarity,
2397 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2398 it so. (thanks to Markus Ziegler)
2399 ** on the Alpha, the compiler succeeds in compiling functions
2400 returning a known number of arguments greater than 63.
2401 ** fixed handling of invalid NIL arguments in keyword position
2403 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2405 ** provided a readably-printable representation for RANDOM-STATE
2407 ** ensured that pathnames, where they have a namestring, always
2408 print using #P"..." syntax.
2410 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2411 * Support for the forthcoming 2.0 version of the NetBSD kernel
2412 running on x86 hardware has been added. (thanks to Perry
2413 E. Metzger most immediately, and others for their past work)
2414 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2415 current release version; SBCL's OpenBSD support had been broken
2416 since about the time of OpenBSD's switch to ELF binary format.)
2417 * [placeholder for DX summary]
2418 ** user code with &REST lists declared dynamic-extent, under high
2419 speed or space and low safety and debug optimization policy.
2420 * The manual has been converted to Texinfo format and the debugger
2421 chapter from the cmucl manual has been added.
2422 * A facility has been added to extract documentation strings from
2423 sbcl and store them as Texinfo-formatted snippets for inclusion in
2424 the manual (via Texinfo's @include directive)
2425 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2426 the test case to Dave Roberts)
2427 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2428 pointers. (reported by Sean Ross)
2429 * bug fix: PROFILE output is printed nicely even for large numerical
2430 values. (thanks to Zach Beane)
2431 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2432 greater than 32 handle EOF correctly.
2433 * bug fix: on X86 an immediate argument of the IMUL instruction is
2434 correctly printed in disassembly. (thanks to Lutz Euler)
2435 * bug fix: class slots in redefined classes preserve their old
2436 values. (thanks to Bruno Haible and Nikodemus Siivola)
2437 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2438 succeeds. (reported by Marco Baringer)
2439 * fixed some bugs revealed by Paul Dietz' test suite:
2440 ** READ-SEQUENCE now works on ECHO-STREAMs.
2441 ** RATIONALIZE works more according to its specification. (thanks
2444 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2445 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2446 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2447 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2448 should matter to you only if you rebind the printer control
2449 variables and then find you want different bindings in the
2450 debugger than in the ordinary execution of your program.)
2451 * The runtime build system has been tweaked to support building
2452 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2453 assembler and linker. (thanks to Nikodemus Siivola)
2454 * Unbound, undefined, undeclared variables now trigger full
2455 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2456 more often programmer error than deliberate exploitation of undefined
2458 * optimization: the hash algorithm for strings has changed to one
2459 that is less vulnerable to spurious collisions. (thanks to Juho
2461 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2462 needless bounds checking. (thanks to Juho Snellman)
2463 * optimization: implemented multiplication as a modular
2464 (UNSIGNED-BYTE 32) operation on the PPC backend.
2465 * fixed some bugs revealed by Paul Dietz' test suite:
2466 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2468 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2469 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2470 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2472 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2474 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2476 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2477 their output stream on EOF from read.
2478 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2479 have been read to end-of-file.
2480 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2482 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2483 description of determination of which consecutive characters
2485 ** Printing the "Space" character with escaping on now yields "#\\ ",
2486 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2487 ** Reading floating-point numbers with *READ-BASE* set to a number
2488 less than 10 works correctly.
2489 ** Reading floating-point numbers with *READ-BASE* set to a number
2490 more than 10 works correctly.
2491 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2492 the readtable currently in effect.
2494 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2495 * minor incompatible change: parsing of namestrings on a physical
2496 (Unix) host has changed; numbers after the final #\. in a
2497 namestring are no longer interpreted as a version field. This is
2498 intented to be largely invisible to the user, except that the
2499 meaning of the namestring "*.*.*" has changed: it now refers to a
2500 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2501 should usually be replaced by
2502 (make-pathname :name :wild :type :wild :version :wild)
2503 with the added benefit that this is more likely to be portable.
2504 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2505 option to OPEN now signals an error if the file being opened
2506 exists; this may have an impact on existing code.
2507 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2508 signals received, on the PowerPC platforms (both Linux and
2509 Darwin). (thanks to Pierre Mai for pointing out the location of
2511 * several fixes on OS X: The system now builds and runs cleanly on
2512 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2513 (thanks to Brian Mastenbrook)
2514 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2515 second-resolution: integer multiples of 1/3600 between -24 and 24.
2516 (thanks to Vincent Arkesteijn)
2517 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2518 type when called with 1 argument; PEEK-CHAR checked type of
2519 PEEK-TYPE only after having read first character from a
2520 stream. (reported by Peter Graves)
2521 * bug fix: the garbage collector now has much better locality
2522 behaviour, and in particular no longer treats all memory as being
2523 exhausted when many small objects point to each other in a deeply
2525 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2526 recognized as being TYPEP their class.
2527 * bug fix: the PUSHNEW documentation string has been corrected.
2528 (thanks to Vincent Arkesteijn)
2529 * bug fix: defaulting of the value for the last of an atypically
2530 large number of multiple values being bound was not being
2531 performed correctly on the Alpha or PPC platforms
2532 * optimization: implemented multiplication as a modular
2533 (UNSIGNED-BYTE 32) operation on the x86 backend.
2534 * optimization: SEARCH on simple-base-strings can now be open-coded.
2535 (see also contrib/compiler-extras.lisp for inspiration for
2536 teaching the compiler about the Boyer-Moore algorithm).
2537 * value, returned by MAX (and MIN) called with several EQUALP, but
2538 not EQL, arguments now does not depend on compiler settings.
2539 * fixed some bugs revealed by Paul Dietz' test suite:
2540 ** in stack analysis liveness information is propagated from
2541 non-local entry points.
2542 ** pathwise CAST removing failed when the CAST node did not start
2544 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2545 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2547 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2548 host is already defined.
2549 ** RENAME-FILE works on streams instead of signalling an internal
2551 ** PEEK-CHAR uses the current readtable when determining whether
2552 or not a character is whitespace.
2553 ** MERGE-PATHNAMES handles the case when the pathname does not
2554 specify a name while the default-pathname specifies a version.
2555 ** Pathnames now stand a better chance of respecting print/read
2557 ** Attempting to use standardized file system operators with a
2558 pathname with invalid :DIRECTORY components signals a
2560 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2562 ** DIRECTORY on logical pathnames is more correct.
2563 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2564 signal a TYPE-ERROR if their argument is not a stream.
2565 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2566 designator argument does not designate a stream.
2567 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2568 examining the synonym.
2569 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2571 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2572 element-type larger than ([UN]SIGNED-BYTE 32).
2574 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2575 * When built with the :SB-FUTEX feature, threaded builds now take
2576 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2577 for faster/more reliable mutex and condition variable support.
2578 * Incompatible change (but one you probably shouldn't have been using
2579 anyway): the interface and code for arbitrating between multiple
2580 threads in the same user session has been redesigned.
2581 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2582 error when called without an explicit environment argument.
2583 (thanks to Nikodemus Siivola)
2584 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2585 confused about their position. (thanks to Adam Warner and Gerd
2587 * bug fix: Pretty printing backquoted forms with unquotations in the
2588 argument list position of various code constructs such as LAMBDA
2589 now works correctly. (reported by Paul Dietz)
2590 * bug fix: Pretty printing unquotations no longer loses all
2591 stream position information.
2592 * optimization: performance of string output streams is now less
2593 poor for multiple small sequence writes.
2594 * optimization: performance of CSUBTYPEP in the presence of complex
2595 expressions involving CONS and NOT many times has been improved.
2596 (reported by Paul Dietz)
2597 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2599 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2600 streams for the REPL, for the debugger, and so on. (thanks to
2602 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2603 * fixed some bugs revealed by Paul Dietz' test suite:
2604 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2605 is ignored if it is too small, rather than propagating through
2606 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2607 ** extremely complex negations of CONS types were not being
2608 sufficiently canonicalized, leading to inconsistencies in
2610 ** VALUES tranformer lost derived type.
2612 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2613 * fixed a bootstrapping bug: the build process no longer assumes
2614 that the various BOOLE-related constants have the same value in
2615 host and target lisps. (noted by Paul Dietz' test suite on an
2616 SBCL binary built from CLISP)
2617 * The system can now be dynamically linked on the MIPS platform,
2618 which enables dynamic loading of foreign code from Lisp. (thanks
2619 to Ralf Baechle for discussions on the MIPS ABI)
2620 * The system now records debugging information for its own source
2621 files in a filesystem-position-independent manner, relative to
2622 the "SYS" logical host.
2623 * fixed a compiler bug: MV-LET convertion did not check references
2624 to the "max args" entry point. (reported by Brian Downing)
2625 * tweaked disassembly notes to be less confident about proclaiming
2626 some instruction as an LRA. (thanks to Brian Downing)
2627 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2628 now each have their own history, command character, and other
2629 characteristics. (thanks to David Lichteblau)
2630 * fixed some bugs revealed by Paul Dietz' test suite:
2631 ** compiler failure in compiling LOGAND expressions including a
2633 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2634 shift greater than 32.
2635 ** FLUSH-DEST did not mark blocks for type check regeneration.
2636 ** HANDLER-CASE failed to accept declarations in handler clauses
2637 in some circumstances.
2639 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2640 * New code in contrib/sb-introspect (still probably not entirely
2641 stable yet) provides some support for smart Lisp development
2642 environments like SLIME.
2643 * The conditions signalled for errors occurring when loading .fasl
2644 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2645 in a way which should help ASDF recover gracefully.
2646 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2647 stuff has been cleaned up. If you code contrib/ stuff, this might
2648 affect you, and you can look at contrib/README, contrib/STANDARDS,
2649 and/or the 0.8.4.27 diff to check.
2650 * In full calls the compiler now does not generate checks for declared
2651 argument types for all arguments.
2652 * various threading fixes
2653 ** and some experimental patches which didn't make it into
2654 the main tree for this release, but which are shipped in
2655 contrib/experimental-thread.patch as a possible fix for some
2656 failures (deadlock, spinning...) in GC-intensive multithreaded
2658 * fixed PPC build problem (source code incompatibility of different
2659 library versions): added offsetof() hackery which attempts to divine
2660 where glibc maintainers put uc_mcontext today
2661 * fixed bug 282: compiler does not trust type assertions while passing
2662 arguments to a full call.
2663 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2664 &OPTIONAL VALUES type parameter.
2665 * fix bug 214: algorithm for noting rejected templates is now more
2666 similar to that of template seletion. (also reported by rydis on
2668 * fixed bug 141b: printing backquoted information readably and prettily
2669 inserts a space where necessary.
2670 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2671 (CHARACTER 10) are now reported as errors, rather than propagated
2672 as unknown types. (reported by piso on #lisp)
2673 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2674 with values NIL and :ERROR. (thanks to Milan Zamazal)
2675 * fixed bug 191c: CLOS now does proper keyword argument checking as
2676 described in CLHS 7.6.5 and 7.6.5.1.
2677 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2678 longer raise an error; further, using a list as a for-as-arithmetic
2679 counter now raises a meaningful error.
2680 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2681 now understood by sequence creation functions such as MAKE-SEQUENCE
2683 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2684 a STRING-INPUT-STREAM.
2685 * compiler enhancement: SIGNUM is now better able to derive the type
2687 * type declarations inside WITH-SLOTS are checked. (reported by
2689 * fixed some bugs revealed by Paul Dietz' test suite:
2690 ** incorrect optimization of TRUNCATE for a positive first
2691 argument and negative second.
2692 ** compiler failure in let-convertion during flushing dead code.
2693 ** compiler failure while deriving type of TRUNCATE on an
2694 interval, containing 0.
2695 ** ASH of a negative bignum by a negative bignum count now returns
2697 ** intersection of CONS types now canonicalizes properly, fixing
2698 inconsistencies in SUBTYPEP.
2700 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2701 * incompatible change: The --disable-debugger command line
2702 option now clobbers the debugger at a more fundamental
2703 level, by redefining #'INVOKE-DEBUGGER instead of by
2704 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2705 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2706 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2707 enter the debugger REPL and then suffer recursive errors
2708 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2709 while under the new system, BREAK in --disable-debugger mode
2710 terminates the system just as an unhandled error would.
2711 * fixed compiler performance when processing loops with a step >1;
2712 * bug fix: DOCUMENTATION now retrieves generic function
2713 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2714 support has been systematized, and now supports the methods
2715 specified by ANSI, along with a default method and a method for
2716 slot documentation. (reported by Nathan Froyd)
2717 * bug fix: effective methods associated with a generic function are
2718 no longer cached over a change of that generic function's method
2719 combination. (reported by Andreas Fuchs)
2720 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2721 names a non-existent directory. (thanks to Andreas Fuchs)
2722 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2723 the PPC platform, lead to stack corruption; this has been fixed.
2724 (reported by Rainer Joswig)
2725 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2726 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2727 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2728 arithmetic in general, is now much more reliable on the PPC
2730 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2731 platform now returns the right answer.
2732 * optimization: restored some effective method precomputation in
2733 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2734 precomputation is now tunable.
2735 * optimization: compiler-internal data structure use has been
2736 reviewed, and changes have been made that should improve the
2737 performance of the compiler by about 20%.
2738 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2739 simple-base-strings and simple-bit-vectors is improved.
2740 * optimization: the optimization of 32-bit logical and arithmetic
2741 functions introduced in version 0.8.3 on the x86 has been
2742 implemented on the mips, ppc and sparc platforms; an
2743 implementation of the same facility, but for 64-bit arithmetic,
2744 has been added for the alpha.
2745 * microoptimization: the compiler is better able to make use of the
2746 x86 LEA instruction for multiplication by constants.
2747 * bug fix: in some situations compiler did not report usage of
2748 generic arithmetic in (SPEED 3) policy.
2749 * bug 145b fix: compiler used wrong type specifier while converting
2750 MEMBER-types to numeric.
2751 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2753 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2754 index argument now works properly on the Alpha platform.
2755 * bug fix: floating point exception treatment on the Alpha platform
2757 * bug fix: FILE-POSITION works much better on string input and
2758 output streams. (thanks to Nikodemus Siivola)
2759 * bug fix: many threading/garbage collection symptoms sorted.
2760 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2761 might be pseudo-atomic.
2762 * internal change: Stopping for GC is now done with signals not
2763 ptrace. GC is now done in whichever thread wanted it, instead of
2765 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2767 * bug fix: result form in DO is not contained in the implicit
2769 * incompatible change: ICR structure is changed; the value part of
2770 CONTINUATION is now called LVAR; corresponding functions are
2771 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2772 * added type deriver for ISQRT (thanks to Robert E. Brown).
2773 * bug fix: better support for loading from the command line when an
2774 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2776 * fixed some bugs revealed by Paul Dietz' test suite:
2777 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2778 ** ROUND and FROUND now give the right answer when given very
2779 small float arguments.
2780 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2782 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2783 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2784 ** LCM should return a non-negative integer.
2785 ** PARSE-INTEGER returned the index of a terminator instead of the
2786 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2787 ** PARSE-INTEGER returned an incorrect index being applied to a
2789 ** LCM with two arguments of 0 returns 0 rather than signalling
2791 ** unsigned addition of a 32-bit constant with the high bit set no
2792 longer causes an internal compiler error on the x86.
2793 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2794 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2795 with negative last argument.
2796 ** byte specifiers with zero size and position no longer cause
2797 an error during type derivation.
2798 ** bignum multiplication on the Alpha platform now returns the
2800 * porting: The system now builds on SuSE AMD64, although it still
2801 generates a 32-bit binary.
2802 * .fasl file incompatibility: The fasl file version number has
2803 been incremented (because of the changes to internal compiler
2804 data structures referred to above).
2806 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2807 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2808 more accurately, on the Darwin kernel running on PowerPC hardware.
2809 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2810 * Compiler code deletion notes now signal a condition of type
2811 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2812 an associated MUFFLE-WARNING restart.
2813 * The compiler now performs limited argument count validation of
2814 constant format strings in FORMAT, and where appropriate in ERROR,
2815 CERROR and WARN. (thanks to Gerd Moellmann)
2816 * New ASDF-INSTALL contrib can be used for automatic download and
2817 installation of third-party Lisp code from CCLAN or other sites
2819 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2820 which forces another thread to execute a function supplied by the
2822 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2823 now accept and act upon their :ELEMENT-TYPE keyword argument.
2824 (reported by Martin Atzmueller, Edi Weitz)
2825 * bug fix: FILE-POSITION now accepts position designators up to
2826 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2827 is the greater. (thanks to Patrik Nordebo)
2828 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2829 Kalle Olavi Niemitalo)
2830 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2831 printable. (reported by Eric Marsden)
2832 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2833 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2834 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2835 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2836 (thanks to Stig Sandoe)
2837 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2838 UNREAD-CHAR now correctly includes the unread character in the
2839 target sequence. (thanks to Gerd Moellmann)
2840 * bug fix in threaded builds: the system can now be suspended and
2841 resumed by shell job control with minimal disruption.
2842 * bug fixes in times and timezones >2038AD
2843 * better handling of "where is GNU make?" problem in build scripts
2844 (thanks to Nikodemus Siivola)
2845 * new optimization: inside a named function any reference to a
2846 function with the same name is considered to be a self-reference;
2847 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2848 optimization quality.
2849 * new optimization on x86: logical functions and + now have
2850 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2851 used when the result is truncated to 32 bits.
2852 * VALUES declaration is partially enabled.
2853 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2854 * fixed some bugs revealed by Paul Dietz' test suite:
2855 ** The system now obeys the constraint imposed by
2856 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2857 types form a lattice under type intersection.
2858 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2859 ** ASSOC now ignores NIL elements in an alist.
2860 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2861 and (1+ MOST-POSITIVE-FIXNUM) answers.
2862 ** The addition of a method with invalid qualifiers to a generic
2863 function does not cause an error to be signalled immediately;
2864 a warning is signalled, and the error is generated only on
2865 calling the generic function.
2866 * changed .fasl file version number, in part to add type codes for
2867 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2868 obscure ANSI requirements
2870 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2871 * fixed bug 148: failure to inline-expand a local function left
2872 garbage, confusing the compiler.
2873 * fixed bugs 3cd: structure slot readers perform type check if the
2874 slot can have an invalid value (i.e. it is either not initialized
2875 or can be written with a less specific slot writer).
2876 * bug fix: the compiler now traps array references to elements off
2877 the end of an array; previously, the bounds checking in some
2878 circumstances could go off-by-one.
2879 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2881 * type declarations for array element types now obey the description
2882 on the CLHS page "Declaration TYPE", as per discussions on
2883 sbcl-help around 2003-05-08. This means that a declaration
2884 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2885 declaration, all references to BAR will be asserted or assumed
2886 (with THE, so dependent on compiler policy) to involve objects of
2887 type FOO. Note that no such declaration is implied in
2888 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2889 * declared types of functions from the "Conditions"
2890 chapter. (reported by Paul Dietz)
2891 * bug fix: CERROR accepts a function as its first argument.
2892 * bug fix: NTH an NTHCDR accept a bignum as index
2893 arguments. (reported by Adam Warner)
2894 * optimization: character compare routines now optimize comparing
2895 against a constant character. (reported by Gilbert Baumann)
2896 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2897 argument now works properly on the MIPS platform.
2898 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2899 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2900 declared to have a functional type, does not cause an error
2902 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2903 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2904 * fixed bug: if last continuation of a deleted block has a
2905 destination, this destination should be deleted too. (reported by
2907 * fixed a bug in the bootstrap process: the host compiler's values
2908 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2909 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2910 test case from Patrik Nordebo)
2911 * improved the ability of the disassembler on the PPC platform to
2912 provide helpful disassembly notes.
2913 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2914 the class in more cases than previously.
2915 * bug fix: FILE-POSITION now understands :START and :END for
2916 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2917 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2918 8) data. (thanks to David Lichteblau)
2919 * bug fix: it is possible to add a method to a generic function
2920 without lambda list.
2921 * bug fix: reader failed to signal END-OF-FILE inside an
2922 object representation. (reported by Nikodemus Siivola)
2923 * fixed some bugs revealed by Paul Dietz' test suite:
2924 ** LAST and [N]BUTLAST should accept a bignum.
2925 ** condition slot accessors are methods.
2926 ** (VECTOR NIL) is a subtype of STRING.
2928 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2929 * minor incompatible change: some nonsensical specialized lambda
2930 lists (used in DEFMETHOD) which were previously ignored now signal
2932 * minor incompatible change: the system is now aware of the types of
2933 variables in the COMMON-LISP package, and will signal errors for
2934 most violations of these type constraints (where previously they
2935 were silently accepted).
2936 * minor incompatible change: COMPILE-FILE now uses the freedom
2937 afforded (ANSI 3.2.2.3) to use derived function types for
2938 functions defined in the same file. This also permits the system
2939 to warn on static type mismatches and function
2940 redefinition. (Currently it does not work with high DEBUG level.)
2941 * minor incompatible change: VALUES declaration is disabled.
2942 * When issuing notes, the compiler now signals a condition of type
2943 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2944 restart for use in user handlers. It is expected that the
2945 COMPILER-NOTE condition will eventually become a condition
2946 supertype to a hierarchy of note types, which will then be
2947 handleable in a similar fashion. However, at the moment, no such
2948 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2949 but it's an implementation detail, not a classification for the
2951 * Changes in type checking closed the following bugs:
2952 ** type checking of unused values (192b, 194d, 203);
2953 ** template selection based on unsafe type assertions (192c, 236);
2954 ** type checking in branches (194bc).
2955 * A short form of VALUES type specifier has ANSI meaning (it has
2956 increased the number of situations when SBCL cannot perform type
2958 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2959 keywords or constants is permissible.
2960 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2961 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2962 classes. (thanks to Antonio Martinez)
2963 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2964 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2965 argument to a pretty-print stream. Instead, it leaves any such
2966 operations to DESCRIBE-OBJECT methods.
2967 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2969 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2970 lambda lists. (thanks to David Lichteblau)
2971 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2972 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2973 (thanks to Gerd Moellmann)
2974 * fixed bug 246: increased compilation speed of long
2975 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2977 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2979 * DEFCONSTANT now throws a condition of type
2980 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2981 constant to a non-EQL value; CONTINUE and ABORT restarts
2982 respectively change and preserve the value.
2983 * fixed bug 63: The code walker, part of the implementation of CLOS,
2984 is now better at handling symbol macros.
2985 * bug fix: There is no longer an internal implementation type named
2986 CL:LENGTH. (reported by Raymond Toy)
2987 * bug fix: In macro-like defining macros/special operators the
2988 implicit block does not enclose the lambda list.
2989 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2990 enclosed in parentheses) are not suitable as type specifiers, and
2991 their use properly signals an error now.
2992 * bug fix: An argument count mismatch for a type specifier in code
2993 being compiled no longer causes an unhandled error at compile
2994 time, but signals a compile-time warning.
2995 * fixed simple vector readable printing
2996 * bug fix: DESCRIBE takes more care over whether the class
2997 precedence list slot of a class is bound before accessing it.
2998 (reported by Markus Krummenacker)
2999 * bug fix: FORMATTER can successfully compile pretty-printer format
3000 strings which use variants of the ~* directive inside.
3001 * bug fix: SEARCH now applies its TEST predicate to the elements of
3002 the arguments in the correct order. (thanks to Wolfhard Buss)
3003 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3004 in type inference. (thanks to Robert E. Brown)
3005 * bug fix: Reading in symbols with an explicit package name of ""
3006 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3007 * fixed some bugs revealed by Paul Dietz' test suite:
3008 ** NIL is now allowed as a structure slot name.
3009 ** Arbitrary numbers, not just REALs, are allowed in certain
3010 circumstances in LOOP for-as-arithmetic clauses.
3011 ** Multiple class redefinitions before slot access no longer
3012 causes a type error.
3013 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3014 association between the name and a class.
3015 ** Generic functions with non-standard method-combination and over
3016 five methods all of which return constants no longer return NIL
3017 after the first few invocations. (thanks to Gerd Moellmann)
3018 ** CALL-NEXT-METHOD with no arguments now passes the original
3019 values of the arguments, even in the presence of assignment.
3020 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3022 ** DISASSEMBLE works with closures and funcallable instances.
3023 ** ADD-METHOD now returns the generic function, not the new method.
3024 ** FIND-METHOD signals an error if the lengths of the specializers
3025 is incompatible with the generic function, even if the ERRORP
3027 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3028 which its argument is a member.
3029 ** DEFCLASS only redefines the class named by its class-name
3030 argument if that name is the proper name of the class;
3031 otherwise, it creates a new class.
3032 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3033 of the UNBOUND-SLOT condition to the name of the slot.
3034 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3035 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3036 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3037 SLOT-MAKUNBOUND in the specified fashion.
3039 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3040 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3041 cross-compilation host. As a consequence, we can now bootstrap our
3042 way up to SBCL starting with a bare gcc toolchain and human-readable
3043 source code (first the source to CLISP, then the source to SBCL).
3044 * A contributed module containing a partial implementation of the
3045 simple-streams interface has been included. (thanks to Rudi
3047 * A contributed module implementing the RFC1321 Message Digest
3048 Algorithm, known as MD5, has been included.
3049 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3050 no longer has any effect, as the code controlled by this feature
3051 has been deleted. (As far as we know, no-one has ever built using
3052 this feature, and its semantics were confused in any case).
3053 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3054 work on conditions (as required by the ANSI specification),
3055 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3056 expected behaviour on conditions. Users should note, however,
3057 that such behaviour is not required by the ANSI specification,
3058 and so use of this behaviour may render their code unportable.
3059 * fixed some bugs revealed by Paul Dietz' test suite:
3060 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3062 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3063 ** SXHASH on condition objects no longer returns NIL.
3064 ** :ALLOCATION :CLASS slots are better treated; their values are
3065 updated on class redefinition, and initforms inherited from
3066 superclasses are applied.
3067 ** REMOVE-METHOD returns its generic function argument even when
3068 no method was removed.
3069 ** SHARED-INITIALIZE now initializes the values of the requested
3070 slots, including those with :ALLOCATION :CLASS.
3071 ** ALLOCATE-INSTANCE now works on structure classes defined via
3072 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3074 ** SLOT-EXISTS-P now works on conditions, as well as structures
3076 ** MAKE-LOAD-FORM now has the required methods on
3077 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3078 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3079 keyword, and now implements the SLOT-NAMES argument.
3080 ** methods with &OPTIONAL arguments no longer allow too many
3081 arguments to be passed in the call without error.
3082 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3083 option is consistent with the required arguments of the generic
3084 function lambda list.
3085 * bug fix: REQUIRE accepts a string designator. (Thanks to
3087 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3088 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3089 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3090 * bug fix: The compiler now checks for duplicated variables in macro
3092 * bug fix: SETQ on globals returns the correct value.
3093 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3094 not a COMPILER-ERROR (followed by some other strange error on
3095 choosing the CONTINUE restart).
3096 * bug fix: make.sh and friends are now more consistent in the way that
3097 they look for GNU "make".
3099 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3100 * experimental native threads support (on x86 Linux >=2.4 only).
3101 This is not compiled in by default: you need to add :SB-THREAD to
3102 the target features. See the "Beyond ANSI" chapter of the manual
3104 * fix for longstanding nonANSIism: The old distinction between
3105 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3106 The return value from CL:FIND-CLASS is now a CLOS class, and
3107 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3108 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3109 * An interface to the MetaObject Protocol, as described in Kiczales,
3110 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3111 MIT Press, 1991, is available from the SB-MOP package.
3112 * incompatible change: the SB-PCL package should now be considered
3113 a private implementation detail, and no longer a semi-private MOP
3115 * minor incompatible change: due to rearrangement for threads, the
3116 control stack and binding stack are now allocated at arbitrary
3117 addresses instead of being hardcoded per-port. Users affected by
3118 this probably have to be doing advanced things with shared
3119 libraries, and will know who they are.
3120 * minor incompatible change: Previously, all --eval forms used were
3121 processed with READ before any of them were processed with EVAL.
3122 Now each --eval form is processed with both READ and EVAL before
3123 the next --eval form is processed. (Thus package operations like
3124 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3125 work as the user might reasonably expect.)
3126 * minor incompatible change: *STANDARD-INPUT* is now only an
3127 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3129 * minor incompatible change: Y-OR-N-P is now character-oriented,
3130 not line oriented. Also, YES-OR-NO-P now works without errors.
3131 (thanks to Antonio Martinez)
3132 * sb-aclrepl module improvements: an integrated inspector, added
3133 repl features, and a bug fix to :trace command.
3134 * Known functions, which cannot be open coded by the backend, are
3135 considered to be able to check types of their arguments. (fixing
3136 a bug report by Nathan J. Froyd)
3137 * fixed a bug in computing method discriminating functions: It is
3138 now possible to define methods specialized on classes which have
3139 forward-referenced superclasses. (thanks to Gerd Moellmann)
3140 * fixed evaluation order in optional entries (reported by Gilbert
3142 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3143 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3144 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3145 required-by-AMOP NAME argument, as well as CLASS and
3146 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3147 * fixed bug 20: DEFMETHOD can define methods using names that are
3148 not the proper names of classes to designate class specializers.
3149 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3150 Unix file descriptors, instead of blowing up. (thanks to Antonio
3152 * Garbage collection refactoring: user-visible change is that a
3153 call to the GC function during WITHOUT-GCING will not do garbage
3154 collection until the end of the WITHOUT-GCING. If you were doing
3155 this you were probably losing anyway.
3156 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3157 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3158 TYPEP the latter but not the former.
3159 * The compiler issues a full WARNING for calls to undefined functions
3160 with names from the CL package.
3161 * MAP-INTO for a vector destination is open coded. (reported by
3162 Brian Downing on c.l.l)
3163 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3164 documentation string.
3165 * fixed some bugs revealed by Paul Dietz' test suite:
3166 ** COPY-ALIST now signals an error if its argument is a dotted
3168 ** Condition slots are now accessed more correctly in the presence
3169 of multiple initargs for a given slot.
3170 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3171 exclude restarts of the same name associated with a different
3173 ** DEFCLASS of forward-referenced classes with another
3174 forward-referenced class in the superclasses list no longer
3176 ** Condition slots are now initialized once each, not multiple
3177 times. (thanks to Gerd Moellmann)
3178 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3180 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3181 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3182 arguments contain duplicated elements.
3183 ** RESTART-CASE understands local macros.
3184 ** RESTART-CASE associates exactly its own restarts with a condition.
3185 ** ENDP in safe mode checks its argument to be of type LIST.
3186 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3187 in question is unbound.
3188 ** Optimized MAKE-INSTANCE functions no longer cause internal
3189 assertion failures in the presence of duplicate initargs.
3190 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3191 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3192 propagated even in the ctor optimized implementation of
3194 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3196 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3197 * a better implementation of SXHASH on (simple) bit vectors,
3198 measured both in execution speed and in distribution of results
3199 over the positive fixnums, has been installed. Likewise, a better
3200 implementation of EQUAL for simple bit vectors is now available.
3201 * fixed CEILING optimization for a divisor of form 2^k.
3202 * fixed bug 240 (emitting extra style warnings "using the lexical
3203 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3204 by Antonio Martinez)
3205 * fixed SXHASH, giving different results for NIL depending on type
3206 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3207 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3208 variable. (found by Rolf Wester)
3209 * SBCL does not ignore type declarations for special
3210 variables. (reported by rif on c.l.l 2003-03-05)
3211 * some bug fixes in contrib/sb-aclrepl/
3212 * fixed some bugs revealed by Paul Dietz' test suite:
3213 ** a bug in the CONS type specifier, whereby the CAR and CDR
3214 types got intertwined, has been fixed;
3215 ** the type system is now able to reason about the interaction
3216 between INTEGER and RATIO types more completely;
3217 ** APPEND, [N]REVERSE and NRECONC check that those their
3218 arguments, which must be proper lists, are really so;
3219 ** An array specialized to be unable to hold elements has been
3220 implemented, as required -- yes, really -- by ANSI;
3221 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3222 on malformed property lists;
3224 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3225 * incompatible packaging change: in line with Unix convention,
3226 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3227 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3228 * REQUIRE and PROVIDE are now optionally capable of doing something
3229 useful. See the documentation string for REQUIRE.
3230 * infrastructure for a managed SBCL contrib system: contributed
3231 modules in this release include:
3232 ** the ASDF system definition facility;
3233 ** an interface to the BSD Sockets API;
3234 ** an ACL-like convenience interface to the repl;
3235 (thanks to Kevin Rosenberg)
3236 ** an implementation of ROTATE-BYTE, with efficient implementation
3238 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3239 gives a closer approximation to the right answer than previously.
3240 (thanks to Raymond Toy)
3241 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3242 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3243 optional environment arguments, as required by ANSI.
3244 * fixed bugs in other functions taking environment objects, allowing
3245 calls with an explicit NIL environment argument to be compiled
3247 * fixed bug 228: primary return values from
3248 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3249 COMPILE or FUNCTION.
3250 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3251 :TYPE will no longer signal errors on innocuous objects.
3252 * fixed bug 231b: SETQ is better at respecting type declarations in
3253 the lexical environment.
3254 * fixed a bug in DEFCLASS: classes named by symbols with no or
3255 unprintable packages can now be defined.
3256 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3257 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3258 * fixed compiler failure related to checking types of functions.
3259 (reported by Robert E. Brown)
3260 * the compiler is now much more consistent in its error-checking
3261 treatment of bounding index arguments to sequence functions: in
3262 (SAFETY 3) code, errors will be signalled in almost all cases if
3263 invalid sequence bounding indices are passed to functions defined
3264 by ANSI to operate on sequences.
3265 * fixed a bug in the build procedure: documentation of SBCL-specific
3266 packages is now preserved and available in the final Lisp image.
3267 * lifted FDEFINITION lookup out of loops in the implementation of
3268 many list operations. (thanks to Robert E. Brown)
3269 * fixed a bug in the reader: the #n# reader macro now works for
3270 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3271 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3272 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3273 E. Brown and Tony Martinez respectively)
3274 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3275 SLOT-BOUNDP: in particular, it is now able to optimize them much
3276 better, and is now not vulnerable to having packages renamed.
3277 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3278 always called when a slot is not present in an instance. (thanks
3280 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3281 longer signal an error on unreading a character following EOF on
3282 the previous constituent stream. (thanks to Tony Martinez)
3283 * fixed some bugs revealed by Paul Dietz' test suite:
3284 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3285 not just nonnegative fixnums;
3286 ** the logical bit-array operators such as BIT-AND now accept an
3287 explicit NIL for their "opt-arg" argument (to indicate a
3288 freshly-consed result bit-array);
3289 ** ELT now signals an error on an invalid sequence index in safe
3291 ** the type system is now cleverer about negations of numeric
3292 types, and consequently understands the BIGNUM and RATIO types
3294 ** the type system is now cleverer about the interaction between
3295 INTEGER and RATIO types: while bugs still remain, many more
3296 cases are accurately computed;
3297 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3298 if it is in the last clause;
3299 ** CONSTANTLY now correctly returns a side-effect-free function in
3301 ** DECLARE is no longer treated as a special-operator; in
3302 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3303 * incremented fasl file version number due to the change in the
3304 DEFSTRUCT-SLOT-DESCRIPTION structure.
3306 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3307 * minor incompatible change: code processed by the "interpreter" or
3308 EVAL now has a compilation optimization policy of (DEBUG 2)
3309 (changed from (DEBUG 1)) to improve debuggability of interactive
3310 development, and to allow the use of the debug RETURN command in
3312 * an experimental implementation of the RETURN command for the
3313 debugger has been included. (thanks to Frederik Kuivinen)
3314 * fixed bug 62: constraints were not propagated into a loop.
3315 * fixed bug in embedded calls of SORT (reported and investigated by
3317 * fixed some bugs revealed by Paul Dietz' test suite:
3318 ** printing and reading of arrays with some dimensions having
3319 length 0 (thanks to Gerd Moellmann);
3320 ** BOA constructor with &AUX argument without a default value does
3321 not cause a type error;
3322 ** CONSTANTP now returns true for all self-evaluating objects.
3324 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3325 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3326 accessors that are related by inheritance, as specified in the
3327 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3328 Valtteri Vuorikoski)
3329 * The compiler is now able to inline functions that were defined in
3330 a complex lexical environment (e.g. inside a MACROLET).
3331 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3333 * Support for the upcoming FreeBSD-5.0 release has been included.
3334 (thanks to Dag-Erling Smorgrav)
3335 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3336 effect when it is not in a toplevel context.
3337 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3338 stand a better chance of being correct. (thanks to Gerd
3340 * fixed bug in COERCE, which now signals an error on coercing a
3341 rational to a bounded real type which excludes the expected
3343 * The compiler is now able to derive types more accurately from the
3344 COERCE and COMPILE functions.
3345 * fixed bug 223: functional binding is considered to be constant
3346 only for symbols in the CL package.
3347 * fixed bug 231: SETQ did not check the type of a variable being set
3348 (reported by Robert E. Brown)
3349 * A new optimization for MAKE-INSTANCE has been included, fixing
3350 various bugs (including relating to :ALLOCATION :CLASS slots and
3351 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3353 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3354 ** As required by ANSI, LOOP now disallows anonymous collection
3355 clauses such as COLLECT I in conjunction with aggregate boolean
3356 clauses such as THEREIS (= I 1);
3357 ** LOOP now signals an error when any variable is reused in the
3358 same loop (including the potentially useful construct analogous
3359 to WITH A = 1 WITH A = (1+ A);
3360 ** IT is only a special loop symbol within the first clause of a
3361 conditional loop clause;
3362 ** LOOP with a typed iteration variable over a hashtable now
3363 signals a type error iff it should.
3364 * fixed some other bugs revealed by Paul Dietz' test suite:
3365 ** FILE-STREAM now names the class previously known as FD-STREAM;
3366 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3367 argument) no longer signals an error;
3368 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3369 of the slot symbol, rather than using the current package
3370 ((:CONC-NAME "") continues to intern the slot's name in the
3372 * incremented fasl file version number, because of the incompatible
3373 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3374 of the new implementation of DEFINE-COMPILER-MACRO.
3376 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3377 * Support for building SBCL for MIPS platforms running in
3378 little-endian mode has now been checked in, and basic
3379 functionality on said platforms verified.
3380 * minor incompatible change: PCL now records the pathname of a file
3381 in which methods and the like are defined, rather than its
3383 * minor incompatible change: TRUENAME now considers the truename of
3384 a file naming a directory to be the pathname with :DIRECTORY
3385 component indicating that directory.
3386 * minor incompatible change: a NAMED clause in the extended form of
3387 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3388 reason for the previous behaviour is unclear.
3389 * more systematization and improvement of CLOS and MOP conformance
3390 in PCL (thanks to Gerd Moellman and Pierre Mai):
3391 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3393 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3394 primary methods with no specializers;
3395 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3397 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3398 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3399 FORWARD-REFERENCED-CLASSes; error reporting on
3400 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3402 ** SXHASH on CLOS instances now uses a slot internal to the
3403 instance to return different numbers on distinct instances,
3404 while preserving the same return value through invocations of
3406 ** DEFMETHOD signals errors when methods with longer incongruent
3407 lambda lists are added to generic functions;
3408 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3409 CLASS, as specified in AMOP;
3410 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3411 on the order returned by the primary method for classes of
3412 class STANDARD-CLASS;
3413 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3414 * fixed some bugs shown by Paul Dietz' test suite:
3415 ** DOLIST puts its body in TAGBODY;
3416 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3418 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3419 value producing form;
3420 ** if more variables are given to PROGV than values, extra
3421 variables are bound and made to have no value;
3422 ** NSUBSTITUTE on list arguments gets the right answer with
3424 ** ELT signals an error of type TYPE-ERROR when the index argument
3425 is not a valid sequence index;
3426 ** LOOP signals (at macroexpansion time) an error of type
3427 PROGRAM-ERROR when duplicate variable names are found;
3428 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3429 ** FUNCALL of special-operators now cause an error of type
3431 ** PSETQ now works as required in the presence of side-effecting
3432 symbol-macro places;
3433 ** NCONC accepts any object as its last argument;
3434 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3436 ** loop-for-as-package does not require a package to be explicitely
3438 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3440 * fixed bug 166: compiler preserves "there is a way to go"
3441 invariant when deleting code.
3442 * fixed bug 172: macro lambda lists with required arguments after
3443 &REST arguments now cause an error to be signalled. (thanks to
3445 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3446 bug. (thanks to Matthew Danish)
3447 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3449 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3450 arguments are lexically rebound. (thanks to Gerd Moellmann and
3452 * fixed bug 194: error messages are now more informative when there
3453 is no primary method applicable in a call to a generic
3454 function. (thanks to Gerd Moellmann)
3455 * fixed bug in command line argument checking (thanks to Julian
3457 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3458 (thanks to Matthew Danish)
3459 * incremented fasl file version number, because of the
3460 SXHASH-related changes in the layout of CLOS data structures
3462 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3463 * minor incompatible change: The runtime (the Unix executable named
3464 "sbcl") is now much pickier about the .core files it will load.
3465 Essentially it now requires .core files to descend from the same
3466 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3467 as the runtime does. (The intent is to prevent the crashes which
3468 can occur, and which can even be reported as mysterious failures,
3469 when people patch the sources or change the build parameters
3470 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3471 sbcl and .core files.)
3472 * fixed bug: VALUES-LIST is no longer optimized away.
3473 * fixed bug 142: The FFI conversion of C string values to Lisp
3474 string values no longer conses excessively. (thanks to Nathan
3475 Froyd porting Raymond Toy's fix to CMU CL)
3476 * began to systematize and improve MOP conformance in PCL (thanks to
3477 Nathan Froyd, Gerd Moellman and Pierre Mai):
3478 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3480 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3481 ** FINALIZE-INHERITANCE is now called on class finalization;
3482 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3483 argument precedence order.
3484 * fixed bug 202: The compiler no longer fails on functions whose
3485 derived types contradict their declared type.
3486 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3487 so it can be non-toplevel.
3488 * The fasl file version number has changed (because of the new
3489 implementation of DEFMACRO).
3490 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3491 safe code, that any length requirement by their type-specifier
3492 argument is valid. The exceptions to this are described in bug
3494 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3495 functions now check the types of their inputs as required by ANSI.
3496 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3497 introduced symbol is DECLAREd to be SPECIAL.
3498 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3499 * fixed bug: :COUNT argument to sequence functions may be negative
3500 * fixed bug: body of DO-SYMBOLS may contain declarations
3501 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3502 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3503 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3504 symbol macro only once
3505 * fixed printing of call frame when argument list is unavailable
3506 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3507 * compiler no longer signals WARNING on unknown keyword
3510 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3511 * A beta-quality port to the mips architecture running Linux,
3512 based on the old CMUCL backend, has been made. It has been tested
3513 on a big-endian kernel, and works sufficiently well to be able to
3514 rebuild itself; it has not been tested in little-endian mode.
3515 * fixed an inconsistency between gencgc.c and purify.c which made
3516 dumping/loading .core files unreliable
3517 * fixed bug 120a: The compiler now deals correctly with IFs where
3518 the consequent is the same as the alternative, instead of
3519 misderiving the return type. (thanks to Alexey Dejneka)
3520 * fixed bug 113: Logical pathnames are now dumpable (the logical
3521 host is resolved at load-time, throwing an error if it is not
3523 * fixed bug 174: FORMAT's error message is slightly clearer when a
3524 non-printing character is used in a format directive.
3525 * fixed several bugs in compiler checking of type declarations, i.e.
3526 violations of the Python "declarations are assertions" principle
3527 (thanks to Alexey Dejneka)
3528 * fixed several bugs in PCL's error checking (thanks to Gerd
3530 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3532 * fixed bug in compilation of functions as first class values
3533 (thanks to Antonio Martinez-Shotton)
3534 * The compiler's handling TYPE-ERRORs which it can prove will
3535 inevitably happen at runtime has been cleaned up and corrected
3536 in several ways. (thanks to Alexey Dejneka)
3537 * improved argument type checking for various basic arithmetic
3538 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3539 lavished on them in the past that they can be compiled in many
3540 ways in different special cases
3541 * fixed bug 181: compiler checks validity of user supplied type
3543 * cleaned up code flushing in optimization: Function calls which
3544 should signal errors for safety purposes (e.g. which ANSI says
3545 should signal errors when their arguments are of incorrect type)
3546 are no longer optimized away.
3547 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3548 * incremented fasl file version number, because changes in the
3549 implementation of sequence functions like COERCE caused
3550 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3551 in old inline expansions) to become undefined. (Actually these
3552 changes were later undone, so we might very well be binary
3553 compatible with 0.7.7 after all, but leaving the version number
3554 incremented seemed like the simplest and most conservative
3557 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3558 * An alpha-quality port to the parisc architecture running Linux,
3559 based on the old CMUCL backend, has been made. This, even more so
3560 than the other backends, should be considered still a work in
3561 progress; known problems include that the Linux kernel in 64-bit
3562 mode does not propagate the correct sigcontext structure to
3563 userspace, and consequently SBCL on a parisc64 kernel will not
3565 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3566 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3567 suppressed.) Also now that the compiler is looking at declarations
3568 in the environment, it checks optimization declarations as well,
3569 and suppresses inlining when (> DEBUG SPEED).
3570 * More fixes have been made to treatment of floating point exception
3571 treatment and other Unix signals. In particular, floating point
3572 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3573 * The detection and handling of control stack exhaustion (infinite
3574 or very deeply nested recursion) has changed. Stack exhaustion
3575 detection is now done by write-protecting pages at the OS level
3576 and applies at all optimization settings; when found, a
3577 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3578 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3579 be caught using IGNORE-ERRORS.
3580 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3581 correctly when transforming recursive LABELS functions to LETs.
3582 (thanks to Alexey Dejneka)
3583 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3584 that are names of constants or global variables.
3585 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3586 alien routines with docstrings.
3587 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3588 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3590 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3591 (thanks to Eric Marsden)
3592 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3593 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3594 * Bug fix: The SPARC backend can now compile functions involving
3595 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3596 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3597 to a functions expecting a PATHNAME-DESIGNATOR.
3598 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3599 lambda lists. (thanks to Alexey Dejneka)
3600 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3601 functions so that EXPORTing the name of the class doesn't cause
3602 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3603 to Antonio Martinez for reporting this)
3604 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3605 constant-folded without failing an assertion. (thanks to Einar
3606 Floystad Dorum for reporting this)
3607 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3608 types are now dealt with more correctly by the compiler.
3609 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3610 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3611 behaviour analogous to RENAME-FILE. This puts its behaviour more
3612 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3614 * The fasl file version number has changed again. (because of the
3615 bug fix involving the names of PCL MAKE-INSTANCE functions)
3617 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3618 * bug fix: Floating point exceptions are treated much more
3619 consistently on the x86/Linux and PPC/Linux platforms.
3620 * Array initialization with :INITIAL-ELEMENT is now much faster for
3621 cases when the compiler cannot open code the array creation, but
3622 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3623 array accesses have also seen a speed increase.
3624 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3625 specified. (This was at the root of some bad interactions between
3626 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3628 * bug fix: Internal error arguments for undefined functions are now
3629 computed correctly on the PPC/Linux platform.
3630 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3631 Raymond Toy's patch for CMU CL)
3632 * Support for the Solaris 9 operating environment has been included
3633 (thanks to Daniel Merritt)
3634 * A very ugly but hopefully complete draft of the missing FFI chapter
3635 of the manual has been created by reformatting the corresponding
3636 CMU CL manual chapter into (currently very ugly and incoherent)
3637 DocBook and bringing it up to date for SBCL behavior. Thus, the
3638 manual is now essentially complete, at least by my extreme
3639 once-and-only-once standards, whereby it's acceptable to refer to
3640 the doc strings of SB-EXT functions as the primary documentation.
3641 * The fasl file version number has changed again, due to cleanup of
3642 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3643 is no longer a static symbol.)
3645 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3646 * SBCL now builds with OpenMCL (version 0.12) as the
3647 cross-compilation host; also, more progress has been made toward
3648 bootstrapping under CLISP.
3649 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3651 * bug 158 fixed: The compiler can now deal with integer loop
3652 increments different from 1; fixing this turned out also to fix
3654 * bug 169 fixed: no more bogus warnings about using lexical bindings
3655 despite the presence of perfectly good SPECIAL declarations (thanks
3656 to David Lichteblau)
3657 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3658 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3659 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3661 * bug 180 fixed: Method combination specifications no longer ignore
3662 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3663 * bug fix: Structure type predicate functions now check their argument
3664 count as they should.
3665 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3666 correctly. (thanks to Pierre Mai)
3667 * minor incompatible change: The --noprogrammer option is deprecated
3668 in favor of the new --disable-debugger option, which is very similar.
3669 (The major difference is that it takes effect at a slightly different
3670 time at startup, causing handling of errors in --sysinit and
3671 --userinit files will be affected differently.) The
3672 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3673 been added to allow this functionality to be controlled from ordinary
3674 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3675 maintainers, who might want to run non-interactive scripts to
3676 build SBCL cores which will later be used interactively.)
3677 * minor incompatible change: The LOAD function no longer, when given
3678 a wild pathname to load, loads all files matching that pathname.
3679 Instead, an error of type FILE-ERROR is signalled.
3681 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3682 * bug 147 fixed: The compiler preserves its block link/count
3683 invariants more correctly now so that it doesn't crash. (thanks
3685 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3687 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3688 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3689 no-value-for-this-character as NIL (as specified by ANSI).
3690 * HOST-NAMESTRING on physical pathnames now returns a string that is
3691 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3692 (thanks to Christophe Rhodes)
3693 * The Alpha port handles icache flushing more correctly. (thanks to
3695 * More progress has been made toward bootstrapping under CLISP. (thanks
3696 to Christophe Rhodes)
3697 * The fasl file format has changed again, because dynamic loading
3698 on OpenBSD (which has non-ELF object files) motivated some cleanups
3699 in the way that foreign symbols are transformed and passed around.
3700 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3701 is no longer treated as whitespace by the reader, but instead as
3702 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3703 |A
\7fB|, instead of A as it used to.
3705 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3706 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3707 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3708 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3709 this as of 0.7.1.45, but was left out of the previous news section
3710 (thanks to Dan Barlow)
3711 * SBCL now runs on the Solaris operating system on SPARC architectures
3712 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3713 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3714 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3715 * SPARC backend cleanups, allowing builds of cores optimized for V8
3716 and V9 SPARCS, and also emission of code targeted to a particular
3717 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3719 * SBCL is closer to bootstrapping under CLISP, thanks to various
3720 fixes by Christophe Rhodes.
3721 * The fasl file format has changed again, to allow the compiler's
3722 INFO database to support symbol macros.
3723 * The user manual (in doc/) is formatted into HTML more nicely.
3724 (thanks to coreythomas)
3725 * The system is smarter about SUBTYPEP relationships, especially
3726 those involving NOT types (including types such as ATOM which are
3727 represented internally using NOT types). Thus SUBTYPEP is less
3728 likely to return (VALUES NIL NIL) in general, and in particular
3729 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3731 * The fasl file format has changed again, because the internal
3732 representation of types now includes a new slot to support the new
3733 SUBTYPEP-of-NOT-types logic.
3734 * (not a change in the main branch of SBCL, but a related prototype
3735 which can hopefully be merged into the main branch of SBCL in the
3736 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3737 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3738 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3739 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3740 dispatch macro character combinations. (thanks to Alexey Dejneka)
3741 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3742 that superficially look like logical namestrings correctly.
3743 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3745 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3746 default supplied-p) for &optional and &key arguments. (thanks to
3749 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3750 * incompatible change: The compiler is now less aggressive about
3751 tail call optimization, doing it only when (> SPACE DEBUG) or
3752 (> SPEED DEBUG). (This is an incompatible change because there are
3753 programs which relied on the old CMU-CL-style behavior to optimize
3754 away their unbounded recursion which will now die of stack overflow.)
3755 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3756 for non-GENCGC systems has been increased to 20M (since that
3757 seems much closer to the likely performance optimum for modern
3758 systems than the old 4M value was)
3759 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3760 and *DEBUG-PRINT-LEVEL*
3761 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3762 of CMU CL's support for SPARC, and various endianness and other
3763 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3764 * new syntactic sugar for the Unix command line: --load foo.bar is now
3765 an alternate notation for --eval '(load "foo.bar")'.
3767 ** The system now detects stack overflow and handles it gracefully,
3768 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3769 optimization settings. (This is a good thing in general, and
3770 its introduction in this version should be particularly timely
3771 for anyone whose code fails because of suppression of tail
3773 ** The system now hunts for the C variable "environ" in a more
3774 devious way, to avoid segfaults when the C library version differs
3775 between compile time and run time. (thanks to Christophe Rhodes)
3776 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3777 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3778 ** The compiler no longer issues bogus style warnings for undefined
3779 classes in the same source file as the DEFCLASSes which defined
3780 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3782 ** fixes in CONDITION class precedence list for undefined function
3783 errors (thanks to Alexei Dejneka)
3784 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3785 correctly. (thanks to Dan Barlow)
3786 ** portability fixes aiming at bootstrapping under CLISP (thanks
3787 to Dave McDonald and Christophe Rhodes)
3788 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3789 ** fixes in type translation and and type inference (thanks to
3791 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3792 ** various fixes in the new ports (thanks to Dan Barlow)
3793 * several changes related to debugging:
3794 ** suppression of tail recursion, as noted above
3795 ** stack overflow detection, as noted above
3796 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3797 is now the default. (For some time encapsulation has been more
3798 reliable than the breakpoint-based :ENCAPSULATE NIL
3799 implementation, at least on X86 systems; and I just noticed that
3800 encapsulation also seems closer to the spirit of the ANSI
3803 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3805 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3806 up properly again. (There was a packaging bug in 0.7.0 which
3807 left their definitions in SB-SYS::LOAD-FOREIGN and
3808 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3809 vital for most things which interface to C-level interfaces,
3810 like extensions working with sockets or databases or
3811 Perl-compatible regexes or whatever, and the need to fix
3812 this bug is the main reason that 0.7.1 was released so
3814 ** DEFGENERIC is now choosier about the methods it redefines, so that
3815 reLOADing a previously-LOADed file containing DEFGENERICs does
3816 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3817 cycle now works as it should. (thanks to Alexey Dejneka)
3818 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3819 fixed by Christophe Rhodes. (He actually submitted this patch
3820 months ago, and I delayed until after 0.7.0.)
3821 ** Bug 111 (internal compiler confusion about runtime checks on
3822 FUNCTION types) was fixed by Alexey Dejneka.
3823 * Some internal cleanups (getting rid of variables which aren't
3824 needed now that the byte interpreter is gone) caused the fasl
3825 file format number to change again.
3827 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3828 * major incompatible change: The default fasl file extension, i.e. the
3829 default extension for files produced by COMPILE-FILE, has changed
3830 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3832 ** There are many changes in the implementation of the compiler.
3833 SBCL is now essentially a compiler-only implementation of ANSI
3834 Common Lisp. EVAL still "interprets" a few special cases, but
3835 almost all the interesting cases are handled by creating
3836 a LAMBDA expression, calling COMPILE on it, then calling
3837 FUNCALL on the result.
3838 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3839 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3840 Since the code is newer, there might still be some new bugs
3841 (though not as many as before Martin Atzmueller's fixes:-). But
3842 the new code is substantially simpler and clearer, and hopefully
3843 any remaining bugs will be simpler, less fundamental, and more
3844 fixable then the bugs in the old code.
3845 ** The revised compiler is still a little unsteady on its feet.
3847 *** The debugging information it produces (particularly the names
3848 of FUNCTION objects) is sometimes much less useful than what
3849 the old compiler produced.
3850 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3851 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3852 within a MACROLET) has been temporarily weakened.
3853 ** There are new compiler optimizations for various functions:
3854 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3855 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3856 *** the math functions TRUNCATE, FLOOR, and CEILING
3857 *** the function-of-all-trades COERCE
3858 Mostly these should be transparent, but there's one
3859 potentially-annoying problem (bug 117): when the compiler
3860 inline-expands a function and does type analysis on the result,
3861 it can create control paths which have type mismatches, and
3862 when it can't prove that those control paths aren't taken,
3863 it will issue WARNINGs about the type mismatches. This is
3864 a particular problem in practice for the new sequence functions.
3865 It's not clear how this should be fixed, and for now, a
3866 workaround is given in the entry for 117 in the BUGS file.
3867 ** (Because of the interaction between the two previous items --
3868 occasional inlining problems and new inline expansions -- some
3869 of the new sequence function optimizations won't really kick in
3870 completely until debugging information, and then inlining, are
3871 straightened out in some future version.)
3872 * minor incompatible changes:
3873 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3874 DIRECTORY no longer implicitly promotes NIL slots of its
3875 pathname argument to :WILD. In particular, when you ask for the
3876 contents of a directory (which you used to be able to do without
3877 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3878 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3879 ** changes in behavior that ANSI explicitly defines to be
3880 implementation dependent:
3881 *** The new compiler-only implementation still conforms with ANSI,
3882 but acts a little different than before. Besides the obvious
3883 changes in performance tradeoffs (that the cost per form passed
3884 to EVAL has gone up, and the cost per form executed by EVAL
3885 has gone down), the behavior of the system changes a little
3886 because there are no longer any interpreted function objects.
3887 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3888 e.g. doing COMPILE on the output of interactive DEFUN is
3890 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3892 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3893 has changed from (:CL) to NIL.
3894 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3895 e.g. #\Newline and #\Tab, don't have names specified in the
3896 ANSI Common Lisp standard, is now based on their ASCII symbolic
3897 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3898 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3899 are no longer used for output.
3900 ** changes in internal implementation constants:
3901 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3902 4 million. (If your application spends a lot of time GCing and
3903 you have a lot of RAM, you might want to experiment with
3904 increasing it even more.)
3905 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3906 However, almost all old code should still continue to work without
3907 immediate update, as SB-C-CALL is now a (deprecated) nickname
3909 ** Old operator names in the style DEF-FOO are now deprecated in
3910 favor of new corresponding names DEFINE-FOO, for consistency with
3911 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3912 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3913 internal symbols, but a few supported extensions like
3914 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3915 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3916 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3917 etc. as you get deeper into recursive calls to the debugger
3918 command loop, instead of the old "5]", "5]]", "5]]]"
3919 sequence. (I was motivated to do this when squabbles between
3920 ILISP and SBCL left me very deeply nested in the debugger. In the
3921 short term, this change will probably provoke more ILISP/SBCL
3922 squabbles, but hopefully it will be an improvement in the long run.)
3923 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3924 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3925 function, since declaiming return types involving aliens is
3926 (1) annoyingly messy to do by hand and (2) vital to efficient
3927 compilation of code which calls such functions.
3928 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3929 longer reexported by the SB-EXT package. They're solely useful
3930 for alien code, so it seems more logical that you should get
3931 them from the SB-ALIEN package, not in SB-EXT.
3932 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3933 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3934 features. Instead, the code that they used to control is always
3935 built into the system.
3936 * many other bug fixes
3937 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3938 advantage of the new EVAL-WHEN stuff and to clean them up in
3939 general, and they are now more ANSI-compliant in a number of
3940 ways. Martin Atzmueller is responsible for a lot of this.
3941 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3943 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3944 *** correct ERROR type for various file operations
3945 *** some fixes for Lisp streams
3946 *** DEFMETHOD syntax checking
3947 *** changing old weird representation of debug information as
3948 strings (which, among their other deficiencies, don't transform
3949 correctly when you rename packages, and don't change their
3950 print representation when you change things like *PACKAGE*
3951 and *PRINT-LENGTH*) to symbols and lists of symbols
3952 He also made several improvements and fixed several bugs in DESCRIBE.
3953 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3955 *** misbehavior of WRITE-STRING/WRITE-LINE
3956 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3957 and several other LOOP problems as well
3958 *** DIRECTORY when similar filenames are present
3959 *** DEFGENERIC with :METHOD options
3960 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3961 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3962 *** argument ordering in FIND with :TEST option
3963 *** mishandled package designator argument in APROPOS-LIST
3964 *** various problems in the backquote readmacro
3965 *** a bug in APROPOS
3966 *** probably some others that I'm not describing very well here,
3967 since the CVS log documents them by reference to sbcl-devel
3968 messages, and the SourceForge archives aren't working well.:-(
3969 ** Dan Barlow improved the Alpha port (and is making progress on the
3970 PPC port, for those of you who think different).
3971 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3972 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3973 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3974 continues to work on the SPARC port (for those of you in a position
3975 to look down upon our little PC-compatible boxes from a great height).
3976 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3977 and uses that copy, rather than the current dynamic value, when
3978 it's trying to decide whether to truncate output. Thus e.g.
3979 (let ((*print-lines* 50))
3980 (pprint-logical-block (stream nil)
3982 (let ((*print-lines* 8))
3983 (print (aref possiblybigthings i) stream)))))
3984 should now truncate the logical block only at 50 lines, instead of
3985 often truncating it at 8 lines, as it did before.
3986 * The doc/cmucl/ directory, containing old CMU CL documentation
3987 from the time of the fork, is no longer part of the base system.
3988 SourceForge has shut down its anonymous FTP service, and with it
3989 my original plan for distributing the old CMU CL documentation
3990 there. For now, if you need these files you can download an old
3991 SBCL source release and extract them from it.
3992 * The fasl file version number changed again, for dozens of reasons,
3993 some of which are apparent above.
3995 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3996 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3997 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3998 patch, so that DISASSEMBLE output is much nicer.
3999 * The code in the SB-PROFILE package now seems reasonably stable.
4000 I still haven't decided what the final interface should look like
4001 (I'd like PROFILE to interact cleanly with TRACE, since both
4002 facilities use function encapsulation) but if you have a need
4003 for profiling now, you can probably use it successfully with
4004 the current CMU-CL-style interface.
4005 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4006 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4007 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4008 link now returns the dangling link itself, and for similar
4009 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4010 link itself. (In these cases the old code signalled an error and
4011 looped endlessly, respectively.) Thus, DIRECTORY now works even
4012 in the presence of dangling and cyclic symbolic links.
4013 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4014 is now a supported extension again, since the consensus on
4015 sbcl-devel was that it can be useful for ordinary development
4016 work, not just for debugging SBCL itself.
4017 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4018 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4019 that currently-defined functions might be redefined later with
4020 different return types.
4021 * Hash tables can be printed readably, as inspired by CMU CL code
4022 of Eric Marsden and SBCL code of Martin Atzmueller.
4023 * better error handling in CLOS method combination, thanks to
4024 Martin Atzmueller porting Pierre Mai's CMU CL patches
4025 * more overflow fixes for >16Mbyte I/O buffers
4026 * A bug in READ has been fixed, so that now a single Ctrl-D
4027 character suffices to cause end-of-file on character streams.
4028 In particular, now you only need one Ctrl-D at the command
4029 line (not two) to exit SBCL.
4030 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4032 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4033 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4034 does the right thing.
4035 * The implementation of some type tests, especially for CONDITION
4036 types, is now tidier and maybe faster, due to CMU CL code
4037 originally by Douglas Crosher, ported by Martin Atzmueller.
4038 * Some math functions have been fixed, and there are new
4039 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4040 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4041 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4042 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4043 likely to be built into sbcl-0.7.0.)
4044 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4045 CL, ported to SBCL by Martin Atzmueller)
4046 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4047 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4048 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4049 * There's a new slam.sh hack to shorten the edit/compile/debug
4050 cycle for low-level changes to SBCL itself, and a new
4051 :SB-AFTER-XC-CORE target feature to control the generation of
4052 the after-xc.core file needed by slam.sh.
4053 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4054 COMPILE-FILE is no longer supported, so that now every function
4055 gets an entry point, so that block compilation looks a little
4056 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4057 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4058 returns the number of bytes consed since the system started,
4059 rather than the number consed since the first time the function
4060 was called. (The new definition parallels ANSI functions like
4061 CL:GET-INTERNAL-RUN-TIME.)
4062 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4063 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4064 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4065 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4066 * Fasl file version numbers are now independent of the target CPU,
4067 since historically most system changes which required version
4068 number changes have affected all CPUs equally. Similarly,
4069 the byte fasl file version is now equal to the ordinary
4072 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4073 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4074 is no longer recognized. I apologize for this, because it was
4075 listed in SB-EXT as a supported extension, but I found that
4076 its existing behavior was poorly specified, as well as incorrectly
4077 specified, and it looked like too much of a mess to straighten it
4078 out. I have enough on my hands trying to get ANSI stuff to work..
4079 * many patches ported from CMU CL by Martin Atzmueller, with
4080 half a dozen bug fixes in pretty-printing and the debugger, and
4081 half a dozen others elsewhere
4082 * fixed bug 13: Floating point infinities are now supported again.
4083 They might still be a little bit flaky, but thanks to bug reports
4084 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4085 as flaky as they were.
4086 * The --noprogrammer command line option is now supported. (Its
4087 behavior is slightly different in detail from what the old man
4088 page claimed it would do, but it's still appropriate under the
4089 same circumstances that the man page talks about.)
4090 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4091 are now supported, and enabled by default. Thus, the compiler can
4092 handle many floating point and complex operations much less
4093 inefficiently. (Thus e.g. you can implement a complex FFT
4095 * The compiler now detects type mismatches between DECLAIM FTYPE
4096 and DEFUN better, and implements CHECK-TYPE more correctly, and
4097 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4098 thanks to patches from Martin Atzmueller.
4099 * various fixes to make the cross-compiler more portable to
4100 ANSI-conforming-but-different cross-compilation hosts (notably
4101 Lispworks for Windows, following bug reports from Arthur Lemmens)
4102 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4103 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4104 fixed thanks to Pierre Mai's CMU CL patches.
4105 * fixes to keep the system from overflowing internal counters when
4106 it tries to use i/o buffers larger than 16M bytes
4107 * fixed bug 45a: Various internal functions required to support
4108 complex special functions have been merged from CMU CL sources.
4109 (When I was first setting up SBCL, I misunderstood a compile-time
4110 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4111 * improved support for type intersection and union, fixing bug 12
4112 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4113 more obscure bugs as well
4114 * some steps toward byte-compiling non-performance-critical
4115 parts of the system, courtesy of patches from Martin Atzmueller
4116 * Christophe Rhodes has made some debian packages of sbcl at
4117 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4118 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4119 stable, but are nonetheless usable. When he's ready, I'd be happy
4120 to add them to the SourceForge "File Releases" section. (And if
4121 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4122 * new fasl file format version number (because of changes in
4123 internal representation of (OR ..) types to accommodate the new
4124 support for (AND ..) types, among other things)
4126 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4127 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4129 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4131 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4132 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4133 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4134 * DESCRIBE now works on structure objects again.
4135 * Most function call argument type mismatches are now handled as
4136 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4137 know whether the function will be redefined before the call is
4138 executed. (The compiler could flag local calls with full WARNINGs,
4139 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4140 it doesn't keep track of enough information to know whether calls
4141 are local in this sense.)
4142 * Compiler output is now more verbose, with messages truncated
4143 later than before. (There should be some supported way for users
4144 to override the default verbosity, but I haven't decided how to
4145 provide it yet, so this behavior is still controlled by the internal
4146 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4147 src/compiler/ir1util.lisp.)
4148 * Fasl file format version numbers have increased again, because
4149 support for the Gray streams extension changes the layout of the
4150 system's STREAM objects.
4151 * The Gray subclassable streams extension now works, thanks to a
4152 patch from Martin Atzmueller.
4153 * The full LOAD-FOREIGN extension (not just the primitive
4154 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4155 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4156 but like most other programs, it defaults to copying the Unix
4157 environment from the original process instead of starting the
4158 new process in an empty environment.
4159 * Extensions which manipulate the Unix environment now support
4160 an :ENVIRONMENT keyword option which doesn't smash case or
4161 do other bad things. The CMU-CL-style :ENV option is retained
4162 for porting convenience.
4163 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4164 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4166 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4168 * A patch from Martin Atzmueller seems to have solved the SIGINT
4169 problem, and as far as we know, signal-handling now works cleanly.
4170 (If you find any new bugs, please report them!)
4171 * The system no longer defaults Lisp source file names to types
4172 ".l", ".cl", or ".lsp", but only to ".lisp".
4173 * The compiler no longer uses special default file extensions for
4174 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4175 seems to expect a single default extension for all compiled code,
4176 and there's no compelling reason to try to stretch the standard
4177 to allow two different extensions.) Instead, byte-compiled files
4178 default to the same extension as native-compiled files.
4179 * Fasl file format version numbers have increased again, because
4180 a rearrangement of internal implementation packages made some
4181 dumped symbols in old fasl files unreadable in new cores.
4182 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4183 many fewer weird special cases.
4184 * Bug #17 (differing COMPILE-FILE behavior between logical and
4185 physical pathnames) has been fixed, and some related misbehavior too,
4186 thanks to a patch from Martin Atzmueller.
4187 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4188 by Tim Moore, ported to SBCL by Martin Atzmueller.
4189 * Martin Atzmueller fixed several filesystem-related problems,
4190 including bug #36, in part by porting CMU CL patches, which were
4191 written in part by Paul Werkowski.
4192 * More compiler warnings in src/runtime/ are gone, thanks to
4193 more patches from Martin Atzmueller.
4194 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4197 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4199 * DESCRIBE now works on CONDITION objects.
4200 * The debugger now handles errors which arise when trying to print
4201 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4203 * The build system now uses an additional file, customize-target-features.lisp,
4204 to allow local modifications to the target *FEATURES* list. (The point of
4205 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4206 debugging features enabled, without having to worry about propagating it
4207 into everyone's system when I do a "cvs update".) When no
4208 customize-target-features.lisp file exists, the target *FEATURES* list
4209 should be constructed the same way as before.
4210 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4211 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4212 warn about behavior which is undefined under the ANSI spec. Note: This
4213 is specified by ANSI, but it's not very popular with programmers.
4214 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4215 macro in the SBCL sources for an example of a workaround which you
4216 might use to make portable ANSI-standard code which does what you want.
4217 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4218 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4219 when it's used as a non-toplevel form. (This is required in order
4220 to implement the DEFCONSTANT-EQX macro.)
4221 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4222 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4223 funny things, due to bugs in the implementation of EVAL-WHEN
4224 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4225 non-toplevel DEFCONSTANTs are uncommon.)
4226 * The core file version number and fasl file version number have been
4227 incremented, because the old noncompliant DEFCONSTANT behavior involved
4228 calling functions which no longer exist, and because I also took the
4229 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4230 * fixed bug 1 (error handling before read-eval-print loop starts), and
4231 redid debugger restarts and related debugger commands somewhat while
4233 ** The QUIT debugger command is gone, since it did something
4234 rather different than the SB-EXT:QUIT command, and since it never
4235 worked properly outside the main toplevel read/eval/print loop.
4236 Invoking the new TOPLEVEL restart provides the same functionality.
4237 ** The GO debugger command is also gone, since you can just invoke
4238 the CONTINUE restart directly instead.
4239 ** The TOP debugger command is also gone, since it's redundant with the
4240 FRAME 0 command, and since it interfered with abbreviations for the
4242 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4244 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4245 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4246 some porting work by Martin Atzmueller)
4247 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4248 with a boatload of patches which he ported to SBCL. Now that those
4250 ** The system tries to make sure that its low-priority messages
4251 are prefixed by semicolons, to help people who like to use
4252 syntax highlighting in their ILISP buffer. (This patch
4253 was originally due to Raymond Toy.)
4254 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4255 patches originally written by Raymond Toy.
4256 ** The compiler understands coercion between single-value and
4257 multiple-VALUES type expressions better, getting rid of some very
4258 weird behavior, thanks to patches originally by Robert MacLachlan
4259 and Douglas Crosher.
4260 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4261 lambda lists, thanks to a patch originally by Pierre Mai.
4262 ** The system no longer bogusly warns about "abbreviated type
4264 ** The compiler gets less confused by inlining and RETURN-FROM,
4265 thanks to some patches originally by Tim Moore.
4266 ** The system no longer hangs when dumping circular lists to fasl
4267 files, thanks to a patch originally from Douglas Crosher.
4268 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4269 undefined function error.
4270 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4271 in structure slot initforms) for now, documented workaround instead:-|
4272 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4273 * fixed bug 5: added stubs for various Gray stream functions called
4274 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4275 installed, at least appropriate type errors are generated
4276 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4277 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4278 consistently in DEFMETHOD forms.
4279 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4280 it doesn't seem to affect SBCL after all
4281 * The C runtime system now builds with better optimization and many
4282 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4284 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4286 * The system is now under CVS at SourceForge (instead of the
4287 CVS repository on my home machine).
4288 * The new signal handling code has been tweaked to treat register
4289 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4290 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4291 downstream, e.g. in debug-int.lisp, has implicit dependencies
4292 on the unsignedness of integer representation of machine words,
4293 and that caused the system to bomb out with infinite regress
4294 when trying to recover from type errors involving signed values,
4295 e.g. (BUTLAST '(1 2 3) -1).)
4296 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4297 (This was one of the bugs Peter Van Eynde reported back in July.)
4298 * The system now uses code inspired by Colin Walters' O(N)
4299 implementation of MAP (from the cmucl-imp@cons.org mailing
4300 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4301 inline the MAP operation, and there is more than one
4302 sequence argument to the MAP call (so that it can't just
4303 do ETYPECASE once and for all based on the type of the
4304 single sequence argument). (The old non-inline implementation
4305 of the general M-argument sequence-of-length-N case required
4306 O(M*N*N) time when any of the sequence arguments were LISTs.)
4307 * The QUIT :UNIX-CODE keyword argument has been renamed to
4308 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4309 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4311 * Raymond Wiker's patches to port dynamic loading from Linux to
4312 FreeBSD have been added.
4313 * The BUGS file is now more nearly up to date, thanks in large part
4314 to Martin Atzmueller's review of it.
4315 * The debugger now flushes standard output streams before it begins
4316 its output ("debugger invoked" and so forth).
4317 * The core version number and fasl file version number have both
4318 been incremented, because of incompatible changes in the layout
4320 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4321 used to not be called for a saved Lisp image.)
4322 * Martin Atzmueller's version of a patch to fix a compiler crash,
4323 as posted on sbcl-devel 13 September 2000, has been installed.
4324 * Instead of installing Martin Atzmueller's patch for the
4325 compiler transform for SUBSEQ, I deleted the compiler transform,
4326 and transforms for some similar consing operations.
4327 * A bug in signal handling which kept TRACE from working on OpenBSD
4329 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4330 away by constant folding
4331 * The system now defines its address space constants in one place
4332 (in the Lisp sources), and propagates them automatically elsewhere
4333 (through GENESIS and the sbcl.h file). Therefore, patching the
4334 address map is less unnecessarily tedious and error-prone. The
4335 Lisp names of address space constants have also been systematized.
4336 * CVS tags like dollar-Header-dollar have been removed from
4337 the sources, because they have never saved me trouble and
4338 they've been source of trouble working with patches and other
4339 diff-related operations.
4340 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4341 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4343 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4345 * The system has been ported to OpenBSD.
4346 * The system now compiles with a simple "sh make.sh" on the systems
4347 that it's supported on. I.e., now you no longer need to tweak
4348 text in the target-features.lisp-expr and symlinks in src/runtime/
4349 by hand, the make.sh takes care of it for you.
4350 * The system is no longer so grossly inefficient when compiling code
4351 involving vectors implemented as general (not simple) vectors (VECTOR T),
4352 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4353 which dares to use the various sequence functions on non-simple
4354 vectors, takes less of a performance hit.
4355 * There is now a primitive type predicate VECTOR-T-P
4356 to test for the (VECTOR T) type, so that e.g.
4357 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4358 can now be compiled with some semblance of efficiency. (The old code
4359 turned the type declaration into a full call to %TYPEP at runtime!)
4360 * AREF on (VECTOR T) is still not fast, since it's still compiled
4361 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4362 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4363 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4364 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4365 without making full calls to SUBTYPEP at runtime.
4366 (Some analogous efficiency issues for non-simple vectors specialized to
4367 element types other than T, or for non-simple multidimensional arrays,
4368 have not been addressed. They could almost certainly be handled the
4369 same way if anyone is motivated to do so.)
4370 * The changes in array handling break binary compatibility, so
4371 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4372 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4373 instead of (VALUES T T).
4374 * By following the instructions that Dan Barlow posted to sbcl-devel
4375 on 2 July 2000, I was able to enable primitive dynamic object
4376 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4377 functionality is not implemented (since it calls ld to resolve
4378 library references automatically, requiring RUN-PROGRAM for its
4379 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4380 try to resolve library references) is now supported.
4381 * The system now flushes the standard output streams when it terminates,
4382 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4383 them at several other probably-convenient times, e.g. in each pass of
4384 the toplevel read-eval-print loop, and after evaluating a form given
4385 as an "--eval" command-line option. (These changes were motivated by a
4386 discussion of stream flushing issues on cmucl-imp in August 2000.)
4387 * The source transform for TYPEP of array types no longer assumes
4388 that an array whose element type is a not-yet-defined type
4389 is implemented as an array of T, but instead punts, so that the
4390 type will be interpreted at runtime.
4391 * There is now some support for cross-compiling in make.sh: each of
4392 the phases of make.sh has its own script. (This should be transparent
4393 to people doing ordinary, non-cross-compile builds.)
4394 * Since my laptop doesn't have hundreds of megabytes of memory like
4395 my desktop machine, I became more motivated to do some items on
4396 my to-do list in order to reduce the size of the system a little:
4397 ** Arrange for various needed-only-at-cold-init things to be
4398 uninterned after cold init. To support this, those things have
4399 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4400 symbols with such names are now uninterned after cold init).
4401 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4402 fasl files for cold load.
4403 ** Remove the old compiler structure pooling code (which used to
4404 be conditional on the target feature :SB-ALLOC) completely.
4405 ** Redo the representation of some data in cold init to be more compact.
4406 (I also looked into supporting byte compiled code at bootstrap time,
4407 which would probably reduce the size of the system a lot, but that
4408 looked too complicated, so I punted for now.)
4409 * The maximum signal nesting depth in the src/runtime/ support code has
4410 been reduced from 4096 to 256. (I don't know any reason for the very
4411 large old value. If the new smaller value turns out to break something,
4412 I'll probably just bump it back up.)
4413 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4415 * Many, many bugs reported by Peter Van Eynde have been added to
4416 the BUGS list; some have even been fixed.
4417 * While enabling dynamic object file loading, I tried to make the
4418 code easier to understand, renaming various functions and variables
4419 with less ambiguous names, and changing some function calling
4420 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4421 * While trying to figure out how to do the OpenBSD port, I tried to
4422 clean up some of the code in src/runtime/. In particular, I dropped
4423 support for non-POSIX signal handling, added various comments,
4424 tweaked the code to reduce the number of compilation warnings, and
4425 renamed some files to increase consistency.
4426 * To support the new automatic configuration functionality in make.sh,
4427 the source file target-features.lisp-expr has been replaced with the
4428 source file base-target-features.lisp-expr and the machine-generated
4429 file local-target-features.lisp-expr.
4430 * fixed a stupid quoting error in make.sh so that using CMU CL
4431 "lisp -batch" as cross-compilation host works again
4433 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4435 * DESCRIBE no longer tries to call itself recursively to describe
4436 bound/fbound values, so that it no longer fails on symbols which are
4437 bound to themselves (like keywords, T, and NIL).
4438 * DESCRIBE now works on generic functions.
4439 * The printer now prints less-screwed-up representations of closures
4440 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4441 * A private symbol is used instead of the :EMPTY keyword previously
4442 used to mark empty slots in hash tables. Thus
4443 (DEFVAR *HT* (MAKE-HASH-TABLE))
4444 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4445 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4446 now does what ANSI says that it should. (You can still get
4447 similar noncompliant behavior if bang on the hash table
4448 implementation with all the symbols you get back from
4449 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4450 This breaks binary compatibility, since tests for equality to
4451 :EMPTY are wired into things like the macroexpansion of
4452 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4454 * There's now a minimal placeholder implementation for CL:STEP,
4455 as required by ANSI.
4456 * An obscure bug in the interaction of the normal compiler, the byte
4457 compiler, inlining, and structure predicates has been patched
4458 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4459 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4461 * Missing ordinary arguments in a macro call are now detected even
4462 when the macro lambda list contains &KEY or &REST.
4463 * The debugger no longer complains about encountering the top of the
4464 stack when you type "FRAME 0" to explicitly instruct it to go to
4465 the top of the stack. And it now prints the frame you request even
4466 if it's the current frame (instead of saying "You are here.").
4467 * As specified by ANSI, the system now always prints keywords
4468 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4470 * The default initial SIZE of HASH-TABLEs is now smaller.
4471 * Type information from CLOS class dispatch is now propagated
4472 into DEFMETHOD bodies, so that e.g.
4473 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4475 is now basically equivalent to
4476 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4477 (DECLARE (TYPE SINGLE-FLOAT X))
4479 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4480 operation, without having to do run-time type dispatch.
4481 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4482 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4483 * Since I don't seem to be making big file reorganizations very often
4484 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4485 is overflowing my ability to conveniently back them up), I've finally
4486 checked the system into CVS. (The CVS repository is on my home system,
4487 not at SourceForge -- putting it on SourceForge might come later.)
4488 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4489 high-level GC-NOTIFY-FOO functions send their output. (There's
4490 still very little control of where low-level verbose GC functions
4491 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4492 less than it used to -- the GC-NOTIFY-FOO functions are now under
4493 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4494 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4495 in only one place in the source code, and propagates it automatically
4496 everywhere that it's needed. Thus e.g. when I bump the version from
4497 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4498 * The C source files now include boilerplate legalese and documentation
4499 at the head of each file (just as the Lisp source files already did).
4500 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4501 to his page will be replaced with a link to his new CLiki service.
4503 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4505 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4506 * The build process now looks for GNU make under the default name "gmake",
4507 instead of "make" as it used to. If GNU make is not available as "gmake"
4508 on your system, you can change this default behavior by setting the
4509 GNUMAKE environment variable.
4510 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4511 can build without error under CMU CL.
4513 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4515 * There is now a partial SBCL user manual (with some new text and some
4516 text cribbed from the CMU CL manual).
4517 * The beginnings of a profiler have been added (starting with the
4518 CMU CL profiler and simplifying and cleaning up). Eventually the
4519 main interface should be through the TRACE macro, but for now,
4520 it's still accessed through vaguely CMU-CL-style functions and macros
4521 exported from the package SB-PROFILE.
4522 * Some problems left over from porting CMU CL to the new
4523 cross-compilation bootstrap process have been cleaned up:
4524 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4525 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4526 part when getting rid of the compiler *BACKEND* stuff.)
4527 ** The value of *NULL-TYPE* was screwed up, because it was
4528 being initialized before the type system knew the final
4529 definition of the 'NULL type. This screwed up several key
4530 optimizations in the compiler, causing inefficiency in all sorts
4531 of places. (I found it because I wanted to understand why
4532 GET-INTERNAL-RUN-TIME was consing.)
4533 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4534 PROCLAIM FTYPE information. Unfortunately this broke binary
4535 compatibility again, since now the forms output by DEFGENERIC
4536 to refer to functions which didn't exist in 0.6.3.
4537 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4538 can use the new (as of 0.6.3) transform for SOME into MAP into
4540 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4541 values to (INTEGER 0 1000000), so that the time code will no longer
4542 occasionally get blown up by Linux returning 1000000 microseconds
4543 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4544 its output conform to the ANSI spec. (Alas, this makes its output
4545 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4546 * A full call to MAP NIL with a single sequence argument no longer conses.
4547 * fixes to problems pointed out by Martin Atzmueller:
4548 * The manual page no longer talks about multiprocessing as though
4549 it were currently supported.
4550 * The ILISP support patches have been removed from the distribution,
4551 because as of version 5.10.1, ILISP now supports SBCL without us
4552 having to maintain patches.
4553 * added a modified version of Raymond Toy's recent CMU CL patch for
4554 EQUALP comparison of HASH-TABLE
4556 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4558 * The system still can't cross-compile itself with
4559 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4560 entails), but at least it can compile more of itself that way
4561 than it used to be able to, and various buglets which were uncovered
4562 by trying to cross-compile itself that way have now been fixed.
4563 * This release breaks binary compatibility again. This time
4564 at least I've incremented the FASL file format version to 2, so that the
4565 problem can be detected reliably instead of just causing weird errors.
4566 * various new style warnings:
4567 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4568 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4569 ** using the lexical binding of a variable named in the *FOO* style
4570 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4571 as specified by ANSI.
4572 * *RANDOM-STATE* is no longer automatically initialized from
4573 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4574 default behavior of the system is to repeat its behavior every time
4575 it's run. If you'd like to change this behavior, you can always
4576 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4577 old convention there was no comparably easy way to get the system to
4578 repeat its behavior every time it was run.
4579 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4580 FTYPE declarations has been removed, in favor of their ANSI
4581 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4582 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4583 the types of their sequence arguments can be determined at compile time.
4584 This is done through a new open code expansion for MAP which eliminates
4585 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4586 argument types can be determined at compile time.
4587 * The optimizer now transforms COERCE into an identity operation when it
4588 can prove that the coerced object is already of the correct type. (This
4589 can be a win for machine generated code, including the output of other
4590 optimization transforms, such as the MAP transform above.)
4591 * Credit information has been moved from source file headers into CREDITS.
4592 * Source file headers have been made more standard.
4593 * The CASE macro now compiles without complaining even when it has
4596 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4598 * (Note that the way that the PCL macroexpansions were rewritten
4599 to accommodate the change in DEFGENERIC below breaks binary
4600 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4601 not run under sbcl-0.6.2. Once we get out of alpha releases,
4602 i.e. hit release 1.0.0, we'll probably try to maintain binary
4603 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4604 and sbcl-1.4.4. Until then, however, it might be fairly common
4605 for maintenance releases to break binary compatibility.)
4606 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4608 * The reporting of unhandled signals has been changed to print some
4609 explanatory text as well as the report form. (Previously only
4610 the report form was printed.)
4611 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4612 it defines, so that the compiler no longer issues undefined function
4613 warnings for compiled-but-not-yet-loaded generic functions.
4614 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4615 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4616 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4617 section of the ANSI spec are supported.
4618 * The "" nickname for the "KEYWORD" package has been removed.
4619 The reader still handles symbol tokens which begin with a package marker
4620 as keywords, but it doesn't expose its mechanism for doing so in the
4621 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4622 * The system now issues STYLE-WARNINGs for contradictory TYPE
4623 proclamations. (Warnings for contradictory FTYPE proclamations would
4624 be nice too, but those can't be done usefully unless the type system
4625 is made smarter about FUNCTION types.)
4626 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4627 systematized, so that "*target-*.lisp is supposed to exist only on the
4628 target and imply that there's a related file which exists on the
4629 host, and *host-*.lisp is supposed to exist only on the host and imply
4630 that there's a related file which exists on the target. This involves a
4631 lot of renaming. Hopefully the acute confusion caused by the renaming
4632 will be justified by the reduction in chronic confusion..
4633 ** runtime-type.lisp -> early-target-type.lisp
4634 ** target-type.lisp -> late-target-type.lisp
4635 ** early-host-format.lisp -> early-format.lisp
4636 ** late-host-format.lisp -> late-format.lisp
4637 ** host-error.lisp -> misc-error.lisp
4638 ** early-error.lisp -> early-target-error.lisp
4639 ** late-error.lisp -> late-target-error.lisp
4640 ** host-defboot.lisp -> early-defboot.lisp
4641 ** code/misc.lisp -> code/target-misc.lisp
4642 ** code/host-misc.lisp -> code/misc.lisp
4643 ** code/numbers.lisp -> code/target-numbers.lisp
4644 ** code/early-numbers.lisp -> numbers.lisp
4645 ** early-host-type.lisp -> early-type.lisp
4646 ** late-host-type.lisp -> late-type.lisp
4647 ** host-typep.lisp -> typep.lisp
4648 ** load.lisp -> target-load.lisp
4649 ** host-load.lisp -> load.lisp
4650 ** host-disassem.lisp -> disassem.lisp
4651 ** host-insts.lisp -> insts.lisp
4652 ** byte-comp.lisp -> target-byte-comp.lisp
4653 ** host-byte-comp.lisp -> byte-comp.lisp
4654 ** host-signal.lisp -> signal.lisp
4655 ** host-defstruct.lisp -> defstruct.lisp
4656 ** late-target-type.lisp -> deftypes-for-target.lisp
4657 Furthermore, several other previously target-only files foo.lisp (e.g.
4658 hash-table.lisp and random.lisp) have been split into a target-and-host
4659 foo.lisp file and a target-only target-foo.lisp file, with their key type
4660 definitions in the target-and-host part, so that the cross-compiler will
4661 know more about target types.
4662 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4663 gone away. In their place are various *BACKEND-FOO* variables
4664 corresponding to the slots of the old structure.
4665 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4666 into the target SBCL has been fixed.
4667 * The chill.lisp system for loading cold code into a running SBCL
4669 * Support for the CMU CL "scavenger hook" extension has been removed.
4670 (It was undocumented and unused in the CMU CL sources that SBCL was
4671 derived from, and stale in sbcl-0.6.1.)
4672 * Various errors in the cross-compiler type system were detected
4673 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4674 (enabling various consistency checks). Many of them were fixed,
4675 but some hard problems remain, so the compiler is back to
4676 running without *TYPE-SYSTEM-INITIALIZED* for now.
4677 * As part of the cross-compiler type system cleanup, I implemented
4678 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4679 * I have started adding UNCROSS calls throughout the type system
4680 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4681 on cross-compiler input files will be able to go away, and various
4683 * CONSTANTP now returns true for quoted forms (as explicitly required
4686 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4688 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4689 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4690 (MAKE-LIST -1). In the longer term, it ought to become true
4691 that declarations are assertions even at SAFETY 1. For now, it's not
4692 quite true even at SAFETY 3, but it's at least more nearly true..
4693 (Note that this change seems to increases the size of the system by
4694 O(5%) and to decrease the speed of the compiler by 20% or more.)
4695 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4696 for the problem of printing dozens of lines of distracting information
4697 about low-level system machinery as part of the top stack frame
4698 on entry to the debugger when an undefined function was called.
4699 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4700 is not reset to COMMON-LISP-USER.
4701 * Compilation of stuff related to dyncount.lisp has been made conditional
4702 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4703 smaller. The various dyncount-related symbols have been moved into
4704 a new "SB-DYNCOUNT" package.
4705 * tty-inspect.lisp has been renamed to inspect.lisp.
4706 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4707 feature has gone away. (When we eventually port to other flavors of
4708 libc and/or Unix, we'll try to make the differences between flavors
4709 invisible at the user level.)
4710 * Various other *FEATURES* tags, and/or their associated conditionals,
4711 have been removed if obsolescent, or given better documentation, or
4712 sometimes given more-mnemonic names.
4714 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4716 * tidied up "make.sh" script
4717 * tidied up system directory structure
4718 * better "clean.sh" behavior
4719 * added doc/FOR-CMUCL-DEVELOPERS
4720 * many many small tweaks to output format, e.g. removing possibly-confusing
4721 trailing #\. character in DESCRIBE-INSTANCE
4722 * (EQUALP #\A 'A) no longer signals an error.
4723 * new hashing code, including EQUALP hashing
4724 * tidied up Lisp initialization and toplevel
4725 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4726 * command line argument processing
4727 * added POSIX-GETENV function to deal with Unix-ish environment variables
4728 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4729 terminating SBCL on EOF
4730 * non-verbose GC by default
4731 * There is no more "sbcl" shell script; the sbcl file is now the C
4732 runtime executable (just like CMU CL).
4733 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4735 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4736 symbol and package information as Lisp native symbol and package objects
4737 instead of strings naming symbols and strings naming packages. This way,
4738 whenever packages are renamed (as in warm init), debug information is
4739 transformed along with everything else.
4740 * tweaked the optimization policy declarations which control the building
4741 of SBCL itself. Now, among other things, the system no longer saves
4742 source location debugging information. (This helps two problems at once
4743 by reducing SBCL size and by keeping SBCL from trying to look for its
4744 sources -- which may not exist -- when reporting errors.)
4745 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4746 debugging and testing purposes
4747 * cleaned up printing, making the printer call PRINT-OBJECT for
4748 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4749 methods, giving nearly-ANSI behavior
4750 * converted almost all special variables to use *FOO* naming convention
4751 * deleted PARSE-TIME functionality, since it can be done portably
4752 * moved some files out of cold init into warm init
4753 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4755 * regularized formatting of source files
4756 * added an install.sh script
4757 * fixed ridiculous memory usage of cross-compiler by making
4758 compiler/alloc.lisp not try to do pooling unless it can hook
4759 itself into the GC of the cross-compilation host. Now the system
4760 builds nicely on my old laptop.
4761 * added :SB-ALLOC in target-features.lisp-expr
4762 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4763 was not implemented)
4764 * re-did condition handling and note reporting in the compiler. Notes
4765 are no longer handled by signalling conditions. Style warnings
4766 and warnings are handled more correctly and reported in such a way
4767 that it's easy to find one or the other in your output (so that you
4768 can e.g. figure out which of many problems caused COMPILE-FILE to
4770 * changed the severity of several compiler warnings from full WARNING
4771 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4772 compiler note reporting so that it doesn't use the condition system
4773 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4774 COMPILE-FILE command)
4775 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4776 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4777 no longer supported, since their implementation was deeply intertwingled
4778 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4779 * removed bogus "support" for compiler macros named (SETF FOO), and
4780 removed the compiler macro for SETF INFO (but only after making a fool
4781 of myself on the cmucl-imp mailing list by posting a bogus patch for
4782 DEFINE-COMPILER-MACRO..)
4783 * Compiled files containing forms which have side effects on the Lisp
4784 reader (such as DEFPACKAGE forms) are now handled more correctly.
4785 (Compiler queuing of top level lambdas has been suppressed by setting
4786 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4787 * deleted various currently-unused source files, e.g. gengc.lisp. They
4788 may be added back at some point e.g. when porting to other architectures,
4789 but until they are it's distracting to distribute them and to try to
4791 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4792 there were so many of them they're just distractions, and UNCROSS is
4793 known to be able to handle the current sources
4794 * moved PROFILE functionality into TRACE, so that it will be clear
4795 how the wrapping and unwrapping of functions when you profile them
4796 interacts with the wrapping and unwrapping of functions when you
4797 trace them. (Actually, the functionality isn't there yet, but at least
4798 the interface specification is there. Hopefully, the functionality will
4799 arrive with some future maintenance release.)
4800 * removed host-oops.lisp
4801 * changed signature of QUIT function to allow UNIX-CODE argument
4802 * fixed READ-SEQUENCE bug
4803 * tweaked verbose GC output so that it looks more like the progress
4804 output that ANSI specifies for functions like LOAD
4805 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4806 * added <http://sbcl.sourceforge.com> to the banner information printed by