1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.19 relative to 1.0.18:
3 * bug fix: compiler no longer makes erronous assumptions in the
4 presense of non-foldable SATISFIES types.
5 * bug fix: stack analysis missed cleanups of dynamic-extent
6 arguments in non-let-converted calls to local functions.
7 * fixed some bugs revealed by Paul Dietz' test suite:
8 ** interval arithmetic during type derivation used inexact integer
9 to single-float coercions.
10 ** arithmetic operations involving large integers and single
11 floats give the same results in compiled and interpreted code.
12 ** deriving the result type of COERCE no longer signals an error
13 if the derived type of the second argument is a MEMBER type
14 containing invalid type specifiers.
15 ** ADJOIN with constant NIL as second argument works correctly.
16 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
18 changes in sbcl-1.0.18 relative to 1.0.17:
19 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
20 profiles only the current thread.
21 * minor incompatible change: changes to SYMBOL-VALUE of constants
22 defined with DEFCONSTANT now signal an error.
23 * enhancement: SB-SPROF now has support for wallclock profiling,
24 and is also able to profile specific threads. REPORT output
25 has also additional sorting options.
26 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
28 * optimization: structure allocation has been improved
29 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
30 ** out of line constructors are ~10% faster.
31 ** inline constructors are ~15% faster.
32 ** inline constructors are capable of dynamic extent allocation
33 (generally on x86 and x86-64, in some cases on other platforms
35 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
37 * optimization: file compiler is now able to coalesce non-circular
38 lists, non-base strings, and bit-vectors. Additionally, constants
39 are never referenced through SYMBOL-VALUE at runtime.
40 * optimization: code defining methods on PRINT-OBJECT (and other
41 generic functions in the COMMON-LISP package) now loads faster.
42 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
44 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
45 fasl is now deleted. (reported by Attila Lendvai)
46 * bug fix: READ-LINE always returned NIL for the last line in files.
47 (reported by Yoshinori Tahara)
48 * bug fix: more accurate disassembly annotations of foreign function
49 calls. (thanks to Andy Hefner)
50 * bug fix: trimming non-simple strings and non-string string
51 designators when the there is nothing to trim works properly.
52 (thanks to James Knight)
53 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
54 (thanks to Travis Cross)
55 * fixed some bugs revealed by Paul Dietz' test suite:
56 ** NIL is a valid function name (regression at 1.0.13.38)
57 ** FILL on lists was missing its return value (regression at 1.0.12.27)
58 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
59 fill pointers properly (regression at 1.0.12.23)
60 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
61 displacement indices properly (regression at 1.0.12.23)
63 changes in sbcl-1.0.17 relative to 1.0.16:
64 * temporary regression: user code can no longer allocate closure
65 variable storage on stack, due to bug 419 without explicitly
66 requesting it. Please consult sbcl-devel for advice if you need to
67 use this feature in the meanwhile.
68 * new feature: runtime argument --control-stack-size can be used to
69 adjust thread default control stack size.
70 * enhancement: improved TIME output
71 ** all times are reported using the measured accuracy (milliseconds
72 for real and GC times, microseconds for everything else.)
73 ** processor cycle counts on x86 and x86-64.
74 ** interpreted forms are counted for both evaluator modes.
75 ** number of lambdas converted by the compiler is reported.
76 ** CPU percentage report (computed from real and total run time.)
77 ** more comprehensive run time reporting, using a condenced format
78 ** interperted form, lambda, and page fault counts are omitted
80 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
82 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
83 * optimization: two argument forms of LAST are upto ~10% faster
84 in normal SPEED policies.
85 * optimization: NCONC no longer needs to heap cons its &REST list
86 in normal SPEED policies.
87 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
89 * bug fix: UNION and NUNION work with :TEST-NOT once more,
90 regression since 1.0.9.1. (thanks to Eric Marsden)
91 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
92 since 1.0.15.36. (thanks to Paul Khuong)
93 * bug fix: LAST when always returned the whole list when given a bignum
94 as the second argument.
95 * bug fix: dynamic extent allocation of nested lists and vectors
96 could leak to otherwise accessible parts.
97 * bug fix: invalid optimization of heap-allocated alien variable
99 * bug fix: fasl header checking is less vulnerable to different
100 platform word lengths.
101 * bug fix: more correct assembler syntax for GNU binutils
102 2.18.50.0.4 support. (thanks to Marijn Schouten)
103 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
105 * bug fix: the fopcompiler can handle LOCALLY forms (with no
106 declarations) successfully. (reported by Attila Lendvai)
108 changes in sbcl-1.0.16 relative to 1.0.15:
109 * minor incompatible change: revert the changes to sb-posix's error
110 signaling added in 1.0.14.
111 * minor incompatible change: change PROBE-FILE back to returning
112 NIL whenever we can't get a truename, as was the case before 1.0.14.
113 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
114 inherits from ERROR instead of just CONDITION.
115 * new feature: SB-INTROSPECT can provide source locations for instances
116 as well. (thanks to Tobian Ritterweiler)
117 * optimization: binding special variables now generates smaller code
118 on threaded platforms.
119 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
120 and cases where no :TEST is given but the compiler can infer that
121 the element to search is of type (OR FIXNUM (NOT NUMBER)).
122 * optimization: better LOGNOT on fixnums.
123 * optimization: modular arithmetic for a particular requested width
124 is implemented using a tagged representation unless a better
125 representation is available.
126 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
127 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
128 for wellformedness and returns a specialized vector. (reported by
129 Francois-Rene Rideau)
130 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
131 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
132 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
133 unscheduling timer at the same time as another timer fires.
134 * bug fix: don't reschedule timers for dead threads.
135 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
136 * bug fix: copying output from RUN-PROGRAM to a stream signalled
137 bogus errors if select() was interrupted.
138 * enhancement: add support for fcntl's struct flock to SB-POSIX.
140 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
141 * enhancement: cleaner backtraces for interactive interrupts, as
142 well as other cases where the interesting frames used to be
143 obscured by interrupt handling frames.
144 * enhancement: untracing a whole package using (UNTRACE "FOO") is
145 now supported, and tracing a whole package using (TRACE "FOO") now
146 traces SETF-functions as well.
147 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
148 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
149 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
150 * unreadably printed representation of hash-tables now includes
152 * bug fix: partially fixed #188: type propagation from assignments
153 is now more efficient.
154 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
155 'DOUBLE-FLOAT) are not flushable.
156 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
157 full address of the object, and none of the tag bits.
158 * bug fix: readably printing hash-tables now respects other printer
159 control variables. (reported by Cedric St-Jean)
160 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
161 keyword with MAKE-HASH-TABLE.
162 * bug fix: export SB-POSIX:MKSTEMP.
163 * bug fix: SORT was not interrupt safe.
164 * bug fix: XREF accounts for the last node of each basic-block as
166 * bug fix: MAKE-INSTANCE optimizations interacted badly with
167 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
168 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
169 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
170 create obsolete instances in the case of redefinition or
171 obsoletion of a superclass. (thanks to Andy Hefner)
172 * bug fix: Support for the Alpha architecture has been revived; it had
173 suffered somewhat from lack of maintenance since sbcl-1.0.
174 * improvements to the Windows port:
175 ** The system detects the codepage to use at startup, instead of
176 using the value from the saved core. (thanks to Kei Suzuki)
178 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
179 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
180 (see documentation for details.)
181 * revived support for OpenBSD (contributed by Josh Elsasser)
182 * partially fixed bug #108: ROOM no longer suffers from occasional
183 (AVER (SAP= CURRENT END)) failures.
184 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
185 bodies no longer cause a WARNING to be signalled. (reported by
187 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
188 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
189 the bug in TRUNCATE.) (reported by Michael Weber)
190 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
191 no samples. (reported by Andy Hefner)
192 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
193 no longer appear as (NIL ...) frames in backtraces.
194 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
195 is now actually exported from SB-EXT as documented. (reported by
197 * bug fix: sb-aclrepl now correctly understands how to inspect
198 single-floats on 64-bit platforms where single-floats are not boxed.
199 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
200 yet been finalized. (reported by Levente Meszaros)
201 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
202 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
203 SBCL-specific optimize qualities.
205 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
206 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
207 an executable in the search path, and does so in the child
208 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
209 been removed; it can be found in the CVS history, for somebody who
210 needs that search behavior (see the manual).
211 * minor incompatible change: compiler policy re. weakening type
212 checks has changed: now type checks are weakened only if SAFETY < 2
214 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
215 unparsing of directory pathnames as files. Analogously,
216 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
217 filename to parse into a directory pathname.
218 * enhancement: implicit generic function creation now signals a
219 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
220 users can bind handlers for and muffle around calls to LOAD.
221 * enhancement: RUN-PROGRAM allows unicode arguments and environments
222 to be used (using the default stream external format), and allows
223 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
224 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
225 strings and vectors whose element-type or simplicity is not fully
226 known at compile-time.
227 * optimization: STRING-TRIM and related functions no longer allocate
228 a new string when no trimming needs to be performed. These functions
229 are also faster than before when the input string has been declared
231 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
232 * optimization: READ-LINE is significantly faster for files containing
234 * optimization: non-open coded uses of character comparison operators
235 (e.g. char=) no longer cons when called with more than one parameter
236 on platforms supporting dynamic-extent allocation.
237 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
238 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
239 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
241 * bug fix: some sequence functions elided bounds checking when
243 * bug fix: too liberal weakening of union-type checks when SPEED >
245 * bug fix: more bogus fixnum declarations in ROOM implementation
248 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
249 * new feature: MAKE-HASH-TABLE now experimentally accepts a
250 :SYNCHRONIZED argument, which makes the hash-table safe for
251 concurrent accesses (but not iteration.) See also:
252 SB-EXT:WITH-LOCKED-HASH-TABLE, and
253 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
254 * optimization: CONCATENATE on strings is an order of magnitude faster
255 in code compiled with (> SPEED SPACE).
256 * optimization: SUBSEQ is ~50% faster on lists.
257 * bug fix: bug 417 fixed -- source location reporting is now more robust.
258 * bug fix: SUBSEQ on a list will now correctly signal an error if if
259 END is smaller then START.
260 * bug fix: SB-PROFILE will no longer report extra consing for nested
261 calls to profiled functions.
262 * bug fix: ROOM implementation had bogus fixnum declarations which
263 could cause type-errors when calling ROOM on large images.
264 * bug fix: if file compilation is aborted, the partial fasl is now
265 deleted, and COMPILE-FILE returns NIL as the primary value.
266 * bug fix: number of thread safety issues relating to SBCL's internal
267 hash-table usage have been fixed.
268 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
269 be returned from its body when the values were being returned
270 using unknown-values return convection and the W-P-O was wrapped
271 inside an UNWIND-PROTECT.
272 * bug fix: sb-posix should now compile again under Windows, enabling
275 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
276 * incompatible change: hash-table accessor functions are no longer
277 automatically protected by locks. Concurrent accesses on the same hash-table
278 from multiple threads can give inconsistent results or even corrupt the
279 hash-table completely. Multi-threaded applications should do their own
280 locking at the correct granularity. In the current implementation it is
281 still safe to have multiple readers access the same table, but it's not
282 guaranteed that this property will be maintained in future releases.
283 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
284 and will signal an error at runtime.
285 * enhancement: SB-THREAD package now exports a semaphore interface.
286 * enhancement: CONS can now stack-allocate on x86 and
287 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
289 * enhancement: nested lists can now be stack allocated on
290 platforms providing stack allocation support.
291 * enhancement: dynamic-extent support has been extended to support
292 cases where there are multiple possible sources for the stack
294 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
295 if the mutex is uncontested on Linux.
296 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
297 as specified by the CLHS. (thanks to Tobias Rittweiler)
298 * bug fix: NaN comparison now works on x86-64.
299 * bug fix: CLOSE :ABORT T on a stream with pending output now
301 * bug fix: instances of non-standard metaclasses using standard
302 instance structure protocol sometimes missed the slot type checks
304 * bug fix: known functions can be inlined in high-debug code.
306 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
307 * minor incompatible change: the MSI installer on Windows no longer
308 associates .lisp and .fasl files with the installed SBCL.
309 * minor incompatible change: :UNIX is no longer present in *FEATURES*
310 on Windows. (thanks to Luis Oliviera)
311 * new feature: SB-CLTL2 contrib module now implements
312 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
313 * optimization: scavenging weak pointers is now more efficient,
314 requiring O(1) instead of O(N) per weak pointer to identify
315 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
316 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
317 method bodies using constant slot-names when the first argument is
318 a specializer parameter for the method.
319 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
320 names, when the class of the instance is a direct instance
321 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
322 as slow as the constant slot-name case.)
323 * optimization: member type construction is now O(N) instead
325 * optimization: UNION and NUNION are now O(N+M) for large
326 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
328 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
329 now more readable in environments like Slime which display it.
330 (thanks to Tobias C. Rittweiler)
331 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
332 was able to infer the correct argument type for the object on which
333 the CAS operation was being performed.
334 * bug fix: copy propagation interfered with parallel assignment
335 semantics in local calls. (reported by Paul Khuong)
336 * bug fix: the signed modular fixnum shift compiled to wrong code on
337 x86 and x86-64. (spotted by a slight modification to some of
340 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
341 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
342 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
343 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
345 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
346 and generic function now print the object identity for anonymous
347 instances. (thanks to Attila Lendvai)
348 * enhancement: as an extension to MOP, SBCL now supports SETF of
349 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
350 (thanks to Attila Lendvai)
351 * enhancement: sb-cover has an alternate annotation mode which puts
352 the form annotations at the car of the form, instead of annotating
353 it completely. The benefit of this mode is that it shows better
354 which forms were instrumented by the compiler.
355 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
356 by printer control variables. (reported by Dan Corkill)
357 * bug fix: system leaked memory when delayed output was performed by
358 the OS in smaller chunks then expected. (thanks to David Smith)
359 * bug fix: system leaked memory when file streams were not closed
361 * bug fix: large objects written to slow streams that were modified
362 after the write could end up with the modified state written to
363 the underlying file descriptor.
364 * bug fix: multiple threads operating in parallel on the same stream
365 could cause buffer-overflows.
366 * bug fix: source location information is stored correctly
367 (broken since 1.0.6). This bug would generally show up as the
368 Slime debugger highlighting the wrong form when the sldb-show-source
370 * bug fix: Branch forms are again annotated as branches in the sb-cover
372 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
373 on x86-64. (reported by Gregory Vanuxem)
374 * bug fix: EQUALP could return wrong results for structures with raw
375 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
376 word). (reported by Vjacheslav Fyodorov)
377 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
380 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
381 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
382 atomic compare-and-swap operations on threaded platforms.
383 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
384 allows assigning a global minimum value to optimization qualities
385 (overriding proclamations and declarations).
386 * enhancement: closed over variables can be stack-allocated on x86
388 * performance bug fix: GETHASH and (SETF GETHASH) are once again
390 * optimization: slot definition lookup is now O(1). This speeds up
391 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
392 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
393 * optimization: ASSOC and MEMBER can now be open-coded for all
394 combinations of keyword arguments when second argument is constant
395 and SPEED >= SPACE. In other cases a specialized version is
397 * bug fix: using obsoleted structure instances with TYPEP and
398 generic functions now signals a sensible error.
399 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
400 (reported by Kristoffer Kvello)
401 * bug fix: backtrace construction is now more careful when making
402 lisp-objects from pointers on the stack, to avoid creating bogus
403 objects that can be seen by the GC.
404 * bug fix: defaulting of values in contexts expecting more than 7
405 variables now works on x86-64. (reported by Christopher Laux)
406 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
408 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
409 as the property-list of a symbol.
410 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
411 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
412 situations requested, are once again file-compileable. (reported
415 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
416 * MOP improvement: support for user-defined subclasses of
417 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
418 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
419 the expansion of DEFMETHOD, is responsible for generating a form
420 which creates a list of specializers when evaluated. Additional
421 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
422 debugging and introspective support.
423 * minor incompatible change: the (unsupported) spinlock interface
424 has changed: free spinlock now has the value NIL, and a held spinlock
425 has the owning thread as its value.
426 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
427 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
429 * enhancement: name of a socket-stream is now "a socket" instead of
431 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
432 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
433 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
434 for generic functions. (Thanks to Tobias C. Rittweiler)
435 * optimization: bignum printing speed has been improved by 20-40%
436 (depending on the bignum size.)
437 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
439 * bug fix: the cache used by the CLOS to store precomputed effective
440 methods, slot offsets, and constant return values is now thread and
442 * bug fix: generic function dispatch function updating is now thread
443 and interrupt safe (in the sense that the known issues have been
445 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
446 * bug fix: interning EQL-specializers is now thread and interrupt safe.
447 * bug fix: asdf systems with dependencies to the SB-POSIX or
448 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
449 * bug fix: interrupt safety of applicable method computation has been
452 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
453 * new contrib: sb-cover, an experimental code coverage tool, is included
455 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
456 significantly faster.
457 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
458 produce more efficient code on x86-64 if the compiler can determine
459 that (+ INDEX <constant>) does not require a bounds check and FOO
460 has an element type at least 8 bits wide.
461 * enhancement: a new, experimental synchronous timeout facility is
462 provided. Refer to SB-SYS:WITH-DEADLINE for details.
463 * enhancement: when a symbol name conflict error arises, the
464 conflicting symbols are always printed with a package prefix.
465 (thanks to Kevin Reid)
466 * enhancement: stepping is now once again supported on the SPARC and
467 MIPS platforms. (It is also now more likely to work on CheneyGC
469 * enhancement: sb-sprof can now also track and report accurate call
471 * bug fixes: the treatment of non-standard subclasses of
472 SB-MOP:SPECIALIZER is more correct.
473 * incompatible change: PURIFY no longer copies the data from the
474 dynamic space into the static and read-only spaces on platforms
475 that use the generational garbage collector
476 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
478 * bug fix: GC race condition occasionally resulting in crashes with
479 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
481 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
483 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
484 a STYLE-WARNING to be signalled (regression from 1.0.4.)
485 * bug fix: an asynchronous interrupt could previously leave the
486 system running with GC inhibited.
487 * bug fix: a DECLARE form evaluated at top-level now causes an error
488 rather than silently (or verbosely) returning NIL.
489 * bug fix: trying to dissassemble functions compiled at high DEBUG could
490 result in a error being signalled due to source form lookup errors.
491 (reported by Peter Graves)
493 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
494 * incompatible change: removed writer methods for host-ent-name,
495 host-ent-addresses -- changing the values did not update the DNS
497 * minor incompatible change: changed experimental JOIN-THREAD interface
498 * documentation: the manual now lists reader and writer methods
499 in class slot documentation sections. (thanks to Richard M Kreuter)
500 * documentation: unwinding from asyncronous events has been
501 documented as unsafe.
502 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
503 in multithreaded application code.
504 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
505 platforms. (thanks to James Anderson for the optimization hint)
506 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
508 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
509 variants no longer cons.
510 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
511 their NOT- variants no longer cons.
512 * optimization: EQUAL hash tables no longer use SXHASH for objects
513 of all data types, but instead use an EQL hash for types for which
514 EQUAL is the same as EQL.
515 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
516 are significantly faster.
517 * optimization: new STRING-OUTPUT-STREAM implementation is much
518 faster for large outputs and conses 30% less on average.
519 * enhancement: XREF information is now collected for references made
520 to global variables using SYMBOL-VALUE and a constant argument.
521 * enhancement: SIGINT now causes a specific condition
522 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
523 * bug fix: FILE-POSITION used to signal an error in some cases where
524 ANSI requires it to return NIL.
525 * bug fix: ADJUST-ARRAY is now interrupt-safe.
526 * bug fix: adding and removing fd-handlers is now interrupt-safe.
527 * bug fix: inlined calls to C now ensure 16byte stack alignment on
529 * bug fix: bad type declaration in the CLOS implementation has
530 been fixed. (reported by James Anderson)
531 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
532 (thanks to Lutz Euler)
533 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
534 required by the CLHS. (reported by Samium Gromoff)
535 * bug fix: dead unbound variable references now signal an error.
536 (reported by Marco Monteiro)
537 * bug fix: / with an unused value was being deleted in safe code.
538 (thanks to Marco Monteiro and Kevin Reid)
539 * bug fix: number of characters that can be written onto a single
540 line in a file is unlimited.
541 * bug fix: some GC deadlocks caused by asynchronous interrupts have
542 been fixed by inhibiting interrupts for when GC is disabled.
543 * bug fix: some interrupt safety issues with GETHASH, (SETF
544 GETHASH), CLRHASH and REMHASH have been fixed.
545 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
546 specifier no longer causes infinite recursion.
547 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
548 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
549 * bug fix: an &environment argument with the correct variable information
550 is passed to macros that are expanded during byte compilation.
551 (reported by Samium Gromoff)
552 * bug fix: a logic error was causing the hash function for bignums to
553 have excessive amounts of collisions. (reported by Faré Rideau)
554 * bug fix: modifying the contents of an array could change the return
555 value of SXHASH on that array, which is only allowed for strings
556 and bit vectors. (bug introduced in 0.9.16)
557 * improvement: the x86-64/darwin port now passes all tests (except
558 for the debugger tests) but should still be considered
559 experimental until this is fixed.
560 * improvement: a style-warning is signaled for CASE (etc) clauses with
561 duplicate keys. (thanks to Kevin Reid)
562 * improvement: macos/x86 and macos/x86-64 now use mach exception
563 handlers for dealing with illegal instructions (for trapping and
564 error handling) and memory protection violations (for GC).
566 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
567 * new platform: experimental support for x86-64/darwin (MacOS).
568 * incompatible change: the thread-safe (on most platforms) getaddrinfo
569 and getnameinfo sockets functions are used instead of gethostbyaddr
570 and gethostbyname, on platforms where the newer functions are available.
571 As a result, the ALIASES field of HOST-ENT will always be NIL on these
573 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
574 don't print the startup banner, but behave as if --noinform was passed
575 as a command line argument. (thanks to Kevin Reid)
576 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
577 * optimization: code using alien values with undeclared types is much faster.
578 * optimization: the compiler is now able to open code SEARCH in more cases.
579 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
580 * bug fix: using standardized COMMON-LISP special variables as loop
581 variables no longer signals bogus package lock violations. (reported
583 * bug fix: declaring local loop variables to be of a range-limited type
584 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
585 (reported by Andras Simon)
586 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
587 bugs remain on x86-64.)
588 * bug fix: the #= and ## reader macros now interact reasonably with
589 funcallable instances.
590 * bug fix: type-checks for function arguments were compiled using the
591 compiler policy settings of the wrong lexical environment (in 1.0.2
593 * bug fix: SHADOW accepts characters as string designators, as required
594 by the spec (thanks to Eric Marsden)
595 * bug fix: fixed GC safety issues when foreign functions are called with
596 non-base strings as arguments
597 * bug fix: more consistent error output for fatal-compiler-warnings, like
599 * bug fix: functions with proper names were showing up as NIL in
600 backtraces (reported by Edi Weitz, regression in 1.0.2)
602 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
603 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
604 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
605 produce more efficient code on the x86 if the compiler can determine
606 that (+ INDEX <constant>) does not require a bounds check and FOO
607 has an element type at least 8 bits wide.
608 * bug fix: references to undefined variables are handled the same way
609 in toplevel forms as in the normal compiler
610 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
612 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
613 (thanks to Jon Buller)
614 * improvement: faster compilation times for complex functions
615 * improvement: added readlink support to SB-POSIX (thanks to Richard
618 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
619 * improvement: experimental support for mach exception handling on
620 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
622 * improvement: support for GBK external format.
623 (thanks to Chun Tian (binghe))
624 * improvement: the debugger now displays variables that have been closed
625 over, in code compiled with (DEBUG 2) or higher
626 * improvement: support for executable cores on NetBSD (thanks to
628 * new feature: added a RESTART-FRAME debugger command
629 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
630 be used to provide an implementation for FILE-POSITION on Gray streams
631 (thanks to Eric Marsden)
632 * new feature: add syslog, openlog and closelog support to SB-POSIX
633 (thanks to Richard Kreuter)
634 * optimization: the function call overhead in code compiled with
635 a high DEBUG optimization setting is significantly
636 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
637 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
638 (thanks to Tony Martinez)
639 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
640 evaluated once (reported by Kevin Reid)
641 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
642 works again (thanks to Kilian Sprotte)
643 * bug fix: an error is signaled for tagbodies with duplicate tags
644 (thanks to Stephen Wilson)
645 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
646 * bug fix: Win32 port can now handle foreign code unwinding Lisp
647 stack frames from alien callbacks.
648 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
649 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
650 65k pages (thanks to David Woodhouse)
651 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
653 changes in sbcl-1.0.1 relative to sbcl-1.0:
654 * new platform: FreeBSD/x86-64, including support for threading.
655 * new feature: the compiler stores cross-referencing information
656 about function calls (who-calls), macroexpansion (who-macroexpands)
657 and special variables (who-binds, who-sets, who-references) for code
658 compiled with (< SPACE 3). This information is available through the
659 sb-introspect contrib.
660 * new feature: users may subclass SEQUENCE, and have instances of
661 these classes interoperate with standard Common Lisp functions if
662 a number of methods are defined. (This feature is experimental
663 and the interface subject to change based on feedback from SBCL
664 users and the general community)
665 * improvement: sb-sprof traces call stacks to an arbitrary depth on
666 x86/x86-64, rather than the previous fixed depth of 8
667 * improvement: another pthread back-end of mutex "pthread-futex".
668 * bug fix: non-ascii command-line arguments are processed correctly
669 (thanks to Yaroslav Kavenchuk)
670 * bug fix: non-required arguments were not passed correctly when a method
671 defined using DEFMETHOD was called from a mop-generated method using
672 CALL-NEXT-METHOD (reported by Pascal Costanza)
673 * bug fix: recursion is now permitted in accessors through
674 SLOT-UNBOUND. (reported by Pascal Costanza)
675 * bug fix: an error was signaled at startup if the HOME environment
676 variable was defined, but had an empty value (reported by Peter Van Eynde)
677 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
678 signal a PROGRAM-ERROR, not a BUG.
679 * bug fix: SB-POSIX:READDIR works when built with large file support.
680 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
682 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
683 for better automatic indentation support. (thanks to Matt Pillsbury)
684 * optimization: loading generic functions no longer takes O(n^2) time,
685 proportional to the amount of methods in the generic function
686 (reported by Todd Sabin and Jeremy Brown)
687 * optimization: the FIND and POSITION family of sequence functions
688 are significantly faster on arrays whose element types have been
690 * improvements to the Windows port:
691 ** Intermittent heap corruption problems have been fixed. (thanks
692 to Alastair Bridgewater)
693 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
695 ** Lisp is able to unwind foreign exception frames from alien
696 callbacks. (thanks to Alastair Bridgewater)
698 changes in sbcl-1.0 relative to sbcl-0.9.18:
699 * improvement: experimental support for threading on FreeBSD/x86.
700 (thanks to NIIMI Satoshi)
701 * improvement: runtime option --dynamic-space-size can be used
702 to set the size of the dynamic space reserved on startup.
703 * improvement: floating point modes in effect are now saved in
704 core, and restored on startup.
705 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
706 startup, not time since first call to GET-INTERNAL-REAL-TIME.
707 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
708 threads are not running after *SAVE-HOOKS* have run.
709 * improvement: writes to CLOS instance slots are type-checked in code
710 compiled with (SAFETY 3)
711 * improvement: floating-point exception handling on FreeBSD (thanks to
713 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
714 (thanks to Zach Beane)
715 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
717 * improvement: added support for the Shift-JIS external format.
718 (contributed by NIIMI Satoshi)
719 * improvement: callbacks are supported on Linux/PPC. (thanks to
721 * bug fix: compiler bug triggered by a (non-standard) VALUES
722 declaration in a LET* was fixed. (reported by Kaersten Poeck)
723 * bug fix: file compiler no longer confuses validated and already
724 dumped structurres. (reported by Kaersten Poeck)
725 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
726 fill-pointer signals a type-error as required. (thanks to
728 * bug fix: disassemly of funcallable instances works.
729 * bug fix: single stepping on PPC.
730 * bug fix: fix thread-safety problems in the type system (generally
731 manifesting as nonsensical errors like "STRING is a bad type specifier
732 for sequences" or "The value 1 is not of type FIXNUM")
733 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
734 (reported by Josip Gracin)
735 * bug fix: an error is signaled for attempts to displace arrays with
736 incompatible element types (thanks to Mario Mommer)
737 * bug fix: more correct handling of wide characters in debug info
738 (bug reported by Attila Lendvai and fixed by Juho Snellman)
739 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
740 and don't cause extra consing
741 * optimization: MAP and MAP-INTO are significantly faster on vectors
742 whose elements types have been declared.
743 * Improvements to SB-SPROF:
744 ** Support for allocation profiling
745 ** Reduced profiling overhead, especially for long profiling runs
746 * Improvements to the Windows port:
747 ** floating point exceptions are now reported correctly.
748 ** stack exhaustion detection works partially.
749 ** more accurate GET-INTERNAL-REAL-TIME.
750 ** better RUN-PROGRAM behaviour in case of errors in the child
752 ** PROBE-FILE now simplifies pathnames correctly.
753 ** DIRECTORY now works correctly with :WILD-INFERIORS.
754 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
755 1st 1970 (although time timezone data will be unreliable).
756 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
758 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
759 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
760 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
762 * bug fix: two potential GC deadlocks affecting threaded builds.
763 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
764 returns T (reported by Anton Kazennikov)
765 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
766 with non-variable places
767 * bug fix: remove a race condition in the setting of
768 funcallable-instance functions, this should make threaded CLOS
769 code more stable against memory faults.
770 * bug fix: corruption of specials when unbinding is interrupted by an
771 asynchronous unwind (reported by Hannu Koivisto)
772 * improvement: the debugger will now also display local variables that
773 are only used once, for code compiled with a DEBUG optimization quality
776 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
777 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
778 * incompatible change: External-format support for FFI calls. The
779 SB-ALIEN:C-STRING no longer implies an ASCII
780 external-format. Instead the string is subject to external-format
781 conversion. Additionally return values of type SB-ALIEN:C-STRING
782 are by default of type (SIMPLE-ARRAY CHARACTER), not
783 SIMPLE-BASE-STRING. For an alien type that behaves like the old
784 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
785 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
786 * incompatible change: SB-EXT package no longer contains the
787 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
788 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
790 * incompatible change: the single-stepper is no longer available
791 on Alpha, Mips and Sparc platforms.
792 * minor incompatible change: the direct superclasses of
793 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
794 not (STANDARD-OBJECT FUNCTION). This makes the
795 class-precedence-lists of GENERIC-FUNCTION and
796 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
798 * minor incompatible change: the default stream external format on
799 non-unicode SBCL is no longer detected from the locale, but is always
801 * new feature: Add a version of evaluator that uses an interpreter instead
802 of the compiler. EVAL still uses the compiler by default, to switch it
803 to use the interpreter, set the value of the variable
804 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
805 * minor incompatible change: the single-stepper REPL has been merged
806 with the normal debugger (see the "Stepping" heading of the debugger help
807 for more details). The debugger command STEP will no longer switch
808 to the single-stepper REPL.
809 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
810 for a type now works.
811 * bug fix: loading of default sysinit file works. (thanks to Leonid
813 * bug fix: better detection of circularities in the file-compiler.
814 (reported by Marco Monteiro)
815 * bug fix: the CL pathname functions now work with files that have
816 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
817 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
818 systems with Unix98 pty semantics.
819 * bug fix: ASDF-INSTALL will now work with bsd tar.
820 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
822 * bug fix: timers expiring in dead threads no longer cause a
823 type-error (reported by Paul "Nonny Mouse").
824 * bug fix: thanks to more lightweight single-stepper instrumentation,
825 code compiled with (DEBUG 3) will compile and execute significantly faster,
826 and will have more accurate type-inferencing than before
827 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
828 whose bindings are modified
829 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
830 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
831 ** replace dummy implementations of CL:MACHINE-INSTANCE and
832 CL:SOFTWARE-VERSION with proper version
834 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
835 * feature: implemented the READER-METHOD-CLASS and
836 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
837 as specified by AMOP.
838 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
840 * optimization: faster LOGCOUNT implementation on x86 and x86-64
841 (thanks to Lutz Euler)
842 * optimization: hashing of general arrays and vectors has been
843 improved. (reported by Any Fingerhut)
844 * enhancement: SB-INTROSPECT is now able to find definitions of
845 profiled functions. (thanks to Troels Henriksen)
846 * enhancement: compiler-macro expansion applies now to FUNCALL forms
847 as well, allowing compiler-macros for SETF-functions to expand.
848 * enhancement: step-instrumentation no longer wraps known
849 single-value functions with multiple-value context, allowing
850 better type inference.
851 * fixed bug #337: use of MAKE-METHOD in method combination now works
852 even in the presence of user-defined method classes. (reported by
853 Bruno Haible and Pascal Costanza)
854 * fixed bug #339(c): if there are applicable methods not part of any
855 long-form method-combination group, call INVALID-METHOD-ERROR.
856 (reported by Bruno Haible)
857 * fixed bug #361: the :FUNCTION initarg in the protocol for
858 initialization of methods can now be used to override
859 internally-produced optimized functions. (reported by Bruno
861 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
862 system-provided lambda expression no longer cause warnings about
863 unbound #:|pv-table| symbols.
864 * bug fix: improved the handling of type declarations and the
865 detection of violations for keyword arguments with non-constant
867 * bug fix: potentially erronous calls to PATHNAME and
868 MERGE-PATHNAMES were being flushed in some cases.
869 (reported by Richard Kreuter)
870 * bug fix: compiled calls to TYPEP were mishandling obsolete
871 instances. (reported by James Bielman and Attila Lendvai)
872 * bug fix: format strings with ~> without matching ~< no longer
873 trigger an AVER, but signal an understandable error instead.
874 (reported by Antonio Martinez)
875 * bug fix: specifying an output-file without a pathname-type for
876 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
877 of input-file instead of "fasl". (reported by Robert Dodier)
878 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
879 (reported by James Y Knight).
880 * bug fix: compiler-macros-function did not consider the environment
881 argument for shadowing by local functions.
882 * bug fix: compiler-macros expansion was inhibited by local INLINE
884 * bug fix: inline expansions of known functions were subject to
885 step-instrumentation in high debug policies, leading to problems
887 * bug fix: compiler failed to differentiate between different CONS
889 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
890 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
891 * bug fix: #368: incorrect use of expressed vs. upgraded array
893 * bug fix: #306a: more precise unions of array types.
894 * thread-safety improvements:
895 ** CONDITION-WAIT could return early on Linux, if the thread was
896 interrupted and subsequently continued with SIGCONT.
897 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
899 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
900 * added support for the ucs-2 external format. (contributed by Ivan
902 * minor incompatible change: pretty printing of objects of type
903 (cons symbol) is, in the default pprint-dispatch-table, now
904 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
906 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
907 called later in a class's lifetime, possibly as late as when the
908 first instance of the class is created. Previously,
909 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
910 class became finalizeable.
911 * fixed bug: FILE-POSITION sometimes returned inconsistent results
912 for multibyte external-format streams. (thanks to "vbzoli")
913 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
914 with :ALLOCATION :CLASS inherited from superclasses of the
916 * fixed bug: anonymous classes can now be created using the :NAME
917 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
918 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
919 * fixed bug: core-files saved with :EXECUTABLE T can again be
920 executed when SBCL_HOME isn't set. (reported by James Knight)
921 * fixed bug: toplevel LOCALLY forms with declarations could
922 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
923 * fixed bug: printing from several different threads using different
924 values of *print-case* could cause invalid output, due to
925 some internal special variables of the printer not being bound
926 thread-locally (reported by Max Mikhanosha)
927 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
928 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
929 (reported by Pascal Costanza)
930 * fixed bug: FINALIZE-INHERITANCE is called from
931 REINITIALIZE-INSTANCE on classes when the class has previously
932 been finalized, as required by AMOP.
933 * minor code generation optimizations:
934 ** better register allocation in CLOS dispatching functions
935 ** overflow detection when coercing signed bytes to fixnums on x86-64
936 ** is now implemented with one IMUL instruction instead of three shifts
937 ** more efficient bit-vector access on x86 and x86-64
938 ** more efficient access to raw structure slots on x86 and x86-64
939 * fixed some bugs revealed by Paul Dietz' test suite:
940 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
943 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
944 * feature: thread support on Solaris/x86, and experimental thread support
946 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
948 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
949 default method for SLOT-UNBOUND.
950 * minor incompatible change: prevent the user from specializing the
951 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
952 somewhat counter to the intent of the protocol, I (CSR) think, and
953 additionally it just doesn't work in SBCL as currently
954 implemented, thanks to optimizations (that are always valid for
955 the other three SLOT-VALUEish functions, but not for the setter).
956 * minor incompatible change: the :SB-LDB feature is now enabled by
957 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
958 the low-level debugger.
959 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
960 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
961 down to an SB-SYS:FD-STREAM, allowing direct communication in
962 more cases. Temporary files and pipes are still used as a fallback
964 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
965 on the non-reentrant C functions localtime(3) and gmtime(3).
966 * bug fix: global optimization policy was not visible in LOCALLY and
968 * bug fix: class objects can be used as specializers in methods.
969 (reported by Pascal Costanza)
970 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
971 with a type error. (reported by blitz_ on #lisp)
972 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
973 failed with a type error. (reported by Pascal Bourguignon)
974 * bug fix: merging pathnames against defaults with :DIRECTORY
975 starting with '(:RELATIVE :BACK) should preserve the :BACK.
976 (reported by James Y Knight)
977 * bug fix: saving large (>2GB) cores on x86-64 now works
978 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
979 constant index and a (simple-array (signed-byte 32)) array
980 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
981 error (patch by Robert J. Macomber)
982 * feature: TIME now displays the amount of run-time spent in GC
983 * bug fix: The debugger now does a better job of respecting
984 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
985 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
986 workaround for bug 403.)
987 * bug fix: floating point exception handling now works on FreeBSD
988 (thanks to NIIMI Satoshi)
989 * fixed some bugs revealed by Paul Dietz' test suite:
990 ** MISC.641: LET-conversion were not supposed to work in late
992 * improvements to the Win32/x86 port:
993 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
994 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
996 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
997 * new feature: source path information is generated for macro-expansion
998 errors for use in IDE's like Slime (thanks to Helmut Eller)
999 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1000 * bug fix: compiler does not loop forever on an invalid type in
1002 * improvement: compilation of most CLOS applications is significantly
1004 * optimization: added a limited bytecode compiler for simple toplevel
1005 forms, speeding up compilation and FASL loading
1006 * bug fix: the statistical profiler now properly distinguishes anonymous
1009 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1010 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1011 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1012 into subclasses of the finalized class. Now user calls to
1013 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1014 system return before any subclasses are finalized.
1015 * minor incompatible change: The reader no longer ignores errors
1016 regarding non-existent packages in #+ and #- feature tests.
1017 * new feature: command line options --no-sysinit, --no-userinit to
1018 inhibit loading the corresponding init files
1019 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1020 for sending data through UDP sockets (thanks to François-René Rideau)
1021 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1022 error is signalled instead (thanks to François-René Rideau)
1023 * minor incompatible change: Error signalling behaviour of lexical
1024 operations violating package locks has changed slightly. Refer to
1025 documentation on package locks for details.
1026 * enhancement: EVAL can process IF-expressions without resorting to the
1028 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1029 constant-folded properly due to missing out-of-line functions.
1030 (reported by Mika Pihlajamäki)
1031 * bug fix: LISTEN sometimes returned T even in cases where no data was
1032 immediately available from the stream
1033 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1034 were reversed. (reported by Levente Mészáros)
1035 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1036 when given a vector displaced to an adjustable array. (reported
1038 * bug fix: derivation of float boundaries from numbers outside the
1039 appropriate float range (reported by John Wiseman)
1040 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1042 * bug fix: precomputing cache entries for generic functions with
1043 some subclasses of specializers as yet invalid does not attempt to
1044 fill a cache line with a negative offset. (reported by Levente
1046 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1047 allow condition class objects as arguments to DOCUMENTATION and
1048 (SETF DOCUMENTATION); only find and set documentation for
1049 structure names for the STRUCTURE doc-type. (suggested by Gary
1051 * improvements to the Win32/x86 port:
1052 ** (user-homedir-pathname) and default initialization file
1053 locations now know about the user's "Documents and Settings"
1054 directory (thanks to Yaroslav Kavenchuk)
1055 ** run-program is implemented (thanks to Mike Thomas)
1056 ** sockets support (thanks to Timothy Ritchey)
1057 ** better backtrace support (thanks to Alastair Bridgewater)
1058 ** sb-grovel supported
1059 ** asdf-install and sb-posix work somewhat
1060 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1061 * improvements to the Solaris/x86 port:
1062 ** works on Solaris 11/Solaris Express
1063 ** floating-point exception handling support
1064 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1065 * fixed some bugs revealed by Paul Dietz' test suite:
1066 ** REMOVE-METHOD returns its generic function argument even when
1067 the method is not one of the generic functions' methods.
1068 ** objects declared as MEMBER types can be admissible arguments to
1069 structure accessors.
1070 ** printing characters should simply be printed by the FORMAT ~:C
1072 ** compiler failure when compiling functions with hairy constant
1073 defaults for optional parameters.
1074 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1075 function, which is already optimized.
1077 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1078 * new platform: experimental support for SBCL x86/Darwin, including
1079 MacOS X 10.4.5 on Intel.
1080 * new feature: Unicode character names are now known to the system
1081 (through CHAR-NAME and NAME-CHAR).
1082 * new feature: the filesystem location of SBCL's core file is
1083 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1084 * minor incompatible change: the contrib modules SB-POSIX and
1085 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1086 this change is to make it easier to distribute
1087 location-independent binaries.
1088 * bug fix: as implied by AMOP, standardized classes no longer have
1089 slots named by external symbols of public packages. (reported by
1091 * optimization: calling functions via constant symbols -- as in
1092 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1093 via the function object as in (FUNCALL #'FOO).
1094 * optimization: on x86 and x86-64, the calling convention for the
1095 case of dealing with an unknown number of values has been altered
1096 to be friendlier to the prediction heuristics implemented,
1097 particularly with reference to CALL and RET pairing. (thanks to
1098 Alastair Bridgewater)
1099 * optimization: on x86, the code for access to thread-local symbol
1100 values and binding/unbinding of thread-local symbols is smaller.
1101 (thanks to Alastair Bridgewater)
1102 * enhancement: CONSTANTP is now able to determine constantness of
1103 more complex forms, including calls to constant-foldable standardized
1104 functions and some special forms beyond QUOTE.
1105 * fixed bug: occasional GC crashes on Solaris/x86
1106 * optimization: x86-64 supports stack allocation of results of simple
1107 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1108 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1109 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1110 (thanks to James Knight)
1111 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1112 by the standard (thanks to Stephen Compall)
1114 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1115 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1116 be used for bundling the runtime and the core file into one
1117 executable binary. This feature is not currently supported on all SBCL
1118 platforms. (thanks to James Bielman and NIIMI Satoshi)
1119 * new feature: a generational or ephemeral garbage collector is now
1120 the default on the PowerPC platform (both Linux and Darwin). The
1121 old Cheney (stop and copy) collector is a build-time option.
1122 (thanks to Cyrus Harmon, after Raymond Toy)
1123 * minor incompatible change: the method by which SBCL finds its
1124 contributed modules has changed; it no longer relies on symbolic
1125 links from an $SBCL_HOME/systems directory, but searches directly
1126 in the subdirectories of $SBCL_HOME.
1127 * enhancement: the dynamic heap size on the Linux/PPC platform is
1128 markedly larger, even using the older Cheney garbage collector.
1129 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1130 certain complicated string types. (reported by Gary King)
1131 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1132 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1133 * fixed bug: DOCUMENTATION on structures no longer signals an
1134 error if no structure type of the right name exists. (reported by
1136 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1137 applicable non-standard primary or :AROUND method on
1138 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1139 no longer cause unbound variable errors. (reported by Kevin Reid)
1140 * fixed bug: metacircle resolution in cases where methods have slots
1141 added before the slots from STANDARD-METHOD. (reported by Jean
1143 * fixed bug: the Power PC assembly code for calling into lisp
1144 sometimes computed its return address wrongly (depending on the
1145 15th bit of the address, fixed at link time). This used to
1146 manifest itself in a segmentation violation while building PCL.
1147 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1148 many others over the years)
1149 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1150 the stack pointer (treated as a signed integer) is negative.
1151 (thanks to Peter van Eynde)
1153 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1154 * new platform: experimental support for the Windows operating
1155 system has been added. (thanks to Alastair Bridgewater)
1156 * fixed several bugs in and robustified the PPC FFI (including
1157 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1158 * bug fix: allow non-simple string symbol names (reported by Paul
1160 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1162 * bug fix: saving a core corrupted callbacks on x86/x86-64
1163 * bug fix: closed a loophole in metacircularity detection and
1164 grounding in the PCL implementation of CLOS.
1165 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1166 (reported by tomppa on #lisp)
1167 * bug fix: clear the direction flag on Lisp -> C transitions, as
1168 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1169 (reported by Andrej Grozin and Hendrik Maryns)
1170 * optimization: major improvements to GC efficiency on GENCGC platforms
1171 * optimization: faster implementation of EQUAL
1172 * optimization: emit more efficient opcodes for some common
1173 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1174 * optimization: several other minor code-generation improvements on x86-64
1175 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1177 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1178 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1179 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1180 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1181 by ANSI as a generic function, and for consistency (SETF
1182 GENERIC-FUNCTION-NAME) is treated likewise.
1183 * fixed bug #233.b: make constraint propagation notice when a variable
1184 value is changed after it is referenced but before it is used
1185 * fixed bug #296: no more arbitrary behaviour from filenames with
1186 odd characters as --load arguments. (reported by Adam Warner)
1187 * fixed bug #390: :CHARACTER-SET pathname components now work as
1188 expected. (reported by Tim Daly Jr)
1189 * fixed bug #391: complicated :TYPE intersections in slot
1190 definitions no longer cause an error in PCL internals.
1191 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1192 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1193 returning the number of octets which would be written to the
1194 file-stream. (thanks to Robert J. Macomber)
1195 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1196 arguments to the internal stream functions. (thanks to David
1198 * bug fix: stack exhaustion is now handled on the x86/SunOS
1199 platform. (thanks to Daisuke Homma)
1200 * bug fix: more accurate ROOM results on GENCGC platforms
1201 * optimization: improved type inference for arithmetic-for
1202 index variables in LOOP
1203 * optimization: faster floating-point SQRT on x86-64
1204 * fixed some bugs revealed by Paul Dietz' test suite:
1205 ** DOCUMENTATION returns NIL instead of "" for method combinations
1206 that don't have a docstring
1208 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1209 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1210 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1211 therefore are not customizeable by user code (as seems to be at
1212 least permitted and maybe required by AMOP). As a consolation,
1213 however, the SBCL implementation of these functions now calls
1214 REINITIALIZE-INSTANCE as specified by AMOP.
1215 * bug fix: slot-definition documentation is propagated as per ANSI
1216 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1217 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1218 argument, as do eslotd initialization methods. (from Pascal
1219 Costanza's "Closer" project)
1220 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1221 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1223 * bug fix: it is now possible to have more than one subclass of
1224 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1225 (reported by Bruno Haible, Pascal Costanza and others)
1226 * bug fix: the dependent update protocol now works for generic
1227 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1228 and Pascal Costanza)
1229 * bug fix: condition-class instances corresponding to
1230 DEFINE-CONDITION forms are now created eagerly. (reported by
1231 Kalle Olavi Niemitalo on comp.lang.lisp)
1232 * bug fix: floating point printing is more accurate in some
1233 circumstances. (thanks to Simon Alexander)
1234 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1235 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1236 * bug fix: callbacks on OS X now preserve stack-alignment.
1237 (thanks to Cyrus Harmon)
1238 * enhancement: the x86-64 disassembler is much better at
1239 disassembling SSE instructions. (thanks to Lutz Euler)
1240 * enhancement: CHECK-TYPE on a variable now allows type-inference
1241 (thanks to Vincent Arkesteijn)
1242 * optimization: improved performance of EUC-JP external format.
1243 (thanks to NIIMI Satoshi)
1244 * optimization: performance improvements to IO on file streams of
1245 :ELEMENT-TYPE CHARACTER
1246 * optimization: much faster memory allocation on x86-64
1247 * optimization: faster garbage collections (latency and throughput)
1249 * optimization: faster bignum subtraction and division on x86 and x86-64
1250 (thanks to Lutz Euler)
1251 * optimization: more accurate type-inference for loops that use a
1252 floating point index variable or a negative step.
1254 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1255 * bug fix: add a workaround to SBCL looping infinitely at startup on
1256 Linux kernels with apparently buggy implementations of personality().
1257 (thanks to Svein Ove Aas)
1258 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1259 (thanks to David Lichteblau)
1260 * bug fix: local bindings shadow global symbol macros for
1261 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1262 * optimization: non-open coded uses of numeric comparison operators
1263 (e.g. >) no longer cons when called with more than one parameter
1264 on platforms supporting dynamic-extent allocation.
1265 * enhancement: saving cores with foreign code loaded is now supported
1266 on MIPS/Linux in addition to the previously supported platforms.
1267 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1268 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1269 * bug fix: division by zero in sb-sprof when no samples were collected
1270 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1271 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1272 (thanks to Alexander Kjeldaas)
1273 * new feature: ignoring the return values of destructive functions
1274 like NREVERSE signals a compile-time style-warning.
1275 (thanks to Kevin Reid)
1276 * new feature: the alignment of alien structure fields can be
1277 explicitly specified. (thanks to Cyrus Harmon)
1278 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1279 and dump core on SIGQUIT
1281 ** incompatible change: threads do not inherit values of specials
1282 from their parents (see manual)
1283 ** bug fix: threads stacks belonging to dead threads are freed by the
1284 next exiting thread, no need to gc to collect thread stacks anymore
1285 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1286 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1287 ** bug fix: dynamic variable and thread start related gc lossage
1288 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1290 * fixed some bugs revealed by Paul Dietz' test suite:
1291 ** SUBTYPEP is slightly more accurate on heinously complicated
1292 CONS types where some of the members have uncertain (in the
1293 NIL, NIL sense) type relationships to each other.
1295 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1296 no longer purified unless :PURIFY T is explicitly specified.
1297 ** Non-purified cores are significantly smaller than before
1299 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1300 * new feature: timers based on Zach Beane's excellent timer package
1301 * added support for the following external formats: koi8-u,
1302 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1303 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1304 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1305 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1306 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1307 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1308 * incompatible change: a threaded SBCL will no longer revert to
1309 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1310 * bug fix: interrupts are disabled until startup is complete; no
1311 more sigsegvs when receiving a signal to soon
1312 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1314 * bug fix: add a workaround for the memory randomization features in
1315 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1316 workaround will only be in effect on systems with the proc filesystem
1318 * bug fix: printing objects of type HASH-TABLE signals a
1319 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1321 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1322 have been running for over 50 days. (reported by Gilbert Baumann)
1323 * bug fix: the logic for getting names of functions gets less
1324 confused when confronded with alternate-metaclass
1325 funcallable-instances. (reported by Cyrus Harmon)
1326 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1327 consistent, even on internal alternate-metaclass objects.
1328 * bug fix: flush closure information collected by physical
1329 environment analysis prepass before the main pass. (bug reported
1331 * bug fix: compiler pack phase does not modify a hashtable, which is
1332 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1333 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1334 may be expanded. (reported by Alexander <malishev> on c.l.l)
1335 * bug fix: no more style-warnings when compiling code using LAST
1336 under high-SPEED low-SPACE settings. (reported by David Wallin
1338 * fixed some incompatibilities between SBCL's MOP and the MOP
1340 ** the METAOBJECT class is now implemented;
1341 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1342 STANDARD-OBJECT, as required;
1343 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1344 now compatible (as required by VALIDATE-SUPERCLASS); there
1345 remains a consistency requirement over the presence or absence
1346 of the FUNCTION class in the superclass list of finalized
1347 classes; see the manual for more details;
1348 ** the :AROUND method for COMPUTE-SLOTS on
1349 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1350 requested slot ordering.
1352 ** bug fix: parent thread now can be gc'ed even with a live
1354 ** bug fix: binding a special with PROGV to no value is not
1355 the same as that symbol not having been bound (thanks to
1357 ** bug fix: binding specials is thread safe (thanks to
1359 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1360 and SB-SPROF do not die with 'no handler for signal XX in
1361 interrupt_handle_now(..)' anymore
1362 ** bug fix: WITH-TIMEOUT works with multiple running threads
1363 * fixed some bugs revealed by Paul Dietz' test suite:
1364 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1365 the :method-class keyword argument.
1367 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1368 * new port: the Solaris operating system on x86 processors is now
1369 mostly supported, though some rough edges in the environment
1370 remain. (thanks to Daisuke Homma)
1371 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1372 space for its heap. (thanks to Thiemo Seufer)
1373 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1374 closures. (thanks to Thiemo Seufer)
1375 * minor incompatible change: eof selects abort in the debugger.
1376 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1377 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1379 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1380 deprecated, and will go away in a future revision of SBCL.
1381 * minor incompatible change: GC-ON and GC-OFF are no longer
1382 implemented with a counter, it does not matter how many times gc
1383 is switched on or off
1384 * bug fix: discriminating functions for generic function classes
1385 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1386 make invalid assumptions about method precedence order. (reported
1388 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1389 platforms. (thanks to Thiemo Seufer)
1390 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1391 * bug fix: correct pathname printing with printer escaping is on.
1392 (thanks to Kevin Reid)
1393 * bug fix: complex VOP definitions in "user-space" no longer trigger
1394 package locks. (reported by Zach Beane)
1395 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1396 for funcallable-instance functions, and is no different from
1398 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1399 characters in the current readtable. (reported by Nicholas Neuss)
1400 * bug fix: SUBTYPEP on various CONS types returns more a more
1401 accurate acknowledgment of its certainty.
1402 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1403 lists in some cases. This partially fixes bug 384.
1404 * flush all standard streams before prompting in the REPL and the
1406 * bug fix: signal handling and triggering gc do not conflict
1407 directly anymore, in particular a high frequency sb-sprof does
1408 not prevent gc from running
1409 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1410 approximation for timezone and DST information between the
1411 universal time epoch and the smallest negative 32-bit time_t.
1412 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1413 year 1899 so long as the encoded time is non-negative.
1414 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1415 the correct answer for non-constant rotate amounts when performing
1416 an inline 32-bit rotation.
1418 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1419 there is only one thread in the session
1420 ** bug fix: memory leak for streams created in one thread and
1421 written to in another
1422 ** bug fix: lockup when compiled with gcc4
1423 ** bug fix: race that allows the gc to be triggered when gc is
1425 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1426 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1428 ** bug fix: finished threads are gc'ed properly
1429 * fixed some bugs revealed by Paul Dietz' test suite:
1430 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1431 has arguments with a different set of applicable methods from
1432 the orignal arguments.
1433 ** The type error thrown by MAP now has a correct expected-type
1435 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1436 name a compiled function.
1437 ** Three MISC tests where a large bignum was improperly coerced to
1438 a float that couldn't represent that bignum during type
1439 derivation were fixed.
1440 ** SUBTYPEP can now handle types involving the negation of a
1441 list-form FUNCTION type.
1442 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1443 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1444 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1446 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1447 * New feature: Experimental support for bivalent streams: streams
1448 opened with :element-type :default now allow character and binary
1449 (unsigned-byte 8) I/O
1450 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1451 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1453 * bug fix: no more highly sporadic "couldn't check whether ~S is
1454 readable" when reading a stream and an interrupt hits in the middle
1455 of a select system call
1456 * compiler better recognizes complex arrays (reported by Eduardo
1458 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1459 for variables on SBCL built with linkage-tables. (reported by Luis
1461 * various error reporting improvements.
1462 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1463 (Thanks to Thiemo Seufer)
1464 * bug fix: nested reader invokations work correctly
1465 * bug fix: it is possible to have simultaneous references to foreign
1466 code and foreign data with the same name.
1468 ** added x86-64 support
1469 ** incompatible change: the threading api now works with thread
1470 objects instead of thread ids
1471 ** bug fix: threads are protected from signals and interruption when
1472 starting up or going down
1473 ** bug fix: a race where an exiting thread could lose its stack to gc
1474 ** bug fix: don't halt on infinite error in threads if possible
1475 ** fixed numerous gc deadlocks introduced in the pthread merge
1476 ** bug fix: fixed thread safety issues in read and print
1477 ** bug fix: debugger doesn't hang on session lock if interrupted at
1478 an inappropriate moment
1479 ** bug fix: run-program is now thread safe(r)
1480 ** bug fix: inner with-recursive-lock no longer releases the mutex
1481 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1482 * fixed some bugs revealed by Paul Dietz' test suite:
1483 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1484 EXPECTED-TYPE slots filled.
1485 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1487 ** MAKE-CONDITION accepts classes as type-designators.
1488 ** COMPILE may never return NIL.
1489 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1490 range before calling Unix time functions
1492 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1493 * numerous signal handling fixes to increase stability
1494 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1495 * minor incompatible change: we now correctly canonize default
1496 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1497 as per the MOP, rather than the historical (INITARG INITFUNCTION
1498 INITFORM). (reported by Bruno Haible)
1499 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1500 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1501 for more information.
1502 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1503 pathname is a directory pathname.
1504 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1505 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1507 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1508 :METACLASS argument in addition to a class name. (reported by
1509 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1510 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1511 (reported by Bruno Haible; patch from Gerd Moellmann)
1512 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1514 * bug fix: more cleanups to the floating point exception handling on
1515 x86-64 (thanks to James Knight)
1516 * bug fix: the compiler does not try to constant fold calls of
1517 COERCE and BIT-* functions when they return freshly constructed
1518 objects. (reported by Nikodemus Siivola and Paul Dietz)
1519 * optimization: improved the MIPS versions of generic (in the
1520 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1521 * optimization: direct conversion of (unsigned-byte 32) to floats on
1522 the PowerPC platform.
1523 * optimization: structure instances with raw slots now use less
1524 memory, and probably show better memory locality. (thanks to
1526 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1527 are treated as requests for stack allocation on the x86-64,
1528 Alpha-32, PPC and SPARC platforms.
1529 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1530 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1531 (thanks to Tony Martinez)
1533 ** gcing a dead thread can no longer lead to lockups
1534 ** threads block signals until they are set up properly
1535 ** errno is no longer shared by threads
1536 ** interrupt-thread restores the eflags register on x86
1537 ** fixed some lockups due to gc/thread interaction
1538 * fixed some bugs revealed by Paul Dietz' test suite:
1539 ** invalid dotted lists no longer raise a read error when
1540 *READ-SUPPRESS* is T
1541 ** signal an error if a symbol that names a declaration is used
1542 as the name of a type, or vice versa
1543 ** allow using the (declare (typespec var*)) abbreviation for
1544 (declare (type typespec var*)) with all type specifiers
1545 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1546 ** always return NIL from PPRINT-POP when OBJECT is NIL
1547 ** don't signal errors when pretty-printing malformed LABELS,
1548 FLET or MACROLET forms
1549 ** declarations in a DOLIST body are also in scope for the
1551 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1553 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1556 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1557 * fixed cross-compiler leakages that prevented building a 32-bit
1558 target with a 64-bit host compiler.
1559 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1560 opened with :IF-EXISTS :OVERWRITE.
1561 * fixed bug 281: error for an invalid qualifier in a short-form method
1562 combination method is not signalled until the faulty method is called.
1563 * bug fix: iteration variable type inferrer failed to deal with open
1564 intervals. (reported by Alan Shields)
1565 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1566 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1567 or more runtime options were provided to the sbcl binary.
1568 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1569 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1570 * optimization: there's now a fast-path for fixnum arguments in the
1571 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1572 * optimization: the code generated on x86-64 is more compact thanks
1573 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1574 * bug fix: floating-point exception handling now partly works on
1575 x86-64. (Thanks to James Knight)
1576 * improvement to the MIPS backend from Thiemo Seufer: C-style
1577 64-bit long long arguments and return values to alien functions
1579 * fixed some bugs revealed by Paul Dietz' test suite:
1580 ** the type-error signalled from WARN has a filled-in DATUM slot.
1581 ** the type-error required when a stream is not associated with
1582 a file has the stream as its datum.
1583 ** type-errors on single-floats on x86-64 no longer have
1584 :INVALID-OBJECT as the datum
1585 ** the type-errors signalled for invalid function names now have
1586 a correct expected type
1587 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1588 for structures defined with a :TYPE. Documentation strings for
1589 typed structures are no longer immediately discarded
1590 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1591 broadcast streams and synonym streams. FILE-LENGTH now also works
1592 on broadcast streams.
1594 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1595 * incompatible change: the --noprogrammer option, deprecated since
1596 version 0.7.5, has been removed. Please use the equivalent
1597 --disable-debugger option instead.
1598 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1600 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1601 inoperational for a while now) has been completely removed.
1602 * null lexical environments are now printed as #<NULL-LEXENV>,
1603 significantly reducing the amount of clutter in typical backtraces.
1604 * documentation on weak pointers, finalization, and after GC hooks
1605 has been added to the manual.
1606 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1607 as other specialized array types, is much faster. SUBSEQ and
1608 COPY-SEQ on such arrays have also been sped up.
1609 * optimization: EQL is now more efficient when at least other argument
1610 is known to be of type (OR FIXNUM (NOT NUMBER)).
1611 * fixed bug: the runtime is now less vulnerable to changes in the
1612 size of the SBCL object on OS X, and virtual memory is reserved for
1613 all spaces that need to be at a fixed address.
1614 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1615 * fixed bug: finalizers and after GC hooks that cause consing are now
1617 * fixed bug: compiler error messages and summaries are now printed to
1618 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1619 * fixed inference of the upper bound of an iteration variable.
1620 (reported by Rajat Datta).
1621 * fixed bug 211e: calling local functions with duplicated constant
1622 keyword argument no longer causes a bogus style warning about an
1624 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1625 ftype declaration to be disregarded. (reported by Dave Roberts)
1626 * fixed bug 373: caused by erronous compilation of references to alien
1627 variables in the runtime on ppc/darwin.
1628 * fixed bug 376: CONJUGATE type deriver.
1629 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1630 (reported by Baughn on #lisp)
1631 * a cross-compiler bug on non-x86 platforms has been identified and
1632 fixed. (thanks to Bruno Haible)
1633 * improvements to the MIPS runtime code for increased stability.
1634 (thanks to Thiemo Seufer)
1635 * increased the maximimum compact environment size to allow
1636 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1637 * partial workaround for bug 135: don't name the function we're
1638 calling for hairy cases of EVAL, so as not to accumulate one environment
1639 entry per call to EVAL. (reported by Kevin Reid)
1640 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1641 * optimization: type testing for non-vector arrays should be faster.
1642 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1644 * bug fix: setting 31st element of a bit vector to zero did not work
1646 * bug fix: redefining a class definition which failed due to a
1647 previous accessor / function clash now works (but see BUGS entry
1648 #380 for more problems in this area). (thanks to Zach Beane)
1649 * the long form of DEFINE-METHOD-COMBINATION disables method group
1650 checking when given a single method group with pattern *. (thanks
1652 * on x86 compiler supports stack allocation of results of simple
1653 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1654 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1655 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1656 * fixed some bugs related to Unicode integration:
1657 ** the restarts for recovering from input and output encoding
1658 errors only appear when there is in fact such an error to
1660 * fixed some bugs revealed by Paul Dietz' test suite:
1661 ** MISC.549 and similar: late transformation of unsafe type
1662 assertions into derived types caused unexpected code
1664 ** SCALE-FLOAT type deriver is less wrong.
1665 ** type derivers for EXP, LOG and similar functions compute result
1666 types for complex arguments better.
1667 ** (MISC.563) CONJUGATE type deriver works for very restricted
1669 ** out-of-line type testers for character strings are available.
1670 ** EQUAL compiler transform understands specialness of objects
1672 ** accessing double-floats stored on the stack now works on x86-64.
1673 ** debugger internals could sometimes create invalid lispobjs,
1674 resulting in GC crashes.
1675 ** MISC.548: type check weakening can convert required type into
1677 ** initialization forms for bindings are not in scope of free special
1680 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1681 * incompatible change: thread support for non-NPTL systems has
1682 been removed - locking is buggy and unreliable. A threaded
1683 SBCL build will now warn at startup and refuse to create
1684 new threads, unless futex support is detected in the kernel
1685 * incompatible change: the top level REPL now has only an ABORT
1686 restart associated with it, not TOPLEVEL and ABORT as it used to.
1687 TOP and TOPLEVEL are now available as debugger commands for
1688 returning to the top level.
1689 * incompatible change: forms evaluated in the REPL now use the
1690 global optimization policy.
1691 * incompatible change: user- and system-initialization files are
1692 no longer processed with LOAD, but by READ and EVAL; hence the
1693 global optimization policy, startup package, readtable, etc,
1695 * The .fasl file format number has been incremented because of
1696 various incompatible changes.
1697 * internal entry point details and argument counts no longer appear
1698 in backtraces unless explicitly requested by setting
1699 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1700 * built-in and standard functions no longer have names like "top
1701 level local call to FOO".
1702 * fixed bug 32: functions defined in non-null lexical environments
1703 now have more legible printed representation
1704 * fixed bug 33: functions defined in non-null lexical environemnts
1705 are now more amenable to inspection by INSPECT.
1706 * workaround for bug 354: XEPs no longer appear in backtraces unless
1707 explicitly requested.
1708 * fixed bug: receiving the signal which results from attempting to
1709 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1710 SIGBUS on other *BSDs) on architectures where the C stack is also
1711 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1712 to debugger confusion or wild instability. (reported by Cheuksan
1714 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1715 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1716 (reported by Lutz Euler)
1717 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1718 compiled code. (reported by Paul Dietz)
1719 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1720 specialized parameter to a method to something that is not TYPEP
1721 the specializer is now possible.
1722 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1723 face of package deletion.
1724 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1725 pathname. (reported by David Tolpin on c.l.l)
1726 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1727 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1728 STRING-STREAM and FILE-STREAM, respectively.
1729 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1730 than previously; in addition, its test suite is now run on build.
1731 (thanks to Luke Gorrie)
1732 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1733 * optimization: added a immediate representation for single-floats
1735 * fixed some bugs revealed by Paul Dietz' test suite:
1736 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1737 ** attempting to create a package with a colliding nickname causes
1738 correctable errors to be signalled.
1739 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1740 ** several x86-64 backend bugs related to sign-extension of immediate
1743 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1744 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1745 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1747 * fixed loading of multiply forward-referenced layouts.
1748 (thanks to Cheuksan Wang)
1749 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1750 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1751 (see COMPILE-FILE documentation for details of :PRINT option)
1752 * fixed bugs 19 and 317: fixed-format floating point printing is
1753 more accurate. This also fixes a bug reported by Adam Warner
1754 related to the ~@F format directive.
1755 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1757 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1758 dispatch table if the from-char is a dispatch macro character.
1759 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1760 by the wordsize no longer ignore the last word. (reported by Lutz
1762 * fixed bug in type checking of dynamic-extent variables. (reported
1764 * optimization: sequence traversal functions use their freedom to
1765 coerce function designators to functions.
1766 * optimization: code with many calls to CLOS methods specialized on
1767 CLOS classes has had redundant type checks removed.
1768 * optimization: added declarations to speed up operations that access
1769 the internal character database (for example STRING-UPCASE)
1770 * optimization: comparison operations between floats and sufficiently small
1771 fixnums no longer create extra rationals
1772 * fixed some bugs related to Unicode integration:
1773 ** portions of multibyte characters at the end of buffers for
1774 character-based file input are correctly transferred to the
1775 start of the buffer at the next read.
1776 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1777 passing it through to OPEN.
1778 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1779 argument given to internal calls to OPEN.
1780 ** the built-in comment readers (introduced by character sequences
1781 ";" and "#|") are more forgiving to encoding errors; they will
1782 STYLE-WARN and then attempt to resync the stream at a character
1783 boundary. (thanks to Teemu Kalvas)
1784 * fixed some bugs in the x86-64 port:
1785 ** Negative short int return values from c-calls are sign-extended
1787 ** The stack is aligned to 16-bytes for c-calls, as required by
1788 the ABI. (thanks to Cheuksan Wang)
1789 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1790 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1791 for enums. (thanks to Vincent Arkesteijn)
1792 ** Multiple small optimizations and bugfixes for floating point
1794 * fixed some bugs revealed by Paul Dietz' test suite:
1795 ** Space, Tab, Linefeed, Return and Page have the invalid
1796 secondary constituent character trait.
1797 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1799 ** WITH-INPUT-FROM-STRING should only update the index place on
1801 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1802 works more reliably.
1803 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1804 with one-dimensional arrays with fill pointers.
1805 ** TYPE-OF failed on a complex with an integer realpart and a
1807 ** compiler failure during type inference for the code of form
1808 (IF (EQL X (THE ...)) ...) (MISC.535).
1810 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1811 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1812 platform. The port supports 61-bit fixnums, large memory spaces
1813 and reloading shared object files.
1814 * enhancement: saving cores with foreign code loaded is now
1815 supported on ppc/Darwin in addition to the previously supported
1817 * enhancement: the statistical profiler now walks deeper into the
1818 call stack for more meaningful call-graphs and accrued time
1819 reports (x86/x86-64 only). It also now reports time spent in
1821 * enhancement: it is now possible to trace most individual methods
1822 of a generic function in addition to tracing the generic function
1824 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1826 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1827 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1828 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1829 produces an error. (thanks to Vincent Arkesteijn)
1830 * bug fix: NAMESTRING on pathnames with :WILD components in their
1831 directories works correctly. (thanks to Artem V. Andreev)
1832 * fixed bug 125: compiler preserves identity of closures. (reported
1834 * bug fixed: functions with &REST arguments sometimes failed with
1835 "Undefined function" when compiled with (DEBUG 3). (reported by
1837 * bug fix: overflow during compiling of setting element of a bit
1838 vector with constant index and value. (reported by Timmy Douglas)
1839 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1840 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1841 * on x86 compiler partially supports stack allocation of dynamic-extent
1843 * GO and RETURN-FROM do not check the extent of their exit points
1844 when compiled with SAFETY 0.
1845 * fixed some bugs related to Unicode integration:
1846 ** encoding and decoding errors are now much more robustly
1847 handled; it should now be possible to recover even from invalid
1848 input or output to the terminal. (thanks to Teemu Kalvas)
1849 ** provided a first cut at implementing STRING-TO-OCTETS and
1850 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1851 ** altered the SB-MD5 contributed module slightly, changing the
1852 interface just enough for it to be supportable for builds where
1853 lisp characters are not eight bits.
1854 * fixed some bugs revealed by Paul Dietz' test suite:
1855 ** the FORMATTER-generated functions for ~V[ conditionals require
1856 the correct number of arguments.
1857 ** READ-FROM-STRING returns the mandated second value when applied
1858 to displaced strings.
1859 ** the #\Rubout and #\Backspace characters are treated as invalid
1860 constituent characters by the tokenizer.
1862 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1863 * new feature: reloading changed shared object files with
1864 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1865 * new feature: references to foreign variables and functions
1866 can now be compiled and loaded before the corresponding shared
1867 object file is loaded, as long as the foreign definitions are
1868 available at runtime.
1869 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1870 supported. (thanks to Dan Debertin)
1871 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1872 just keywords. (thanks to Vincent Arkesteijn)
1873 * fixed bug #331: structure-class instances corresponding to
1874 DEFSTRUCT forms are now created eagerly.
1875 * fixed bug #345: backtraces from calls to undefined functions work
1876 on x86 as well. Related bug #61 is now also partially fixed on x86
1877 (backtraces from throws to unknown catch tags.)
1878 * bug fix: lambda-list parsing is now stricter vrt. order and number
1879 of lambda-list keywords.
1880 * bug fix: as specified by AMOP, an error is signalled if a
1881 class-option appears multiple times in a DEFCLASS form. (reported
1883 * bug fix: RANDOM can be compiled when the compiler derives the type
1884 of its numeric argument as a disjoint set of small integers.
1885 (reported by Paul Dietz)
1886 * bug fix: starting a core saved with shared objects loaded when
1887 those objects are not available no longer causes threaded SBCL to
1888 hang. (reported by Sean Ross)
1889 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1891 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1892 argument, which must always be NIL. (reported by Kalle Niemitalo)
1893 * bug fix: printing 1.0d+23 no longer results in an error.
1894 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1895 * bug fix: structure slot setters preserve evaluation order. (thanks
1897 * fixed some bugs related to Unicode integration:
1898 ** RUN-PROGRAM can allow its child to take input from a Lisp
1899 stream. (reported by Stefan Scholl)
1900 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1901 locales. (reported by Ken Causey)
1902 * fixed some bugs revealed by Paul Dietz' test suite:
1903 ** INCF, DECF and REMF evaluate their place form as specified in
1905 ** FORMATTER expands ~{ ~} iteration directives with V or #
1906 parameters correctly.
1907 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1908 consequent uses no arguments correctly.
1909 ** the system has a partial understanding of the (COMPLEX RATIO)
1911 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1914 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1915 * new feature: a build-time option (controlled by the :SB-UNICODE
1916 keyword feature, enabled by default) for building the system with
1917 support for the entire 21-bit character space defined by the
1919 * new feature: the system now has rudimentary external-format
1920 support; the primary user-visible change at this time is that
1921 characters with the high bit set (such as lower-case-e-acute) will
1922 print correctly to a terminal in a UTF-8 environment.
1923 * minor incompatible change: BASE-CHAR no longer names a class;
1924 however, CHARACTER continues to do so, as required by ANSI.
1925 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1926 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1927 has been moved to the SB-EXT package (temporarily re-exported from
1929 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1930 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1932 * fixed bug #351: better error-handlind and reporting for malformed
1934 * fixed bug #350: bignum-printing is now more memory-efficient,
1935 allowing printing of very large bignums, eg. (expt 2 10000000).
1936 (reported by Bruno Haible)
1937 * fixed bug #302: better primitive-type selection for intersection
1939 * fixed bug #308: non-graphic characters now all have names, as
1940 required. (reported by Bruno Haible)
1941 * bug fix: redefining a class with different superclasses now correctly
1942 removes it from the direct-subclasses of its previous superclasses.
1943 (reported by David Morse)
1944 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1945 new value now works. (reported by Bruno Haible)
1946 * bug fix: correct canonicalization of multiple non-standard slot
1947 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1948 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1949 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1950 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1951 now exists, an signals an error.
1952 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1953 during macroexpansion for non-iteration variables as well. (reported
1954 by Bruno Haible for CMUCL)
1955 * bug fix: Cyclic structures and unprintable objects in compiler
1956 messages no longer cause errors. (reported by Bruno Haible)
1957 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1958 and READ-FROM-STRING all now return a primary value of NIL if
1959 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1960 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1961 (reported by Bruno Haible for CMUCL)
1962 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1963 do not designate a condition. (reported by Bruno Haible for
1965 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1966 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1967 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1968 permitting resolution in favour of any of the conflicting symbols.
1969 (reported by Bruno Haible for CMUCL)
1970 * bug fix: EQUAL compiler optimizations is less aggressive on
1971 strings which can potentially compare true despite having distinct
1972 specialized array element types.
1973 * bug fix: unit enumerations can be defined without dividing by
1974 zero. (thanks to Vincent Arkesteijn)
1975 * FORMAT compile-time argument count checking has been enhanced.
1976 (report from Bruno Haible for CMUCL)
1977 * a partial workaround for the bug 262: the compiler does not try to
1978 inline-expand a local function doing RETURN-FROM from a deleted
1979 BLOCK. (thanks to Peter Denno for the bug report and to David
1980 Wragg for the simple test case)
1981 * fixed some bugs revealed by Paul Dietz' test suite:
1982 ** RENAME-PACKAGE allows all package designators as new package
1984 ** constraint propagation and lambda variable substitution are
1985 more cautious in dealing with partially deleted code.
1986 ** compiler sometimes lost reoptimization passes.
1987 ** CERROR, given a condition as condition designator, passes its
1988 remaining arguments to the continue format control without
1990 ** Case-altering FORMAT directives work correctly on non-ASCII
1992 ** The REAL type specifier handles bounds outside the floating
1993 point ranges without signalling FLOATING-POINT-OVERFLOW.
1994 ** Functions with IR1-transformations can create intercomponent
1995 references to global functions.
1996 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1998 ** FORMAT ~R treats a nil value for its first parameter correctly.
2000 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2001 * enhancement: saving cores with foreign code loaded is now
2002 supported on x86/NetBSD and sparc/Linux in addition to the previously
2003 supported platforms.
2004 * bug fix: on some platforms repeated installations caused multiple
2005 copies of HTML documentation to be installed -- should not happen
2006 any more. (reported by Stefan Scholl)
2007 * bug fix: parsing self-recursive alien record types multiple times
2008 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2009 original patch by Helmut Eller for CMUCL)
2010 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2011 (thanks to Richard Kreuter)
2012 * bug fix: defining classes whose accessors are methods on existing
2013 generic functions in other (locked) packages no longer signals
2014 bogus package lock violations. (reported by François-René Rideau)
2015 * bug fix: special variables as DEFMETHOD parameters no longer have
2016 associated bogus type declarations. (reported by David Wragg and
2018 * bug fix: read-write consistency on streams of element-type
2019 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2020 * bug fix: redefiniton of the only method of a generic function with
2021 no DEFGENERIC no longer emits a full WARNING. In addition,
2022 redefinition of generic functions with no DEFGENERIC to an
2023 incompatible lambda list now signals an error. (thanks to Zach
2025 * bug fix: DEFGENERIC now works even when there's a function of the
2026 same name in an enclosing lexical environment. (thanks to Zach
2028 * fixed compiler failure, caused by instrumenting code during
2029 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2030 * optimization: added loop analysis and improved register allocation
2031 * fixed some bugs revealed by Paul Dietz' test suite:
2032 ** POSITION on displaced vectors with non-zero displacement
2033 returns the right answer.
2034 ** (SIMPLE-STRING) is a valid type specifier for sequence
2036 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2038 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2039 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2041 ** The justification version of the FORMAT ~< directive treats
2042 non-zero minpad parameter correctly.
2044 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2045 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2046 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2047 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2048 the supported interface.
2049 * new feature: Single-stepping of code compiled with DEBUG 2 or
2050 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2051 * new feature: saving cores with foreign code loaded is now
2052 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2053 Timothy Moore's work for CMUCL)
2054 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2055 parameters to * instead of NIL if no initform is supplied.
2056 (reported by Johan Bockgaard)
2057 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2058 correctly when FUNCALL appears as the car of the form. Note:
2059 despite this FUNCALL forms are not currently subject to
2060 compiler-macro expansion. (port of Raymond Toy's fix for the
2061 same from CMUCL, reported by Johan Bockgaard)
2062 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2063 (thanks for Teemu Kalvas)
2064 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2065 case instead of terminating the iteration (thanks for Julian
2066 Squires, Sean Champ and Raymond Toy)
2067 * bug fix: incorrect expansion of defgeneric that caused a style
2068 warning. (thanks for Zach Beane)
2069 * bug fix: slot accessor effective method computation works properly
2070 for classes with multiple non-standard applicable methods on
2071 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2072 * on x86 compiler supports stack allocation of results of LIST and
2073 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2074 CMUCL implementation by Gerd Moellmann)
2075 * fixed some bugs revealed by Paul Dietz' test suite:
2076 ** FORMAT strings with both the ~<~:;~> form of the justification
2077 directive and pretty-printing directives cause an error.
2079 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2080 * incompatible change: the internal functions
2081 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2082 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2083 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2084 instead of the old functions.
2085 * new feature: on platforms where "dladdr" is available foreign
2086 function names now appear in backtraces. (based on Helmut Eller's
2088 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2089 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2091 * documentation: ASDF manual is now distributed with SBCL.
2092 * bug fix: SBCL can now load its contributed modules with REQUIRE
2093 even if the system-provided entries have been removed from
2094 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2096 * bug fix: backtraces involving undefined functions or assembly
2097 routines are more informative. (thanks to Brian Downing)
2098 * bug fix: mutually referent alien structures now work correctly.
2099 (reported by Rick Taube)
2100 * bug fix: structures defined by WITH-ALIEN can be referred to
2101 within other definitions in the same WITH-ALIEN.
2102 * bug fix: division operators (MOD, TRUNCATE and the like) with
2103 constant zero divisors and integer dividends no longer generate
2105 * bug fix: provide default methods for INPUT-STREAM-P and
2106 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2107 * bug fix: improve the detection and resolution of MOP metacycles
2108 (where effective method computation depends on the generic function
2109 having its effective method computed). (reported by Bruno Haible)
2110 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2111 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2112 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2113 * bug fix: pathnames with relative directory components can be
2114 represented relative to default pathnames.
2115 * optimization: in taking the GCD of bignums, reduce the two bignums
2116 to approximately the same size (using Euclid's algorithm) before
2117 applying the more sophisticated binary GCD. (thanks to Juho
2119 * optimization: COUNT on bitvectors now operates word-at-a-time.
2120 * optimization: ASH with a positive, but not necessarily constant,
2121 (leftwards) shift, when in a modular context, is compiled to a
2123 * fixed some bugs revealed by Paul Dietz' test suite:
2124 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2125 if the corresponding argument is NIL.
2126 ** FORMAT directives accepting a minpad parameter treat negative
2128 ** literal commas as character parameters to FORMAT directives are
2130 ** literal spaces directly after ~<Newline> directives within a
2131 format-logical-block (~:< ~@:>) do not induce :FILL-style
2132 conditional newlines.
2133 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2134 as specified: it no longer includes conditional newlines).
2135 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2137 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2138 :PER-LINE-PREFIX argument does not evaluate to a string.
2140 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2141 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2142 the manual for details; for now, package locks can be disabled by
2143 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2144 if no major problems are found then it is likely that they will be
2145 compiled in unconditionally.
2146 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2147 unsupported operators on all platforms. To load a shared library
2148 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2149 object file, link it into a shared library outside of SBCL and
2150 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2152 * bug workaround: ROOM T and the GC were not working together
2153 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2154 reporting facility aren't preserved. That reporting has been
2155 disabled, so now until and unless someone figures out how to make it
2156 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2157 * minor incompatible change: as threatened around sbcl-0.8.0, the
2158 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2159 The symbols are also exported from SB-PCL for backwards
2160 compatibility, but more so than before SB-PCL should be treated as
2161 an implementation-internal package.
2162 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2164 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2165 It is known as of this release not to work on the Alpha, however.
2166 * fixed bug #167: errors signalled due to illegal syntax in method
2167 bodies are now more legible.
2168 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2169 designators and can hence be used with TYPEP.
2170 * fixed bug #333: CHECK-TYPE now ensures that the type error
2171 signalled, if any, has the right object to be accessed by
2172 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2173 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2174 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2176 * fixed bug #334: programmatic addition of slots using specialized
2177 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2178 and :ALLOCATION :CLASS slots.
2179 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2180 float exponents. (rereported by Peter Seibel)
2181 * fixed a bug: #\Space (and other whitespace characters) are no
2182 longer considered to be macro characters in standard syntax by
2183 GET-MACRO-CHARACTER.
2184 * fixed bug: initialization of condition class metaobjects no longer
2185 causes an instance of the condition to be created. (reported by
2187 * fixed bug: it is now possible to have slots such that
2188 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2189 description is neither :INSTANCE nor :CLASS.
2190 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2191 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2192 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2193 non-standard SLOT-DEFINITION-ALLOCATION values.
2194 * fixed bugs: various race conditions handling exiting threads.
2195 CL-PPCRE's thread torture test now passes.
2196 * fixed arguably-a-bug: GC time proportional to number of threads in
2197 system even when most of them are idle
2198 * optimization: improved performance of BIT and SBIT on bit-vectors.
2199 * .fasl file incompatibility: The fasl file version number has
2200 been incremented because of changes associated with package locks.
2202 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2203 * minor incompatible change: the system no longer provides
2204 optimization hints (conditions which are TYPEP
2205 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2206 mode; these hints are still emitted when higher SPEED optimization
2208 * new contrib module: a sampling profiler (profiling by statistical
2209 sampling, rather than by instrumenting functions) is available as
2210 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2211 * the behaviour of the standard function ED is now customizeable by
2212 third parties through a hook variable: see ED's documentation
2213 string for information on the protocol.
2214 * the compiler no longer emits efficiency notes for (FUNCALL X)
2215 when the type of X is uncertain under default optimization
2217 * fixed bug 276: mutating a binding of a specialized parameter to a
2218 method to something that is not TYPEP the specializer is now
2220 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2221 exhibit the required behaviour on the x86 platform. (thanks to
2222 Peter van Eynde, Eric Marsden and Bruno Haible)
2223 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2224 even for difficult arguments. (reported by Peter Graves)
2225 * fixed bug 141a: the backquote printer now descends quoted
2227 * fixed another bug in backquote printing: no more destructive
2228 modification of the form's list structure. (reported by Brian
2230 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2231 move between its address being taken and the call to
2232 interrupt_thread, fixing a crashing race condition.
2233 * the SB-POSIX contrib implementation has been adjusted so that it
2234 no longer exhibits ridiculously poor performance when constructing
2235 instances corresponding to C structs.
2237 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2238 * minor incompatible change: the sb-grovel contrib now treats C
2239 structures as alien (in the sense of SB-ALIEN) objects rather than
2240 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2241 has implications for memory management of client code
2242 (sb-grovel-returned objects must now be manually managed) and for
2243 type safety (alien objects now have full types).
2244 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2245 used to control emission of compiler diagnostics, rather than the
2246 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2247 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2248 quality should be considered deprecated.
2249 * install.sh now installs the user manual as well
2250 * (not quite a new documentable feature, but worth considering in
2251 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2252 beginnings of a semantically meaningful condition hierarchy is
2253 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2255 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2256 designator as the defaults argument.
2257 * fixed bug: Displaced arrays whose displaced-to array has become
2258 too small now cause ARRAY-DIMENSION to signal an error, providing
2259 for safer bounds-checking. (reported by Bruno Haible)
2260 * fixed bug: DEFCLASS slot definitions with identical :READER and
2261 :WRITER names now signal a reasonable error. (reported by Thomas
2263 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2265 * fixed bug: Invalid slot specification errors now print correctly.
2266 (thanks to Zach Beane)
2267 * fixed bug 320: Shared to local slot value transfers in class
2268 redefinitions now happen corrently from superclasses as
2269 well. (reported by Bruno Haible)
2270 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2272 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2273 improper lists correctly. (reported by Bruno Haible)
2274 * fixed bug 313: source-transform for <fun-name> was erroneously
2275 applied to a call of a value of a variable with name <fun-name>.
2276 (reported by Antonio Menezes Leitao)
2277 * fixed bug 307: The obsolete instance protocol ensures that
2278 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2279 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2280 remove dead unknown-values globs from the middle of the stack.
2281 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2282 easier to resume long computations after using *BREAK-ON-SIGNALS*
2283 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2284 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2285 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2286 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2287 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2288 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2290 * fixed a (fairly theoretical) bug in string printing: if
2291 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2292 does not have array-element-type equal to the most general string
2294 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2295 function name. (thanks to Nikodemus Siivola)
2296 * fixed bug: compile-time format string checker failed on
2297 non-closed ~{. (reported by Thomas F Burdick)
2298 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2299 don't warn on legal loop code involving more than one aggregate
2300 boolean. (thanks to Nikodemus Siivola)
2301 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2302 in keyword position to MAKE-PACKAGE was not being reported as
2304 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2305 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2307 * fixed bugs 280 and 312: the checking for multiple definitions in a
2308 file is less likely to become confused by uses of inline
2310 * fixed bug: the #S reader macro performs the keyword coercion
2311 specified for slot names. (reported by Kalle Niemitalo)
2312 * fixed bug: lambda lists may contain symbols whose names start with
2313 & but are not lambda-list-keywords; their occurrence triggers a
2315 * fixed bug 321: define-method-combination argument lambda lists do
2316 not cause invalid code to be generated when &optional or &aux
2317 variables have default values. (reported by Bruno Haible)
2318 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2319 have CLOS classes; this appears also to have fixed reported
2320 instability in stack exhaustion detection.
2321 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2322 MUFFLE-WARNING when no associated restart is present is now
2324 * optimization: rearranged the expansion of various defining macros
2325 so that each expands into only one top-level form in a
2326 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2328 * optimization: used a previously little-used slot in symbols to
2329 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2331 * fixed some bugs revealed by Paul Dietz' test suite:
2332 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2334 ** Fixed an optimization bug related to inheritance of initforms
2335 from local to shared slots.
2336 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2337 ** CAST optimizer forgot to flush argument derived type.
2338 ** print/read consistency on floats is now orders of magnitude
2339 more likely. (thanks also to Bruno Haible for a similar report
2341 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2342 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2343 for objects of type REAL. Make it so.
2344 ** SXHASH is specified (infelicitously) to respect similarity,
2345 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2346 it so. (thanks to Markus Ziegler)
2347 ** on the Alpha, the compiler succeeds in compiling functions
2348 returning a known number of arguments greater than 63.
2349 ** fixed handling of invalid NIL arguments in keyword position
2351 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2353 ** provided a readably-printable representation for RANDOM-STATE
2355 ** ensured that pathnames, where they have a namestring, always
2356 print using #P"..." syntax.
2358 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2359 * Support for the forthcoming 2.0 version of the NetBSD kernel
2360 running on x86 hardware has been added. (thanks to Perry
2361 E. Metzger most immediately, and others for their past work)
2362 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2363 current release version; SBCL's OpenBSD support had been broken
2364 since about the time of OpenBSD's switch to ELF binary format.)
2365 * [placeholder for DX summary]
2366 ** user code with &REST lists declared dynamic-extent, under high
2367 speed or space and low safety and debug optimization policy.
2368 * The manual has been converted to Texinfo format and the debugger
2369 chapter from the cmucl manual has been added.
2370 * A facility has been added to extract documentation strings from
2371 sbcl and store them as Texinfo-formatted snippets for inclusion in
2372 the manual (via Texinfo's @include directive)
2373 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2374 the test case to Dave Roberts)
2375 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2376 pointers. (reported by Sean Ross)
2377 * bug fix: PROFILE output is printed nicely even for large numerical
2378 values. (thanks to Zach Beane)
2379 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2380 greater than 32 handle EOF correctly.
2381 * bug fix: on X86 an immediate argument of the IMUL instruction is
2382 correctly printed in disassembly. (thanks to Lutz Euler)
2383 * bug fix: class slots in redefined classes preserve their old
2384 values. (thanks to Bruno Haible and Nikodemus Siivola)
2385 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2386 succeeds. (reported by Marco Baringer)
2387 * fixed some bugs revealed by Paul Dietz' test suite:
2388 ** READ-SEQUENCE now works on ECHO-STREAMs.
2389 ** RATIONALIZE works more according to its specification. (thanks
2392 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2393 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2394 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2395 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2396 should matter to you only if you rebind the printer control
2397 variables and then find you want different bindings in the
2398 debugger than in the ordinary execution of your program.)
2399 * The runtime build system has been tweaked to support building
2400 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2401 assembler and linker. (thanks to Nikodemus Siivola)
2402 * Unbound, undefined, undeclared variables now trigger full
2403 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2404 more often programmer error than deliberate exploitation of undefined
2406 * optimization: the hash algorithm for strings has changed to one
2407 that is less vulnerable to spurious collisions. (thanks to Juho
2409 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2410 needless bounds checking. (thanks to Juho Snellman)
2411 * optimization: implemented multiplication as a modular
2412 (UNSIGNED-BYTE 32) operation on the PPC backend.
2413 * fixed some bugs revealed by Paul Dietz' test suite:
2414 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2416 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2417 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2418 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2420 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2422 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2424 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2425 their output stream on EOF from read.
2426 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2427 have been read to end-of-file.
2428 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2430 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2431 description of determination of which consecutive characters
2433 ** Printing the "Space" character with escaping on now yields "#\\ ",
2434 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2435 ** Reading floating-point numbers with *READ-BASE* set to a number
2436 less than 10 works correctly.
2437 ** Reading floating-point numbers with *READ-BASE* set to a number
2438 more than 10 works correctly.
2439 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2440 the readtable currently in effect.
2442 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2443 * minor incompatible change: parsing of namestrings on a physical
2444 (Unix) host has changed; numbers after the final #\. in a
2445 namestring are no longer interpreted as a version field. This is
2446 intented to be largely invisible to the user, except that the
2447 meaning of the namestring "*.*.*" has changed: it now refers to a
2448 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2449 should usually be replaced by
2450 (make-pathname :name :wild :type :wild :version :wild)
2451 with the added benefit that this is more likely to be portable.
2452 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2453 option to OPEN now signals an error if the file being opened
2454 exists; this may have an impact on existing code.
2455 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2456 signals received, on the PowerPC platforms (both Linux and
2457 Darwin). (thanks to Pierre Mai for pointing out the location of
2459 * several fixes on OS X: The system now builds and runs cleanly on
2460 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2461 (thanks to Brian Mastenbrook)
2462 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2463 second-resolution: integer multiples of 1/3600 between -24 and 24.
2464 (thanks to Vincent Arkesteijn)
2465 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2466 type when called with 1 argument; PEEK-CHAR checked type of
2467 PEEK-TYPE only after having read first character from a
2468 stream. (reported by Peter Graves)
2469 * bug fix: the garbage collector now has much better locality
2470 behaviour, and in particular no longer treats all memory as being
2471 exhausted when many small objects point to each other in a deeply
2473 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2474 recognized as being TYPEP their class.
2475 * bug fix: the PUSHNEW documentation string has been corrected.
2476 (thanks to Vincent Arkesteijn)
2477 * bug fix: defaulting of the value for the last of an atypically
2478 large number of multiple values being bound was not being
2479 performed correctly on the Alpha or PPC platforms
2480 * optimization: implemented multiplication as a modular
2481 (UNSIGNED-BYTE 32) operation on the x86 backend.
2482 * optimization: SEARCH on simple-base-strings can now be open-coded.
2483 (see also contrib/compiler-extras.lisp for inspiration for
2484 teaching the compiler about the Boyer-Moore algorithm).
2485 * value, returned by MAX (and MIN) called with several EQUALP, but
2486 not EQL, arguments now does not depend on compiler settings.
2487 * fixed some bugs revealed by Paul Dietz' test suite:
2488 ** in stack analysis liveness information is propagated from
2489 non-local entry points.
2490 ** pathwise CAST removing failed when the CAST node did not start
2492 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2493 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2495 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2496 host is already defined.
2497 ** RENAME-FILE works on streams instead of signalling an internal
2499 ** PEEK-CHAR uses the current readtable when determining whether
2500 or not a character is whitespace.
2501 ** MERGE-PATHNAMES handles the case when the pathname does not
2502 specify a name while the default-pathname specifies a version.
2503 ** Pathnames now stand a better chance of respecting print/read
2505 ** Attempting to use standardized file system operators with a
2506 pathname with invalid :DIRECTORY components signals a
2508 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2510 ** DIRECTORY on logical pathnames is more correct.
2511 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2512 signal a TYPE-ERROR if their argument is not a stream.
2513 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2514 designator argument does not designate a stream.
2515 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2516 examining the synonym.
2517 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2519 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2520 element-type larger than ([UN]SIGNED-BYTE 32).
2522 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2523 * When built with the :SB-FUTEX feature, threaded builds now take
2524 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2525 for faster/more reliable mutex and condition variable support.
2526 * Incompatible change (but one you probably shouldn't have been using
2527 anyway): the interface and code for arbitrating between multiple
2528 threads in the same user session has been redesigned.
2529 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2530 error when called without an explicit environment argument.
2531 (thanks to Nikodemus Siivola)
2532 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2533 confused about their position. (thanks to Adam Warner and Gerd
2535 * bug fix: Pretty printing backquoted forms with unquotations in the
2536 argument list position of various code constructs such as LAMBDA
2537 now works correctly. (reported by Paul Dietz)
2538 * bug fix: Pretty printing unquotations no longer loses all
2539 stream position information.
2540 * optimization: performance of string output streams is now less
2541 poor for multiple small sequence writes.
2542 * optimization: performance of CSUBTYPEP in the presence of complex
2543 expressions involving CONS and NOT many times has been improved.
2544 (reported by Paul Dietz)
2545 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2547 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2548 streams for the REPL, for the debugger, and so on. (thanks to
2550 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2551 * fixed some bugs revealed by Paul Dietz' test suite:
2552 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2553 is ignored if it is too small, rather than propagating through
2554 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2555 ** extremely complex negations of CONS types were not being
2556 sufficiently canonicalized, leading to inconsistencies in
2558 ** VALUES tranformer lost derived type.
2560 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2561 * fixed a bootstrapping bug: the build process no longer assumes
2562 that the various BOOLE-related constants have the same value in
2563 host and target lisps. (noted by Paul Dietz' test suite on an
2564 SBCL binary built from CLISP)
2565 * The system can now be dynamically linked on the MIPS platform,
2566 which enables dynamic loading of foreign code from Lisp. (thanks
2567 to Ralf Baechle for discussions on the MIPS ABI)
2568 * The system now records debugging information for its own source
2569 files in a filesystem-position-independent manner, relative to
2570 the "SYS" logical host.
2571 * fixed a compiler bug: MV-LET convertion did not check references
2572 to the "max args" entry point. (reported by Brian Downing)
2573 * tweaked disassembly notes to be less confident about proclaiming
2574 some instruction as an LRA. (thanks to Brian Downing)
2575 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2576 now each have their own history, command character, and other
2577 characteristics. (thanks to David Lichteblau)
2578 * fixed some bugs revealed by Paul Dietz' test suite:
2579 ** compiler failure in compiling LOGAND expressions including a
2581 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2582 shift greater than 32.
2583 ** FLUSH-DEST did not mark blocks for type check regeneration.
2584 ** HANDLER-CASE failed to accept declarations in handler clauses
2585 in some circumstances.
2587 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2588 * New code in contrib/sb-introspect (still probably not entirely
2589 stable yet) provides some support for smart Lisp development
2590 environments like SLIME.
2591 * The conditions signalled for errors occurring when loading .fasl
2592 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2593 in a way which should help ASDF recover gracefully.
2594 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2595 stuff has been cleaned up. If you code contrib/ stuff, this might
2596 affect you, and you can look at contrib/README, contrib/STANDARDS,
2597 and/or the 0.8.4.27 diff to check.
2598 * In full calls the compiler now does not generate checks for declared
2599 argument types for all arguments.
2600 * various threading fixes
2601 ** and some experimental patches which didn't make it into
2602 the main tree for this release, but which are shipped in
2603 contrib/experimental-thread.patch as a possible fix for some
2604 failures (deadlock, spinning...) in GC-intensive multithreaded
2606 * fixed PPC build problem (source code incompatibility of different
2607 library versions): added offsetof() hackery which attempts to divine
2608 where glibc maintainers put uc_mcontext today
2609 * fixed bug 282: compiler does not trust type assertions while passing
2610 arguments to a full call.
2611 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2612 &OPTIONAL VALUES type parameter.
2613 * fix bug 214: algorithm for noting rejected templates is now more
2614 similar to that of template seletion. (also reported by rydis on
2616 * fixed bug 141b: printing backquoted information readably and prettily
2617 inserts a space where necessary.
2618 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2619 (CHARACTER 10) are now reported as errors, rather than propagated
2620 as unknown types. (reported by piso on #lisp)
2621 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2622 with values NIL and :ERROR. (thanks to Milan Zamazal)
2623 * fixed bug 191c: CLOS now does proper keyword argument checking as
2624 described in CLHS 7.6.5 and 7.6.5.1.
2625 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2626 longer raise an error; further, using a list as a for-as-arithmetic
2627 counter now raises a meaningful error.
2628 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2629 now understood by sequence creation functions such as MAKE-SEQUENCE
2631 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2632 a STRING-INPUT-STREAM.
2633 * compiler enhancement: SIGNUM is now better able to derive the type
2635 * type declarations inside WITH-SLOTS are checked. (reported by
2637 * fixed some bugs revealed by Paul Dietz' test suite:
2638 ** incorrect optimization of TRUNCATE for a positive first
2639 argument and negative second.
2640 ** compiler failure in let-convertion during flushing dead code.
2641 ** compiler failure while deriving type of TRUNCATE on an
2642 interval, containing 0.
2643 ** ASH of a negative bignum by a negative bignum count now returns
2645 ** intersection of CONS types now canonicalizes properly, fixing
2646 inconsistencies in SUBTYPEP.
2648 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2649 * incompatible change: The --disable-debugger command line
2650 option now clobbers the debugger at a more fundamental
2651 level, by redefining #'INVOKE-DEBUGGER instead of by
2652 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2653 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2654 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2655 enter the debugger REPL and then suffer recursive errors
2656 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2657 while under the new system, BREAK in --disable-debugger mode
2658 terminates the system just as an unhandled error would.
2659 * fixed compiler performance when processing loops with a step >1;
2660 * bug fix: DOCUMENTATION now retrieves generic function
2661 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2662 support has been systematized, and now supports the methods
2663 specified by ANSI, along with a default method and a method for
2664 slot documentation. (reported by Nathan Froyd)
2665 * bug fix: effective methods associated with a generic function are
2666 no longer cached over a change of that generic function's method
2667 combination. (reported by Andreas Fuchs)
2668 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2669 names a non-existent directory. (thanks to Andreas Fuchs)
2670 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2671 the PPC platform, lead to stack corruption; this has been fixed.
2672 (reported by Rainer Joswig)
2673 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2674 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2675 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2676 arithmetic in general, is now much more reliable on the PPC
2678 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2679 platform now returns the right answer.
2680 * optimization: restored some effective method precomputation in
2681 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2682 precomputation is now tunable.
2683 * optimization: compiler-internal data structure use has been
2684 reviewed, and changes have been made that should improve the
2685 performance of the compiler by about 20%.
2686 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2687 simple-base-strings and simple-bit-vectors is improved.
2688 * optimization: the optimization of 32-bit logical and arithmetic
2689 functions introduced in version 0.8.3 on the x86 has been
2690 implemented on the mips, ppc and sparc platforms; an
2691 implementation of the same facility, but for 64-bit arithmetic,
2692 has been added for the alpha.
2693 * microoptimization: the compiler is better able to make use of the
2694 x86 LEA instruction for multiplication by constants.
2695 * bug fix: in some situations compiler did not report usage of
2696 generic arithmetic in (SPEED 3) policy.
2697 * bug 145b fix: compiler used wrong type specifier while converting
2698 MEMBER-types to numeric.
2699 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2701 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2702 index argument now works properly on the Alpha platform.
2703 * bug fix: floating point exception treatment on the Alpha platform
2705 * bug fix: FILE-POSITION works much better on string input and
2706 output streams. (thanks to Nikodemus Siivola)
2707 * bug fix: many threading/garbage collection symptoms sorted.
2708 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2709 might be pseudo-atomic.
2710 * internal change: Stopping for GC is now done with signals not
2711 ptrace. GC is now done in whichever thread wanted it, instead of
2713 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2715 * bug fix: result form in DO is not contained in the implicit
2717 * incompatible change: ICR structure is changed; the value part of
2718 CONTINUATION is now called LVAR; corresponding functions are
2719 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2720 * added type deriver for ISQRT (thanks to Robert E. Brown).
2721 * bug fix: better support for loading from the command line when an
2722 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2724 * fixed some bugs revealed by Paul Dietz' test suite:
2725 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2726 ** ROUND and FROUND now give the right answer when given very
2727 small float arguments.
2728 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2730 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2731 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2732 ** LCM should return a non-negative integer.
2733 ** PARSE-INTEGER returned the index of a terminator instead of the
2734 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2735 ** PARSE-INTEGER returned an incorrect index being applied to a
2737 ** LCM with two arguments of 0 returns 0 rather than signalling
2739 ** unsigned addition of a 32-bit constant with the high bit set no
2740 longer causes an internal compiler error on the x86.
2741 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2742 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2743 with negative last argument.
2744 ** byte specifiers with zero size and position no longer cause
2745 an error during type derivation.
2746 ** bignum multiplication on the Alpha platform now returns the
2748 * porting: The system now builds on SuSE AMD64, although it still
2749 generates a 32-bit binary.
2750 * .fasl file incompatibility: The fasl file version number has
2751 been incremented (because of the changes to internal compiler
2752 data structures referred to above).
2754 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2755 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2756 more accurately, on the Darwin kernel running on PowerPC hardware.
2757 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2758 * Compiler code deletion notes now signal a condition of type
2759 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2760 an associated MUFFLE-WARNING restart.
2761 * The compiler now performs limited argument count validation of
2762 constant format strings in FORMAT, and where appropriate in ERROR,
2763 CERROR and WARN. (thanks to Gerd Moellmann)
2764 * New ASDF-INSTALL contrib can be used for automatic download and
2765 installation of third-party Lisp code from CCLAN or other sites
2767 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2768 which forces another thread to execute a function supplied by the
2770 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2771 now accept and act upon their :ELEMENT-TYPE keyword argument.
2772 (reported by Martin Atzmueller, Edi Weitz)
2773 * bug fix: FILE-POSITION now accepts position designators up to
2774 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2775 is the greater. (thanks to Patrik Nordebo)
2776 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2777 Kalle Olavi Niemitalo)
2778 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2779 printable. (reported by Eric Marsden)
2780 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2781 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2782 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2783 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2784 (thanks to Stig Sandoe)
2785 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2786 UNREAD-CHAR now correctly includes the unread character in the
2787 target sequence. (thanks to Gerd Moellmann)
2788 * bug fix in threaded builds: the system can now be suspended and
2789 resumed by shell job control with minimal disruption.
2790 * bug fixes in times and timezones >2038AD
2791 * better handling of "where is GNU make?" problem in build scripts
2792 (thanks to Nikodemus Siivola)
2793 * new optimization: inside a named function any reference to a
2794 function with the same name is considered to be a self-reference;
2795 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2796 optimization quality.
2797 * new optimization on x86: logical functions and + now have
2798 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2799 used when the result is truncated to 32 bits.
2800 * VALUES declaration is partially enabled.
2801 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2802 * fixed some bugs revealed by Paul Dietz' test suite:
2803 ** The system now obeys the constraint imposed by
2804 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2805 types form a lattice under type intersection.
2806 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2807 ** ASSOC now ignores NIL elements in an alist.
2808 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2809 and (1+ MOST-POSITIVE-FIXNUM) answers.
2810 ** The addition of a method with invalid qualifiers to a generic
2811 function does not cause an error to be signalled immediately;
2812 a warning is signalled, and the error is generated only on
2813 calling the generic function.
2814 * changed .fasl file version number, in part to add type codes for
2815 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2816 obscure ANSI requirements
2818 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2819 * fixed bug 148: failure to inline-expand a local function left
2820 garbage, confusing the compiler.
2821 * fixed bugs 3cd: structure slot readers perform type check if the
2822 slot can have an invalid value (i.e. it is either not initialized
2823 or can be written with a less specific slot writer).
2824 * bug fix: the compiler now traps array references to elements off
2825 the end of an array; previously, the bounds checking in some
2826 circumstances could go off-by-one.
2827 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2829 * type declarations for array element types now obey the description
2830 on the CLHS page "Declaration TYPE", as per discussions on
2831 sbcl-help around 2003-05-08. This means that a declaration
2832 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2833 declaration, all references to BAR will be asserted or assumed
2834 (with THE, so dependent on compiler policy) to involve objects of
2835 type FOO. Note that no such declaration is implied in
2836 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2837 * declared types of functions from the "Conditions"
2838 chapter. (reported by Paul Dietz)
2839 * bug fix: CERROR accepts a function as its first argument.
2840 * bug fix: NTH an NTHCDR accept a bignum as index
2841 arguments. (reported by Adam Warner)
2842 * optimization: character compare routines now optimize comparing
2843 against a constant character. (reported by Gilbert Baumann)
2844 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2845 argument now works properly on the MIPS platform.
2846 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2847 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2848 declared to have a functional type, does not cause an error
2850 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2851 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2852 * fixed bug: if last continuation of a deleted block has a
2853 destination, this destination should be deleted too. (reported by
2855 * fixed a bug in the bootstrap process: the host compiler's values
2856 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2857 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2858 test case from Patrik Nordebo)
2859 * improved the ability of the disassembler on the PPC platform to
2860 provide helpful disassembly notes.
2861 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2862 the class in more cases than previously.
2863 * bug fix: FILE-POSITION now understands :START and :END for
2864 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2865 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2866 8) data. (thanks to David Lichteblau)
2867 * bug fix: it is possible to add a method to a generic function
2868 without lambda list.
2869 * bug fix: reader failed to signal END-OF-FILE inside an
2870 object representation. (reported by Nikodemus Siivola)
2871 * fixed some bugs revealed by Paul Dietz' test suite:
2872 ** LAST and [N]BUTLAST should accept a bignum.
2873 ** condition slot accessors are methods.
2874 ** (VECTOR NIL) is a subtype of STRING.
2876 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2877 * minor incompatible change: some nonsensical specialized lambda
2878 lists (used in DEFMETHOD) which were previously ignored now signal
2880 * minor incompatible change: the system is now aware of the types of
2881 variables in the COMMON-LISP package, and will signal errors for
2882 most violations of these type constraints (where previously they
2883 were silently accepted).
2884 * minor incompatible change: COMPILE-FILE now uses the freedom
2885 afforded (ANSI 3.2.2.3) to use derived function types for
2886 functions defined in the same file. This also permits the system
2887 to warn on static type mismatches and function
2888 redefinition. (Currently it does not work with high DEBUG level.)
2889 * minor incompatible change: VALUES declaration is disabled.
2890 * When issuing notes, the compiler now signals a condition of type
2891 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2892 restart for use in user handlers. It is expected that the
2893 COMPILER-NOTE condition will eventually become a condition
2894 supertype to a hierarchy of note types, which will then be
2895 handleable in a similar fashion. However, at the moment, no such
2896 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2897 but it's an implementation detail, not a classification for the
2899 * Changes in type checking closed the following bugs:
2900 ** type checking of unused values (192b, 194d, 203);
2901 ** template selection based on unsafe type assertions (192c, 236);
2902 ** type checking in branches (194bc).
2903 * A short form of VALUES type specifier has ANSI meaning (it has
2904 increased the number of situations when SBCL cannot perform type
2906 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2907 keywords or constants is permissible.
2908 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2909 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2910 classes. (thanks to Antonio Martinez)
2911 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2912 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2913 argument to a pretty-print stream. Instead, it leaves any such
2914 operations to DESCRIBE-OBJECT methods.
2915 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2917 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2918 lambda lists. (thanks to David Lichteblau)
2919 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2920 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2921 (thanks to Gerd Moellmann)
2922 * fixed bug 246: increased compilation speed of long
2923 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2925 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2927 * DEFCONSTANT now throws a condition of type
2928 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2929 constant to a non-EQL value; CONTINUE and ABORT restarts
2930 respectively change and preserve the value.
2931 * fixed bug 63: The code walker, part of the implementation of CLOS,
2932 is now better at handling symbol macros.
2933 * bug fix: There is no longer an internal implementation type named
2934 CL:LENGTH. (reported by Raymond Toy)
2935 * bug fix: In macro-like defining macros/special operators the
2936 implicit block does not enclose the lambda list.
2937 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2938 enclosed in parentheses) are not suitable as type specifiers, and
2939 their use properly signals an error now.
2940 * bug fix: An argument count mismatch for a type specifier in code
2941 being compiled no longer causes an unhandled error at compile
2942 time, but signals a compile-time warning.
2943 * fixed simple vector readable printing
2944 * bug fix: DESCRIBE takes more care over whether the class
2945 precedence list slot of a class is bound before accessing it.
2946 (reported by Markus Krummenacker)
2947 * bug fix: FORMATTER can successfully compile pretty-printer format
2948 strings which use variants of the ~* directive inside.
2949 * bug fix: SEARCH now applies its TEST predicate to the elements of
2950 the arguments in the correct order. (thanks to Wolfhard Buss)
2951 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2952 in type inference. (thanks to Robert E. Brown)
2953 * bug fix: Reading in symbols with an explicit package name of ""
2954 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2955 * fixed some bugs revealed by Paul Dietz' test suite:
2956 ** NIL is now allowed as a structure slot name.
2957 ** Arbitrary numbers, not just REALs, are allowed in certain
2958 circumstances in LOOP for-as-arithmetic clauses.
2959 ** Multiple class redefinitions before slot access no longer
2960 causes a type error.
2961 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2962 association between the name and a class.
2963 ** Generic functions with non-standard method-combination and over
2964 five methods all of which return constants no longer return NIL
2965 after the first few invocations. (thanks to Gerd Moellmann)
2966 ** CALL-NEXT-METHOD with no arguments now passes the original
2967 values of the arguments, even in the presence of assignment.
2968 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2970 ** DISASSEMBLE works with closures and funcallable instances.
2971 ** ADD-METHOD now returns the generic function, not the new method.
2972 ** FIND-METHOD signals an error if the lengths of the specializers
2973 is incompatible with the generic function, even if the ERRORP
2975 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2976 which its argument is a member.
2977 ** DEFCLASS only redefines the class named by its class-name
2978 argument if that name is the proper name of the class;
2979 otherwise, it creates a new class.
2980 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2981 of the UNBOUND-SLOT condition to the name of the slot.
2982 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2983 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2984 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2985 SLOT-MAKUNBOUND in the specified fashion.
2987 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2988 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2989 cross-compilation host. As a consequence, we can now bootstrap our
2990 way up to SBCL starting with a bare gcc toolchain and human-readable
2991 source code (first the source to CLISP, then the source to SBCL).
2992 * A contributed module containing a partial implementation of the
2993 simple-streams interface has been included. (thanks to Rudi
2995 * A contributed module implementing the RFC1321 Message Digest
2996 Algorithm, known as MD5, has been included.
2997 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2998 no longer has any effect, as the code controlled by this feature
2999 has been deleted. (As far as we know, no-one has ever built using
3000 this feature, and its semantics were confused in any case).
3001 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3002 work on conditions (as required by the ANSI specification),
3003 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3004 expected behaviour on conditions. Users should note, however,
3005 that such behaviour is not required by the ANSI specification,
3006 and so use of this behaviour may render their code unportable.
3007 * fixed some bugs revealed by Paul Dietz' test suite:
3008 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3010 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3011 ** SXHASH on condition objects no longer returns NIL.
3012 ** :ALLOCATION :CLASS slots are better treated; their values are
3013 updated on class redefinition, and initforms inherited from
3014 superclasses are applied.
3015 ** REMOVE-METHOD returns its generic function argument even when
3016 no method was removed.
3017 ** SHARED-INITIALIZE now initializes the values of the requested
3018 slots, including those with :ALLOCATION :CLASS.
3019 ** ALLOCATE-INSTANCE now works on structure classes defined via
3020 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3022 ** SLOT-EXISTS-P now works on conditions, as well as structures
3024 ** MAKE-LOAD-FORM now has the required methods on
3025 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3026 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3027 keyword, and now implements the SLOT-NAMES argument.
3028 ** methods with &OPTIONAL arguments no longer allow too many
3029 arguments to be passed in the call without error.
3030 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3031 option is consistent with the required arguments of the generic
3032 function lambda list.
3033 * bug fix: REQUIRE accepts a string designator. (Thanks to
3035 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3036 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3037 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3038 * bug fix: The compiler now checks for duplicated variables in macro
3040 * bug fix: SETQ on globals returns the correct value.
3041 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3042 not a COMPILER-ERROR (followed by some other strange error on
3043 choosing the CONTINUE restart).
3044 * bug fix: make.sh and friends are now more consistent in the way that
3045 they look for GNU "make".
3047 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3048 * experimental native threads support (on x86 Linux >=2.4 only).
3049 This is not compiled in by default: you need to add :SB-THREAD to
3050 the target features. See the "Beyond ANSI" chapter of the manual
3052 * fix for longstanding nonANSIism: The old distinction between
3053 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3054 The return value from CL:FIND-CLASS is now a CLOS class, and
3055 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3056 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3057 * An interface to the MetaObject Protocol, as described in Kiczales,
3058 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3059 MIT Press, 1991, is available from the SB-MOP package.
3060 * incompatible change: the SB-PCL package should now be considered
3061 a private implementation detail, and no longer a semi-private MOP
3063 * minor incompatible change: due to rearrangement for threads, the
3064 control stack and binding stack are now allocated at arbitrary
3065 addresses instead of being hardcoded per-port. Users affected by
3066 this probably have to be doing advanced things with shared
3067 libraries, and will know who they are.
3068 * minor incompatible change: Previously, all --eval forms used were
3069 processed with READ before any of them were processed with EVAL.
3070 Now each --eval form is processed with both READ and EVAL before
3071 the next --eval form is processed. (Thus package operations like
3072 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3073 work as the user might reasonably expect.)
3074 * minor incompatible change: *STANDARD-INPUT* is now only an
3075 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3077 * minor incompatible change: Y-OR-N-P is now character-oriented,
3078 not line oriented. Also, YES-OR-NO-P now works without errors.
3079 (thanks to Antonio Martinez)
3080 * sb-aclrepl module improvements: an integrated inspector, added
3081 repl features, and a bug fix to :trace command.
3082 * Known functions, which cannot be open coded by the backend, are
3083 considered to be able to check types of their arguments. (fixing
3084 a bug report by Nathan J. Froyd)
3085 * fixed a bug in computing method discriminating functions: It is
3086 now possible to define methods specialized on classes which have
3087 forward-referenced superclasses. (thanks to Gerd Moellmann)
3088 * fixed evaluation order in optional entries (reported by Gilbert
3090 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3091 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3092 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3093 required-by-AMOP NAME argument, as well as CLASS and
3094 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3095 * fixed bug 20: DEFMETHOD can define methods using names that are
3096 not the proper names of classes to designate class specializers.
3097 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3098 Unix file descriptors, instead of blowing up. (thanks to Antonio
3100 * Garbage collection refactoring: user-visible change is that a
3101 call to the GC function during WITHOUT-GCING will not do garbage
3102 collection until the end of the WITHOUT-GCING. If you were doing
3103 this you were probably losing anyway.
3104 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3105 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3106 TYPEP the latter but not the former.
3107 * The compiler issues a full WARNING for calls to undefined functions
3108 with names from the CL package.
3109 * MAP-INTO for a vector destination is open coded. (reported by
3110 Brian Downing on c.l.l)
3111 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3112 documentation string.
3113 * fixed some bugs revealed by Paul Dietz' test suite:
3114 ** COPY-ALIST now signals an error if its argument is a dotted
3116 ** Condition slots are now accessed more correctly in the presence
3117 of multiple initargs for a given slot.
3118 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3119 exclude restarts of the same name associated with a different
3121 ** DEFCLASS of forward-referenced classes with another
3122 forward-referenced class in the superclasses list no longer
3124 ** Condition slots are now initialized once each, not multiple
3125 times. (thanks to Gerd Moellmann)
3126 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3128 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3129 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3130 arguments contain duplicated elements.
3131 ** RESTART-CASE understands local macros.
3132 ** RESTART-CASE associates exactly its own restarts with a condition.
3133 ** ENDP in safe mode checks its argument to be of type LIST.
3134 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3135 in question is unbound.
3136 ** Optimized MAKE-INSTANCE functions no longer cause internal
3137 assertion failures in the presence of duplicate initargs.
3138 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3139 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3140 propagated even in the ctor optimized implementation of
3142 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3144 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3145 * a better implementation of SXHASH on (simple) bit vectors,
3146 measured both in execution speed and in distribution of results
3147 over the positive fixnums, has been installed. Likewise, a better
3148 implementation of EQUAL for simple bit vectors is now available.
3149 * fixed CEILING optimization for a divisor of form 2^k.
3150 * fixed bug 240 (emitting extra style warnings "using the lexical
3151 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3152 by Antonio Martinez)
3153 * fixed SXHASH, giving different results for NIL depending on type
3154 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3155 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3156 variable. (found by Rolf Wester)
3157 * SBCL does not ignore type declarations for special
3158 variables. (reported by rif on c.l.l 2003-03-05)
3159 * some bug fixes in contrib/sb-aclrepl/
3160 * fixed some bugs revealed by Paul Dietz' test suite:
3161 ** a bug in the CONS type specifier, whereby the CAR and CDR
3162 types got intertwined, has been fixed;
3163 ** the type system is now able to reason about the interaction
3164 between INTEGER and RATIO types more completely;
3165 ** APPEND, [N]REVERSE and NRECONC check that those their
3166 arguments, which must be proper lists, are really so;
3167 ** An array specialized to be unable to hold elements has been
3168 implemented, as required -- yes, really -- by ANSI;
3169 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3170 on malformed property lists;
3172 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3173 * incompatible packaging change: in line with Unix convention,
3174 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3175 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3176 * REQUIRE and PROVIDE are now optionally capable of doing something
3177 useful. See the documentation string for REQUIRE.
3178 * infrastructure for a managed SBCL contrib system: contributed
3179 modules in this release include:
3180 ** the ASDF system definition facility;
3181 ** an interface to the BSD Sockets API;
3182 ** an ACL-like convenience interface to the repl;
3183 (thanks to Kevin Rosenberg)
3184 ** an implementation of ROTATE-BYTE, with efficient implementation
3186 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3187 gives a closer approximation to the right answer than previously.
3188 (thanks to Raymond Toy)
3189 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3190 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3191 optional environment arguments, as required by ANSI.
3192 * fixed bugs in other functions taking environment objects, allowing
3193 calls with an explicit NIL environment argument to be compiled
3195 * fixed bug 228: primary return values from
3196 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3197 COMPILE or FUNCTION.
3198 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3199 :TYPE will no longer signal errors on innocuous objects.
3200 * fixed bug 231b: SETQ is better at respecting type declarations in
3201 the lexical environment.
3202 * fixed a bug in DEFCLASS: classes named by symbols with no or
3203 unprintable packages can now be defined.
3204 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3205 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3206 * fixed compiler failure related to checking types of functions.
3207 (reported by Robert E. Brown)
3208 * the compiler is now much more consistent in its error-checking
3209 treatment of bounding index arguments to sequence functions: in
3210 (SAFETY 3) code, errors will be signalled in almost all cases if
3211 invalid sequence bounding indices are passed to functions defined
3212 by ANSI to operate on sequences.
3213 * fixed a bug in the build procedure: documentation of SBCL-specific
3214 packages is now preserved and available in the final Lisp image.
3215 * lifted FDEFINITION lookup out of loops in the implementation of
3216 many list operations. (thanks to Robert E. Brown)
3217 * fixed a bug in the reader: the #n# reader macro now works for
3218 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3219 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3220 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3221 E. Brown and Tony Martinez respectively)
3222 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3223 SLOT-BOUNDP: in particular, it is now able to optimize them much
3224 better, and is now not vulnerable to having packages renamed.
3225 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3226 always called when a slot is not present in an instance. (thanks
3228 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3229 longer signal an error on unreading a character following EOF on
3230 the previous constituent stream. (thanks to Tony Martinez)
3231 * fixed some bugs revealed by Paul Dietz' test suite:
3232 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3233 not just nonnegative fixnums;
3234 ** the logical bit-array operators such as BIT-AND now accept an
3235 explicit NIL for their "opt-arg" argument (to indicate a
3236 freshly-consed result bit-array);
3237 ** ELT now signals an error on an invalid sequence index in safe
3239 ** the type system is now cleverer about negations of numeric
3240 types, and consequently understands the BIGNUM and RATIO types
3242 ** the type system is now cleverer about the interaction between
3243 INTEGER and RATIO types: while bugs still remain, many more
3244 cases are accurately computed;
3245 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3246 if it is in the last clause;
3247 ** CONSTANTLY now correctly returns a side-effect-free function in
3249 ** DECLARE is no longer treated as a special-operator; in
3250 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3251 * incremented fasl file version number due to the change in the
3252 DEFSTRUCT-SLOT-DESCRIPTION structure.
3254 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3255 * minor incompatible change: code processed by the "interpreter" or
3256 EVAL now has a compilation optimization policy of (DEBUG 2)
3257 (changed from (DEBUG 1)) to improve debuggability of interactive
3258 development, and to allow the use of the debug RETURN command in
3260 * an experimental implementation of the RETURN command for the
3261 debugger has been included. (thanks to Frederik Kuivinen)
3262 * fixed bug 62: constraints were not propagated into a loop.
3263 * fixed bug in embedded calls of SORT (reported and investigated by
3265 * fixed some bugs revealed by Paul Dietz' test suite:
3266 ** printing and reading of arrays with some dimensions having
3267 length 0 (thanks to Gerd Moellmann);
3268 ** BOA constructor with &AUX argument without a default value does
3269 not cause a type error;
3270 ** CONSTANTP now returns true for all self-evaluating objects.
3272 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3273 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3274 accessors that are related by inheritance, as specified in the
3275 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3276 Valtteri Vuorikoski)
3277 * The compiler is now able to inline functions that were defined in
3278 a complex lexical environment (e.g. inside a MACROLET).
3279 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3281 * Support for the upcoming FreeBSD-5.0 release has been included.
3282 (thanks to Dag-Erling Smorgrav)
3283 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3284 effect when it is not in a toplevel context.
3285 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3286 stand a better chance of being correct. (thanks to Gerd
3288 * fixed bug in COERCE, which now signals an error on coercing a
3289 rational to a bounded real type which excludes the expected
3291 * The compiler is now able to derive types more accurately from the
3292 COERCE and COMPILE functions.
3293 * fixed bug 223: functional binding is considered to be constant
3294 only for symbols in the CL package.
3295 * fixed bug 231: SETQ did not check the type of a variable being set
3296 (reported by Robert E. Brown)
3297 * A new optimization for MAKE-INSTANCE has been included, fixing
3298 various bugs (including relating to :ALLOCATION :CLASS slots and
3299 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3301 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3302 ** As required by ANSI, LOOP now disallows anonymous collection
3303 clauses such as COLLECT I in conjunction with aggregate boolean
3304 clauses such as THEREIS (= I 1);
3305 ** LOOP now signals an error when any variable is reused in the
3306 same loop (including the potentially useful construct analogous
3307 to WITH A = 1 WITH A = (1+ A);
3308 ** IT is only a special loop symbol within the first clause of a
3309 conditional loop clause;
3310 ** LOOP with a typed iteration variable over a hashtable now
3311 signals a type error iff it should.
3312 * fixed some other bugs revealed by Paul Dietz' test suite:
3313 ** FILE-STREAM now names the class previously known as FD-STREAM;
3314 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3315 argument) no longer signals an error;
3316 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3317 of the slot symbol, rather than using the current package
3318 ((:CONC-NAME "") continues to intern the slot's name in the
3320 * incremented fasl file version number, because of the incompatible
3321 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3322 of the new implementation of DEFINE-COMPILER-MACRO.
3324 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3325 * Support for building SBCL for MIPS platforms running in
3326 little-endian mode has now been checked in, and basic
3327 functionality on said platforms verified.
3328 * minor incompatible change: PCL now records the pathname of a file
3329 in which methods and the like are defined, rather than its
3331 * minor incompatible change: TRUENAME now considers the truename of
3332 a file naming a directory to be the pathname with :DIRECTORY
3333 component indicating that directory.
3334 * minor incompatible change: a NAMED clause in the extended form of
3335 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3336 reason for the previous behaviour is unclear.
3337 * more systematization and improvement of CLOS and MOP conformance
3338 in PCL (thanks to Gerd Moellman and Pierre Mai):
3339 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3341 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3342 primary methods with no specializers;
3343 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3345 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3346 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3347 FORWARD-REFERENCED-CLASSes; error reporting on
3348 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3350 ** SXHASH on CLOS instances now uses a slot internal to the
3351 instance to return different numbers on distinct instances,
3352 while preserving the same return value through invocations of
3354 ** DEFMETHOD signals errors when methods with longer incongruent
3355 lambda lists are added to generic functions;
3356 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3357 CLASS, as specified in AMOP;
3358 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3359 on the order returned by the primary method for classes of
3360 class STANDARD-CLASS;
3361 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3362 * fixed some bugs shown by Paul Dietz' test suite:
3363 ** DOLIST puts its body in TAGBODY;
3364 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3366 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3367 value producing form;
3368 ** if more variables are given to PROGV than values, extra
3369 variables are bound and made to have no value;
3370 ** NSUBSTITUTE on list arguments gets the right answer with
3372 ** ELT signals an error of type TYPE-ERROR when the index argument
3373 is not a valid sequence index;
3374 ** LOOP signals (at macroexpansion time) an error of type
3375 PROGRAM-ERROR when duplicate variable names are found;
3376 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3377 ** FUNCALL of special-operators now cause an error of type
3379 ** PSETQ now works as required in the presence of side-effecting
3380 symbol-macro places;
3381 ** NCONC accepts any object as its last argument;
3382 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3384 ** loop-for-as-package does not require a package to be explicitely
3386 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3388 * fixed bug 166: compiler preserves "there is a way to go"
3389 invariant when deleting code.
3390 * fixed bug 172: macro lambda lists with required arguments after
3391 &REST arguments now cause an error to be signalled. (thanks to
3393 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3394 bug. (thanks to Matthew Danish)
3395 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3397 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3398 arguments are lexically rebound. (thanks to Gerd Moellmann and
3400 * fixed bug 194: error messages are now more informative when there
3401 is no primary method applicable in a call to a generic
3402 function. (thanks to Gerd Moellmann)
3403 * fixed bug in command line argument checking (thanks to Julian
3405 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3406 (thanks to Matthew Danish)
3407 * incremented fasl file version number, because of the
3408 SXHASH-related changes in the layout of CLOS data structures
3410 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3411 * minor incompatible change: The runtime (the Unix executable named
3412 "sbcl") is now much pickier about the .core files it will load.
3413 Essentially it now requires .core files to descend from the same
3414 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3415 as the runtime does. (The intent is to prevent the crashes which
3416 can occur, and which can even be reported as mysterious failures,
3417 when people patch the sources or change the build parameters
3418 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3419 sbcl and .core files.)
3420 * fixed bug: VALUES-LIST is no longer optimized away.
3421 * fixed bug 142: The FFI conversion of C string values to Lisp
3422 string values no longer conses excessively. (thanks to Nathan
3423 Froyd porting Raymond Toy's fix to CMU CL)
3424 * began to systematize and improve MOP conformance in PCL (thanks to
3425 Nathan Froyd, Gerd Moellman and Pierre Mai):
3426 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3428 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3429 ** FINALIZE-INHERITANCE is now called on class finalization;
3430 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3431 argument precedence order.
3432 * fixed bug 202: The compiler no longer fails on functions whose
3433 derived types contradict their declared type.
3434 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3435 so it can be non-toplevel.
3436 * The fasl file version number has changed (because of the new
3437 implementation of DEFMACRO).
3438 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3439 safe code, that any length requirement by their type-specifier
3440 argument is valid. The exceptions to this are described in bug
3442 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3443 functions now check the types of their inputs as required by ANSI.
3444 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3445 introduced symbol is DECLAREd to be SPECIAL.
3446 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3447 * fixed bug: :COUNT argument to sequence functions may be negative
3448 * fixed bug: body of DO-SYMBOLS may contain declarations
3449 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3450 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3451 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3452 symbol macro only once
3453 * fixed printing of call frame when argument list is unavailable
3454 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3455 * compiler no longer signals WARNING on unknown keyword
3458 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3459 * A beta-quality port to the mips architecture running Linux,
3460 based on the old CMUCL backend, has been made. It has been tested
3461 on a big-endian kernel, and works sufficiently well to be able to
3462 rebuild itself; it has not been tested in little-endian mode.
3463 * fixed an inconsistency between gencgc.c and purify.c which made
3464 dumping/loading .core files unreliable
3465 * fixed bug 120a: The compiler now deals correctly with IFs where
3466 the consequent is the same as the alternative, instead of
3467 misderiving the return type. (thanks to Alexey Dejneka)
3468 * fixed bug 113: Logical pathnames are now dumpable (the logical
3469 host is resolved at load-time, throwing an error if it is not
3471 * fixed bug 174: FORMAT's error message is slightly clearer when a
3472 non-printing character is used in a format directive.
3473 * fixed several bugs in compiler checking of type declarations, i.e.
3474 violations of the Python "declarations are assertions" principle
3475 (thanks to Alexey Dejneka)
3476 * fixed several bugs in PCL's error checking (thanks to Gerd
3478 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3480 * fixed bug in compilation of functions as first class values
3481 (thanks to Antonio Martinez-Shotton)
3482 * The compiler's handling TYPE-ERRORs which it can prove will
3483 inevitably happen at runtime has been cleaned up and corrected
3484 in several ways. (thanks to Alexey Dejneka)
3485 * improved argument type checking for various basic arithmetic
3486 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3487 lavished on them in the past that they can be compiled in many
3488 ways in different special cases
3489 * fixed bug 181: compiler checks validity of user supplied type
3491 * cleaned up code flushing in optimization: Function calls which
3492 should signal errors for safety purposes (e.g. which ANSI says
3493 should signal errors when their arguments are of incorrect type)
3494 are no longer optimized away.
3495 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3496 * incremented fasl file version number, because changes in the
3497 implementation of sequence functions like COERCE caused
3498 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3499 in old inline expansions) to become undefined. (Actually these
3500 changes were later undone, so we might very well be binary
3501 compatible with 0.7.7 after all, but leaving the version number
3502 incremented seemed like the simplest and most conservative
3505 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3506 * An alpha-quality port to the parisc architecture running Linux,
3507 based on the old CMUCL backend, has been made. This, even more so
3508 than the other backends, should be considered still a work in
3509 progress; known problems include that the Linux kernel in 64-bit
3510 mode does not propagate the correct sigcontext structure to
3511 userspace, and consequently SBCL on a parisc64 kernel will not
3513 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3514 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3515 suppressed.) Also now that the compiler is looking at declarations
3516 in the environment, it checks optimization declarations as well,
3517 and suppresses inlining when (> DEBUG SPEED).
3518 * More fixes have been made to treatment of floating point exception
3519 treatment and other Unix signals. In particular, floating point
3520 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3521 * The detection and handling of control stack exhaustion (infinite
3522 or very deeply nested recursion) has changed. Stack exhaustion
3523 detection is now done by write-protecting pages at the OS level
3524 and applies at all optimization settings; when found, a
3525 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3526 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3527 be caught using IGNORE-ERRORS.
3528 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3529 correctly when transforming recursive LABELS functions to LETs.
3530 (thanks to Alexey Dejneka)
3531 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3532 that are names of constants or global variables.
3533 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3534 alien routines with docstrings.
3535 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3536 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3538 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3539 (thanks to Eric Marsden)
3540 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3541 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3542 * Bug fix: The SPARC backend can now compile functions involving
3543 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3544 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3545 to a functions expecting a PATHNAME-DESIGNATOR.
3546 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3547 lambda lists. (thanks to Alexey Dejneka)
3548 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3549 functions so that EXPORTing the name of the class doesn't cause
3550 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3551 to Antonio Martinez for reporting this)
3552 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3553 constant-folded without failing an assertion. (thanks to Einar
3554 Floystad Dorum for reporting this)
3555 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3556 types are now dealt with more correctly by the compiler.
3557 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3558 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3559 behaviour analogous to RENAME-FILE. This puts its behaviour more
3560 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3562 * The fasl file version number has changed again. (because of the
3563 bug fix involving the names of PCL MAKE-INSTANCE functions)
3565 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3566 * bug fix: Floating point exceptions are treated much more
3567 consistently on the x86/Linux and PPC/Linux platforms.
3568 * Array initialization with :INITIAL-ELEMENT is now much faster for
3569 cases when the compiler cannot open code the array creation, but
3570 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3571 array accesses have also seen a speed increase.
3572 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3573 specified. (This was at the root of some bad interactions between
3574 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3576 * bug fix: Internal error arguments for undefined functions are now
3577 computed correctly on the PPC/Linux platform.
3578 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3579 Raymond Toy's patch for CMU CL)
3580 * Support for the Solaris 9 operating environment has been included
3581 (thanks to Daniel Merritt)
3582 * A very ugly but hopefully complete draft of the missing FFI chapter
3583 of the manual has been created by reformatting the corresponding
3584 CMU CL manual chapter into (currently very ugly and incoherent)
3585 DocBook and bringing it up to date for SBCL behavior. Thus, the
3586 manual is now essentially complete, at least by my extreme
3587 once-and-only-once standards, whereby it's acceptable to refer to
3588 the doc strings of SB-EXT functions as the primary documentation.
3589 * The fasl file version number has changed again, due to cleanup of
3590 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3591 is no longer a static symbol.)
3593 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3594 * SBCL now builds with OpenMCL (version 0.12) as the
3595 cross-compilation host; also, more progress has been made toward
3596 bootstrapping under CLISP.
3597 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3599 * bug 158 fixed: The compiler can now deal with integer loop
3600 increments different from 1; fixing this turned out also to fix
3602 * bug 169 fixed: no more bogus warnings about using lexical bindings
3603 despite the presence of perfectly good SPECIAL declarations (thanks
3604 to David Lichteblau)
3605 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3606 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3607 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3609 * bug 180 fixed: Method combination specifications no longer ignore
3610 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3611 * bug fix: Structure type predicate functions now check their argument
3612 count as they should.
3613 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3614 correctly. (thanks to Pierre Mai)
3615 * minor incompatible change: The --noprogrammer option is deprecated
3616 in favor of the new --disable-debugger option, which is very similar.
3617 (The major difference is that it takes effect at a slightly different
3618 time at startup, causing handling of errors in --sysinit and
3619 --userinit files will be affected differently.) The
3620 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3621 been added to allow this functionality to be controlled from ordinary
3622 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3623 maintainers, who might want to run non-interactive scripts to
3624 build SBCL cores which will later be used interactively.)
3625 * minor incompatible change: The LOAD function no longer, when given
3626 a wild pathname to load, loads all files matching that pathname.
3627 Instead, an error of type FILE-ERROR is signalled.
3629 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3630 * bug 147 fixed: The compiler preserves its block link/count
3631 invariants more correctly now so that it doesn't crash. (thanks
3633 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3635 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3636 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3637 no-value-for-this-character as NIL (as specified by ANSI).
3638 * HOST-NAMESTRING on physical pathnames now returns a string that is
3639 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3640 (thanks to Christophe Rhodes)
3641 * The Alpha port handles icache flushing more correctly. (thanks to
3643 * More progress has been made toward bootstrapping under CLISP. (thanks
3644 to Christophe Rhodes)
3645 * The fasl file format has changed again, because dynamic loading
3646 on OpenBSD (which has non-ELF object files) motivated some cleanups
3647 in the way that foreign symbols are transformed and passed around.
3648 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3649 is no longer treated as whitespace by the reader, but instead as
3650 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3651 |A
\x7fB|, instead of A as it used to.
3653 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3654 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3655 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3656 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3657 this as of 0.7.1.45, but was left out of the previous news section
3658 (thanks to Dan Barlow)
3659 * SBCL now runs on the Solaris operating system on SPARC architectures
3660 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3661 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3662 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3663 * SPARC backend cleanups, allowing builds of cores optimized for V8
3664 and V9 SPARCS, and also emission of code targeted to a particular
3665 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3667 * SBCL is closer to bootstrapping under CLISP, thanks to various
3668 fixes by Christophe Rhodes.
3669 * The fasl file format has changed again, to allow the compiler's
3670 INFO database to support symbol macros.
3671 * The user manual (in doc/) is formatted into HTML more nicely.
3672 (thanks to coreythomas)
3673 * The system is smarter about SUBTYPEP relationships, especially
3674 those involving NOT types (including types such as ATOM which are
3675 represented internally using NOT types). Thus SUBTYPEP is less
3676 likely to return (VALUES NIL NIL) in general, and in particular
3677 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3679 * The fasl file format has changed again, because the internal
3680 representation of types now includes a new slot to support the new
3681 SUBTYPEP-of-NOT-types logic.
3682 * (not a change in the main branch of SBCL, but a related prototype
3683 which can hopefully be merged into the main branch of SBCL in the
3684 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3685 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3686 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3687 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3688 dispatch macro character combinations. (thanks to Alexey Dejneka)
3689 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3690 that superficially look like logical namestrings correctly.
3691 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3693 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3694 default supplied-p) for &optional and &key arguments. (thanks to
3697 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3698 * incompatible change: The compiler is now less aggressive about
3699 tail call optimization, doing it only when (> SPACE DEBUG) or
3700 (> SPEED DEBUG). (This is an incompatible change because there are
3701 programs which relied on the old CMU-CL-style behavior to optimize
3702 away their unbounded recursion which will now die of stack overflow.)
3703 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3704 for non-GENCGC systems has been increased to 20M (since that
3705 seems much closer to the likely performance optimum for modern
3706 systems than the old 4M value was)
3707 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3708 and *DEBUG-PRINT-LEVEL*
3709 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3710 of CMU CL's support for SPARC, and various endianness and other
3711 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3712 * new syntactic sugar for the Unix command line: --load foo.bar is now
3713 an alternate notation for --eval '(load "foo.bar")'.
3715 ** The system now detects stack overflow and handles it gracefully,
3716 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3717 optimization settings. (This is a good thing in general, and
3718 its introduction in this version should be particularly timely
3719 for anyone whose code fails because of suppression of tail
3721 ** The system now hunts for the C variable "environ" in a more
3722 devious way, to avoid segfaults when the C library version differs
3723 between compile time and run time. (thanks to Christophe Rhodes)
3724 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3725 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3726 ** The compiler no longer issues bogus style warnings for undefined
3727 classes in the same source file as the DEFCLASSes which defined
3728 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3730 ** fixes in CONDITION class precedence list for undefined function
3731 errors (thanks to Alexei Dejneka)
3732 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3733 correctly. (thanks to Dan Barlow)
3734 ** portability fixes aiming at bootstrapping under CLISP (thanks
3735 to Dave McDonald and Christophe Rhodes)
3736 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3737 ** fixes in type translation and and type inference (thanks to
3739 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3740 ** various fixes in the new ports (thanks to Dan Barlow)
3741 * several changes related to debugging:
3742 ** suppression of tail recursion, as noted above
3743 ** stack overflow detection, as noted above
3744 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3745 is now the default. (For some time encapsulation has been more
3746 reliable than the breakpoint-based :ENCAPSULATE NIL
3747 implementation, at least on X86 systems; and I just noticed that
3748 encapsulation also seems closer to the spirit of the ANSI
3751 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3753 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3754 up properly again. (There was a packaging bug in 0.7.0 which
3755 left their definitions in SB-SYS::LOAD-FOREIGN and
3756 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3757 vital for most things which interface to C-level interfaces,
3758 like extensions working with sockets or databases or
3759 Perl-compatible regexes or whatever, and the need to fix
3760 this bug is the main reason that 0.7.1 was released so
3762 ** DEFGENERIC is now choosier about the methods it redefines, so that
3763 reLOADing a previously-LOADed file containing DEFGENERICs does
3764 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3765 cycle now works as it should. (thanks to Alexey Dejneka)
3766 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3767 fixed by Christophe Rhodes. (He actually submitted this patch
3768 months ago, and I delayed until after 0.7.0.)
3769 ** Bug 111 (internal compiler confusion about runtime checks on
3770 FUNCTION types) was fixed by Alexey Dejneka.
3771 * Some internal cleanups (getting rid of variables which aren't
3772 needed now that the byte interpreter is gone) caused the fasl
3773 file format number to change again.
3775 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3776 * major incompatible change: The default fasl file extension, i.e. the
3777 default extension for files produced by COMPILE-FILE, has changed
3778 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3780 ** There are many changes in the implementation of the compiler.
3781 SBCL is now essentially a compiler-only implementation of ANSI
3782 Common Lisp. EVAL still "interprets" a few special cases, but
3783 almost all the interesting cases are handled by creating
3784 a LAMBDA expression, calling COMPILE on it, then calling
3785 FUNCALL on the result.
3786 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3787 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3788 Since the code is newer, there might still be some new bugs
3789 (though not as many as before Martin Atzmueller's fixes:-). But
3790 the new code is substantially simpler and clearer, and hopefully
3791 any remaining bugs will be simpler, less fundamental, and more
3792 fixable then the bugs in the old code.
3793 ** The revised compiler is still a little unsteady on its feet.
3795 *** The debugging information it produces (particularly the names
3796 of FUNCTION objects) is sometimes much less useful than what
3797 the old compiler produced.
3798 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3799 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3800 within a MACROLET) has been temporarily weakened.
3801 ** There are new compiler optimizations for various functions:
3802 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3803 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3804 *** the math functions TRUNCATE, FLOOR, and CEILING
3805 *** the function-of-all-trades COERCE
3806 Mostly these should be transparent, but there's one
3807 potentially-annoying problem (bug 117): when the compiler
3808 inline-expands a function and does type analysis on the result,
3809 it can create control paths which have type mismatches, and
3810 when it can't prove that those control paths aren't taken,
3811 it will issue WARNINGs about the type mismatches. This is
3812 a particular problem in practice for the new sequence functions.
3813 It's not clear how this should be fixed, and for now, a
3814 workaround is given in the entry for 117 in the BUGS file.
3815 ** (Because of the interaction between the two previous items --
3816 occasional inlining problems and new inline expansions -- some
3817 of the new sequence function optimizations won't really kick in
3818 completely until debugging information, and then inlining, are
3819 straightened out in some future version.)
3820 * minor incompatible changes:
3821 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3822 DIRECTORY no longer implicitly promotes NIL slots of its
3823 pathname argument to :WILD. In particular, when you ask for the
3824 contents of a directory (which you used to be able to do without
3825 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3826 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3827 ** changes in behavior that ANSI explicitly defines to be
3828 implementation dependent:
3829 *** The new compiler-only implementation still conforms with ANSI,
3830 but acts a little different than before. Besides the obvious
3831 changes in performance tradeoffs (that the cost per form passed
3832 to EVAL has gone up, and the cost per form executed by EVAL
3833 has gone down), the behavior of the system changes a little
3834 because there are no longer any interpreted function objects.
3835 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3836 e.g. doing COMPILE on the output of interactive DEFUN is
3838 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3840 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3841 has changed from (:CL) to NIL.
3842 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3843 e.g. #\Newline and #\Tab, don't have names specified in the
3844 ANSI Common Lisp standard, is now based on their ASCII symbolic
3845 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3846 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3847 are no longer used for output.
3848 ** changes in internal implementation constants:
3849 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3850 4 million. (If your application spends a lot of time GCing and
3851 you have a lot of RAM, you might want to experiment with
3852 increasing it even more.)
3853 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3854 However, almost all old code should still continue to work without
3855 immediate update, as SB-C-CALL is now a (deprecated) nickname
3857 ** Old operator names in the style DEF-FOO are now deprecated in
3858 favor of new corresponding names DEFINE-FOO, for consistency with
3859 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3860 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3861 internal symbols, but a few supported extensions like
3862 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3863 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3864 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3865 etc. as you get deeper into recursive calls to the debugger
3866 command loop, instead of the old "5]", "5]]", "5]]]"
3867 sequence. (I was motivated to do this when squabbles between
3868 ILISP and SBCL left me very deeply nested in the debugger. In the
3869 short term, this change will probably provoke more ILISP/SBCL
3870 squabbles, but hopefully it will be an improvement in the long run.)
3871 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3872 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3873 function, since declaiming return types involving aliens is
3874 (1) annoyingly messy to do by hand and (2) vital to efficient
3875 compilation of code which calls such functions.
3876 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3877 longer reexported by the SB-EXT package. They're solely useful
3878 for alien code, so it seems more logical that you should get
3879 them from the SB-ALIEN package, not in SB-EXT.
3880 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3881 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3882 features. Instead, the code that they used to control is always
3883 built into the system.
3884 * many other bug fixes
3885 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3886 advantage of the new EVAL-WHEN stuff and to clean them up in
3887 general, and they are now more ANSI-compliant in a number of
3888 ways. Martin Atzmueller is responsible for a lot of this.
3889 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3891 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3892 *** correct ERROR type for various file operations
3893 *** some fixes for Lisp streams
3894 *** DEFMETHOD syntax checking
3895 *** changing old weird representation of debug information as
3896 strings (which, among their other deficiencies, don't transform
3897 correctly when you rename packages, and don't change their
3898 print representation when you change things like *PACKAGE*
3899 and *PRINT-LENGTH*) to symbols and lists of symbols
3900 He also made several improvements and fixed several bugs in DESCRIBE.
3901 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3903 *** misbehavior of WRITE-STRING/WRITE-LINE
3904 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3905 and several other LOOP problems as well
3906 *** DIRECTORY when similar filenames are present
3907 *** DEFGENERIC with :METHOD options
3908 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3909 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3910 *** argument ordering in FIND with :TEST option
3911 *** mishandled package designator argument in APROPOS-LIST
3912 *** various problems in the backquote readmacro
3913 *** a bug in APROPOS
3914 *** probably some others that I'm not describing very well here,
3915 since the CVS log documents them by reference to sbcl-devel
3916 messages, and the SourceForge archives aren't working well.:-(
3917 ** Dan Barlow improved the Alpha port (and is making progress on the
3918 PPC port, for those of you who think different).
3919 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3920 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3921 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3922 continues to work on the SPARC port (for those of you in a position
3923 to look down upon our little PC-compatible boxes from a great height).
3924 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3925 and uses that copy, rather than the current dynamic value, when
3926 it's trying to decide whether to truncate output. Thus e.g.
3927 (let ((*print-lines* 50))
3928 (pprint-logical-block (stream nil)
3930 (let ((*print-lines* 8))
3931 (print (aref possiblybigthings i) stream)))))
3932 should now truncate the logical block only at 50 lines, instead of
3933 often truncating it at 8 lines, as it did before.
3934 * The doc/cmucl/ directory, containing old CMU CL documentation
3935 from the time of the fork, is no longer part of the base system.
3936 SourceForge has shut down its anonymous FTP service, and with it
3937 my original plan for distributing the old CMU CL documentation
3938 there. For now, if you need these files you can download an old
3939 SBCL source release and extract them from it.
3940 * The fasl file version number changed again, for dozens of reasons,
3941 some of which are apparent above.
3943 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3944 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3945 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3946 patch, so that DISASSEMBLE output is much nicer.
3947 * The code in the SB-PROFILE package now seems reasonably stable.
3948 I still haven't decided what the final interface should look like
3949 (I'd like PROFILE to interact cleanly with TRACE, since both
3950 facilities use function encapsulation) but if you have a need
3951 for profiling now, you can probably use it successfully with
3952 the current CMU-CL-style interface.
3953 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3954 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3955 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3956 link now returns the dangling link itself, and for similar
3957 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3958 link itself. (In these cases the old code signalled an error and
3959 looped endlessly, respectively.) Thus, DIRECTORY now works even
3960 in the presence of dangling and cyclic symbolic links.
3961 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3962 is now a supported extension again, since the consensus on
3963 sbcl-devel was that it can be useful for ordinary development
3964 work, not just for debugging SBCL itself.
3965 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3966 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3967 that currently-defined functions might be redefined later with
3968 different return types.
3969 * Hash tables can be printed readably, as inspired by CMU CL code
3970 of Eric Marsden and SBCL code of Martin Atzmueller.
3971 * better error handling in CLOS method combination, thanks to
3972 Martin Atzmueller porting Pierre Mai's CMU CL patches
3973 * more overflow fixes for >16Mbyte I/O buffers
3974 * A bug in READ has been fixed, so that now a single Ctrl-D
3975 character suffices to cause end-of-file on character streams.
3976 In particular, now you only need one Ctrl-D at the command
3977 line (not two) to exit SBCL.
3978 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3980 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3981 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3982 does the right thing.
3983 * The implementation of some type tests, especially for CONDITION
3984 types, is now tidier and maybe faster, due to CMU CL code
3985 originally by Douglas Crosher, ported by Martin Atzmueller.
3986 * Some math functions have been fixed, and there are new
3987 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3988 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3989 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3990 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3991 likely to be built into sbcl-0.7.0.)
3992 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3993 CL, ported to SBCL by Martin Atzmueller)
3994 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3995 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3996 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3997 * There's a new slam.sh hack to shorten the edit/compile/debug
3998 cycle for low-level changes to SBCL itself, and a new
3999 :SB-AFTER-XC-CORE target feature to control the generation of
4000 the after-xc.core file needed by slam.sh.
4001 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4002 COMPILE-FILE is no longer supported, so that now every function
4003 gets an entry point, so that block compilation looks a little
4004 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4005 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4006 returns the number of bytes consed since the system started,
4007 rather than the number consed since the first time the function
4008 was called. (The new definition parallels ANSI functions like
4009 CL:GET-INTERNAL-RUN-TIME.)
4010 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4011 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4012 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4013 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4014 * Fasl file version numbers are now independent of the target CPU,
4015 since historically most system changes which required version
4016 number changes have affected all CPUs equally. Similarly,
4017 the byte fasl file version is now equal to the ordinary
4020 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4021 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4022 is no longer recognized. I apologize for this, because it was
4023 listed in SB-EXT as a supported extension, but I found that
4024 its existing behavior was poorly specified, as well as incorrectly
4025 specified, and it looked like too much of a mess to straighten it
4026 out. I have enough on my hands trying to get ANSI stuff to work..
4027 * many patches ported from CMU CL by Martin Atzmueller, with
4028 half a dozen bug fixes in pretty-printing and the debugger, and
4029 half a dozen others elsewhere
4030 * fixed bug 13: Floating point infinities are now supported again.
4031 They might still be a little bit flaky, but thanks to bug reports
4032 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4033 as flaky as they were.
4034 * The --noprogrammer command line option is now supported. (Its
4035 behavior is slightly different in detail from what the old man
4036 page claimed it would do, but it's still appropriate under the
4037 same circumstances that the man page talks about.)
4038 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4039 are now supported, and enabled by default. Thus, the compiler can
4040 handle many floating point and complex operations much less
4041 inefficiently. (Thus e.g. you can implement a complex FFT
4043 * The compiler now detects type mismatches between DECLAIM FTYPE
4044 and DEFUN better, and implements CHECK-TYPE more correctly, and
4045 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4046 thanks to patches from Martin Atzmueller.
4047 * various fixes to make the cross-compiler more portable to
4048 ANSI-conforming-but-different cross-compilation hosts (notably
4049 Lispworks for Windows, following bug reports from Arthur Lemmens)
4050 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4051 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4052 fixed thanks to Pierre Mai's CMU CL patches.
4053 * fixes to keep the system from overflowing internal counters when
4054 it tries to use i/o buffers larger than 16M bytes
4055 * fixed bug 45a: Various internal functions required to support
4056 complex special functions have been merged from CMU CL sources.
4057 (When I was first setting up SBCL, I misunderstood a compile-time
4058 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4059 * improved support for type intersection and union, fixing bug 12
4060 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4061 more obscure bugs as well
4062 * some steps toward byte-compiling non-performance-critical
4063 parts of the system, courtesy of patches from Martin Atzmueller
4064 * Christophe Rhodes has made some debian packages of sbcl at
4065 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4066 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4067 stable, but are nonetheless usable. When he's ready, I'd be happy
4068 to add them to the SourceForge "File Releases" section. (And if
4069 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4070 * new fasl file format version number (because of changes in
4071 internal representation of (OR ..) types to accommodate the new
4072 support for (AND ..) types, among other things)
4074 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4075 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4077 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4079 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4080 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4081 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4082 * DESCRIBE now works on structure objects again.
4083 * Most function call argument type mismatches are now handled as
4084 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4085 know whether the function will be redefined before the call is
4086 executed. (The compiler could flag local calls with full WARNINGs,
4087 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4088 it doesn't keep track of enough information to know whether calls
4089 are local in this sense.)
4090 * Compiler output is now more verbose, with messages truncated
4091 later than before. (There should be some supported way for users
4092 to override the default verbosity, but I haven't decided how to
4093 provide it yet, so this behavior is still controlled by the internal
4094 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4095 src/compiler/ir1util.lisp.)
4096 * Fasl file format version numbers have increased again, because
4097 support for the Gray streams extension changes the layout of the
4098 system's STREAM objects.
4099 * The Gray subclassable streams extension now works, thanks to a
4100 patch from Martin Atzmueller.
4101 * The full LOAD-FOREIGN extension (not just the primitive
4102 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4103 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4104 but like most other programs, it defaults to copying the Unix
4105 environment from the original process instead of starting the
4106 new process in an empty environment.
4107 * Extensions which manipulate the Unix environment now support
4108 an :ENVIRONMENT keyword option which doesn't smash case or
4109 do other bad things. The CMU-CL-style :ENV option is retained
4110 for porting convenience.
4111 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4112 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4114 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4116 * A patch from Martin Atzmueller seems to have solved the SIGINT
4117 problem, and as far as we know, signal-handling now works cleanly.
4118 (If you find any new bugs, please report them!)
4119 * The system no longer defaults Lisp source file names to types
4120 ".l", ".cl", or ".lsp", but only to ".lisp".
4121 * The compiler no longer uses special default file extensions for
4122 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4123 seems to expect a single default extension for all compiled code,
4124 and there's no compelling reason to try to stretch the standard
4125 to allow two different extensions.) Instead, byte-compiled files
4126 default to the same extension as native-compiled files.
4127 * Fasl file format version numbers have increased again, because
4128 a rearrangement of internal implementation packages made some
4129 dumped symbols in old fasl files unreadable in new cores.
4130 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4131 many fewer weird special cases.
4132 * Bug #17 (differing COMPILE-FILE behavior between logical and
4133 physical pathnames) has been fixed, and some related misbehavior too,
4134 thanks to a patch from Martin Atzmueller.
4135 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4136 by Tim Moore, ported to SBCL by Martin Atzmueller.
4137 * Martin Atzmueller fixed several filesystem-related problems,
4138 including bug #36, in part by porting CMU CL patches, which were
4139 written in part by Paul Werkowski.
4140 * More compiler warnings in src/runtime/ are gone, thanks to
4141 more patches from Martin Atzmueller.
4142 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4145 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4147 * DESCRIBE now works on CONDITION objects.
4148 * The debugger now handles errors which arise when trying to print
4149 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4151 * The build system now uses an additional file, customize-target-features.lisp,
4152 to allow local modifications to the target *FEATURES* list. (The point of
4153 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4154 debugging features enabled, without having to worry about propagating it
4155 into everyone's system when I do a "cvs update".) When no
4156 customize-target-features.lisp file exists, the target *FEATURES* list
4157 should be constructed the same way as before.
4158 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4159 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4160 warn about behavior which is undefined under the ANSI spec. Note: This
4161 is specified by ANSI, but it's not very popular with programmers.
4162 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4163 macro in the SBCL sources for an example of a workaround which you
4164 might use to make portable ANSI-standard code which does what you want.
4165 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4166 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4167 when it's used as a non-toplevel form. (This is required in order
4168 to implement the DEFCONSTANT-EQX macro.)
4169 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4170 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4171 funny things, due to bugs in the implementation of EVAL-WHEN
4172 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4173 non-toplevel DEFCONSTANTs are uncommon.)
4174 * The core file version number and fasl file version number have been
4175 incremented, because the old noncompliant DEFCONSTANT behavior involved
4176 calling functions which no longer exist, and because I also took the
4177 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4178 * fixed bug 1 (error handling before read-eval-print loop starts), and
4179 redid debugger restarts and related debugger commands somewhat while
4181 ** The QUIT debugger command is gone, since it did something
4182 rather different than the SB-EXT:QUIT command, and since it never
4183 worked properly outside the main toplevel read/eval/print loop.
4184 Invoking the new TOPLEVEL restart provides the same functionality.
4185 ** The GO debugger command is also gone, since you can just invoke
4186 the CONTINUE restart directly instead.
4187 ** The TOP debugger command is also gone, since it's redundant with the
4188 FRAME 0 command, and since it interfered with abbreviations for the
4190 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4192 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4193 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4194 some porting work by Martin Atzmueller)
4195 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4196 with a boatload of patches which he ported to SBCL. Now that those
4198 ** The system tries to make sure that its low-priority messages
4199 are prefixed by semicolons, to help people who like to use
4200 syntax highlighting in their ILISP buffer. (This patch
4201 was originally due to Raymond Toy.)
4202 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4203 patches originally written by Raymond Toy.
4204 ** The compiler understands coercion between single-value and
4205 multiple-VALUES type expressions better, getting rid of some very
4206 weird behavior, thanks to patches originally by Robert MacLachlan
4207 and Douglas Crosher.
4208 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4209 lambda lists, thanks to a patch originally by Pierre Mai.
4210 ** The system no longer bogusly warns about "abbreviated type
4212 ** The compiler gets less confused by inlining and RETURN-FROM,
4213 thanks to some patches originally by Tim Moore.
4214 ** The system no longer hangs when dumping circular lists to fasl
4215 files, thanks to a patch originally from Douglas Crosher.
4216 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4217 undefined function error.
4218 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4219 in structure slot initforms) for now, documented workaround instead:-|
4220 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4221 * fixed bug 5: added stubs for various Gray stream functions called
4222 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4223 installed, at least appropriate type errors are generated
4224 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4225 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4226 consistently in DEFMETHOD forms.
4227 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4228 it doesn't seem to affect SBCL after all
4229 * The C runtime system now builds with better optimization and many
4230 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4232 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4234 * The system is now under CVS at SourceForge (instead of the
4235 CVS repository on my home machine).
4236 * The new signal handling code has been tweaked to treat register
4237 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4238 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4239 downstream, e.g. in debug-int.lisp, has implicit dependencies
4240 on the unsignedness of integer representation of machine words,
4241 and that caused the system to bomb out with infinite regress
4242 when trying to recover from type errors involving signed values,
4243 e.g. (BUTLAST '(1 2 3) -1).)
4244 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4245 (This was one of the bugs Peter Van Eynde reported back in July.)
4246 * The system now uses code inspired by Colin Walters' O(N)
4247 implementation of MAP (from the cmucl-imp@cons.org mailing
4248 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4249 inline the MAP operation, and there is more than one
4250 sequence argument to the MAP call (so that it can't just
4251 do ETYPECASE once and for all based on the type of the
4252 single sequence argument). (The old non-inline implementation
4253 of the general M-argument sequence-of-length-N case required
4254 O(M*N*N) time when any of the sequence arguments were LISTs.)
4255 * The QUIT :UNIX-CODE keyword argument has been renamed to
4256 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4257 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4259 * Raymond Wiker's patches to port dynamic loading from Linux to
4260 FreeBSD have been added.
4261 * The BUGS file is now more nearly up to date, thanks in large part
4262 to Martin Atzmueller's review of it.
4263 * The debugger now flushes standard output streams before it begins
4264 its output ("debugger invoked" and so forth).
4265 * The core version number and fasl file version number have both
4266 been incremented, because of incompatible changes in the layout
4268 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4269 used to not be called for a saved Lisp image.)
4270 * Martin Atzmueller's version of a patch to fix a compiler crash,
4271 as posted on sbcl-devel 13 September 2000, has been installed.
4272 * Instead of installing Martin Atzmueller's patch for the
4273 compiler transform for SUBSEQ, I deleted the compiler transform,
4274 and transforms for some similar consing operations.
4275 * A bug in signal handling which kept TRACE from working on OpenBSD
4277 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4278 away by constant folding
4279 * The system now defines its address space constants in one place
4280 (in the Lisp sources), and propagates them automatically elsewhere
4281 (through GENESIS and the sbcl.h file). Therefore, patching the
4282 address map is less unnecessarily tedious and error-prone. The
4283 Lisp names of address space constants have also been systematized.
4284 * CVS tags like dollar-Header-dollar have been removed from
4285 the sources, because they have never saved me trouble and
4286 they've been source of trouble working with patches and other
4287 diff-related operations.
4288 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4289 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4291 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4293 * The system has been ported to OpenBSD.
4294 * The system now compiles with a simple "sh make.sh" on the systems
4295 that it's supported on. I.e., now you no longer need to tweak
4296 text in the target-features.lisp-expr and symlinks in src/runtime/
4297 by hand, the make.sh takes care of it for you.
4298 * The system is no longer so grossly inefficient when compiling code
4299 involving vectors implemented as general (not simple) vectors (VECTOR T),
4300 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4301 which dares to use the various sequence functions on non-simple
4302 vectors, takes less of a performance hit.
4303 * There is now a primitive type predicate VECTOR-T-P
4304 to test for the (VECTOR T) type, so that e.g.
4305 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4306 can now be compiled with some semblance of efficiency. (The old code
4307 turned the type declaration into a full call to %TYPEP at runtime!)
4308 * AREF on (VECTOR T) is still not fast, since it's still compiled
4309 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4310 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4311 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4312 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4313 without making full calls to SUBTYPEP at runtime.
4314 (Some analogous efficiency issues for non-simple vectors specialized to
4315 element types other than T, or for non-simple multidimensional arrays,
4316 have not been addressed. They could almost certainly be handled the
4317 same way if anyone is motivated to do so.)
4318 * The changes in array handling break binary compatibility, so
4319 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4320 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4321 instead of (VALUES T T).
4322 * By following the instructions that Dan Barlow posted to sbcl-devel
4323 on 2 July 2000, I was able to enable primitive dynamic object
4324 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4325 functionality is not implemented (since it calls ld to resolve
4326 library references automatically, requiring RUN-PROGRAM for its
4327 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4328 try to resolve library references) is now supported.
4329 * The system now flushes the standard output streams when it terminates,
4330 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4331 them at several other probably-convenient times, e.g. in each pass of
4332 the toplevel read-eval-print loop, and after evaluating a form given
4333 as an "--eval" command-line option. (These changes were motivated by a
4334 discussion of stream flushing issues on cmucl-imp in August 2000.)
4335 * The source transform for TYPEP of array types no longer assumes
4336 that an array whose element type is a not-yet-defined type
4337 is implemented as an array of T, but instead punts, so that the
4338 type will be interpreted at runtime.
4339 * There is now some support for cross-compiling in make.sh: each of
4340 the phases of make.sh has its own script. (This should be transparent
4341 to people doing ordinary, non-cross-compile builds.)
4342 * Since my laptop doesn't have hundreds of megabytes of memory like
4343 my desktop machine, I became more motivated to do some items on
4344 my to-do list in order to reduce the size of the system a little:
4345 ** Arrange for various needed-only-at-cold-init things to be
4346 uninterned after cold init. To support this, those things have
4347 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4348 symbols with such names are now uninterned after cold init).
4349 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4350 fasl files for cold load.
4351 ** Remove the old compiler structure pooling code (which used to
4352 be conditional on the target feature :SB-ALLOC) completely.
4353 ** Redo the representation of some data in cold init to be more compact.
4354 (I also looked into supporting byte compiled code at bootstrap time,
4355 which would probably reduce the size of the system a lot, but that
4356 looked too complicated, so I punted for now.)
4357 * The maximum signal nesting depth in the src/runtime/ support code has
4358 been reduced from 4096 to 256. (I don't know any reason for the very
4359 large old value. If the new smaller value turns out to break something,
4360 I'll probably just bump it back up.)
4361 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4363 * Many, many bugs reported by Peter Van Eynde have been added to
4364 the BUGS list; some have even been fixed.
4365 * While enabling dynamic object file loading, I tried to make the
4366 code easier to understand, renaming various functions and variables
4367 with less ambiguous names, and changing some function calling
4368 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4369 * While trying to figure out how to do the OpenBSD port, I tried to
4370 clean up some of the code in src/runtime/. In particular, I dropped
4371 support for non-POSIX signal handling, added various comments,
4372 tweaked the code to reduce the number of compilation warnings, and
4373 renamed some files to increase consistency.
4374 * To support the new automatic configuration functionality in make.sh,
4375 the source file target-features.lisp-expr has been replaced with the
4376 source file base-target-features.lisp-expr and the machine-generated
4377 file local-target-features.lisp-expr.
4378 * fixed a stupid quoting error in make.sh so that using CMU CL
4379 "lisp -batch" as cross-compilation host works again
4381 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4383 * DESCRIBE no longer tries to call itself recursively to describe
4384 bound/fbound values, so that it no longer fails on symbols which are
4385 bound to themselves (like keywords, T, and NIL).
4386 * DESCRIBE now works on generic functions.
4387 * The printer now prints less-screwed-up representations of closures
4388 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4389 * A private symbol is used instead of the :EMPTY keyword previously
4390 used to mark empty slots in hash tables. Thus
4391 (DEFVAR *HT* (MAKE-HASH-TABLE))
4392 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4393 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4394 now does what ANSI says that it should. (You can still get
4395 similar noncompliant behavior if bang on the hash table
4396 implementation with all the symbols you get back from
4397 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4398 This breaks binary compatibility, since tests for equality to
4399 :EMPTY are wired into things like the macroexpansion of
4400 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4402 * There's now a minimal placeholder implementation for CL:STEP,
4403 as required by ANSI.
4404 * An obscure bug in the interaction of the normal compiler, the byte
4405 compiler, inlining, and structure predicates has been patched
4406 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4407 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4409 * Missing ordinary arguments in a macro call are now detected even
4410 when the macro lambda list contains &KEY or &REST.
4411 * The debugger no longer complains about encountering the top of the
4412 stack when you type "FRAME 0" to explicitly instruct it to go to
4413 the top of the stack. And it now prints the frame you request even
4414 if it's the current frame (instead of saying "You are here.").
4415 * As specified by ANSI, the system now always prints keywords
4416 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4418 * The default initial SIZE of HASH-TABLEs is now smaller.
4419 * Type information from CLOS class dispatch is now propagated
4420 into DEFMETHOD bodies, so that e.g.
4421 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4423 is now basically equivalent to
4424 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4425 (DECLARE (TYPE SINGLE-FLOAT X))
4427 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4428 operation, without having to do run-time type dispatch.
4429 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4430 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4431 * Since I don't seem to be making big file reorganizations very often
4432 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4433 is overflowing my ability to conveniently back them up), I've finally
4434 checked the system into CVS. (The CVS repository is on my home system,
4435 not at SourceForge -- putting it on SourceForge might come later.)
4436 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4437 high-level GC-NOTIFY-FOO functions send their output. (There's
4438 still very little control of where low-level verbose GC functions
4439 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4440 less than it used to -- the GC-NOTIFY-FOO functions are now under
4441 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4442 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4443 in only one place in the source code, and propagates it automatically
4444 everywhere that it's needed. Thus e.g. when I bump the version from
4445 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4446 * The C source files now include boilerplate legalese and documentation
4447 at the head of each file (just as the Lisp source files already did).
4448 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4449 to his page will be replaced with a link to his new CLiki service.
4451 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4453 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4454 * The build process now looks for GNU make under the default name "gmake",
4455 instead of "make" as it used to. If GNU make is not available as "gmake"
4456 on your system, you can change this default behavior by setting the
4457 GNUMAKE environment variable.
4458 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4459 can build without error under CMU CL.
4461 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4463 * There is now a partial SBCL user manual (with some new text and some
4464 text cribbed from the CMU CL manual).
4465 * The beginnings of a profiler have been added (starting with the
4466 CMU CL profiler and simplifying and cleaning up). Eventually the
4467 main interface should be through the TRACE macro, but for now,
4468 it's still accessed through vaguely CMU-CL-style functions and macros
4469 exported from the package SB-PROFILE.
4470 * Some problems left over from porting CMU CL to the new
4471 cross-compilation bootstrap process have been cleaned up:
4472 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4473 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4474 part when getting rid of the compiler *BACKEND* stuff.)
4475 ** The value of *NULL-TYPE* was screwed up, because it was
4476 being initialized before the type system knew the final
4477 definition of the 'NULL type. This screwed up several key
4478 optimizations in the compiler, causing inefficiency in all sorts
4479 of places. (I found it because I wanted to understand why
4480 GET-INTERNAL-RUN-TIME was consing.)
4481 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4482 PROCLAIM FTYPE information. Unfortunately this broke binary
4483 compatibility again, since now the forms output by DEFGENERIC
4484 to refer to functions which didn't exist in 0.6.3.
4485 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4486 can use the new (as of 0.6.3) transform for SOME into MAP into
4488 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4489 values to (INTEGER 0 1000000), so that the time code will no longer
4490 occasionally get blown up by Linux returning 1000000 microseconds
4491 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4492 its output conform to the ANSI spec. (Alas, this makes its output
4493 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4494 * A full call to MAP NIL with a single sequence argument no longer conses.
4495 * fixes to problems pointed out by Martin Atzmueller:
4496 * The manual page no longer talks about multiprocessing as though
4497 it were currently supported.
4498 * The ILISP support patches have been removed from the distribution,
4499 because as of version 5.10.1, ILISP now supports SBCL without us
4500 having to maintain patches.
4501 * added a modified version of Raymond Toy's recent CMU CL patch for
4502 EQUALP comparison of HASH-TABLE
4504 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4506 * The system still can't cross-compile itself with
4507 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4508 entails), but at least it can compile more of itself that way
4509 than it used to be able to, and various buglets which were uncovered
4510 by trying to cross-compile itself that way have now been fixed.
4511 * This release breaks binary compatibility again. This time
4512 at least I've incremented the FASL file format version to 2, so that the
4513 problem can be detected reliably instead of just causing weird errors.
4514 * various new style warnings:
4515 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4516 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4517 ** using the lexical binding of a variable named in the *FOO* style
4518 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4519 as specified by ANSI.
4520 * *RANDOM-STATE* is no longer automatically initialized from
4521 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4522 default behavior of the system is to repeat its behavior every time
4523 it's run. If you'd like to change this behavior, you can always
4524 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4525 old convention there was no comparably easy way to get the system to
4526 repeat its behavior every time it was run.
4527 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4528 FTYPE declarations has been removed, in favor of their ANSI
4529 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4530 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4531 the types of their sequence arguments can be determined at compile time.
4532 This is done through a new open code expansion for MAP which eliminates
4533 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4534 argument types can be determined at compile time.
4535 * The optimizer now transforms COERCE into an identity operation when it
4536 can prove that the coerced object is already of the correct type. (This
4537 can be a win for machine generated code, including the output of other
4538 optimization transforms, such as the MAP transform above.)
4539 * Credit information has been moved from source file headers into CREDITS.
4540 * Source file headers have been made more standard.
4541 * The CASE macro now compiles without complaining even when it has
4544 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4546 * (Note that the way that the PCL macroexpansions were rewritten
4547 to accommodate the change in DEFGENERIC below breaks binary
4548 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4549 not run under sbcl-0.6.2. Once we get out of alpha releases,
4550 i.e. hit release 1.0.0, we'll probably try to maintain binary
4551 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4552 and sbcl-1.4.4. Until then, however, it might be fairly common
4553 for maintenance releases to break binary compatibility.)
4554 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4556 * The reporting of unhandled signals has been changed to print some
4557 explanatory text as well as the report form. (Previously only
4558 the report form was printed.)
4559 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4560 it defines, so that the compiler no longer issues undefined function
4561 warnings for compiled-but-not-yet-loaded generic functions.
4562 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4563 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4564 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4565 section of the ANSI spec are supported.
4566 * The "" nickname for the "KEYWORD" package has been removed.
4567 The reader still handles symbol tokens which begin with a package marker
4568 as keywords, but it doesn't expose its mechanism for doing so in the
4569 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4570 * The system now issues STYLE-WARNINGs for contradictory TYPE
4571 proclamations. (Warnings for contradictory FTYPE proclamations would
4572 be nice too, but those can't be done usefully unless the type system
4573 is made smarter about FUNCTION types.)
4574 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4575 systematized, so that "*target-*.lisp is supposed to exist only on the
4576 target and imply that there's a related file which exists on the
4577 host, and *host-*.lisp is supposed to exist only on the host and imply
4578 that there's a related file which exists on the target. This involves a
4579 lot of renaming. Hopefully the acute confusion caused by the renaming
4580 will be justified by the reduction in chronic confusion..
4581 ** runtime-type.lisp -> early-target-type.lisp
4582 ** target-type.lisp -> late-target-type.lisp
4583 ** early-host-format.lisp -> early-format.lisp
4584 ** late-host-format.lisp -> late-format.lisp
4585 ** host-error.lisp -> misc-error.lisp
4586 ** early-error.lisp -> early-target-error.lisp
4587 ** late-error.lisp -> late-target-error.lisp
4588 ** host-defboot.lisp -> early-defboot.lisp
4589 ** code/misc.lisp -> code/target-misc.lisp
4590 ** code/host-misc.lisp -> code/misc.lisp
4591 ** code/numbers.lisp -> code/target-numbers.lisp
4592 ** code/early-numbers.lisp -> numbers.lisp
4593 ** early-host-type.lisp -> early-type.lisp
4594 ** late-host-type.lisp -> late-type.lisp
4595 ** host-typep.lisp -> typep.lisp
4596 ** load.lisp -> target-load.lisp
4597 ** host-load.lisp -> load.lisp
4598 ** host-disassem.lisp -> disassem.lisp
4599 ** host-insts.lisp -> insts.lisp
4600 ** byte-comp.lisp -> target-byte-comp.lisp
4601 ** host-byte-comp.lisp -> byte-comp.lisp
4602 ** host-signal.lisp -> signal.lisp
4603 ** host-defstruct.lisp -> defstruct.lisp
4604 ** late-target-type.lisp -> deftypes-for-target.lisp
4605 Furthermore, several other previously target-only files foo.lisp (e.g.
4606 hash-table.lisp and random.lisp) have been split into a target-and-host
4607 foo.lisp file and a target-only target-foo.lisp file, with their key type
4608 definitions in the target-and-host part, so that the cross-compiler will
4609 know more about target types.
4610 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4611 gone away. In their place are various *BACKEND-FOO* variables
4612 corresponding to the slots of the old structure.
4613 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4614 into the target SBCL has been fixed.
4615 * The chill.lisp system for loading cold code into a running SBCL
4617 * Support for the CMU CL "scavenger hook" extension has been removed.
4618 (It was undocumented and unused in the CMU CL sources that SBCL was
4619 derived from, and stale in sbcl-0.6.1.)
4620 * Various errors in the cross-compiler type system were detected
4621 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4622 (enabling various consistency checks). Many of them were fixed,
4623 but some hard problems remain, so the compiler is back to
4624 running without *TYPE-SYSTEM-INITIALIZED* for now.
4625 * As part of the cross-compiler type system cleanup, I implemented
4626 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4627 * I have started adding UNCROSS calls throughout the type system
4628 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4629 on cross-compiler input files will be able to go away, and various
4631 * CONSTANTP now returns true for quoted forms (as explicitly required
4634 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4636 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4637 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4638 (MAKE-LIST -1). In the longer term, it ought to become true
4639 that declarations are assertions even at SAFETY 1. For now, it's not
4640 quite true even at SAFETY 3, but it's at least more nearly true..
4641 (Note that this change seems to increases the size of the system by
4642 O(5%) and to decrease the speed of the compiler by 20% or more.)
4643 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4644 for the problem of printing dozens of lines of distracting information
4645 about low-level system machinery as part of the top stack frame
4646 on entry to the debugger when an undefined function was called.
4647 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4648 is not reset to COMMON-LISP-USER.
4649 * Compilation of stuff related to dyncount.lisp has been made conditional
4650 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4651 smaller. The various dyncount-related symbols have been moved into
4652 a new "SB-DYNCOUNT" package.
4653 * tty-inspect.lisp has been renamed to inspect.lisp.
4654 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4655 feature has gone away. (When we eventually port to other flavors of
4656 libc and/or Unix, we'll try to make the differences between flavors
4657 invisible at the user level.)
4658 * Various other *FEATURES* tags, and/or their associated conditionals,
4659 have been removed if obsolescent, or given better documentation, or
4660 sometimes given more-mnemonic names.
4662 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4664 * tidied up "make.sh" script
4665 * tidied up system directory structure
4666 * better "clean.sh" behavior
4667 * added doc/FOR-CMUCL-DEVELOPERS
4668 * many many small tweaks to output format, e.g. removing possibly-confusing
4669 trailing #\. character in DESCRIBE-INSTANCE
4670 * (EQUALP #\A 'A) no longer signals an error.
4671 * new hashing code, including EQUALP hashing
4672 * tidied up Lisp initialization and toplevel
4673 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4674 * command line argument processing
4675 * added POSIX-GETENV function to deal with Unix-ish environment variables
4676 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4677 terminating SBCL on EOF
4678 * non-verbose GC by default
4679 * There is no more "sbcl" shell script; the sbcl file is now the C
4680 runtime executable (just like CMU CL).
4681 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4683 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4684 symbol and package information as Lisp native symbol and package objects
4685 instead of strings naming symbols and strings naming packages. This way,
4686 whenever packages are renamed (as in warm init), debug information is
4687 transformed along with everything else.
4688 * tweaked the optimization policy declarations which control the building
4689 of SBCL itself. Now, among other things, the system no longer saves
4690 source location debugging information. (This helps two problems at once
4691 by reducing SBCL size and by keeping SBCL from trying to look for its
4692 sources -- which may not exist -- when reporting errors.)
4693 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4694 debugging and testing purposes
4695 * cleaned up printing, making the printer call PRINT-OBJECT for
4696 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4697 methods, giving nearly-ANSI behavior
4698 * converted almost all special variables to use *FOO* naming convention
4699 * deleted PARSE-TIME functionality, since it can be done portably
4700 * moved some files out of cold init into warm init
4701 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4703 * regularized formatting of source files
4704 * added an install.sh script
4705 * fixed ridiculous memory usage of cross-compiler by making
4706 compiler/alloc.lisp not try to do pooling unless it can hook
4707 itself into the GC of the cross-compilation host. Now the system
4708 builds nicely on my old laptop.
4709 * added :SB-ALLOC in target-features.lisp-expr
4710 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4711 was not implemented)
4712 * re-did condition handling and note reporting in the compiler. Notes
4713 are no longer handled by signalling conditions. Style warnings
4714 and warnings are handled more correctly and reported in such a way
4715 that it's easy to find one or the other in your output (so that you
4716 can e.g. figure out which of many problems caused COMPILE-FILE to
4718 * changed the severity of several compiler warnings from full WARNING
4719 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4720 compiler note reporting so that it doesn't use the condition system
4721 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4722 COMPILE-FILE command)
4723 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4724 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4725 no longer supported, since their implementation was deeply intertwingled
4726 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4727 * removed bogus "support" for compiler macros named (SETF FOO), and
4728 removed the compiler macro for SETF INFO (but only after making a fool
4729 of myself on the cmucl-imp mailing list by posting a bogus patch for
4730 DEFINE-COMPILER-MACRO..)
4731 * Compiled files containing forms which have side effects on the Lisp
4732 reader (such as DEFPACKAGE forms) are now handled more correctly.
4733 (Compiler queuing of top level lambdas has been suppressed by setting
4734 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4735 * deleted various currently-unused source files, e.g. gengc.lisp. They
4736 may be added back at some point e.g. when porting to other architectures,
4737 but until they are it's distracting to distribute them and to try to
4739 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4740 there were so many of them they're just distractions, and UNCROSS is
4741 known to be able to handle the current sources
4742 * moved PROFILE functionality into TRACE, so that it will be clear
4743 how the wrapping and unwrapping of functions when you profile them
4744 interacts with the wrapping and unwrapping of functions when you
4745 trace them. (Actually, the functionality isn't there yet, but at least
4746 the interface specification is there. Hopefully, the functionality will
4747 arrive with some future maintenance release.)
4748 * removed host-oops.lisp
4749 * changed signature of QUIT function to allow UNIX-CODE argument
4750 * fixed READ-SEQUENCE bug
4751 * tweaked verbose GC output so that it looks more like the progress
4752 output that ANSI specifies for functions like LOAD
4753 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4754 * added <http://sbcl.sourceforge.com> to the banner information printed by