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 * fixed some bugs revealed by Paul Dietz' test suite:
6 ** interval arithmetic during type derivation used inexact integer
7 to single-float coercions.
8 ** artihmetic operations involving large integers and single
9 floats give the same results in compiled and interpreted code.
10 ** deriving the result type of COERCE no longer signals an error
11 if the derived type of the second argument is a MEMBER type
12 containing invalid type specifiers.
13 ** ADJOIN with constant NIL as second argument works correctly.
15 changes in sbcl-1.0.18 relative to 1.0.17:
16 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
17 profiles only the current thread.
18 * minor incompatible change: changes to SYMBOL-VALUE of constants
19 defined with DEFCONSTANT now signal an error.
20 * enhancement: SB-SPROF now has support for wallclock profiling,
21 and is also able to profile specific threads. REPORT output
22 has also additional sorting options.
23 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
25 * optimization: structure allocation has been improved
26 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
27 ** out of line constructors are ~10% faster.
28 ** inline constructors are ~15% faster.
29 ** inline constructors are capable of dynamic extent allocation
30 (generally on x86 and x86-64, in some cases on other platforms
32 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
34 * optimization: file compiler is now able to coalesce non-circular
35 lists, non-base strings, and bit-vectors. Additionally, constants
36 are never referenced through SYMBOL-VALUE at runtime.
37 * optimization: code defining methods on PRINT-OBJECT (and other
38 generic functions in the COMMON-LISP package) now loads faster.
39 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
41 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
42 fasl is now deleted. (reported by Attila Lendvai)
43 * bug fix: READ-LINE always returned NIL for the last line in files.
44 (reported by Yoshinori Tahara)
45 * bug fix: more accurate disassembly annotations of foreign function
46 calls. (thanks to Andy Hefner)
47 * bug fix: trimming non-simple strings and non-string string
48 designators when the there is nothing to trim works properly.
49 (thanks to James Knight)
50 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
51 (thanks to Travis Cross)
52 * fixed some bugs revealed by Paul Dietz' test suite:
53 ** NIL is a valid function name (regression at 1.0.13.38)
54 ** FILL on lists was missing its return value (regression at 1.0.12.27)
55 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
56 fill pointers properly (regression at 1.0.12.23)
57 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
58 displacement indices properly (regression at 1.0.12.23)
60 changes in sbcl-1.0.17 relative to 1.0.16:
61 * temporary regression: user code can no longer allocate closure
62 variable storage on stack, due to bug 419 without explicitly
63 requesting it. Please consult sbcl-devel for advice if you need to
64 use this feature in the meanwhile.
65 * new feature: runtime argument --control-stack-size can be used to
66 adjust thread default control stack size.
67 * enhancement: improved TIME output
68 ** all times are reported using the measured accuracy (milliseconds
69 for real and GC times, microseconds for everything else.)
70 ** processor cycle counts on x86 and x86-64.
71 ** interpreted forms are counted for both evaluator modes.
72 ** number of lambdas converted by the compiler is reported.
73 ** CPU percentage report (computed from real and total run time.)
74 ** more comprehensive run time reporting, using a condenced format
75 ** interperted form, lambda, and page fault counts are omitted
77 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
79 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
80 * optimization: two argument forms of LAST are upto ~10% faster
81 in normal SPEED policies.
82 * optimization: NCONC no longer needs to heap cons its &REST list
83 in normal SPEED policies.
84 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
86 * bug fix: UNION and NUNION work with :TEST-NOT once more,
87 regression since 1.0.9.1. (thanks to Eric Marsden)
88 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
89 since 1.0.15.36. (thanks to Paul Khuong)
90 * bug fix: LAST when always returned the whole list when given a bignum
91 as the second argument.
92 * bug fix: dynamic extent allocation of nested lists and vectors
93 could leak to otherwise accessible parts.
94 * bug fix: invalid optimization of heap-allocated alien variable
96 * bug fix: fasl header checking is less vulnerable to different
97 platform word lengths.
98 * bug fix: more correct assembler syntax for GNU binutils
99 2.18.50.0.4 support. (thanks to Marijn Schouten)
100 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
102 * bug fix: the fopcompiler can handle LOCALLY forms (with no
103 declarations) successfully. (reported by Attila Lendvai)
105 changes in sbcl-1.0.16 relative to 1.0.15:
106 * minor incompatible change: revert the changes to sb-posix's error
107 signaling added in 1.0.14.
108 * minor incompatible change: change PROBE-FILE back to returning
109 NIL whenever we can't get a truename, as was the case before 1.0.14.
110 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
111 inherits from ERROR instead of just CONDITION.
112 * new feature: SB-INTROSPECT can provide source locations for instances
113 as well. (thanks to Tobian Ritterweiler)
114 * optimization: binding special variables now generates smaller code
115 on threaded platforms.
116 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
117 and cases where no :TEST is given but the compiler can infer that
118 the element to search is of type (OR FIXNUM (NOT NUMBER)).
119 * optimization: better LOGNOT on fixnums.
120 * optimization: modular arithmetic for a particular requested width
121 is implemented using a tagged representation unless a better
122 representation is available.
123 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
124 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
125 for wellformedness and returns a specialized vector. (reported by
126 Francois-Rene Rideau)
127 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
128 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
129 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
130 unscheduling timer at the same time as another timer fires.
131 * bug fix: don't reschedule timers for dead threads.
132 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
133 * bug fix: copying output from RUN-PROGRAM to a stream signalled
134 bogus errors if select() was interrupted.
135 * enhancement: add support for fcntl's struct flock to SB-POSIX.
137 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
138 * enhancement: cleaner backtraces for interactive interrupts, as
139 well as other cases where the interesting frames used to be
140 obscured by interrupt handling frames.
141 * enhancement: untracing a whole package using (UNTRACE "FOO") is
142 now supported, and tracing a whole package using (TRACE "FOO") now
143 traces SETF-functions as well.
144 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
145 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
146 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
147 * unreadably printed representation of hash-tables now includes
149 * bug fix: partially fixed #188: type propagation from assignments
150 is now more efficient.
151 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
152 'DOUBLE-FLOAT) are not flushable.
153 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
154 full address of the object, and none of the tag bits.
155 * bug fix: readably printing hash-tables now respects other printer
156 control variables. (reported by Cedric St-Jean)
157 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
158 keyword with MAKE-HASH-TABLE.
159 * bug fix: export SB-POSIX:MKSTEMP.
160 * bug fix: SORT was not interrupt safe.
161 * bug fix: XREF accounts for the last node of each basic-block as
163 * bug fix: MAKE-INSTANCE optimizations interacted badly with
164 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
165 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
166 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
167 create obsolete instances in the case of redefinition or
168 obsoletion of a superclass. (thanks to Andy Hefner)
169 * bug fix: Support for the Alpha architecture has been revived; it had
170 suffered somewhat from lack of maintenance since sbcl-1.0.
171 * improvements to the Windows port:
172 ** The system detects the codepage to use at startup, instead of
173 using the value from the saved core. (thanks to Kei Suzuki)
175 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
176 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
177 (see documentation for details.)
178 * revived support for OpenBSD (contributed by Josh Elsasser)
179 * partially fixed bug #108: ROOM no longer suffers from occasional
180 (AVER (SAP= CURRENT END)) failures.
181 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
182 bodies no longer cause a WARNING to be signalled. (reported by
184 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
185 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
186 the bug in TRUNCATE.) (reported by Michael Weber)
187 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
188 no samples. (reported by Andy Hefner)
189 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
190 no longer appear as (NIL ...) frames in backtraces.
191 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
192 is now actually exported from SB-EXT as documented. (reported by
194 * bug fix: sb-aclrepl now correctly understands how to inspect
195 single-floats on 64-bit platforms where single-floats are not boxed.
196 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
197 yet been finalized. (reported by Levente Meszaros)
198 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
199 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
200 SBCL-specific optimize qualities.
202 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
203 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
204 an executable in the search path, and does so in the child
205 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
206 been removed; it can be found in the CVS history, for somebody who
207 needs that search behavior (see the manual).
208 * minor incompatible change: compiler policy re. weakening type
209 checks has changed: now type checks are weakened only if SAFETY < 2
211 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
212 unparsing of directory pathnames as files. Analogously,
213 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
214 filename to parse into a directory pathname.
215 * enhancement: implicit generic function creation now signals a
216 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
217 users can bind handlers for and muffle around calls to LOAD.
218 * enhancement: RUN-PROGRAM allows unicode arguments and environments
219 to be used (using the default stream external format), and allows
220 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
221 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
222 strings and vectors whose element-type or simplicity is not fully
223 known at compile-time.
224 * optimization: STRING-TRIM and related functions no longer allocate
225 a new string when no trimming needs to be performed. These functions
226 are also faster than before when the input string has been declared
228 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
229 * optimization: READ-LINE is significantly faster for files containing
231 * optimization: non-open coded uses of character comparison operators
232 (e.g. char=) no longer cons when called with more than one parameter
233 on platforms supporting dynamic-extent allocation.
234 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
235 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
236 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
238 * bug fix: some sequence functions elided bounds checking when
240 * bug fix: too liberal weakening of union-type checks when SPEED >
242 * bug fix: more bogus fixnum declarations in ROOM implementation
245 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
246 * new feature: MAKE-HASH-TABLE now experimentally accepts a
247 :SYNCHRONIZED argument, which makes the hash-table safe for
248 concurrent accesses (but not iteration.) See also:
249 SB-EXT:WITH-LOCKED-HASH-TABLE, and
250 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
251 * optimization: CONCATENATE on strings is an order of magnitude faster
252 in code compiled with (> SPEED SPACE).
253 * optimization: SUBSEQ is ~50% faster on lists.
254 * bug fix: bug 417 fixed -- source location reporting is now more robust.
255 * bug fix: SUBSEQ on a list will now correctly signal an error if if
256 END is smaller then START.
257 * bug fix: SB-PROFILE will no longer report extra consing for nested
258 calls to profiled functions.
259 * bug fix: ROOM implementation had bogus fixnum declarations which
260 could cause type-errors when calling ROOM on large images.
261 * bug fix: if file compilation is aborted, the partial fasl is now
262 deleted, and COMPILE-FILE returns NIL as the primary value.
263 * bug fix: number of thread safety issues relating to SBCL's internal
264 hash-table usage have been fixed.
265 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
266 be returned from its body when the values were being returned
267 using unknown-values return convection and the W-P-O was wrapped
268 inside an UNWIND-PROTECT.
269 * bug fix: sb-posix should now compile again under Windows, enabling
272 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
273 * incompatible change: hash-table accessor functions are no longer
274 automatically protected by locks. Concurrent accesses on the same hash-table
275 from multiple threads can give inconsistent results or even corrupt the
276 hash-table completely. Multi-threaded applications should do their own
277 locking at the correct granularity. In the current implementation it is
278 still safe to have multiple readers access the same table, but it's not
279 guaranteed that this property will be maintained in future releases.
280 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
281 and will signal an error at runtime.
282 * enhancement: SB-THREAD package now exports a semaphore interface.
283 * enhancement: CONS can now stack-allocate on x86 and
284 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
286 * enhancement: nested lists can now be stack allocated on
287 platforms providing stack allocation support.
288 * enhancement: dynamic-extent support has been extended to support
289 cases where there are multiple possible sources for the stack
291 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
292 if the mutex is uncontested on Linux.
293 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
294 as specified by the CLHS. (thanks to Tobias Rittweiler)
295 * bug fix: NaN comparison now works on x86-64.
296 * bug fix: CLOSE :ABORT T on a stream with pending output now
298 * bug fix: instances of non-standard metaclasses using standard
299 instance structure protocol sometimes missed the slot type checks
301 * bug fix: known functions can be inlined in high-debug code.
303 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
304 * minor incompatible change: the MSI installer on Windows no longer
305 associates .lisp and .fasl files with the installed SBCL.
306 * minor incompatible change: :UNIX is no longer present in *FEATURES*
307 on Windows. (thanks to Luis Oliviera)
308 * new feature: SB-CLTL2 contrib module now implements
309 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
310 * optimization: scavenging weak pointers is now more efficient,
311 requiring O(1) instead of O(N) per weak pointer to identify
312 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
313 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
314 method bodies using constant slot-names when the first argument is
315 a specializer parameter for the method.
316 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
317 names, when the class of the instance is a direct instance
318 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
319 as slow as the constant slot-name case.)
320 * optimization: member type construction is now O(N) instead
322 * optimization: UNION and NUNION are now O(N+M) for large
323 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
325 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
326 now more readable in environments like Slime which display it.
327 (thanks to Tobias C. Rittweiler)
328 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
329 was able to infer the correct argument type for the object on which
330 the CAS operation was being performed.
331 * bug fix: copy propagation interfered with parallel assignment
332 semantics in local calls. (reported by Paul Khuong)
333 * bug fix: the signed modular fixnum shift compiled to wrong code on
334 x86 and x86-64. (spotted by a slight modification to some of
337 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
338 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
339 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
340 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
342 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
343 and generic function now print the object identity for anonymous
344 instances. (thanks to Attila Lendvai)
345 * enhancement: as an extension to MOP, SBCL now supports SETF of
346 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
347 (thanks to Attila Lendvai)
348 * enhancement: sb-cover has an alternate annotation mode which puts
349 the form annotations at the car of the form, instead of annotating
350 it completely. The benefit of this mode is that it shows better
351 which forms were instrumented by the compiler.
352 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
353 by printer control variables. (reported by Dan Corkill)
354 * bug fix: system leaked memory when delayed output was performed by
355 the OS in smaller chunks then expected. (thanks to David Smith)
356 * bug fix: system leaked memory when file streams were not closed
358 * bug fix: large objects written to slow streams that were modified
359 after the write could end up with the modified state written to
360 the underlying file descriptor.
361 * bug fix: multiple threads operating in parallel on the same stream
362 could cause buffer-overflows.
363 * bug fix: source location information is stored correctly
364 (broken since 1.0.6). This bug would generally show up as the
365 Slime debugger highlighting the wrong form when the sldb-show-source
367 * bug fix: Branch forms are again annotated as branches in the sb-cover
369 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
370 on x86-64. (reported by Gregory Vanuxem)
371 * bug fix: EQUALP could return wrong results for structures with raw
372 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
373 word). (reported by Vjacheslav Fyodorov)
374 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
377 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
378 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
379 atomic compare-and-swap operations on threaded platforms.
380 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
381 allows assigning a global minimum value to optimization qualities
382 (overriding proclamations and declarations).
383 * enhancement: closed over variables can be stack-allocated on x86
385 * performance bug fix: GETHASH and (SETF GETHASH) are once again
387 * optimization: slot definition lookup is now O(1). This speeds up
388 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
389 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
390 * optimization: ASSOC and MEMBER can now be open-coded for all
391 combinations of keyword arguments when second argument is constant
392 and SPEED >= SPACE. In other cases a specialized version is
394 * bug fix: using obsoleted structure instances with TYPEP and
395 generic functions now signals a sensible error.
396 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
397 (reported by Kristoffer Kvello)
398 * bug fix: backtrace construction is now more careful when making
399 lisp-objects from pointers on the stack, to avoid creating bogus
400 objects that can be seen by the GC.
401 * bug fix: defaulting of values in contexts expecting more than 7
402 variables now works on x86-64. (reported by Christopher Laux)
403 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
405 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
406 as the property-list of a symbol.
407 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
408 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
409 situations requested, are once again file-compileable. (reported
412 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
413 * MOP improvement: support for user-defined subclasses of
414 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
415 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
416 the expansion of DEFMETHOD, is responsible for generating a form
417 which creates a list of specializers when evaluated. Additional
418 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
419 debugging and introspective support.
420 * minor incompatible change: the (unsupported) spinlock interface
421 has changed: free spinlock now has the value NIL, and a held spinlock
422 has the owning thread as its value.
423 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
424 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
426 * enhancement: name of a socket-stream is now "a socket" instead of
428 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
429 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
430 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
431 for generic functions. (Thanks to Tobias C. Rittweiler)
432 * optimization: bignum printing speed has been improved by 20-40%
433 (depending on the bignum size.)
434 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
436 * bug fix: the cache used by the CLOS to store precomputed effective
437 methods, slot offsets, and constant return values is now thread and
439 * bug fix: generic function dispatch function updating is now thread
440 and interrupt safe (in the sense that the known issues have been
442 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
443 * bug fix: interning EQL-specializers is now thread and interrupt safe.
444 * bug fix: asdf systems with dependencies to the SB-POSIX or
445 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
446 * bug fix: interrupt safety of applicable method computation has been
449 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
450 * new contrib: sb-cover, an experimental code coverage tool, is included
452 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
453 significantly faster.
454 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
455 produce more efficient code on x86-64 if the compiler can determine
456 that (+ INDEX <constant>) does not require a bounds check and FOO
457 has an element type at least 8 bits wide.
458 * enhancement: a new, experimental synchronous timeout facility is
459 provided. Refer to SB-SYS:WITH-DEADLINE for details.
460 * enhancement: when a symbol name conflict error arises, the
461 conflicting symbols are always printed with a package prefix.
462 (thanks to Kevin Reid)
463 * enhancement: stepping is now once again supported on the SPARC and
464 MIPS platforms. (It is also now more likely to work on CheneyGC
466 * enhancement: sb-sprof can now also track and report accurate call
468 * bug fixes: the treatment of non-standard subclasses of
469 SB-MOP:SPECIALIZER is more correct.
470 * incompatible change: PURIFY no longer copies the data from the
471 dynamic space into the static and read-only spaces on platforms
472 that use the generational garbage collector
473 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
475 * bug fix: GC race condition occasionally resulting in crashes with
476 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
478 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
480 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
481 a STYLE-WARNING to be signalled (regression from 1.0.4.)
482 * bug fix: an asynchronous interrupt could previously leave the
483 system running with GC inhibited.
484 * bug fix: a DECLARE form evaluated at top-level now causes an error
485 rather than silently (or verbosely) returning NIL.
486 * bug fix: trying to dissassemble functions compiled at high DEBUG could
487 result in a error being signalled due to source form lookup errors.
488 (reported by Peter Graves)
490 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
491 * incompatible change: removed writer methods for host-ent-name,
492 host-ent-addresses -- changing the values did not update the DNS
494 * minor incompatible change: changed experimental JOIN-THREAD interface
495 * documentation: the manual now lists reader and writer methods
496 in class slot documentation sections. (thanks to Richard M Kreuter)
497 * documentation: unwinding from asyncronous events has been
498 documented as unsafe.
499 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
500 in multithreaded application code.
501 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
502 platforms. (thanks to James Anderson for the optimization hint)
503 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
505 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
506 variants no longer cons.
507 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
508 their NOT- variants no longer cons.
509 * optimization: EQUAL hash tables no longer use SXHASH for objects
510 of all data types, but instead use an EQL hash for types for which
511 EQUAL is the same as EQL.
512 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
513 are significantly faster.
514 * optimization: new STRING-OUTPUT-STREAM implementation is much
515 faster for large outputs and conses 30% less on average.
516 * enhancement: XREF information is now collected for references made
517 to global variables using SYMBOL-VALUE and a constant argument.
518 * enhancement: SIGINT now causes a specific condition
519 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
520 * bug fix: FILE-POSITION used to signal an error in some cases where
521 ANSI requires it to return NIL.
522 * bug fix: ADJUST-ARRAY is now interrupt-safe.
523 * bug fix: adding and removing fd-handlers is now interrupt-safe.
524 * bug fix: inlined calls to C now ensure 16byte stack alignment on
526 * bug fix: bad type declaration in the CLOS implementation has
527 been fixed. (reported by James Anderson)
528 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
529 (thanks to Lutz Euler)
530 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
531 required by the CLHS. (reported by Samium Gromoff)
532 * bug fix: dead unbound variable references now signal an error.
533 (reported by Marco Monteiro)
534 * bug fix: / with an unused value was being deleted in safe code.
535 (thanks to Marco Monteiro and Kevin Reid)
536 * bug fix: number of characters that can be written onto a single
537 line in a file is unlimited.
538 * bug fix: some GC deadlocks caused by asynchronous interrupts have
539 been fixed by inhibiting interrupts for when GC is disabled.
540 * bug fix: some interrupt safety issues with GETHASH, (SETF
541 GETHASH), CLRHASH and REMHASH have been fixed.
542 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
543 specifier no longer causes infinite recursion.
544 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
545 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
546 * bug fix: an &environment argument with the correct variable information
547 is passed to macros that are expanded during byte compilation.
548 (reported by Samium Gromoff)
549 * bug fix: a logic error was causing the hash function for bignums to
550 have excessive amounts of collisions. (reported by Faré Rideau)
551 * bug fix: modifying the contents of an array could change the return
552 value of SXHASH on that array, which is only allowed for strings
553 and bit vectors. (bug introduced in 0.9.16)
554 * improvement: the x86-64/darwin port now passes all tests (except
555 for the debugger tests) but should still be considered
556 experimental until this is fixed.
557 * improvement: a style-warning is signaled for CASE (etc) clauses with
558 duplicate keys. (thanks to Kevin Reid)
559 * improvement: macos/x86 and macos/x86-64 now use mach exception
560 handlers for dealing with illegal instructions (for trapping and
561 error handling) and memory protection violations (for GC).
563 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
564 * new platform: experimental support for x86-64/darwin (MacOS).
565 * incompatible change: the thread-safe (on most platforms) getaddrinfo
566 and getnameinfo sockets functions are used instead of gethostbyaddr
567 and gethostbyname, on platforms where the newer functions are available.
568 As a result, the ALIASES field of HOST-ENT will always be NIL on these
570 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
571 don't print the startup banner, but behave as if --noinform was passed
572 as a command line argument. (thanks to Kevin Reid)
573 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
574 * optimization: code using alien values with undeclared types is much faster.
575 * optimization: the compiler is now able to open code SEARCH in more cases.
576 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
577 * bug fix: using standardized COMMON-LISP special variables as loop
578 variables no longer signals bogus package lock violations. (reported
580 * bug fix: declaring local loop variables to be of a range-limited type
581 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
582 (reported by Andras Simon)
583 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
584 bugs remain on x86-64.)
585 * bug fix: the #= and ## reader macros now interact reasonably with
586 funcallable instances.
587 * bug fix: type-checks for function arguments were compiled using the
588 compiler policy settings of the wrong lexical environment (in 1.0.2
590 * bug fix: SHADOW accepts characters as string designators, as required
591 by the spec (thanks to Eric Marsden)
592 * bug fix: fixed GC safety issues when foreign functions are called with
593 non-base strings as arguments
594 * bug fix: more consistent error output for fatal-compiler-warnings, like
596 * bug fix: functions with proper names were showing up as NIL in
597 backtraces (reported by Edi Weitz, regression in 1.0.2)
599 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
600 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
601 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
602 produce more efficient code on the x86 if the compiler can determine
603 that (+ INDEX <constant>) does not require a bounds check and FOO
604 has an element type at least 8 bits wide.
605 * bug fix: references to undefined variables are handled the same way
606 in toplevel forms as in the normal compiler
607 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
609 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
610 (thanks to Jon Buller)
611 * improvement: faster compilation times for complex functions
612 * improvement: added readlink support to SB-POSIX (thanks to Richard
615 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
616 * improvement: experimental support for mach exception handling on
617 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
619 * improvement: support for GBK external format.
620 (thanks to Chun Tian (binghe))
621 * improvement: the debugger now displays variables that have been closed
622 over, in code compiled with (DEBUG 2) or higher
623 * improvement: support for executable cores on NetBSD (thanks to
625 * new feature: added a RESTART-FRAME debugger command
626 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
627 be used to provide an implementation for FILE-POSITION on Gray streams
628 (thanks to Eric Marsden)
629 * new feature: add syslog, openlog and closelog support to SB-POSIX
630 (thanks to Richard Kreuter)
631 * optimization: the function call overhead in code compiled with
632 a high DEBUG optimization setting is significantly
633 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
634 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
635 (thanks to Tony Martinez)
636 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
637 evaluated once (reported by Kevin Reid)
638 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
639 works again (thanks to Kilian Sprotte)
640 * bug fix: an error is signaled for tagbodies with duplicate tags
641 (thanks to Stephen Wilson)
642 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
643 * bug fix: Win32 port can now handle foreign code unwinding Lisp
644 stack frames from alien callbacks.
645 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
646 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
647 65k pages (thanks to David Woodhouse)
648 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
650 changes in sbcl-1.0.1 relative to sbcl-1.0:
651 * new platform: FreeBSD/x86-64, including support for threading.
652 * new feature: the compiler stores cross-referencing information
653 about function calls (who-calls), macroexpansion (who-macroexpands)
654 and special variables (who-binds, who-sets, who-references) for code
655 compiled with (< SPACE 3). This information is available through the
656 sb-introspect contrib.
657 * new feature: users may subclass SEQUENCE, and have instances of
658 these classes interoperate with standard Common Lisp functions if
659 a number of methods are defined. (This feature is experimental
660 and the interface subject to change based on feedback from SBCL
661 users and the general community)
662 * improvement: sb-sprof traces call stacks to an arbitrary depth on
663 x86/x86-64, rather than the previous fixed depth of 8
664 * improvement: another pthread back-end of mutex "pthread-futex".
665 * bug fix: non-ascii command-line arguments are processed correctly
666 (thanks to Yaroslav Kavenchuk)
667 * bug fix: non-required arguments were not passed correctly when a method
668 defined using DEFMETHOD was called from a mop-generated method using
669 CALL-NEXT-METHOD (reported by Pascal Costanza)
670 * bug fix: recursion is now permitted in accessors through
671 SLOT-UNBOUND. (reported by Pascal Costanza)
672 * bug fix: an error was signaled at startup if the HOME environment
673 variable was defined, but had an empty value (reported by Peter Van Eynde)
674 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
675 signal a PROGRAM-ERROR, not a BUG.
676 * bug fix: SB-POSIX:READDIR works when built with large file support.
677 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
679 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
680 for better automatic indentation support. (thanks to Matt Pillsbury)
681 * optimization: loading generic functions no longer takes O(n^2) time,
682 proportional to the amount of methods in the generic function
683 (reported by Todd Sabin and Jeremy Brown)
684 * optimization: the FIND and POSITION family of sequence functions
685 are significantly faster on arrays whose element types have been
687 * improvements to the Windows port:
688 ** Intermittent heap corruption problems have been fixed. (thanks
689 to Alastair Bridgewater)
690 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
692 ** Lisp is able to unwind foreign exception frames from alien
693 callbacks. (thanks to Alastair Bridgewater)
695 changes in sbcl-1.0 relative to sbcl-0.9.18:
696 * improvement: experimental support for threading on FreeBSD/x86.
697 (thanks to NIIMI Satoshi)
698 * improvement: runtime option --dynamic-space-size can be used
699 to set the size of the dynamic space reserved on startup.
700 * improvement: floating point modes in effect are now saved in
701 core, and restored on startup.
702 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
703 startup, not time since first call to GET-INTERNAL-REAL-TIME.
704 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
705 threads are not running after *SAVE-HOOKS* have run.
706 * improvement: writes to CLOS instance slots are type-checked in code
707 compiled with (SAFETY 3)
708 * improvement: floating-point exception handling on FreeBSD (thanks to
710 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
711 (thanks to Zach Beane)
712 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
714 * improvement: added support for the Shift-JIS external format.
715 (contributed by NIIMI Satoshi)
716 * improvement: callbacks are supported on Linux/PPC. (thanks to
718 * bug fix: compiler bug triggered by a (non-standard) VALUES
719 declaration in a LET* was fixed. (reported by Kaersten Poeck)
720 * bug fix: file compiler no longer confuses validated and already
721 dumped structurres. (reported by Kaersten Poeck)
722 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
723 fill-pointer signals a type-error as required. (thanks to
725 * bug fix: disassemly of funcallable instances works.
726 * bug fix: single stepping on PPC.
727 * bug fix: fix thread-safety problems in the type system (generally
728 manifesting as nonsensical errors like "STRING is a bad type specifier
729 for sequences" or "The value 1 is not of type FIXNUM")
730 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
731 (reported by Josip Gracin)
732 * bug fix: an error is signaled for attempts to displace arrays with
733 incompatible element types (thanks to Mario Mommer)
734 * bug fix: more correct handling of wide characters in debug info
735 (bug reported by Attila Lendvai and fixed by Juho Snellman)
736 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
737 and don't cause extra consing
738 * optimization: MAP and MAP-INTO are significantly faster on vectors
739 whose elements types have been declared.
740 * Improvements to SB-SPROF:
741 ** Support for allocation profiling
742 ** Reduced profiling overhead, especially for long profiling runs
743 * Improvements to the Windows port:
744 ** floating point exceptions are now reported correctly.
745 ** stack exhaustion detection works partially.
746 ** more accurate GET-INTERNAL-REAL-TIME.
747 ** better RUN-PROGRAM behaviour in case of errors in the child
749 ** PROBE-FILE now simplifies pathnames correctly.
750 ** DIRECTORY now works correctly with :WILD-INFERIORS.
751 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
752 1st 1970 (although time timezone data will be unreliable).
753 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
755 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
756 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
757 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
759 * bug fix: two potential GC deadlocks affecting threaded builds.
760 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
761 returns T (reported by Anton Kazennikov)
762 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
763 with non-variable places
764 * bug fix: remove a race condition in the setting of
765 funcallable-instance functions, this should make threaded CLOS
766 code more stable against memory faults.
767 * bug fix: corruption of specials when unbinding is interrupted by an
768 asynchronous unwind (reported by Hannu Koivisto)
769 * improvement: the debugger will now also display local variables that
770 are only used once, for code compiled with a DEBUG optimization quality
773 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
774 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
775 * incompatible change: External-format support for FFI calls. The
776 SB-ALIEN:C-STRING no longer implies an ASCII
777 external-format. Instead the string is subject to external-format
778 conversion. Additionally return values of type SB-ALIEN:C-STRING
779 are by default of type (SIMPLE-ARRAY CHARACTER), not
780 SIMPLE-BASE-STRING. For an alien type that behaves like the old
781 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
782 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
783 * incompatible change: SB-EXT package no longer contains the
784 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
785 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
787 * incompatible change: the single-stepper is no longer available
788 on Alpha, Mips and Sparc platforms.
789 * minor incompatible change: the direct superclasses of
790 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
791 not (STANDARD-OBJECT FUNCTION). This makes the
792 class-precedence-lists of GENERIC-FUNCTION and
793 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
795 * minor incompatible change: the default stream external format on
796 non-unicode SBCL is no longer detected from the locale, but is always
798 * new feature: Add a version of evaluator that uses an interpreter instead
799 of the compiler. EVAL still uses the compiler by default, to switch it
800 to use the interpreter, set the value of the variable
801 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
802 * minor incompatible change: the single-stepper REPL has been merged
803 with the normal debugger (see the "Stepping" heading of the debugger help
804 for more details). The debugger command STEP will no longer switch
805 to the single-stepper REPL.
806 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
807 for a type now works.
808 * bug fix: loading of default sysinit file works. (thanks to Leonid
810 * bug fix: better detection of circularities in the file-compiler.
811 (reported by Marco Monteiro)
812 * bug fix: the CL pathname functions now work with files that have
813 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
814 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
815 systems with Unix98 pty semantics.
816 * bug fix: ASDF-INSTALL will now work with bsd tar.
817 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
819 * bug fix: timers expiring in dead threads no longer cause a
820 type-error (reported by Paul "Nonny Mouse").
821 * bug fix: thanks to more lightweight single-stepper instrumentation,
822 code compiled with (DEBUG 3) will compile and execute significantly faster,
823 and will have more accurate type-inferencing than before
824 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
825 whose bindings are modified
826 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
827 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
828 ** replace dummy implementations of CL:MACHINE-INSTANCE and
829 CL:SOFTWARE-VERSION with proper version
831 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
832 * feature: implemented the READER-METHOD-CLASS and
833 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
834 as specified by AMOP.
835 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
837 * optimization: faster LOGCOUNT implementation on x86 and x86-64
838 (thanks to Lutz Euler)
839 * optimization: hashing of general arrays and vectors has been
840 improved. (reported by Any Fingerhut)
841 * enhancement: SB-INTROSPECT is now able to find definitions of
842 profiled functions. (thanks to Troels Henriksen)
843 * enhancement: compiler-macro expansion applies now to FUNCALL forms
844 as well, allowing compiler-macros for SETF-functions to expand.
845 * enhancement: step-instrumentation no longer wraps known
846 single-value functions with multiple-value context, allowing
847 better type inference.
848 * fixed bug #337: use of MAKE-METHOD in method combination now works
849 even in the presence of user-defined method classes. (reported by
850 Bruno Haible and Pascal Costanza)
851 * fixed bug #339(c): if there are applicable methods not part of any
852 long-form method-combination group, call INVALID-METHOD-ERROR.
853 (reported by Bruno Haible)
854 * fixed bug #361: the :FUNCTION initarg in the protocol for
855 initialization of methods can now be used to override
856 internally-produced optimized functions. (reported by Bruno
858 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
859 system-provided lambda expression no longer cause warnings about
860 unbound #:|pv-table| symbols.
861 * bug fix: improved the handling of type declarations and the
862 detection of violations for keyword arguments with non-constant
864 * bug fix: potentially erronous calls to PATHNAME and
865 MERGE-PATHNAMES were being flushed in some cases.
866 (reported by Richard Kreuter)
867 * bug fix: compiled calls to TYPEP were mishandling obsolete
868 instances. (reported by James Bielman and Attila Lendvai)
869 * bug fix: format strings with ~> without matching ~< no longer
870 trigger an AVER, but signal an understandable error instead.
871 (reported by Antonio Martinez)
872 * bug fix: specifying an output-file without a pathname-type for
873 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
874 of input-file instead of "fasl". (reported by Robert Dodier)
875 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
876 (reported by James Y Knight).
877 * bug fix: compiler-macros-function did not consider the environment
878 argument for shadowing by local functions.
879 * bug fix: compiler-macros expansion was inhibited by local INLINE
881 * bug fix: inline expansions of known functions were subject to
882 step-instrumentation in high debug policies, leading to problems
884 * bug fix: compiler failed to differentiate between different CONS
886 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
887 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
888 * bug fix: #368: incorrect use of expressed vs. upgraded array
890 * bug fix: #306a: more precise unions of array types.
891 * thread-safety improvements:
892 ** CONDITION-WAIT could return early on Linux, if the thread was
893 interrupted and subsequently continued with SIGCONT.
894 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
896 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
897 * added support for the ucs-2 external format. (contributed by Ivan
899 * minor incompatible change: pretty printing of objects of type
900 (cons symbol) is, in the default pprint-dispatch-table, now
901 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
903 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
904 called later in a class's lifetime, possibly as late as when the
905 first instance of the class is created. Previously,
906 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
907 class became finalizeable.
908 * fixed bug: FILE-POSITION sometimes returned inconsistent results
909 for multibyte external-format streams. (thanks to "vbzoli")
910 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
911 with :ALLOCATION :CLASS inherited from superclasses of the
913 * fixed bug: anonymous classes can now be created using the :NAME
914 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
915 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
916 * fixed bug: core-files saved with :EXECUTABLE T can again be
917 executed when SBCL_HOME isn't set. (reported by James Knight)
918 * fixed bug: toplevel LOCALLY forms with declarations could
919 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
920 * fixed bug: printing from several different threads using different
921 values of *print-case* could cause invalid output, due to
922 some internal special variables of the printer not being bound
923 thread-locally (reported by Max Mikhanosha)
924 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
925 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
926 (reported by Pascal Costanza)
927 * fixed bug: FINALIZE-INHERITANCE is called from
928 REINITIALIZE-INSTANCE on classes when the class has previously
929 been finalized, as required by AMOP.
930 * minor code generation optimizations:
931 ** better register allocation in CLOS dispatching functions
932 ** overflow detection when coercing signed bytes to fixnums on x86-64
933 ** is now implemented with one IMUL instruction instead of three shifts
934 ** more efficient bit-vector access on x86 and x86-64
935 ** more efficient access to raw structure slots on x86 and x86-64
936 * fixed some bugs revealed by Paul Dietz' test suite:
937 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
940 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
941 * feature: thread support on Solaris/x86, and experimental thread support
943 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
945 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
946 default method for SLOT-UNBOUND.
947 * minor incompatible change: prevent the user from specializing the
948 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
949 somewhat counter to the intent of the protocol, I (CSR) think, and
950 additionally it just doesn't work in SBCL as currently
951 implemented, thanks to optimizations (that are always valid for
952 the other three SLOT-VALUEish functions, but not for the setter).
953 * minor incompatible change: the :SB-LDB feature is now enabled by
954 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
955 the low-level debugger.
956 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
957 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
958 down to an SB-SYS:FD-STREAM, allowing direct communication in
959 more cases. Temporary files and pipes are still used as a fallback
961 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
962 on the non-reentrant C functions localtime(3) and gmtime(3).
963 * bug fix: global optimization policy was not visible in LOCALLY and
965 * bug fix: class objects can be used as specializers in methods.
966 (reported by Pascal Costanza)
967 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
968 with a type error. (reported by blitz_ on #lisp)
969 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
970 failed with a type error. (reported by Pascal Bourguignon)
971 * bug fix: merging pathnames against defaults with :DIRECTORY
972 starting with '(:RELATIVE :BACK) should preserve the :BACK.
973 (reported by James Y Knight)
974 * bug fix: saving large (>2GB) cores on x86-64 now works
975 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
976 constant index and a (simple-array (signed-byte 32)) array
977 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
978 error (patch by Robert J. Macomber)
979 * feature: TIME now displays the amount of run-time spent in GC
980 * bug fix: The debugger now does a better job of respecting
981 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
982 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
983 workaround for bug 403.)
984 * bug fix: floating point exception handling now works on FreeBSD
985 (thanks to NIIMI Satoshi)
986 * fixed some bugs revealed by Paul Dietz' test suite:
987 ** MISC.641: LET-conversion were not supposed to work in late
989 * improvements to the Win32/x86 port:
990 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
991 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
993 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
994 * new feature: source path information is generated for macro-expansion
995 errors for use in IDE's like Slime (thanks to Helmut Eller)
996 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
997 * bug fix: compiler does not loop forever on an invalid type in
999 * improvement: compilation of most CLOS applications is significantly
1001 * optimization: added a limited bytecode compiler for simple toplevel
1002 forms, speeding up compilation and FASL loading
1003 * bug fix: the statistical profiler now properly distinguishes anonymous
1006 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1007 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1008 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1009 into subclasses of the finalized class. Now user calls to
1010 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1011 system return before any subclasses are finalized.
1012 * minor incompatible change: The reader no longer ignores errors
1013 regarding non-existent packages in #+ and #- feature tests.
1014 * new feature: command line options --no-sysinit, --no-userinit to
1015 inhibit loading the corresponding init files
1016 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1017 for sending data through UDP sockets (thanks to François-René Rideau)
1018 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1019 error is signalled instead (thanks to François-René Rideau)
1020 * minor incompatible change: Error signalling behaviour of lexical
1021 operations violating package locks has changed slightly. Refer to
1022 documentation on package locks for details.
1023 * enhancement: EVAL can process IF-expressions without resorting to the
1025 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1026 constant-folded properly due to missing out-of-line functions.
1027 (reported by Mika Pihlajamäki)
1028 * bug fix: LISTEN sometimes returned T even in cases where no data was
1029 immediately available from the stream
1030 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1031 were reversed. (reported by Levente Mészáros)
1032 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1033 when given a vector displaced to an adjustable array. (reported
1035 * bug fix: derivation of float boundaries from numbers outside the
1036 appropriate float range (reported by John Wiseman)
1037 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1039 * bug fix: precomputing cache entries for generic functions with
1040 some subclasses of specializers as yet invalid does not attempt to
1041 fill a cache line with a negative offset. (reported by Levente
1043 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1044 allow condition class objects as arguments to DOCUMENTATION and
1045 (SETF DOCUMENTATION); only find and set documentation for
1046 structure names for the STRUCTURE doc-type. (suggested by Gary
1048 * improvements to the Win32/x86 port:
1049 ** (user-homedir-pathname) and default initialization file
1050 locations now know about the user's "Documents and Settings"
1051 directory (thanks to Yaroslav Kavenchuk)
1052 ** run-program is implemented (thanks to Mike Thomas)
1053 ** sockets support (thanks to Timothy Ritchey)
1054 ** better backtrace support (thanks to Alastair Bridgewater)
1055 ** sb-grovel supported
1056 ** asdf-install and sb-posix work somewhat
1057 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1058 * improvements to the Solaris/x86 port:
1059 ** works on Solaris 11/Solaris Express
1060 ** floating-point exception handling support
1061 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1062 * fixed some bugs revealed by Paul Dietz' test suite:
1063 ** REMOVE-METHOD returns its generic function argument even when
1064 the method is not one of the generic functions' methods.
1065 ** objects declared as MEMBER types can be admissible arguments to
1066 structure accessors.
1067 ** printing characters should simply be printed by the FORMAT ~:C
1069 ** compiler failure when compiling functions with hairy constant
1070 defaults for optional parameters.
1071 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1072 function, which is already optimized.
1074 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1075 * new platform: experimental support for SBCL x86/Darwin, including
1076 MacOS X 10.4.5 on Intel.
1077 * new feature: Unicode character names are now known to the system
1078 (through CHAR-NAME and NAME-CHAR).
1079 * new feature: the filesystem location of SBCL's core file is
1080 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1081 * minor incompatible change: the contrib modules SB-POSIX and
1082 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1083 this change is to make it easier to distribute
1084 location-independent binaries.
1085 * bug fix: as implied by AMOP, standardized classes no longer have
1086 slots named by external symbols of public packages. (reported by
1088 * optimization: calling functions via constant symbols -- as in
1089 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1090 via the function object as in (FUNCALL #'FOO).
1091 * optimization: on x86 and x86-64, the calling convention for the
1092 case of dealing with an unknown number of values has been altered
1093 to be friendlier to the prediction heuristics implemented,
1094 particularly with reference to CALL and RET pairing. (thanks to
1095 Alastair Bridgewater)
1096 * optimization: on x86, the code for access to thread-local symbol
1097 values and binding/unbinding of thread-local symbols is smaller.
1098 (thanks to Alastair Bridgewater)
1099 * enhancement: CONSTANTP is now able to determine constantness of
1100 more complex forms, including calls to constant-foldable standardized
1101 functions and some special forms beyond QUOTE.
1102 * fixed bug: occasional GC crashes on Solaris/x86
1103 * optimization: x86-64 supports stack allocation of results of simple
1104 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1105 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1106 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1107 (thanks to James Knight)
1108 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1109 by the standard (thanks to Stephen Compall)
1111 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1112 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1113 be used for bundling the runtime and the core file into one
1114 executable binary. This feature is not currently supported on all SBCL
1115 platforms. (thanks to James Bielman and NIIMI Satoshi)
1116 * new feature: a generational or ephemeral garbage collector is now
1117 the default on the PowerPC platform (both Linux and Darwin). The
1118 old Cheney (stop and copy) collector is a build-time option.
1119 (thanks to Cyrus Harmon, after Raymond Toy)
1120 * minor incompatible change: the method by which SBCL finds its
1121 contributed modules has changed; it no longer relies on symbolic
1122 links from an $SBCL_HOME/systems directory, but searches directly
1123 in the subdirectories of $SBCL_HOME.
1124 * enhancement: the dynamic heap size on the Linux/PPC platform is
1125 markedly larger, even using the older Cheney garbage collector.
1126 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1127 certain complicated string types. (reported by Gary King)
1128 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1129 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1130 * fixed bug: DOCUMENTATION on structures no longer signals an
1131 error if no structure type of the right name exists. (reported by
1133 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1134 applicable non-standard primary or :AROUND method on
1135 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1136 no longer cause unbound variable errors. (reported by Kevin Reid)
1137 * fixed bug: metacircle resolution in cases where methods have slots
1138 added before the slots from STANDARD-METHOD. (reported by Jean
1140 * fixed bug: the Power PC assembly code for calling into lisp
1141 sometimes computed its return address wrongly (depending on the
1142 15th bit of the address, fixed at link time). This used to
1143 manifest itself in a segmentation violation while building PCL.
1144 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1145 many others over the years)
1146 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1147 the stack pointer (treated as a signed integer) is negative.
1148 (thanks to Peter van Eynde)
1150 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1151 * new platform: experimental support for the Windows operating
1152 system has been added. (thanks to Alastair Bridgewater)
1153 * fixed several bugs in and robustified the PPC FFI (including
1154 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1155 * bug fix: allow non-simple string symbol names (reported by Paul
1157 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1159 * bug fix: saving a core corrupted callbacks on x86/x86-64
1160 * bug fix: closed a loophole in metacircularity detection and
1161 grounding in the PCL implementation of CLOS.
1162 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1163 (reported by tomppa on #lisp)
1164 * bug fix: clear the direction flag on Lisp -> C transitions, as
1165 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1166 (reported by Andrej Grozin and Hendrik Maryns)
1167 * optimization: major improvements to GC efficiency on GENCGC platforms
1168 * optimization: faster implementation of EQUAL
1169 * optimization: emit more efficient opcodes for some common
1170 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1171 * optimization: several other minor code-generation improvements on x86-64
1172 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1174 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1175 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1176 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1177 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1178 by ANSI as a generic function, and for consistency (SETF
1179 GENERIC-FUNCTION-NAME) is treated likewise.
1180 * fixed bug #233.b: make constraint propagation notice when a variable
1181 value is changed after it is referenced but before it is used
1182 * fixed bug #296: no more arbitrary behaviour from filenames with
1183 odd characters as --load arguments. (reported by Adam Warner)
1184 * fixed bug #390: :CHARACTER-SET pathname components now work as
1185 expected. (reported by Tim Daly Jr)
1186 * fixed bug #391: complicated :TYPE intersections in slot
1187 definitions no longer cause an error in PCL internals.
1188 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1189 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1190 returning the number of octets which would be written to the
1191 file-stream. (thanks to Robert J. Macomber)
1192 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1193 arguments to the internal stream functions. (thanks to David
1195 * bug fix: stack exhaustion is now handled on the x86/SunOS
1196 platform. (thanks to Daisuke Homma)
1197 * bug fix: more accurate ROOM results on GENCGC platforms
1198 * optimization: improved type inference for arithmetic-for
1199 index variables in LOOP
1200 * optimization: faster floating-point SQRT on x86-64
1201 * fixed some bugs revealed by Paul Dietz' test suite:
1202 ** DOCUMENTATION returns NIL instead of "" for method combinations
1203 that don't have a docstring
1205 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1206 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1207 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1208 therefore are not customizeable by user code (as seems to be at
1209 least permitted and maybe required by AMOP). As a consolation,
1210 however, the SBCL implementation of these functions now calls
1211 REINITIALIZE-INSTANCE as specified by AMOP.
1212 * bug fix: slot-definition documentation is propagated as per ANSI
1213 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1214 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1215 argument, as do eslotd initialization methods. (from Pascal
1216 Costanza's "Closer" project)
1217 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1218 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1220 * bug fix: it is now possible to have more than one subclass of
1221 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1222 (reported by Bruno Haible, Pascal Costanza and others)
1223 * bug fix: the dependent update protocol now works for generic
1224 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1225 and Pascal Costanza)
1226 * bug fix: condition-class instances corresponding to
1227 DEFINE-CONDITION forms are now created eagerly. (reported by
1228 Kalle Olavi Niemitalo on comp.lang.lisp)
1229 * bug fix: floating point printing is more accurate in some
1230 circumstances. (thanks to Simon Alexander)
1231 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1232 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1233 * bug fix: callbacks on OS X now preserve stack-alignment.
1234 (thanks to Cyrus Harmon)
1235 * enhancement: the x86-64 disassembler is much better at
1236 disassembling SSE instructions. (thanks to Lutz Euler)
1237 * enhancement: CHECK-TYPE on a variable now allows type-inference
1238 (thanks to Vincent Arkesteijn)
1239 * optimization: improved performance of EUC-JP external format.
1240 (thanks to NIIMI Satoshi)
1241 * optimization: performance improvements to IO on file streams of
1242 :ELEMENT-TYPE CHARACTER
1243 * optimization: much faster memory allocation on x86-64
1244 * optimization: faster garbage collections (latency and throughput)
1246 * optimization: faster bignum subtraction and division on x86 and x86-64
1247 (thanks to Lutz Euler)
1248 * optimization: more accurate type-inference for loops that use a
1249 floating point index variable or a negative step.
1251 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1252 * bug fix: add a workaround to SBCL looping infinitely at startup on
1253 Linux kernels with apparently buggy implementations of personality().
1254 (thanks to Svein Ove Aas)
1255 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1256 (thanks to David Lichteblau)
1257 * bug fix: local bindings shadow global symbol macros for
1258 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1259 * optimization: non-open coded uses of numeric comparison operators
1260 (e.g. >) no longer cons when called with more than one parameter
1261 on platforms supporting dynamic-extent allocation.
1262 * enhancement: saving cores with foreign code loaded is now supported
1263 on MIPS/Linux in addition to the previously supported platforms.
1264 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1265 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1266 * bug fix: division by zero in sb-sprof when no samples were collected
1267 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1268 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1269 (thanks to Alexander Kjeldaas)
1270 * new feature: ignoring the return values of destructive functions
1271 like NREVERSE signals a compile-time style-warning.
1272 (thanks to Kevin Reid)
1273 * new feature: the alignment of alien structure fields can be
1274 explicitly specified. (thanks to Cyrus Harmon)
1275 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1276 and dump core on SIGQUIT
1278 ** incompatible change: threads do not inherit values of specials
1279 from their parents (see manual)
1280 ** bug fix: threads stacks belonging to dead threads are freed by the
1281 next exiting thread, no need to gc to collect thread stacks anymore
1282 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1283 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1284 ** bug fix: dynamic variable and thread start related gc lossage
1285 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1287 * fixed some bugs revealed by Paul Dietz' test suite:
1288 ** SUBTYPEP is slightly more accurate on heinously complicated
1289 CONS types where some of the members have uncertain (in the
1290 NIL, NIL sense) type relationships to each other.
1292 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1293 no longer purified unless :PURIFY T is explicitly specified.
1294 ** Non-purified cores are significantly smaller than before
1296 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1297 * new feature: timers based on Zach Beane's excellent timer package
1298 * added support for the following external formats: koi8-u,
1299 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1300 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1301 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1302 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1303 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1304 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1305 * incompatible change: a threaded SBCL will no longer revert to
1306 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1307 * bug fix: interrupts are disabled until startup is complete; no
1308 more sigsegvs when receiving a signal to soon
1309 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1311 * bug fix: add a workaround for the memory randomization features in
1312 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1313 workaround will only be in effect on systems with the proc filesystem
1315 * bug fix: printing objects of type HASH-TABLE signals a
1316 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1318 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1319 have been running for over 50 days. (reported by Gilbert Baumann)
1320 * bug fix: the logic for getting names of functions gets less
1321 confused when confronded with alternate-metaclass
1322 funcallable-instances. (reported by Cyrus Harmon)
1323 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1324 consistent, even on internal alternate-metaclass objects.
1325 * bug fix: flush closure information collected by physical
1326 environment analysis prepass before the main pass. (bug reported
1328 * bug fix: compiler pack phase does not modify a hashtable, which is
1329 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1330 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1331 may be expanded. (reported by Alexander <malishev> on c.l.l)
1332 * bug fix: no more style-warnings when compiling code using LAST
1333 under high-SPEED low-SPACE settings. (reported by David Wallin
1335 * fixed some incompatibilities between SBCL's MOP and the MOP
1337 ** the METAOBJECT class is now implemented;
1338 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1339 STANDARD-OBJECT, as required;
1340 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1341 now compatible (as required by VALIDATE-SUPERCLASS); there
1342 remains a consistency requirement over the presence or absence
1343 of the FUNCTION class in the superclass list of finalized
1344 classes; see the manual for more details;
1345 ** the :AROUND method for COMPUTE-SLOTS on
1346 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1347 requested slot ordering.
1349 ** bug fix: parent thread now can be gc'ed even with a live
1351 ** bug fix: binding a special with PROGV to no value is not
1352 the same as that symbol not having been bound (thanks to
1354 ** bug fix: binding specials is thread safe (thanks to
1356 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1357 and SB-SPROF do not die with 'no handler for signal XX in
1358 interrupt_handle_now(..)' anymore
1359 ** bug fix: WITH-TIMEOUT works with multiple running threads
1360 * fixed some bugs revealed by Paul Dietz' test suite:
1361 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1362 the :method-class keyword argument.
1364 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1365 * new port: the Solaris operating system on x86 processors is now
1366 mostly supported, though some rough edges in the environment
1367 remain. (thanks to Daisuke Homma)
1368 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1369 space for its heap. (thanks to Thiemo Seufer)
1370 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1371 closures. (thanks to Thiemo Seufer)
1372 * minor incompatible change: eof selects abort in the debugger.
1373 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1374 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1376 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1377 deprecated, and will go away in a future revision of SBCL.
1378 * minor incompatible change: GC-ON and GC-OFF are no longer
1379 implemented with a counter, it does not matter how many times gc
1380 is switched on or off
1381 * bug fix: discriminating functions for generic function classes
1382 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1383 make invalid assumptions about method precedence order. (reported
1385 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1386 platforms. (thanks to Thiemo Seufer)
1387 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1388 * bug fix: correct pathname printing with printer escaping is on.
1389 (thanks to Kevin Reid)
1390 * bug fix: complex VOP definitions in "user-space" no longer trigger
1391 package locks. (reported by Zach Beane)
1392 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1393 for funcallable-instance functions, and is no different from
1395 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1396 characters in the current readtable. (reported by Nicholas Neuss)
1397 * bug fix: SUBTYPEP on various CONS types returns more a more
1398 accurate acknowledgment of its certainty.
1399 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1400 lists in some cases. This partially fixes bug 384.
1401 * flush all standard streams before prompting in the REPL and the
1403 * bug fix: signal handling and triggering gc do not conflict
1404 directly anymore, in particular a high frequency sb-sprof does
1405 not prevent gc from running
1406 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1407 approximation for timezone and DST information between the
1408 universal time epoch and the smallest negative 32-bit time_t.
1409 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1410 year 1899 so long as the encoded time is non-negative.
1411 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1412 the correct answer for non-constant rotate amounts when performing
1413 an inline 32-bit rotation.
1415 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1416 there is only one thread in the session
1417 ** bug fix: memory leak for streams created in one thread and
1418 written to in another
1419 ** bug fix: lockup when compiled with gcc4
1420 ** bug fix: race that allows the gc to be triggered when gc is
1422 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1423 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1425 ** bug fix: finished threads are gc'ed properly
1426 * fixed some bugs revealed by Paul Dietz' test suite:
1427 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1428 has arguments with a different set of applicable methods from
1429 the orignal arguments.
1430 ** The type error thrown by MAP now has a correct expected-type
1432 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1433 name a compiled function.
1434 ** Three MISC tests where a large bignum was improperly coerced to
1435 a float that couldn't represent that bignum during type
1436 derivation were fixed.
1437 ** SUBTYPEP can now handle types involving the negation of a
1438 list-form FUNCTION type.
1439 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1440 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1441 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1443 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1444 * New feature: Experimental support for bivalent streams: streams
1445 opened with :element-type :default now allow character and binary
1446 (unsigned-byte 8) I/O
1447 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1448 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1450 * bug fix: no more highly sporadic "couldn't check whether ~S is
1451 readable" when reading a stream and an interrupt hits in the middle
1452 of a select system call
1453 * compiler better recognizes complex arrays (reported by Eduardo
1455 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1456 for variables on SBCL built with linkage-tables. (reported by Luis
1458 * various error reporting improvements.
1459 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1460 (Thanks to Thiemo Seufer)
1461 * bug fix: nested reader invokations work correctly
1462 * bug fix: it is possible to have simultaneous references to foreign
1463 code and foreign data with the same name.
1465 ** added x86-64 support
1466 ** incompatible change: the threading api now works with thread
1467 objects instead of thread ids
1468 ** bug fix: threads are protected from signals and interruption when
1469 starting up or going down
1470 ** bug fix: a race where an exiting thread could lose its stack to gc
1471 ** bug fix: don't halt on infinite error in threads if possible
1472 ** fixed numerous gc deadlocks introduced in the pthread merge
1473 ** bug fix: fixed thread safety issues in read and print
1474 ** bug fix: debugger doesn't hang on session lock if interrupted at
1475 an inappropriate moment
1476 ** bug fix: run-program is now thread safe(r)
1477 ** bug fix: inner with-recursive-lock no longer releases the mutex
1478 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1479 * fixed some bugs revealed by Paul Dietz' test suite:
1480 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1481 EXPECTED-TYPE slots filled.
1482 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1484 ** MAKE-CONDITION accepts classes as type-designators.
1485 ** COMPILE may never return NIL.
1486 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1487 range before calling Unix time functions
1489 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1490 * numerous signal handling fixes to increase stability
1491 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1492 * minor incompatible change: we now correctly canonize default
1493 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1494 as per the MOP, rather than the historical (INITARG INITFUNCTION
1495 INITFORM). (reported by Bruno Haible)
1496 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1497 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1498 for more information.
1499 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1500 pathname is a directory pathname.
1501 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1502 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1504 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1505 :METACLASS argument in addition to a class name. (reported by
1506 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1507 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1508 (reported by Bruno Haible; patch from Gerd Moellmann)
1509 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1511 * bug fix: more cleanups to the floating point exception handling on
1512 x86-64 (thanks to James Knight)
1513 * bug fix: the compiler does not try to constant fold calls of
1514 COERCE and BIT-* functions when they return freshly constructed
1515 objects. (reported by Nikodemus Siivola and Paul Dietz)
1516 * optimization: improved the MIPS versions of generic (in the
1517 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1518 * optimization: direct conversion of (unsigned-byte 32) to floats on
1519 the PowerPC platform.
1520 * optimization: structure instances with raw slots now use less
1521 memory, and probably show better memory locality. (thanks to
1523 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1524 are treated as requests for stack allocation on the x86-64,
1525 Alpha-32, PPC and SPARC platforms.
1526 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1527 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1528 (thanks to Tony Martinez)
1530 ** gcing a dead thread can no longer lead to lockups
1531 ** threads block signals until they are set up properly
1532 ** errno is no longer shared by threads
1533 ** interrupt-thread restores the eflags register on x86
1534 ** fixed some lockups due to gc/thread interaction
1535 * fixed some bugs revealed by Paul Dietz' test suite:
1536 ** invalid dotted lists no longer raise a read error when
1537 *READ-SUPPRESS* is T
1538 ** signal an error if a symbol that names a declaration is used
1539 as the name of a type, or vice versa
1540 ** allow using the (declare (typespec var*)) abbreviation for
1541 (declare (type typespec var*)) with all type specifiers
1542 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1543 ** always return NIL from PPRINT-POP when OBJECT is NIL
1544 ** don't signal errors when pretty-printing malformed LABELS,
1545 FLET or MACROLET forms
1546 ** declarations in a DOLIST body are also in scope for the
1548 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1550 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1553 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1554 * fixed cross-compiler leakages that prevented building a 32-bit
1555 target with a 64-bit host compiler.
1556 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1557 opened with :IF-EXISTS :OVERWRITE.
1558 * fixed bug 281: error for an invalid qualifier in a short-form method
1559 combination method is not signalled until the faulty method is called.
1560 * bug fix: iteration variable type inferrer failed to deal with open
1561 intervals. (reported by Alan Shields)
1562 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1563 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1564 or more runtime options were provided to the sbcl binary.
1565 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1566 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1567 * optimization: there's now a fast-path for fixnum arguments in the
1568 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1569 * optimization: the code generated on x86-64 is more compact thanks
1570 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1571 * bug fix: floating-point exception handling now partly works on
1572 x86-64. (Thanks to James Knight)
1573 * improvement to the MIPS backend from Thiemo Seufer: C-style
1574 64-bit long long arguments and return values to alien functions
1576 * fixed some bugs revealed by Paul Dietz' test suite:
1577 ** the type-error signalled from WARN has a filled-in DATUM slot.
1578 ** the type-error required when a stream is not associated with
1579 a file has the stream as its datum.
1580 ** type-errors on single-floats on x86-64 no longer have
1581 :INVALID-OBJECT as the datum
1582 ** the type-errors signalled for invalid function names now have
1583 a correct expected type
1584 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1585 for structures defined with a :TYPE. Documentation strings for
1586 typed structures are no longer immediately discarded
1587 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1588 broadcast streams and synonym streams. FILE-LENGTH now also works
1589 on broadcast streams.
1591 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1592 * incompatible change: the --noprogrammer option, deprecated since
1593 version 0.7.5, has been removed. Please use the equivalent
1594 --disable-debugger option instead.
1595 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1597 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1598 inoperational for a while now) has been completely removed.
1599 * null lexical environments are now printed as #<NULL-LEXENV>,
1600 significantly reducing the amount of clutter in typical backtraces.
1601 * documentation on weak pointers, finalization, and after GC hooks
1602 has been added to the manual.
1603 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1604 as other specialized array types, is much faster. SUBSEQ and
1605 COPY-SEQ on such arrays have also been sped up.
1606 * optimization: EQL is now more efficient when at least other argument
1607 is known to be of type (OR FIXNUM (NOT NUMBER)).
1608 * fixed bug: the runtime is now less vulnerable to changes in the
1609 size of the SBCL object on OS X, and virtual memory is reserved for
1610 all spaces that need to be at a fixed address.
1611 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1612 * fixed bug: finalizers and after GC hooks that cause consing are now
1614 * fixed bug: compiler error messages and summaries are now printed to
1615 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1616 * fixed inference of the upper bound of an iteration variable.
1617 (reported by Rajat Datta).
1618 * fixed bug 211e: calling local functions with duplicated constant
1619 keyword argument no longer causes a bogus style warning about an
1621 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1622 ftype declaration to be disregarded. (reported by Dave Roberts)
1623 * fixed bug 373: caused by erronous compilation of references to alien
1624 variables in the runtime on ppc/darwin.
1625 * fixed bug 376: CONJUGATE type deriver.
1626 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1627 (reported by Baughn on #lisp)
1628 * a cross-compiler bug on non-x86 platforms has been identified and
1629 fixed. (thanks to Bruno Haible)
1630 * improvements to the MIPS runtime code for increased stability.
1631 (thanks to Thiemo Seufer)
1632 * increased the maximimum compact environment size to allow
1633 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1634 * partial workaround for bug 135: don't name the function we're
1635 calling for hairy cases of EVAL, so as not to accumulate one environment
1636 entry per call to EVAL. (reported by Kevin Reid)
1637 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1638 * optimization: type testing for non-vector arrays should be faster.
1639 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1641 * bug fix: setting 31st element of a bit vector to zero did not work
1643 * bug fix: redefining a class definition which failed due to a
1644 previous accessor / function clash now works (but see BUGS entry
1645 #380 for more problems in this area). (thanks to Zach Beane)
1646 * the long form of DEFINE-METHOD-COMBINATION disables method group
1647 checking when given a single method group with pattern *. (thanks
1649 * on x86 compiler supports stack allocation of results of simple
1650 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1651 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1652 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1653 * fixed some bugs related to Unicode integration:
1654 ** the restarts for recovering from input and output encoding
1655 errors only appear when there is in fact such an error to
1657 * fixed some bugs revealed by Paul Dietz' test suite:
1658 ** MISC.549 and similar: late transformation of unsafe type
1659 assertions into derived types caused unexpected code
1661 ** SCALE-FLOAT type deriver is less wrong.
1662 ** type derivers for EXP, LOG and similar functions compute result
1663 types for complex arguments better.
1664 ** (MISC.563) CONJUGATE type deriver works for very restricted
1666 ** out-of-line type testers for character strings are available.
1667 ** EQUAL compiler transform understands specialness of objects
1669 ** accessing double-floats stored on the stack now works on x86-64.
1670 ** debugger internals could sometimes create invalid lispobjs,
1671 resulting in GC crashes.
1672 ** MISC.548: type check weakening can convert required type into
1674 ** initialization forms for bindings are not in scope of free special
1677 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1678 * incompatible change: thread support for non-NPTL systems has
1679 been removed - locking is buggy and unreliable. A threaded
1680 SBCL build will now warn at startup and refuse to create
1681 new threads, unless futex support is detected in the kernel
1682 * incompatible change: the top level REPL now has only an ABORT
1683 restart associated with it, not TOPLEVEL and ABORT as it used to.
1684 TOP and TOPLEVEL are now available as debugger commands for
1685 returning to the top level.
1686 * incompatible change: forms evaluated in the REPL now use the
1687 global optimization policy.
1688 * incompatible change: user- and system-initialization files are
1689 no longer processed with LOAD, but by READ and EVAL; hence the
1690 global optimization policy, startup package, readtable, etc,
1692 * The .fasl file format number has been incremented because of
1693 various incompatible changes.
1694 * internal entry point details and argument counts no longer appear
1695 in backtraces unless explicitly requested by setting
1696 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1697 * built-in and standard functions no longer have names like "top
1698 level local call to FOO".
1699 * fixed bug 32: functions defined in non-null lexical environments
1700 now have more legible printed representation
1701 * fixed bug 33: functions defined in non-null lexical environemnts
1702 are now more amenable to inspection by INSPECT.
1703 * workaround for bug 354: XEPs no longer appear in backtraces unless
1704 explicitly requested.
1705 * fixed bug: receiving the signal which results from attempting to
1706 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1707 SIGBUS on other *BSDs) on architectures where the C stack is also
1708 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1709 to debugger confusion or wild instability. (reported by Cheuksan
1711 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1712 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1713 (reported by Lutz Euler)
1714 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1715 compiled code. (reported by Paul Dietz)
1716 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1717 specialized parameter to a method to something that is not TYPEP
1718 the specializer is now possible.
1719 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1720 face of package deletion.
1721 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1722 pathname. (reported by David Tolpin on c.l.l)
1723 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1724 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1725 STRING-STREAM and FILE-STREAM, respectively.
1726 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1727 than previously; in addition, its test suite is now run on build.
1728 (thanks to Luke Gorrie)
1729 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1730 * optimization: added a immediate representation for single-floats
1732 * fixed some bugs revealed by Paul Dietz' test suite:
1733 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1734 ** attempting to create a package with a colliding nickname causes
1735 correctable errors to be signalled.
1736 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1737 ** several x86-64 backend bugs related to sign-extension of immediate
1740 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1741 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1742 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1744 * fixed loading of multiply forward-referenced layouts.
1745 (thanks to Cheuksan Wang)
1746 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1747 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1748 (see COMPILE-FILE documentation for details of :PRINT option)
1749 * fixed bugs 19 and 317: fixed-format floating point printing is
1750 more accurate. This also fixes a bug reported by Adam Warner
1751 related to the ~@F format directive.
1752 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1754 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1755 dispatch table if the from-char is a dispatch macro character.
1756 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1757 by the wordsize no longer ignore the last word. (reported by Lutz
1759 * fixed bug in type checking of dynamic-extent variables. (reported
1761 * optimization: sequence traversal functions use their freedom to
1762 coerce function designators to functions.
1763 * optimization: code with many calls to CLOS methods specialized on
1764 CLOS classes has had redundant type checks removed.
1765 * optimization: added declarations to speed up operations that access
1766 the internal character database (for example STRING-UPCASE)
1767 * optimization: comparison operations between floats and sufficiently small
1768 fixnums no longer create extra rationals
1769 * fixed some bugs related to Unicode integration:
1770 ** portions of multibyte characters at the end of buffers for
1771 character-based file input are correctly transferred to the
1772 start of the buffer at the next read.
1773 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1774 passing it through to OPEN.
1775 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1776 argument given to internal calls to OPEN.
1777 ** the built-in comment readers (introduced by character sequences
1778 ";" and "#|") are more forgiving to encoding errors; they will
1779 STYLE-WARN and then attempt to resync the stream at a character
1780 boundary. (thanks to Teemu Kalvas)
1781 * fixed some bugs in the x86-64 port:
1782 ** Negative short int return values from c-calls are sign-extended
1784 ** The stack is aligned to 16-bytes for c-calls, as required by
1785 the ABI. (thanks to Cheuksan Wang)
1786 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1787 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1788 for enums. (thanks to Vincent Arkesteijn)
1789 ** Multiple small optimizations and bugfixes for floating point
1791 * fixed some bugs revealed by Paul Dietz' test suite:
1792 ** Space, Tab, Linefeed, Return and Page have the invalid
1793 secondary constituent character trait.
1794 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1796 ** WITH-INPUT-FROM-STRING should only update the index place on
1798 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1799 works more reliably.
1800 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1801 with one-dimensional arrays with fill pointers.
1802 ** TYPE-OF failed on a complex with an integer realpart and a
1804 ** compiler failure during type inference for the code of form
1805 (IF (EQL X (THE ...)) ...) (MISC.535).
1807 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1808 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1809 platform. The port supports 61-bit fixnums, large memory spaces
1810 and reloading shared object files.
1811 * enhancement: saving cores with foreign code loaded is now
1812 supported on ppc/Darwin in addition to the previously supported
1814 * enhancement: the statistical profiler now walks deeper into the
1815 call stack for more meaningful call-graphs and accrued time
1816 reports (x86/x86-64 only). It also now reports time spent in
1818 * enhancement: it is now possible to trace most individual methods
1819 of a generic function in addition to tracing the generic function
1821 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1823 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1824 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1825 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1826 produces an error. (thanks to Vincent Arkesteijn)
1827 * bug fix: NAMESTRING on pathnames with :WILD components in their
1828 directories works correctly. (thanks to Artem V. Andreev)
1829 * fixed bug 125: compiler preserves identity of closures. (reported
1831 * bug fixed: functions with &REST arguments sometimes failed with
1832 "Undefined function" when compiled with (DEBUG 3). (reported by
1834 * bug fix: overflow during compiling of setting element of a bit
1835 vector with constant index and value. (reported by Timmy Douglas)
1836 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1837 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1838 * on x86 compiler partially supports stack allocation of dynamic-extent
1840 * GO and RETURN-FROM do not check the extent of their exit points
1841 when compiled with SAFETY 0.
1842 * fixed some bugs related to Unicode integration:
1843 ** encoding and decoding errors are now much more robustly
1844 handled; it should now be possible to recover even from invalid
1845 input or output to the terminal. (thanks to Teemu Kalvas)
1846 ** provided a first cut at implementing STRING-TO-OCTETS and
1847 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1848 ** altered the SB-MD5 contributed module slightly, changing the
1849 interface just enough for it to be supportable for builds where
1850 lisp characters are not eight bits.
1851 * fixed some bugs revealed by Paul Dietz' test suite:
1852 ** the FORMATTER-generated functions for ~V[ conditionals require
1853 the correct number of arguments.
1854 ** READ-FROM-STRING returns the mandated second value when applied
1855 to displaced strings.
1856 ** the #\Rubout and #\Backspace characters are treated as invalid
1857 constituent characters by the tokenizer.
1859 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1860 * new feature: reloading changed shared object files with
1861 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1862 * new feature: references to foreign variables and functions
1863 can now be compiled and loaded before the corresponding shared
1864 object file is loaded, as long as the foreign definitions are
1865 available at runtime.
1866 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1867 supported. (thanks to Dan Debertin)
1868 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1869 just keywords. (thanks to Vincent Arkesteijn)
1870 * fixed bug #331: structure-class instances corresponding to
1871 DEFSTRUCT forms are now created eagerly.
1872 * fixed bug #345: backtraces from calls to undefined functions work
1873 on x86 as well. Related bug #61 is now also partially fixed on x86
1874 (backtraces from throws to unknown catch tags.)
1875 * bug fix: lambda-list parsing is now stricter vrt. order and number
1876 of lambda-list keywords.
1877 * bug fix: as specified by AMOP, an error is signalled if a
1878 class-option appears multiple times in a DEFCLASS form. (reported
1880 * bug fix: RANDOM can be compiled when the compiler derives the type
1881 of its numeric argument as a disjoint set of small integers.
1882 (reported by Paul Dietz)
1883 * bug fix: starting a core saved with shared objects loaded when
1884 those objects are not available no longer causes threaded SBCL to
1885 hang. (reported by Sean Ross)
1886 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1888 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1889 argument, which must always be NIL. (reported by Kalle Niemitalo)
1890 * bug fix: printing 1.0d+23 no longer results in an error.
1891 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1892 * bug fix: structure slot setters preserve evaluation order. (thanks
1894 * fixed some bugs related to Unicode integration:
1895 ** RUN-PROGRAM can allow its child to take input from a Lisp
1896 stream. (reported by Stefan Scholl)
1897 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1898 locales. (reported by Ken Causey)
1899 * fixed some bugs revealed by Paul Dietz' test suite:
1900 ** INCF, DECF and REMF evaluate their place form as specified in
1902 ** FORMATTER expands ~{ ~} iteration directives with V or #
1903 parameters correctly.
1904 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1905 consequent uses no arguments correctly.
1906 ** the system has a partial understanding of the (COMPLEX RATIO)
1908 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1911 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1912 * new feature: a build-time option (controlled by the :SB-UNICODE
1913 keyword feature, enabled by default) for building the system with
1914 support for the entire 21-bit character space defined by the
1916 * new feature: the system now has rudimentary external-format
1917 support; the primary user-visible change at this time is that
1918 characters with the high bit set (such as lower-case-e-acute) will
1919 print correctly to a terminal in a UTF-8 environment.
1920 * minor incompatible change: BASE-CHAR no longer names a class;
1921 however, CHARACTER continues to do so, as required by ANSI.
1922 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1923 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1924 has been moved to the SB-EXT package (temporarily re-exported from
1926 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1927 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1929 * fixed bug #351: better error-handlind and reporting for malformed
1931 * fixed bug #350: bignum-printing is now more memory-efficient,
1932 allowing printing of very large bignums, eg. (expt 2 10000000).
1933 (reported by Bruno Haible)
1934 * fixed bug #302: better primitive-type selection for intersection
1936 * fixed bug #308: non-graphic characters now all have names, as
1937 required. (reported by Bruno Haible)
1938 * bug fix: redefining a class with different superclasses now correctly
1939 removes it from the direct-subclasses of its previous superclasses.
1940 (reported by David Morse)
1941 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1942 new value now works. (reported by Bruno Haible)
1943 * bug fix: correct canonicalization of multiple non-standard slot
1944 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1945 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1946 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1947 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1948 now exists, an signals an error.
1949 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1950 during macroexpansion for non-iteration variables as well. (reported
1951 by Bruno Haible for CMUCL)
1952 * bug fix: Cyclic structures and unprintable objects in compiler
1953 messages no longer cause errors. (reported by Bruno Haible)
1954 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1955 and READ-FROM-STRING all now return a primary value of NIL if
1956 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1957 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1958 (reported by Bruno Haible for CMUCL)
1959 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1960 do not designate a condition. (reported by Bruno Haible for
1962 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1963 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1964 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1965 permitting resolution in favour of any of the conflicting symbols.
1966 (reported by Bruno Haible for CMUCL)
1967 * bug fix: EQUAL compiler optimizations is less aggressive on
1968 strings which can potentially compare true despite having distinct
1969 specialized array element types.
1970 * bug fix: unit enumerations can be defined without dividing by
1971 zero. (thanks to Vincent Arkesteijn)
1972 * FORMAT compile-time argument count checking has been enhanced.
1973 (report from Bruno Haible for CMUCL)
1974 * a partial workaround for the bug 262: the compiler does not try to
1975 inline-expand a local function doing RETURN-FROM from a deleted
1976 BLOCK. (thanks to Peter Denno for the bug report and to David
1977 Wragg for the simple test case)
1978 * fixed some bugs revealed by Paul Dietz' test suite:
1979 ** RENAME-PACKAGE allows all package designators as new package
1981 ** constraint propagation and lambda variable substitution are
1982 more cautious in dealing with partially deleted code.
1983 ** compiler sometimes lost reoptimization passes.
1984 ** CERROR, given a condition as condition designator, passes its
1985 remaining arguments to the continue format control without
1987 ** Case-altering FORMAT directives work correctly on non-ASCII
1989 ** The REAL type specifier handles bounds outside the floating
1990 point ranges without signalling FLOATING-POINT-OVERFLOW.
1991 ** Functions with IR1-transformations can create intercomponent
1992 references to global functions.
1993 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1995 ** FORMAT ~R treats a nil value for its first parameter correctly.
1997 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1998 * enhancement: saving cores with foreign code loaded is now
1999 supported on x86/NetBSD and sparc/Linux in addition to the previously
2000 supported platforms.
2001 * bug fix: on some platforms repeated installations caused multiple
2002 copies of HTML documentation to be installed -- should not happen
2003 any more. (reported by Stefan Scholl)
2004 * bug fix: parsing self-recursive alien record types multiple times
2005 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2006 original patch by Helmut Eller for CMUCL)
2007 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2008 (thanks to Richard Kreuter)
2009 * bug fix: defining classes whose accessors are methods on existing
2010 generic functions in other (locked) packages no longer signals
2011 bogus package lock violations. (reported by François-René Rideau)
2012 * bug fix: special variables as DEFMETHOD parameters no longer have
2013 associated bogus type declarations. (reported by David Wragg and
2015 * bug fix: read-write consistency on streams of element-type
2016 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2017 * bug fix: redefiniton of the only method of a generic function with
2018 no DEFGENERIC no longer emits a full WARNING. In addition,
2019 redefinition of generic functions with no DEFGENERIC to an
2020 incompatible lambda list now signals an error. (thanks to Zach
2022 * bug fix: DEFGENERIC now works even when there's a function of the
2023 same name in an enclosing lexical environment. (thanks to Zach
2025 * fixed compiler failure, caused by instrumenting code during
2026 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2027 * optimization: added loop analysis and improved register allocation
2028 * fixed some bugs revealed by Paul Dietz' test suite:
2029 ** POSITION on displaced vectors with non-zero displacement
2030 returns the right answer.
2031 ** (SIMPLE-STRING) is a valid type specifier for sequence
2033 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2035 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2036 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2038 ** The justification version of the FORMAT ~< directive treats
2039 non-zero minpad parameter correctly.
2041 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2042 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2043 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2044 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2045 the supported interface.
2046 * new feature: Single-stepping of code compiled with DEBUG 2 or
2047 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2048 * new feature: saving cores with foreign code loaded is now
2049 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2050 Timothy Moore's work for CMUCL)
2051 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2052 parameters to * instead of NIL if no initform is supplied.
2053 (reported by Johan Bockgaard)
2054 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2055 correctly when FUNCALL appears as the car of the form. Note:
2056 despite this FUNCALL forms are not currently subject to
2057 compiler-macro expansion. (port of Raymond Toy's fix for the
2058 same from CMUCL, reported by Johan Bockgaard)
2059 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2060 (thanks for Teemu Kalvas)
2061 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2062 case instead of terminating the iteration (thanks for Julian
2063 Squires, Sean Champ and Raymond Toy)
2064 * bug fix: incorrect expansion of defgeneric that caused a style
2065 warning. (thanks for Zach Beane)
2066 * bug fix: slot accessor effective method computation works properly
2067 for classes with multiple non-standard applicable methods on
2068 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2069 * on x86 compiler supports stack allocation of results of LIST and
2070 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2071 CMUCL implementation by Gerd Moellmann)
2072 * fixed some bugs revealed by Paul Dietz' test suite:
2073 ** FORMAT strings with both the ~<~:;~> form of the justification
2074 directive and pretty-printing directives cause an error.
2076 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2077 * incompatible change: the internal functions
2078 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2079 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2080 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2081 instead of the old functions.
2082 * new feature: on platforms where "dladdr" is available foreign
2083 function names now appear in backtraces. (based on Helmut Eller's
2085 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2086 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2088 * documentation: ASDF manual is now distributed with SBCL.
2089 * bug fix: SBCL can now load its contributed modules with REQUIRE
2090 even if the system-provided entries have been removed from
2091 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2093 * bug fix: backtraces involving undefined functions or assembly
2094 routines are more informative. (thanks to Brian Downing)
2095 * bug fix: mutually referent alien structures now work correctly.
2096 (reported by Rick Taube)
2097 * bug fix: structures defined by WITH-ALIEN can be referred to
2098 within other definitions in the same WITH-ALIEN.
2099 * bug fix: division operators (MOD, TRUNCATE and the like) with
2100 constant zero divisors and integer dividends no longer generate
2102 * bug fix: provide default methods for INPUT-STREAM-P and
2103 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2104 * bug fix: improve the detection and resolution of MOP metacycles
2105 (where effective method computation depends on the generic function
2106 having its effective method computed). (reported by Bruno Haible)
2107 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2108 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2109 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2110 * bug fix: pathnames with relative directory components can be
2111 represented relative to default pathnames.
2112 * optimization: in taking the GCD of bignums, reduce the two bignums
2113 to approximately the same size (using Euclid's algorithm) before
2114 applying the more sophisticated binary GCD. (thanks to Juho
2116 * optimization: COUNT on bitvectors now operates word-at-a-time.
2117 * optimization: ASH with a positive, but not necessarily constant,
2118 (leftwards) shift, when in a modular context, is compiled to a
2120 * fixed some bugs revealed by Paul Dietz' test suite:
2121 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2122 if the corresponding argument is NIL.
2123 ** FORMAT directives accepting a minpad parameter treat negative
2125 ** literal commas as character parameters to FORMAT directives are
2127 ** literal spaces directly after ~<Newline> directives within a
2128 format-logical-block (~:< ~@:>) do not induce :FILL-style
2129 conditional newlines.
2130 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2131 as specified: it no longer includes conditional newlines).
2132 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2134 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2135 :PER-LINE-PREFIX argument does not evaluate to a string.
2137 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2138 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2139 the manual for details; for now, package locks can be disabled by
2140 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2141 if no major problems are found then it is likely that they will be
2142 compiled in unconditionally.
2143 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2144 unsupported operators on all platforms. To load a shared library
2145 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2146 object file, link it into a shared library outside of SBCL and
2147 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2149 * bug workaround: ROOM T and the GC were not working together
2150 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2151 reporting facility aren't preserved. That reporting has been
2152 disabled, so now until and unless someone figures out how to make it
2153 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2154 * minor incompatible change: as threatened around sbcl-0.8.0, the
2155 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2156 The symbols are also exported from SB-PCL for backwards
2157 compatibility, but more so than before SB-PCL should be treated as
2158 an implementation-internal package.
2159 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2161 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2162 It is known as of this release not to work on the Alpha, however.
2163 * fixed bug #167: errors signalled due to illegal syntax in method
2164 bodies are now more legible.
2165 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2166 designators and can hence be used with TYPEP.
2167 * fixed bug #333: CHECK-TYPE now ensures that the type error
2168 signalled, if any, has the right object to be accessed by
2169 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2170 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2171 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2173 * fixed bug #334: programmatic addition of slots using specialized
2174 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2175 and :ALLOCATION :CLASS slots.
2176 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2177 float exponents. (rereported by Peter Seibel)
2178 * fixed a bug: #\Space (and other whitespace characters) are no
2179 longer considered to be macro characters in standard syntax by
2180 GET-MACRO-CHARACTER.
2181 * fixed bug: initialization of condition class metaobjects no longer
2182 causes an instance of the condition to be created. (reported by
2184 * fixed bug: it is now possible to have slots such that
2185 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2186 description is neither :INSTANCE nor :CLASS.
2187 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2188 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2189 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2190 non-standard SLOT-DEFINITION-ALLOCATION values.
2191 * fixed bugs: various race conditions handling exiting threads.
2192 CL-PPCRE's thread torture test now passes.
2193 * fixed arguably-a-bug: GC time proportional to number of threads in
2194 system even when most of them are idle
2195 * optimization: improved performance of BIT and SBIT on bit-vectors.
2196 * .fasl file incompatibility: The fasl file version number has
2197 been incremented because of changes associated with package locks.
2199 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2200 * minor incompatible change: the system no longer provides
2201 optimization hints (conditions which are TYPEP
2202 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2203 mode; these hints are still emitted when higher SPEED optimization
2205 * new contrib module: a sampling profiler (profiling by statistical
2206 sampling, rather than by instrumenting functions) is available as
2207 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2208 * the behaviour of the standard function ED is now customizeable by
2209 third parties through a hook variable: see ED's documentation
2210 string for information on the protocol.
2211 * the compiler no longer emits efficiency notes for (FUNCALL X)
2212 when the type of X is uncertain under default optimization
2214 * fixed bug 276: mutating a binding of a specialized parameter to a
2215 method to something that is not TYPEP the specializer is now
2217 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2218 exhibit the required behaviour on the x86 platform. (thanks to
2219 Peter van Eynde, Eric Marsden and Bruno Haible)
2220 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2221 even for difficult arguments. (reported by Peter Graves)
2222 * fixed bug 141a: the backquote printer now descends quoted
2224 * fixed another bug in backquote printing: no more destructive
2225 modification of the form's list structure. (reported by Brian
2227 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2228 move between its address being taken and the call to
2229 interrupt_thread, fixing a crashing race condition.
2230 * the SB-POSIX contrib implementation has been adjusted so that it
2231 no longer exhibits ridiculously poor performance when constructing
2232 instances corresponding to C structs.
2234 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2235 * minor incompatible change: the sb-grovel contrib now treats C
2236 structures as alien (in the sense of SB-ALIEN) objects rather than
2237 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2238 has implications for memory management of client code
2239 (sb-grovel-returned objects must now be manually managed) and for
2240 type safety (alien objects now have full types).
2241 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2242 used to control emission of compiler diagnostics, rather than the
2243 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2244 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2245 quality should be considered deprecated.
2246 * install.sh now installs the user manual as well
2247 * (not quite a new documentable feature, but worth considering in
2248 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2249 beginnings of a semantically meaningful condition hierarchy is
2250 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2252 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2253 designator as the defaults argument.
2254 * fixed bug: Displaced arrays whose displaced-to array has become
2255 too small now cause ARRAY-DIMENSION to signal an error, providing
2256 for safer bounds-checking. (reported by Bruno Haible)
2257 * fixed bug: DEFCLASS slot definitions with identical :READER and
2258 :WRITER names now signal a reasonable error. (reported by Thomas
2260 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2262 * fixed bug: Invalid slot specification errors now print correctly.
2263 (thanks to Zach Beane)
2264 * fixed bug 320: Shared to local slot value transfers in class
2265 redefinitions now happen corrently from superclasses as
2266 well. (reported by Bruno Haible)
2267 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2269 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2270 improper lists correctly. (reported by Bruno Haible)
2271 * fixed bug 313: source-transform for <fun-name> was erroneously
2272 applied to a call of a value of a variable with name <fun-name>.
2273 (reported by Antonio Menezes Leitao)
2274 * fixed bug 307: The obsolete instance protocol ensures that
2275 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2276 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2277 remove dead unknown-values globs from the middle of the stack.
2278 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2279 easier to resume long computations after using *BREAK-ON-SIGNALS*
2280 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2281 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2282 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2283 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2284 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2285 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2287 * fixed a (fairly theoretical) bug in string printing: if
2288 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2289 does not have array-element-type equal to the most general string
2291 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2292 function name. (thanks to Nikodemus Siivola)
2293 * fixed bug: compile-time format string checker failed on
2294 non-closed ~{. (reported by Thomas F Burdick)
2295 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2296 don't warn on legal loop code involving more than one aggregate
2297 boolean. (thanks to Nikodemus Siivola)
2298 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2299 in keyword position to MAKE-PACKAGE was not being reported as
2301 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2302 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2304 * fixed bugs 280 and 312: the checking for multiple definitions in a
2305 file is less likely to become confused by uses of inline
2307 * fixed bug: the #S reader macro performs the keyword coercion
2308 specified for slot names. (reported by Kalle Niemitalo)
2309 * fixed bug: lambda lists may contain symbols whose names start with
2310 & but are not lambda-list-keywords; their occurrence triggers a
2312 * fixed bug 321: define-method-combination argument lambda lists do
2313 not cause invalid code to be generated when &optional or &aux
2314 variables have default values. (reported by Bruno Haible)
2315 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2316 have CLOS classes; this appears also to have fixed reported
2317 instability in stack exhaustion detection.
2318 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2319 MUFFLE-WARNING when no associated restart is present is now
2321 * optimization: rearranged the expansion of various defining macros
2322 so that each expands into only one top-level form in a
2323 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2325 * optimization: used a previously little-used slot in symbols to
2326 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2328 * fixed some bugs revealed by Paul Dietz' test suite:
2329 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2331 ** Fixed an optimization bug related to inheritance of initforms
2332 from local to shared slots.
2333 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2334 ** CAST optimizer forgot to flush argument derived type.
2335 ** print/read consistency on floats is now orders of magnitude
2336 more likely. (thanks also to Bruno Haible for a similar report
2338 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2339 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2340 for objects of type REAL. Make it so.
2341 ** SXHASH is specified (infelicitously) to respect similarity,
2342 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2343 it so. (thanks to Markus Ziegler)
2344 ** on the Alpha, the compiler succeeds in compiling functions
2345 returning a known number of arguments greater than 63.
2346 ** fixed handling of invalid NIL arguments in keyword position
2348 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2350 ** provided a readably-printable representation for RANDOM-STATE
2352 ** ensured that pathnames, where they have a namestring, always
2353 print using #P"..." syntax.
2355 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2356 * Support for the forthcoming 2.0 version of the NetBSD kernel
2357 running on x86 hardware has been added. (thanks to Perry
2358 E. Metzger most immediately, and others for their past work)
2359 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2360 current release version; SBCL's OpenBSD support had been broken
2361 since about the time of OpenBSD's switch to ELF binary format.)
2362 * [placeholder for DX summary]
2363 ** user code with &REST lists declared dynamic-extent, under high
2364 speed or space and low safety and debug optimization policy.
2365 * The manual has been converted to Texinfo format and the debugger
2366 chapter from the cmucl manual has been added.
2367 * A facility has been added to extract documentation strings from
2368 sbcl and store them as Texinfo-formatted snippets for inclusion in
2369 the manual (via Texinfo's @include directive)
2370 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2371 the test case to Dave Roberts)
2372 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2373 pointers. (reported by Sean Ross)
2374 * bug fix: PROFILE output is printed nicely even for large numerical
2375 values. (thanks to Zach Beane)
2376 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2377 greater than 32 handle EOF correctly.
2378 * bug fix: on X86 an immediate argument of the IMUL instruction is
2379 correctly printed in disassembly. (thanks to Lutz Euler)
2380 * bug fix: class slots in redefined classes preserve their old
2381 values. (thanks to Bruno Haible and Nikodemus Siivola)
2382 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2383 succeeds. (reported by Marco Baringer)
2384 * fixed some bugs revealed by Paul Dietz' test suite:
2385 ** READ-SEQUENCE now works on ECHO-STREAMs.
2386 ** RATIONALIZE works more according to its specification. (thanks
2389 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2390 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2391 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2392 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2393 should matter to you only if you rebind the printer control
2394 variables and then find you want different bindings in the
2395 debugger than in the ordinary execution of your program.)
2396 * The runtime build system has been tweaked to support building
2397 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2398 assembler and linker. (thanks to Nikodemus Siivola)
2399 * Unbound, undefined, undeclared variables now trigger full
2400 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2401 more often programmer error than deliberate exploitation of undefined
2403 * optimization: the hash algorithm for strings has changed to one
2404 that is less vulnerable to spurious collisions. (thanks to Juho
2406 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2407 needless bounds checking. (thanks to Juho Snellman)
2408 * optimization: implemented multiplication as a modular
2409 (UNSIGNED-BYTE 32) operation on the PPC backend.
2410 * fixed some bugs revealed by Paul Dietz' test suite:
2411 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2413 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2414 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2415 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2417 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2419 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2421 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2422 their output stream on EOF from read.
2423 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2424 have been read to end-of-file.
2425 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2427 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2428 description of determination of which consecutive characters
2430 ** Printing the "Space" character with escaping on now yields "#\\ ",
2431 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2432 ** Reading floating-point numbers with *READ-BASE* set to a number
2433 less than 10 works correctly.
2434 ** Reading floating-point numbers with *READ-BASE* set to a number
2435 more than 10 works correctly.
2436 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2437 the readtable currently in effect.
2439 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2440 * minor incompatible change: parsing of namestrings on a physical
2441 (Unix) host has changed; numbers after the final #\. in a
2442 namestring are no longer interpreted as a version field. This is
2443 intented to be largely invisible to the user, except that the
2444 meaning of the namestring "*.*.*" has changed: it now refers to a
2445 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2446 should usually be replaced by
2447 (make-pathname :name :wild :type :wild :version :wild)
2448 with the added benefit that this is more likely to be portable.
2449 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2450 option to OPEN now signals an error if the file being opened
2451 exists; this may have an impact on existing code.
2452 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2453 signals received, on the PowerPC platforms (both Linux and
2454 Darwin). (thanks to Pierre Mai for pointing out the location of
2456 * several fixes on OS X: The system now builds and runs cleanly on
2457 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2458 (thanks to Brian Mastenbrook)
2459 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2460 second-resolution: integer multiples of 1/3600 between -24 and 24.
2461 (thanks to Vincent Arkesteijn)
2462 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2463 type when called with 1 argument; PEEK-CHAR checked type of
2464 PEEK-TYPE only after having read first character from a
2465 stream. (reported by Peter Graves)
2466 * bug fix: the garbage collector now has much better locality
2467 behaviour, and in particular no longer treats all memory as being
2468 exhausted when many small objects point to each other in a deeply
2470 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2471 recognized as being TYPEP their class.
2472 * bug fix: the PUSHNEW documentation string has been corrected.
2473 (thanks to Vincent Arkesteijn)
2474 * bug fix: defaulting of the value for the last of an atypically
2475 large number of multiple values being bound was not being
2476 performed correctly on the Alpha or PPC platforms
2477 * optimization: implemented multiplication as a modular
2478 (UNSIGNED-BYTE 32) operation on the x86 backend.
2479 * optimization: SEARCH on simple-base-strings can now be open-coded.
2480 (see also contrib/compiler-extras.lisp for inspiration for
2481 teaching the compiler about the Boyer-Moore algorithm).
2482 * value, returned by MAX (and MIN) called with several EQUALP, but
2483 not EQL, arguments now does not depend on compiler settings.
2484 * fixed some bugs revealed by Paul Dietz' test suite:
2485 ** in stack analysis liveness information is propagated from
2486 non-local entry points.
2487 ** pathwise CAST removing failed when the CAST node did not start
2489 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2490 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2492 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2493 host is already defined.
2494 ** RENAME-FILE works on streams instead of signalling an internal
2496 ** PEEK-CHAR uses the current readtable when determining whether
2497 or not a character is whitespace.
2498 ** MERGE-PATHNAMES handles the case when the pathname does not
2499 specify a name while the default-pathname specifies a version.
2500 ** Pathnames now stand a better chance of respecting print/read
2502 ** Attempting to use standardized file system operators with a
2503 pathname with invalid :DIRECTORY components signals a
2505 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2507 ** DIRECTORY on logical pathnames is more correct.
2508 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2509 signal a TYPE-ERROR if their argument is not a stream.
2510 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2511 designator argument does not designate a stream.
2512 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2513 examining the synonym.
2514 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2516 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2517 element-type larger than ([UN]SIGNED-BYTE 32).
2519 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2520 * When built with the :SB-FUTEX feature, threaded builds now take
2521 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2522 for faster/more reliable mutex and condition variable support.
2523 * Incompatible change (but one you probably shouldn't have been using
2524 anyway): the interface and code for arbitrating between multiple
2525 threads in the same user session has been redesigned.
2526 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2527 error when called without an explicit environment argument.
2528 (thanks to Nikodemus Siivola)
2529 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2530 confused about their position. (thanks to Adam Warner and Gerd
2532 * bug fix: Pretty printing backquoted forms with unquotations in the
2533 argument list position of various code constructs such as LAMBDA
2534 now works correctly. (reported by Paul Dietz)
2535 * bug fix: Pretty printing unquotations no longer loses all
2536 stream position information.
2537 * optimization: performance of string output streams is now less
2538 poor for multiple small sequence writes.
2539 * optimization: performance of CSUBTYPEP in the presence of complex
2540 expressions involving CONS and NOT many times has been improved.
2541 (reported by Paul Dietz)
2542 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2544 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2545 streams for the REPL, for the debugger, and so on. (thanks to
2547 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2548 * fixed some bugs revealed by Paul Dietz' test suite:
2549 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2550 is ignored if it is too small, rather than propagating through
2551 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2552 ** extremely complex negations of CONS types were not being
2553 sufficiently canonicalized, leading to inconsistencies in
2555 ** VALUES tranformer lost derived type.
2557 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2558 * fixed a bootstrapping bug: the build process no longer assumes
2559 that the various BOOLE-related constants have the same value in
2560 host and target lisps. (noted by Paul Dietz' test suite on an
2561 SBCL binary built from CLISP)
2562 * The system can now be dynamically linked on the MIPS platform,
2563 which enables dynamic loading of foreign code from Lisp. (thanks
2564 to Ralf Baechle for discussions on the MIPS ABI)
2565 * The system now records debugging information for its own source
2566 files in a filesystem-position-independent manner, relative to
2567 the "SYS" logical host.
2568 * fixed a compiler bug: MV-LET convertion did not check references
2569 to the "max args" entry point. (reported by Brian Downing)
2570 * tweaked disassembly notes to be less confident about proclaiming
2571 some instruction as an LRA. (thanks to Brian Downing)
2572 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2573 now each have their own history, command character, and other
2574 characteristics. (thanks to David Lichteblau)
2575 * fixed some bugs revealed by Paul Dietz' test suite:
2576 ** compiler failure in compiling LOGAND expressions including a
2578 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2579 shift greater than 32.
2580 ** FLUSH-DEST did not mark blocks for type check regeneration.
2581 ** HANDLER-CASE failed to accept declarations in handler clauses
2582 in some circumstances.
2584 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2585 * New code in contrib/sb-introspect (still probably not entirely
2586 stable yet) provides some support for smart Lisp development
2587 environments like SLIME.
2588 * The conditions signalled for errors occurring when loading .fasl
2589 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2590 in a way which should help ASDF recover gracefully.
2591 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2592 stuff has been cleaned up. If you code contrib/ stuff, this might
2593 affect you, and you can look at contrib/README, contrib/STANDARDS,
2594 and/or the 0.8.4.27 diff to check.
2595 * In full calls the compiler now does not generate checks for declared
2596 argument types for all arguments.
2597 * various threading fixes
2598 ** and some experimental patches which didn't make it into
2599 the main tree for this release, but which are shipped in
2600 contrib/experimental-thread.patch as a possible fix for some
2601 failures (deadlock, spinning...) in GC-intensive multithreaded
2603 * fixed PPC build problem (source code incompatibility of different
2604 library versions): added offsetof() hackery which attempts to divine
2605 where glibc maintainers put uc_mcontext today
2606 * fixed bug 282: compiler does not trust type assertions while passing
2607 arguments to a full call.
2608 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2609 &OPTIONAL VALUES type parameter.
2610 * fix bug 214: algorithm for noting rejected templates is now more
2611 similar to that of template seletion. (also reported by rydis on
2613 * fixed bug 141b: printing backquoted information readably and prettily
2614 inserts a space where necessary.
2615 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2616 (CHARACTER 10) are now reported as errors, rather than propagated
2617 as unknown types. (reported by piso on #lisp)
2618 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2619 with values NIL and :ERROR. (thanks to Milan Zamazal)
2620 * fixed bug 191c: CLOS now does proper keyword argument checking as
2621 described in CLHS 7.6.5 and 7.6.5.1.
2622 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2623 longer raise an error; further, using a list as a for-as-arithmetic
2624 counter now raises a meaningful error.
2625 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2626 now understood by sequence creation functions such as MAKE-SEQUENCE
2628 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2629 a STRING-INPUT-STREAM.
2630 * compiler enhancement: SIGNUM is now better able to derive the type
2632 * type declarations inside WITH-SLOTS are checked. (reported by
2634 * fixed some bugs revealed by Paul Dietz' test suite:
2635 ** incorrect optimization of TRUNCATE for a positive first
2636 argument and negative second.
2637 ** compiler failure in let-convertion during flushing dead code.
2638 ** compiler failure while deriving type of TRUNCATE on an
2639 interval, containing 0.
2640 ** ASH of a negative bignum by a negative bignum count now returns
2642 ** intersection of CONS types now canonicalizes properly, fixing
2643 inconsistencies in SUBTYPEP.
2645 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2646 * incompatible change: The --disable-debugger command line
2647 option now clobbers the debugger at a more fundamental
2648 level, by redefining #'INVOKE-DEBUGGER instead of by
2649 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2650 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2651 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2652 enter the debugger REPL and then suffer recursive errors
2653 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2654 while under the new system, BREAK in --disable-debugger mode
2655 terminates the system just as an unhandled error would.
2656 * fixed compiler performance when processing loops with a step >1;
2657 * bug fix: DOCUMENTATION now retrieves generic function
2658 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2659 support has been systematized, and now supports the methods
2660 specified by ANSI, along with a default method and a method for
2661 slot documentation. (reported by Nathan Froyd)
2662 * bug fix: effective methods associated with a generic function are
2663 no longer cached over a change of that generic function's method
2664 combination. (reported by Andreas Fuchs)
2665 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2666 names a non-existent directory. (thanks to Andreas Fuchs)
2667 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2668 the PPC platform, lead to stack corruption; this has been fixed.
2669 (reported by Rainer Joswig)
2670 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2671 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2672 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2673 arithmetic in general, is now much more reliable on the PPC
2675 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2676 platform now returns the right answer.
2677 * optimization: restored some effective method precomputation in
2678 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2679 precomputation is now tunable.
2680 * optimization: compiler-internal data structure use has been
2681 reviewed, and changes have been made that should improve the
2682 performance of the compiler by about 20%.
2683 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2684 simple-base-strings and simple-bit-vectors is improved.
2685 * optimization: the optimization of 32-bit logical and arithmetic
2686 functions introduced in version 0.8.3 on the x86 has been
2687 implemented on the mips, ppc and sparc platforms; an
2688 implementation of the same facility, but for 64-bit arithmetic,
2689 has been added for the alpha.
2690 * microoptimization: the compiler is better able to make use of the
2691 x86 LEA instruction for multiplication by constants.
2692 * bug fix: in some situations compiler did not report usage of
2693 generic arithmetic in (SPEED 3) policy.
2694 * bug 145b fix: compiler used wrong type specifier while converting
2695 MEMBER-types to numeric.
2696 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2698 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2699 index argument now works properly on the Alpha platform.
2700 * bug fix: floating point exception treatment on the Alpha platform
2702 * bug fix: FILE-POSITION works much better on string input and
2703 output streams. (thanks to Nikodemus Siivola)
2704 * bug fix: many threading/garbage collection symptoms sorted.
2705 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2706 might be pseudo-atomic.
2707 * internal change: Stopping for GC is now done with signals not
2708 ptrace. GC is now done in whichever thread wanted it, instead of
2710 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2712 * bug fix: result form in DO is not contained in the implicit
2714 * incompatible change: ICR structure is changed; the value part of
2715 CONTINUATION is now called LVAR; corresponding functions are
2716 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2717 * added type deriver for ISQRT (thanks to Robert E. Brown).
2718 * bug fix: better support for loading from the command line when an
2719 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2721 * fixed some bugs revealed by Paul Dietz' test suite:
2722 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2723 ** ROUND and FROUND now give the right answer when given very
2724 small float arguments.
2725 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2727 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2728 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2729 ** LCM should return a non-negative integer.
2730 ** PARSE-INTEGER returned the index of a terminator instead of the
2731 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2732 ** PARSE-INTEGER returned an incorrect index being applied to a
2734 ** LCM with two arguments of 0 returns 0 rather than signalling
2736 ** unsigned addition of a 32-bit constant with the high bit set no
2737 longer causes an internal compiler error on the x86.
2738 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2739 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2740 with negative last argument.
2741 ** byte specifiers with zero size and position no longer cause
2742 an error during type derivation.
2743 ** bignum multiplication on the Alpha platform now returns the
2745 * porting: The system now builds on SuSE AMD64, although it still
2746 generates a 32-bit binary.
2747 * .fasl file incompatibility: The fasl file version number has
2748 been incremented (because of the changes to internal compiler
2749 data structures referred to above).
2751 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2752 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2753 more accurately, on the Darwin kernel running on PowerPC hardware.
2754 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2755 * Compiler code deletion notes now signal a condition of type
2756 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2757 an associated MUFFLE-WARNING restart.
2758 * The compiler now performs limited argument count validation of
2759 constant format strings in FORMAT, and where appropriate in ERROR,
2760 CERROR and WARN. (thanks to Gerd Moellmann)
2761 * New ASDF-INSTALL contrib can be used for automatic download and
2762 installation of third-party Lisp code from CCLAN or other sites
2764 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2765 which forces another thread to execute a function supplied by the
2767 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2768 now accept and act upon their :ELEMENT-TYPE keyword argument.
2769 (reported by Martin Atzmueller, Edi Weitz)
2770 * bug fix: FILE-POSITION now accepts position designators up to
2771 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2772 is the greater. (thanks to Patrik Nordebo)
2773 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2774 Kalle Olavi Niemitalo)
2775 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2776 printable. (reported by Eric Marsden)
2777 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2778 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2779 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2780 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2781 (thanks to Stig Sandoe)
2782 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2783 UNREAD-CHAR now correctly includes the unread character in the
2784 target sequence. (thanks to Gerd Moellmann)
2785 * bug fix in threaded builds: the system can now be suspended and
2786 resumed by shell job control with minimal disruption.
2787 * bug fixes in times and timezones >2038AD
2788 * better handling of "where is GNU make?" problem in build scripts
2789 (thanks to Nikodemus Siivola)
2790 * new optimization: inside a named function any reference to a
2791 function with the same name is considered to be a self-reference;
2792 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2793 optimization quality.
2794 * new optimization on x86: logical functions and + now have
2795 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2796 used when the result is truncated to 32 bits.
2797 * VALUES declaration is partially enabled.
2798 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2799 * fixed some bugs revealed by Paul Dietz' test suite:
2800 ** The system now obeys the constraint imposed by
2801 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2802 types form a lattice under type intersection.
2803 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2804 ** ASSOC now ignores NIL elements in an alist.
2805 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2806 and (1+ MOST-POSITIVE-FIXNUM) answers.
2807 ** The addition of a method with invalid qualifiers to a generic
2808 function does not cause an error to be signalled immediately;
2809 a warning is signalled, and the error is generated only on
2810 calling the generic function.
2811 * changed .fasl file version number, in part to add type codes for
2812 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2813 obscure ANSI requirements
2815 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2816 * fixed bug 148: failure to inline-expand a local function left
2817 garbage, confusing the compiler.
2818 * fixed bugs 3cd: structure slot readers perform type check if the
2819 slot can have an invalid value (i.e. it is either not initialized
2820 or can be written with a less specific slot writer).
2821 * bug fix: the compiler now traps array references to elements off
2822 the end of an array; previously, the bounds checking in some
2823 circumstances could go off-by-one.
2824 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2826 * type declarations for array element types now obey the description
2827 on the CLHS page "Declaration TYPE", as per discussions on
2828 sbcl-help around 2003-05-08. This means that a declaration
2829 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2830 declaration, all references to BAR will be asserted or assumed
2831 (with THE, so dependent on compiler policy) to involve objects of
2832 type FOO. Note that no such declaration is implied in
2833 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2834 * declared types of functions from the "Conditions"
2835 chapter. (reported by Paul Dietz)
2836 * bug fix: CERROR accepts a function as its first argument.
2837 * bug fix: NTH an NTHCDR accept a bignum as index
2838 arguments. (reported by Adam Warner)
2839 * optimization: character compare routines now optimize comparing
2840 against a constant character. (reported by Gilbert Baumann)
2841 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2842 argument now works properly on the MIPS platform.
2843 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2844 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2845 declared to have a functional type, does not cause an error
2847 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2848 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2849 * fixed bug: if last continuation of a deleted block has a
2850 destination, this destination should be deleted too. (reported by
2852 * fixed a bug in the bootstrap process: the host compiler's values
2853 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2854 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2855 test case from Patrik Nordebo)
2856 * improved the ability of the disassembler on the PPC platform to
2857 provide helpful disassembly notes.
2858 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2859 the class in more cases than previously.
2860 * bug fix: FILE-POSITION now understands :START and :END for
2861 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2862 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2863 8) data. (thanks to David Lichteblau)
2864 * bug fix: it is possible to add a method to a generic function
2865 without lambda list.
2866 * bug fix: reader failed to signal END-OF-FILE inside an
2867 object representation. (reported by Nikodemus Siivola)
2868 * fixed some bugs revealed by Paul Dietz' test suite:
2869 ** LAST and [N]BUTLAST should accept a bignum.
2870 ** condition slot accessors are methods.
2871 ** (VECTOR NIL) is a subtype of STRING.
2873 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2874 * minor incompatible change: some nonsensical specialized lambda
2875 lists (used in DEFMETHOD) which were previously ignored now signal
2877 * minor incompatible change: the system is now aware of the types of
2878 variables in the COMMON-LISP package, and will signal errors for
2879 most violations of these type constraints (where previously they
2880 were silently accepted).
2881 * minor incompatible change: COMPILE-FILE now uses the freedom
2882 afforded (ANSI 3.2.2.3) to use derived function types for
2883 functions defined in the same file. This also permits the system
2884 to warn on static type mismatches and function
2885 redefinition. (Currently it does not work with high DEBUG level.)
2886 * minor incompatible change: VALUES declaration is disabled.
2887 * When issuing notes, the compiler now signals a condition of type
2888 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2889 restart for use in user handlers. It is expected that the
2890 COMPILER-NOTE condition will eventually become a condition
2891 supertype to a hierarchy of note types, which will then be
2892 handleable in a similar fashion. However, at the moment, no such
2893 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2894 but it's an implementation detail, not a classification for the
2896 * Changes in type checking closed the following bugs:
2897 ** type checking of unused values (192b, 194d, 203);
2898 ** template selection based on unsafe type assertions (192c, 236);
2899 ** type checking in branches (194bc).
2900 * A short form of VALUES type specifier has ANSI meaning (it has
2901 increased the number of situations when SBCL cannot perform type
2903 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2904 keywords or constants is permissible.
2905 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2906 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2907 classes. (thanks to Antonio Martinez)
2908 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2909 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2910 argument to a pretty-print stream. Instead, it leaves any such
2911 operations to DESCRIBE-OBJECT methods.
2912 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2914 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2915 lambda lists. (thanks to David Lichteblau)
2916 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2917 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2918 (thanks to Gerd Moellmann)
2919 * fixed bug 246: increased compilation speed of long
2920 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2922 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2924 * DEFCONSTANT now throws a condition of type
2925 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2926 constant to a non-EQL value; CONTINUE and ABORT restarts
2927 respectively change and preserve the value.
2928 * fixed bug 63: The code walker, part of the implementation of CLOS,
2929 is now better at handling symbol macros.
2930 * bug fix: There is no longer an internal implementation type named
2931 CL:LENGTH. (reported by Raymond Toy)
2932 * bug fix: In macro-like defining macros/special operators the
2933 implicit block does not enclose the lambda list.
2934 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2935 enclosed in parentheses) are not suitable as type specifiers, and
2936 their use properly signals an error now.
2937 * bug fix: An argument count mismatch for a type specifier in code
2938 being compiled no longer causes an unhandled error at compile
2939 time, but signals a compile-time warning.
2940 * fixed simple vector readable printing
2941 * bug fix: DESCRIBE takes more care over whether the class
2942 precedence list slot of a class is bound before accessing it.
2943 (reported by Markus Krummenacker)
2944 * bug fix: FORMATTER can successfully compile pretty-printer format
2945 strings which use variants of the ~* directive inside.
2946 * bug fix: SEARCH now applies its TEST predicate to the elements of
2947 the arguments in the correct order. (thanks to Wolfhard Buss)
2948 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2949 in type inference. (thanks to Robert E. Brown)
2950 * bug fix: Reading in symbols with an explicit package name of ""
2951 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2952 * fixed some bugs revealed by Paul Dietz' test suite:
2953 ** NIL is now allowed as a structure slot name.
2954 ** Arbitrary numbers, not just REALs, are allowed in certain
2955 circumstances in LOOP for-as-arithmetic clauses.
2956 ** Multiple class redefinitions before slot access no longer
2957 causes a type error.
2958 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2959 association between the name and a class.
2960 ** Generic functions with non-standard method-combination and over
2961 five methods all of which return constants no longer return NIL
2962 after the first few invocations. (thanks to Gerd Moellmann)
2963 ** CALL-NEXT-METHOD with no arguments now passes the original
2964 values of the arguments, even in the presence of assignment.
2965 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2967 ** DISASSEMBLE works with closures and funcallable instances.
2968 ** ADD-METHOD now returns the generic function, not the new method.
2969 ** FIND-METHOD signals an error if the lengths of the specializers
2970 is incompatible with the generic function, even if the ERRORP
2972 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2973 which its argument is a member.
2974 ** DEFCLASS only redefines the class named by its class-name
2975 argument if that name is the proper name of the class;
2976 otherwise, it creates a new class.
2977 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2978 of the UNBOUND-SLOT condition to the name of the slot.
2979 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2980 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2981 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2982 SLOT-MAKUNBOUND in the specified fashion.
2984 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2985 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2986 cross-compilation host. As a consequence, we can now bootstrap our
2987 way up to SBCL starting with a bare gcc toolchain and human-readable
2988 source code (first the source to CLISP, then the source to SBCL).
2989 * A contributed module containing a partial implementation of the
2990 simple-streams interface has been included. (thanks to Rudi
2992 * A contributed module implementing the RFC1321 Message Digest
2993 Algorithm, known as MD5, has been included.
2994 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2995 no longer has any effect, as the code controlled by this feature
2996 has been deleted. (As far as we know, no-one has ever built using
2997 this feature, and its semantics were confused in any case).
2998 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2999 work on conditions (as required by the ANSI specification),
3000 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3001 expected behaviour on conditions. Users should note, however,
3002 that such behaviour is not required by the ANSI specification,
3003 and so use of this behaviour may render their code unportable.
3004 * fixed some bugs revealed by Paul Dietz' test suite:
3005 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3007 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3008 ** SXHASH on condition objects no longer returns NIL.
3009 ** :ALLOCATION :CLASS slots are better treated; their values are
3010 updated on class redefinition, and initforms inherited from
3011 superclasses are applied.
3012 ** REMOVE-METHOD returns its generic function argument even when
3013 no method was removed.
3014 ** SHARED-INITIALIZE now initializes the values of the requested
3015 slots, including those with :ALLOCATION :CLASS.
3016 ** ALLOCATE-INSTANCE now works on structure classes defined via
3017 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3019 ** SLOT-EXISTS-P now works on conditions, as well as structures
3021 ** MAKE-LOAD-FORM now has the required methods on
3022 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3023 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3024 keyword, and now implements the SLOT-NAMES argument.
3025 ** methods with &OPTIONAL arguments no longer allow too many
3026 arguments to be passed in the call without error.
3027 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3028 option is consistent with the required arguments of the generic
3029 function lambda list.
3030 * bug fix: REQUIRE accepts a string designator. (Thanks to
3032 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3033 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3034 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3035 * bug fix: The compiler now checks for duplicated variables in macro
3037 * bug fix: SETQ on globals returns the correct value.
3038 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3039 not a COMPILER-ERROR (followed by some other strange error on
3040 choosing the CONTINUE restart).
3041 * bug fix: make.sh and friends are now more consistent in the way that
3042 they look for GNU "make".
3044 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3045 * experimental native threads support (on x86 Linux >=2.4 only).
3046 This is not compiled in by default: you need to add :SB-THREAD to
3047 the target features. See the "Beyond ANSI" chapter of the manual
3049 * fix for longstanding nonANSIism: The old distinction between
3050 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3051 The return value from CL:FIND-CLASS is now a CLOS class, and
3052 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3053 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3054 * An interface to the MetaObject Protocol, as described in Kiczales,
3055 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3056 MIT Press, 1991, is available from the SB-MOP package.
3057 * incompatible change: the SB-PCL package should now be considered
3058 a private implementation detail, and no longer a semi-private MOP
3060 * minor incompatible change: due to rearrangement for threads, the
3061 control stack and binding stack are now allocated at arbitrary
3062 addresses instead of being hardcoded per-port. Users affected by
3063 this probably have to be doing advanced things with shared
3064 libraries, and will know who they are.
3065 * minor incompatible change: Previously, all --eval forms used were
3066 processed with READ before any of them were processed with EVAL.
3067 Now each --eval form is processed with both READ and EVAL before
3068 the next --eval form is processed. (Thus package operations like
3069 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3070 work as the user might reasonably expect.)
3071 * minor incompatible change: *STANDARD-INPUT* is now only an
3072 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3074 * minor incompatible change: Y-OR-N-P is now character-oriented,
3075 not line oriented. Also, YES-OR-NO-P now works without errors.
3076 (thanks to Antonio Martinez)
3077 * sb-aclrepl module improvements: an integrated inspector, added
3078 repl features, and a bug fix to :trace command.
3079 * Known functions, which cannot be open coded by the backend, are
3080 considered to be able to check types of their arguments. (fixing
3081 a bug report by Nathan J. Froyd)
3082 * fixed a bug in computing method discriminating functions: It is
3083 now possible to define methods specialized on classes which have
3084 forward-referenced superclasses. (thanks to Gerd Moellmann)
3085 * fixed evaluation order in optional entries (reported by Gilbert
3087 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3088 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3089 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3090 required-by-AMOP NAME argument, as well as CLASS and
3091 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3092 * fixed bug 20: DEFMETHOD can define methods using names that are
3093 not the proper names of classes to designate class specializers.
3094 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3095 Unix file descriptors, instead of blowing up. (thanks to Antonio
3097 * Garbage collection refactoring: user-visible change is that a
3098 call to the GC function during WITHOUT-GCING will not do garbage
3099 collection until the end of the WITHOUT-GCING. If you were doing
3100 this you were probably losing anyway.
3101 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3102 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3103 TYPEP the latter but not the former.
3104 * The compiler issues a full WARNING for calls to undefined functions
3105 with names from the CL package.
3106 * MAP-INTO for a vector destination is open coded. (reported by
3107 Brian Downing on c.l.l)
3108 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3109 documentation string.
3110 * fixed some bugs revealed by Paul Dietz' test suite:
3111 ** COPY-ALIST now signals an error if its argument is a dotted
3113 ** Condition slots are now accessed more correctly in the presence
3114 of multiple initargs for a given slot.
3115 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3116 exclude restarts of the same name associated with a different
3118 ** DEFCLASS of forward-referenced classes with another
3119 forward-referenced class in the superclasses list no longer
3121 ** Condition slots are now initialized once each, not multiple
3122 times. (thanks to Gerd Moellmann)
3123 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3125 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3126 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3127 arguments contain duplicated elements.
3128 ** RESTART-CASE understands local macros.
3129 ** RESTART-CASE associates exactly its own restarts with a condition.
3130 ** ENDP in safe mode checks its argument to be of type LIST.
3131 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3132 in question is unbound.
3133 ** Optimized MAKE-INSTANCE functions no longer cause internal
3134 assertion failures in the presence of duplicate initargs.
3135 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3136 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3137 propagated even in the ctor optimized implementation of
3139 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3141 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3142 * a better implementation of SXHASH on (simple) bit vectors,
3143 measured both in execution speed and in distribution of results
3144 over the positive fixnums, has been installed. Likewise, a better
3145 implementation of EQUAL for simple bit vectors is now available.
3146 * fixed CEILING optimization for a divisor of form 2^k.
3147 * fixed bug 240 (emitting extra style warnings "using the lexical
3148 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3149 by Antonio Martinez)
3150 * fixed SXHASH, giving different results for NIL depending on type
3151 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3152 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3153 variable. (found by Rolf Wester)
3154 * SBCL does not ignore type declarations for special
3155 variables. (reported by rif on c.l.l 2003-03-05)
3156 * some bug fixes in contrib/sb-aclrepl/
3157 * fixed some bugs revealed by Paul Dietz' test suite:
3158 ** a bug in the CONS type specifier, whereby the CAR and CDR
3159 types got intertwined, has been fixed;
3160 ** the type system is now able to reason about the interaction
3161 between INTEGER and RATIO types more completely;
3162 ** APPEND, [N]REVERSE and NRECONC check that those their
3163 arguments, which must be proper lists, are really so;
3164 ** An array specialized to be unable to hold elements has been
3165 implemented, as required -- yes, really -- by ANSI;
3166 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3167 on malformed property lists;
3169 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3170 * incompatible packaging change: in line with Unix convention,
3171 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3172 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3173 * REQUIRE and PROVIDE are now optionally capable of doing something
3174 useful. See the documentation string for REQUIRE.
3175 * infrastructure for a managed SBCL contrib system: contributed
3176 modules in this release include:
3177 ** the ASDF system definition facility;
3178 ** an interface to the BSD Sockets API;
3179 ** an ACL-like convenience interface to the repl;
3180 (thanks to Kevin Rosenberg)
3181 ** an implementation of ROTATE-BYTE, with efficient implementation
3183 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3184 gives a closer approximation to the right answer than previously.
3185 (thanks to Raymond Toy)
3186 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3187 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3188 optional environment arguments, as required by ANSI.
3189 * fixed bugs in other functions taking environment objects, allowing
3190 calls with an explicit NIL environment argument to be compiled
3192 * fixed bug 228: primary return values from
3193 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3194 COMPILE or FUNCTION.
3195 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3196 :TYPE will no longer signal errors on innocuous objects.
3197 * fixed bug 231b: SETQ is better at respecting type declarations in
3198 the lexical environment.
3199 * fixed a bug in DEFCLASS: classes named by symbols with no or
3200 unprintable packages can now be defined.
3201 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3202 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3203 * fixed compiler failure related to checking types of functions.
3204 (reported by Robert E. Brown)
3205 * the compiler is now much more consistent in its error-checking
3206 treatment of bounding index arguments to sequence functions: in
3207 (SAFETY 3) code, errors will be signalled in almost all cases if
3208 invalid sequence bounding indices are passed to functions defined
3209 by ANSI to operate on sequences.
3210 * fixed a bug in the build procedure: documentation of SBCL-specific
3211 packages is now preserved and available in the final Lisp image.
3212 * lifted FDEFINITION lookup out of loops in the implementation of
3213 many list operations. (thanks to Robert E. Brown)
3214 * fixed a bug in the reader: the #n# reader macro now works for
3215 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3216 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3217 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3218 E. Brown and Tony Martinez respectively)
3219 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3220 SLOT-BOUNDP: in particular, it is now able to optimize them much
3221 better, and is now not vulnerable to having packages renamed.
3222 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3223 always called when a slot is not present in an instance. (thanks
3225 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3226 longer signal an error on unreading a character following EOF on
3227 the previous constituent stream. (thanks to Tony Martinez)
3228 * fixed some bugs revealed by Paul Dietz' test suite:
3229 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3230 not just nonnegative fixnums;
3231 ** the logical bit-array operators such as BIT-AND now accept an
3232 explicit NIL for their "opt-arg" argument (to indicate a
3233 freshly-consed result bit-array);
3234 ** ELT now signals an error on an invalid sequence index in safe
3236 ** the type system is now cleverer about negations of numeric
3237 types, and consequently understands the BIGNUM and RATIO types
3239 ** the type system is now cleverer about the interaction between
3240 INTEGER and RATIO types: while bugs still remain, many more
3241 cases are accurately computed;
3242 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3243 if it is in the last clause;
3244 ** CONSTANTLY now correctly returns a side-effect-free function in
3246 ** DECLARE is no longer treated as a special-operator; in
3247 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3248 * incremented fasl file version number due to the change in the
3249 DEFSTRUCT-SLOT-DESCRIPTION structure.
3251 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3252 * minor incompatible change: code processed by the "interpreter" or
3253 EVAL now has a compilation optimization policy of (DEBUG 2)
3254 (changed from (DEBUG 1)) to improve debuggability of interactive
3255 development, and to allow the use of the debug RETURN command in
3257 * an experimental implementation of the RETURN command for the
3258 debugger has been included. (thanks to Frederik Kuivinen)
3259 * fixed bug 62: constraints were not propagated into a loop.
3260 * fixed bug in embedded calls of SORT (reported and investigated by
3262 * fixed some bugs revealed by Paul Dietz' test suite:
3263 ** printing and reading of arrays with some dimensions having
3264 length 0 (thanks to Gerd Moellmann);
3265 ** BOA constructor with &AUX argument without a default value does
3266 not cause a type error;
3267 ** CONSTANTP now returns true for all self-evaluating objects.
3269 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3270 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3271 accessors that are related by inheritance, as specified in the
3272 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3273 Valtteri Vuorikoski)
3274 * The compiler is now able to inline functions that were defined in
3275 a complex lexical environment (e.g. inside a MACROLET).
3276 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3278 * Support for the upcoming FreeBSD-5.0 release has been included.
3279 (thanks to Dag-Erling Smorgrav)
3280 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3281 effect when it is not in a toplevel context.
3282 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3283 stand a better chance of being correct. (thanks to Gerd
3285 * fixed bug in COERCE, which now signals an error on coercing a
3286 rational to a bounded real type which excludes the expected
3288 * The compiler is now able to derive types more accurately from the
3289 COERCE and COMPILE functions.
3290 * fixed bug 223: functional binding is considered to be constant
3291 only for symbols in the CL package.
3292 * fixed bug 231: SETQ did not check the type of a variable being set
3293 (reported by Robert E. Brown)
3294 * A new optimization for MAKE-INSTANCE has been included, fixing
3295 various bugs (including relating to :ALLOCATION :CLASS slots and
3296 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3298 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3299 ** As required by ANSI, LOOP now disallows anonymous collection
3300 clauses such as COLLECT I in conjunction with aggregate boolean
3301 clauses such as THEREIS (= I 1);
3302 ** LOOP now signals an error when any variable is reused in the
3303 same loop (including the potentially useful construct analogous
3304 to WITH A = 1 WITH A = (1+ A);
3305 ** IT is only a special loop symbol within the first clause of a
3306 conditional loop clause;
3307 ** LOOP with a typed iteration variable over a hashtable now
3308 signals a type error iff it should.
3309 * fixed some other bugs revealed by Paul Dietz' test suite:
3310 ** FILE-STREAM now names the class previously known as FD-STREAM;
3311 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3312 argument) no longer signals an error;
3313 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3314 of the slot symbol, rather than using the current package
3315 ((:CONC-NAME "") continues to intern the slot's name in the
3317 * incremented fasl file version number, because of the incompatible
3318 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3319 of the new implementation of DEFINE-COMPILER-MACRO.
3321 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3322 * Support for building SBCL for MIPS platforms running in
3323 little-endian mode has now been checked in, and basic
3324 functionality on said platforms verified.
3325 * minor incompatible change: PCL now records the pathname of a file
3326 in which methods and the like are defined, rather than its
3328 * minor incompatible change: TRUENAME now considers the truename of
3329 a file naming a directory to be the pathname with :DIRECTORY
3330 component indicating that directory.
3331 * minor incompatible change: a NAMED clause in the extended form of
3332 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3333 reason for the previous behaviour is unclear.
3334 * more systematization and improvement of CLOS and MOP conformance
3335 in PCL (thanks to Gerd Moellman and Pierre Mai):
3336 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3338 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3339 primary methods with no specializers;
3340 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3342 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3343 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3344 FORWARD-REFERENCED-CLASSes; error reporting on
3345 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3347 ** SXHASH on CLOS instances now uses a slot internal to the
3348 instance to return different numbers on distinct instances,
3349 while preserving the same return value through invocations of
3351 ** DEFMETHOD signals errors when methods with longer incongruent
3352 lambda lists are added to generic functions;
3353 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3354 CLASS, as specified in AMOP;
3355 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3356 on the order returned by the primary method for classes of
3357 class STANDARD-CLASS;
3358 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3359 * fixed some bugs shown by Paul Dietz' test suite:
3360 ** DOLIST puts its body in TAGBODY;
3361 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3363 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3364 value producing form;
3365 ** if more variables are given to PROGV than values, extra
3366 variables are bound and made to have no value;
3367 ** NSUBSTITUTE on list arguments gets the right answer with
3369 ** ELT signals an error of type TYPE-ERROR when the index argument
3370 is not a valid sequence index;
3371 ** LOOP signals (at macroexpansion time) an error of type
3372 PROGRAM-ERROR when duplicate variable names are found;
3373 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3374 ** FUNCALL of special-operators now cause an error of type
3376 ** PSETQ now works as required in the presence of side-effecting
3377 symbol-macro places;
3378 ** NCONC accepts any object as its last argument;
3379 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3381 ** loop-for-as-package does not require a package to be explicitely
3383 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3385 * fixed bug 166: compiler preserves "there is a way to go"
3386 invariant when deleting code.
3387 * fixed bug 172: macro lambda lists with required arguments after
3388 &REST arguments now cause an error to be signalled. (thanks to
3390 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3391 bug. (thanks to Matthew Danish)
3392 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3394 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3395 arguments are lexically rebound. (thanks to Gerd Moellmann and
3397 * fixed bug 194: error messages are now more informative when there
3398 is no primary method applicable in a call to a generic
3399 function. (thanks to Gerd Moellmann)
3400 * fixed bug in command line argument checking (thanks to Julian
3402 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3403 (thanks to Matthew Danish)
3404 * incremented fasl file version number, because of the
3405 SXHASH-related changes in the layout of CLOS data structures
3407 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3408 * minor incompatible change: The runtime (the Unix executable named
3409 "sbcl") is now much pickier about the .core files it will load.
3410 Essentially it now requires .core files to descend from the same
3411 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3412 as the runtime does. (The intent is to prevent the crashes which
3413 can occur, and which can even be reported as mysterious failures,
3414 when people patch the sources or change the build parameters
3415 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3416 sbcl and .core files.)
3417 * fixed bug: VALUES-LIST is no longer optimized away.
3418 * fixed bug 142: The FFI conversion of C string values to Lisp
3419 string values no longer conses excessively. (thanks to Nathan
3420 Froyd porting Raymond Toy's fix to CMU CL)
3421 * began to systematize and improve MOP conformance in PCL (thanks to
3422 Nathan Froyd, Gerd Moellman and Pierre Mai):
3423 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3425 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3426 ** FINALIZE-INHERITANCE is now called on class finalization;
3427 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3428 argument precedence order.
3429 * fixed bug 202: The compiler no longer fails on functions whose
3430 derived types contradict their declared type.
3431 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3432 so it can be non-toplevel.
3433 * The fasl file version number has changed (because of the new
3434 implementation of DEFMACRO).
3435 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3436 safe code, that any length requirement by their type-specifier
3437 argument is valid. The exceptions to this are described in bug
3439 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3440 functions now check the types of their inputs as required by ANSI.
3441 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3442 introduced symbol is DECLAREd to be SPECIAL.
3443 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3444 * fixed bug: :COUNT argument to sequence functions may be negative
3445 * fixed bug: body of DO-SYMBOLS may contain declarations
3446 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3447 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3448 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3449 symbol macro only once
3450 * fixed printing of call frame when argument list is unavailable
3451 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3452 * compiler no longer signals WARNING on unknown keyword
3455 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3456 * A beta-quality port to the mips architecture running Linux,
3457 based on the old CMUCL backend, has been made. It has been tested
3458 on a big-endian kernel, and works sufficiently well to be able to
3459 rebuild itself; it has not been tested in little-endian mode.
3460 * fixed an inconsistency between gencgc.c and purify.c which made
3461 dumping/loading .core files unreliable
3462 * fixed bug 120a: The compiler now deals correctly with IFs where
3463 the consequent is the same as the alternative, instead of
3464 misderiving the return type. (thanks to Alexey Dejneka)
3465 * fixed bug 113: Logical pathnames are now dumpable (the logical
3466 host is resolved at load-time, throwing an error if it is not
3468 * fixed bug 174: FORMAT's error message is slightly clearer when a
3469 non-printing character is used in a format directive.
3470 * fixed several bugs in compiler checking of type declarations, i.e.
3471 violations of the Python "declarations are assertions" principle
3472 (thanks to Alexey Dejneka)
3473 * fixed several bugs in PCL's error checking (thanks to Gerd
3475 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3477 * fixed bug in compilation of functions as first class values
3478 (thanks to Antonio Martinez-Shotton)
3479 * The compiler's handling TYPE-ERRORs which it can prove will
3480 inevitably happen at runtime has been cleaned up and corrected
3481 in several ways. (thanks to Alexey Dejneka)
3482 * improved argument type checking for various basic arithmetic
3483 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3484 lavished on them in the past that they can be compiled in many
3485 ways in different special cases
3486 * fixed bug 181: compiler checks validity of user supplied type
3488 * cleaned up code flushing in optimization: Function calls which
3489 should signal errors for safety purposes (e.g. which ANSI says
3490 should signal errors when their arguments are of incorrect type)
3491 are no longer optimized away.
3492 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3493 * incremented fasl file version number, because changes in the
3494 implementation of sequence functions like COERCE caused
3495 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3496 in old inline expansions) to become undefined. (Actually these
3497 changes were later undone, so we might very well be binary
3498 compatible with 0.7.7 after all, but leaving the version number
3499 incremented seemed like the simplest and most conservative
3502 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3503 * An alpha-quality port to the parisc architecture running Linux,
3504 based on the old CMUCL backend, has been made. This, even more so
3505 than the other backends, should be considered still a work in
3506 progress; known problems include that the Linux kernel in 64-bit
3507 mode does not propagate the correct sigcontext structure to
3508 userspace, and consequently SBCL on a parisc64 kernel will not
3510 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3511 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3512 suppressed.) Also now that the compiler is looking at declarations
3513 in the environment, it checks optimization declarations as well,
3514 and suppresses inlining when (> DEBUG SPEED).
3515 * More fixes have been made to treatment of floating point exception
3516 treatment and other Unix signals. In particular, floating point
3517 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3518 * The detection and handling of control stack exhaustion (infinite
3519 or very deeply nested recursion) has changed. Stack exhaustion
3520 detection is now done by write-protecting pages at the OS level
3521 and applies at all optimization settings; when found, a
3522 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3523 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3524 be caught using IGNORE-ERRORS.
3525 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3526 correctly when transforming recursive LABELS functions to LETs.
3527 (thanks to Alexey Dejneka)
3528 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3529 that are names of constants or global variables.
3530 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3531 alien routines with docstrings.
3532 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3533 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3535 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3536 (thanks to Eric Marsden)
3537 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3538 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3539 * Bug fix: The SPARC backend can now compile functions involving
3540 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3541 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3542 to a functions expecting a PATHNAME-DESIGNATOR.
3543 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3544 lambda lists. (thanks to Alexey Dejneka)
3545 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3546 functions so that EXPORTing the name of the class doesn't cause
3547 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3548 to Antonio Martinez for reporting this)
3549 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3550 constant-folded without failing an assertion. (thanks to Einar
3551 Floystad Dorum for reporting this)
3552 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3553 types are now dealt with more correctly by the compiler.
3554 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3555 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3556 behaviour analogous to RENAME-FILE. This puts its behaviour more
3557 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3559 * The fasl file version number has changed again. (because of the
3560 bug fix involving the names of PCL MAKE-INSTANCE functions)
3562 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3563 * bug fix: Floating point exceptions are treated much more
3564 consistently on the x86/Linux and PPC/Linux platforms.
3565 * Array initialization with :INITIAL-ELEMENT is now much faster for
3566 cases when the compiler cannot open code the array creation, but
3567 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3568 array accesses have also seen a speed increase.
3569 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3570 specified. (This was at the root of some bad interactions between
3571 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3573 * bug fix: Internal error arguments for undefined functions are now
3574 computed correctly on the PPC/Linux platform.
3575 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3576 Raymond Toy's patch for CMU CL)
3577 * Support for the Solaris 9 operating environment has been included
3578 (thanks to Daniel Merritt)
3579 * A very ugly but hopefully complete draft of the missing FFI chapter
3580 of the manual has been created by reformatting the corresponding
3581 CMU CL manual chapter into (currently very ugly and incoherent)
3582 DocBook and bringing it up to date for SBCL behavior. Thus, the
3583 manual is now essentially complete, at least by my extreme
3584 once-and-only-once standards, whereby it's acceptable to refer to
3585 the doc strings of SB-EXT functions as the primary documentation.
3586 * The fasl file version number has changed again, due to cleanup of
3587 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3588 is no longer a static symbol.)
3590 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3591 * SBCL now builds with OpenMCL (version 0.12) as the
3592 cross-compilation host; also, more progress has been made toward
3593 bootstrapping under CLISP.
3594 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3596 * bug 158 fixed: The compiler can now deal with integer loop
3597 increments different from 1; fixing this turned out also to fix
3599 * bug 169 fixed: no more bogus warnings about using lexical bindings
3600 despite the presence of perfectly good SPECIAL declarations (thanks
3601 to David Lichteblau)
3602 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3603 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3604 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3606 * bug 180 fixed: Method combination specifications no longer ignore
3607 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3608 * bug fix: Structure type predicate functions now check their argument
3609 count as they should.
3610 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3611 correctly. (thanks to Pierre Mai)
3612 * minor incompatible change: The --noprogrammer option is deprecated
3613 in favor of the new --disable-debugger option, which is very similar.
3614 (The major difference is that it takes effect at a slightly different
3615 time at startup, causing handling of errors in --sysinit and
3616 --userinit files will be affected differently.) The
3617 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3618 been added to allow this functionality to be controlled from ordinary
3619 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3620 maintainers, who might want to run non-interactive scripts to
3621 build SBCL cores which will later be used interactively.)
3622 * minor incompatible change: The LOAD function no longer, when given
3623 a wild pathname to load, loads all files matching that pathname.
3624 Instead, an error of type FILE-ERROR is signalled.
3626 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3627 * bug 147 fixed: The compiler preserves its block link/count
3628 invariants more correctly now so that it doesn't crash. (thanks
3630 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3632 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3633 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3634 no-value-for-this-character as NIL (as specified by ANSI).
3635 * HOST-NAMESTRING on physical pathnames now returns a string that is
3636 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3637 (thanks to Christophe Rhodes)
3638 * The Alpha port handles icache flushing more correctly. (thanks to
3640 * More progress has been made toward bootstrapping under CLISP. (thanks
3641 to Christophe Rhodes)
3642 * The fasl file format has changed again, because dynamic loading
3643 on OpenBSD (which has non-ELF object files) motivated some cleanups
3644 in the way that foreign symbols are transformed and passed around.
3645 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3646 is no longer treated as whitespace by the reader, but instead as
3647 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3648 |A
\7fB|, instead of A as it used to.
3650 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3651 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3652 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3653 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3654 this as of 0.7.1.45, but was left out of the previous news section
3655 (thanks to Dan Barlow)
3656 * SBCL now runs on the Solaris operating system on SPARC architectures
3657 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3658 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3659 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3660 * SPARC backend cleanups, allowing builds of cores optimized for V8
3661 and V9 SPARCS, and also emission of code targeted to a particular
3662 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3664 * SBCL is closer to bootstrapping under CLISP, thanks to various
3665 fixes by Christophe Rhodes.
3666 * The fasl file format has changed again, to allow the compiler's
3667 INFO database to support symbol macros.
3668 * The user manual (in doc/) is formatted into HTML more nicely.
3669 (thanks to coreythomas)
3670 * The system is smarter about SUBTYPEP relationships, especially
3671 those involving NOT types (including types such as ATOM which are
3672 represented internally using NOT types). Thus SUBTYPEP is less
3673 likely to return (VALUES NIL NIL) in general, and in particular
3674 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3676 * The fasl file format has changed again, because the internal
3677 representation of types now includes a new slot to support the new
3678 SUBTYPEP-of-NOT-types logic.
3679 * (not a change in the main branch of SBCL, but a related prototype
3680 which can hopefully be merged into the main branch of SBCL in the
3681 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3682 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3683 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3684 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3685 dispatch macro character combinations. (thanks to Alexey Dejneka)
3686 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3687 that superficially look like logical namestrings correctly.
3688 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3690 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3691 default supplied-p) for &optional and &key arguments. (thanks to
3694 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3695 * incompatible change: The compiler is now less aggressive about
3696 tail call optimization, doing it only when (> SPACE DEBUG) or
3697 (> SPEED DEBUG). (This is an incompatible change because there are
3698 programs which relied on the old CMU-CL-style behavior to optimize
3699 away their unbounded recursion which will now die of stack overflow.)
3700 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3701 for non-GENCGC systems has been increased to 20M (since that
3702 seems much closer to the likely performance optimum for modern
3703 systems than the old 4M value was)
3704 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3705 and *DEBUG-PRINT-LEVEL*
3706 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3707 of CMU CL's support for SPARC, and various endianness and other
3708 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3709 * new syntactic sugar for the Unix command line: --load foo.bar is now
3710 an alternate notation for --eval '(load "foo.bar")'.
3712 ** The system now detects stack overflow and handles it gracefully,
3713 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3714 optimization settings. (This is a good thing in general, and
3715 its introduction in this version should be particularly timely
3716 for anyone whose code fails because of suppression of tail
3718 ** The system now hunts for the C variable "environ" in a more
3719 devious way, to avoid segfaults when the C library version differs
3720 between compile time and run time. (thanks to Christophe Rhodes)
3721 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3722 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3723 ** The compiler no longer issues bogus style warnings for undefined
3724 classes in the same source file as the DEFCLASSes which defined
3725 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3727 ** fixes in CONDITION class precedence list for undefined function
3728 errors (thanks to Alexei Dejneka)
3729 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3730 correctly. (thanks to Dan Barlow)
3731 ** portability fixes aiming at bootstrapping under CLISP (thanks
3732 to Dave McDonald and Christophe Rhodes)
3733 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3734 ** fixes in type translation and and type inference (thanks to
3736 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3737 ** various fixes in the new ports (thanks to Dan Barlow)
3738 * several changes related to debugging:
3739 ** suppression of tail recursion, as noted above
3740 ** stack overflow detection, as noted above
3741 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3742 is now the default. (For some time encapsulation has been more
3743 reliable than the breakpoint-based :ENCAPSULATE NIL
3744 implementation, at least on X86 systems; and I just noticed that
3745 encapsulation also seems closer to the spirit of the ANSI
3748 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3750 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3751 up properly again. (There was a packaging bug in 0.7.0 which
3752 left their definitions in SB-SYS::LOAD-FOREIGN and
3753 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3754 vital for most things which interface to C-level interfaces,
3755 like extensions working with sockets or databases or
3756 Perl-compatible regexes or whatever, and the need to fix
3757 this bug is the main reason that 0.7.1 was released so
3759 ** DEFGENERIC is now choosier about the methods it redefines, so that
3760 reLOADing a previously-LOADed file containing DEFGENERICs does
3761 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3762 cycle now works as it should. (thanks to Alexey Dejneka)
3763 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3764 fixed by Christophe Rhodes. (He actually submitted this patch
3765 months ago, and I delayed until after 0.7.0.)
3766 ** Bug 111 (internal compiler confusion about runtime checks on
3767 FUNCTION types) was fixed by Alexey Dejneka.
3768 * Some internal cleanups (getting rid of variables which aren't
3769 needed now that the byte interpreter is gone) caused the fasl
3770 file format number to change again.
3772 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3773 * major incompatible change: The default fasl file extension, i.e. the
3774 default extension for files produced by COMPILE-FILE, has changed
3775 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3777 ** There are many changes in the implementation of the compiler.
3778 SBCL is now essentially a compiler-only implementation of ANSI
3779 Common Lisp. EVAL still "interprets" a few special cases, but
3780 almost all the interesting cases are handled by creating
3781 a LAMBDA expression, calling COMPILE on it, then calling
3782 FUNCALL on the result.
3783 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3784 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3785 Since the code is newer, there might still be some new bugs
3786 (though not as many as before Martin Atzmueller's fixes:-). But
3787 the new code is substantially simpler and clearer, and hopefully
3788 any remaining bugs will be simpler, less fundamental, and more
3789 fixable then the bugs in the old code.
3790 ** The revised compiler is still a little unsteady on its feet.
3792 *** The debugging information it produces (particularly the names
3793 of FUNCTION objects) is sometimes much less useful than what
3794 the old compiler produced.
3795 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3796 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3797 within a MACROLET) has been temporarily weakened.
3798 ** There are new compiler optimizations for various functions:
3799 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3800 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3801 *** the math functions TRUNCATE, FLOOR, and CEILING
3802 *** the function-of-all-trades COERCE
3803 Mostly these should be transparent, but there's one
3804 potentially-annoying problem (bug 117): when the compiler
3805 inline-expands a function and does type analysis on the result,
3806 it can create control paths which have type mismatches, and
3807 when it can't prove that those control paths aren't taken,
3808 it will issue WARNINGs about the type mismatches. This is
3809 a particular problem in practice for the new sequence functions.
3810 It's not clear how this should be fixed, and for now, a
3811 workaround is given in the entry for 117 in the BUGS file.
3812 ** (Because of the interaction between the two previous items --
3813 occasional inlining problems and new inline expansions -- some
3814 of the new sequence function optimizations won't really kick in
3815 completely until debugging information, and then inlining, are
3816 straightened out in some future version.)
3817 * minor incompatible changes:
3818 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3819 DIRECTORY no longer implicitly promotes NIL slots of its
3820 pathname argument to :WILD. In particular, when you ask for the
3821 contents of a directory (which you used to be able to do without
3822 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3823 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3824 ** changes in behavior that ANSI explicitly defines to be
3825 implementation dependent:
3826 *** The new compiler-only implementation still conforms with ANSI,
3827 but acts a little different than before. Besides the obvious
3828 changes in performance tradeoffs (that the cost per form passed
3829 to EVAL has gone up, and the cost per form executed by EVAL
3830 has gone down), the behavior of the system changes a little
3831 because there are no longer any interpreted function objects.
3832 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3833 e.g. doing COMPILE on the output of interactive DEFUN is
3835 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3837 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3838 has changed from (:CL) to NIL.
3839 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3840 e.g. #\Newline and #\Tab, don't have names specified in the
3841 ANSI Common Lisp standard, is now based on their ASCII symbolic
3842 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3843 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3844 are no longer used for output.
3845 ** changes in internal implementation constants:
3846 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3847 4 million. (If your application spends a lot of time GCing and
3848 you have a lot of RAM, you might want to experiment with
3849 increasing it even more.)
3850 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3851 However, almost all old code should still continue to work without
3852 immediate update, as SB-C-CALL is now a (deprecated) nickname
3854 ** Old operator names in the style DEF-FOO are now deprecated in
3855 favor of new corresponding names DEFINE-FOO, for consistency with
3856 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3857 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3858 internal symbols, but a few supported extensions like
3859 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3860 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3861 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3862 etc. as you get deeper into recursive calls to the debugger
3863 command loop, instead of the old "5]", "5]]", "5]]]"
3864 sequence. (I was motivated to do this when squabbles between
3865 ILISP and SBCL left me very deeply nested in the debugger. In the
3866 short term, this change will probably provoke more ILISP/SBCL
3867 squabbles, but hopefully it will be an improvement in the long run.)
3868 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3869 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3870 function, since declaiming return types involving aliens is
3871 (1) annoyingly messy to do by hand and (2) vital to efficient
3872 compilation of code which calls such functions.
3873 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3874 longer reexported by the SB-EXT package. They're solely useful
3875 for alien code, so it seems more logical that you should get
3876 them from the SB-ALIEN package, not in SB-EXT.
3877 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3878 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3879 features. Instead, the code that they used to control is always
3880 built into the system.
3881 * many other bug fixes
3882 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3883 advantage of the new EVAL-WHEN stuff and to clean them up in
3884 general, and they are now more ANSI-compliant in a number of
3885 ways. Martin Atzmueller is responsible for a lot of this.
3886 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3888 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3889 *** correct ERROR type for various file operations
3890 *** some fixes for Lisp streams
3891 *** DEFMETHOD syntax checking
3892 *** changing old weird representation of debug information as
3893 strings (which, among their other deficiencies, don't transform
3894 correctly when you rename packages, and don't change their
3895 print representation when you change things like *PACKAGE*
3896 and *PRINT-LENGTH*) to symbols and lists of symbols
3897 He also made several improvements and fixed several bugs in DESCRIBE.
3898 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3900 *** misbehavior of WRITE-STRING/WRITE-LINE
3901 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3902 and several other LOOP problems as well
3903 *** DIRECTORY when similar filenames are present
3904 *** DEFGENERIC with :METHOD options
3905 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3906 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3907 *** argument ordering in FIND with :TEST option
3908 *** mishandled package designator argument in APROPOS-LIST
3909 *** various problems in the backquote readmacro
3910 *** a bug in APROPOS
3911 *** probably some others that I'm not describing very well here,
3912 since the CVS log documents them by reference to sbcl-devel
3913 messages, and the SourceForge archives aren't working well.:-(
3914 ** Dan Barlow improved the Alpha port (and is making progress on the
3915 PPC port, for those of you who think different).
3916 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3917 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3918 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3919 continues to work on the SPARC port (for those of you in a position
3920 to look down upon our little PC-compatible boxes from a great height).
3921 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3922 and uses that copy, rather than the current dynamic value, when
3923 it's trying to decide whether to truncate output. Thus e.g.
3924 (let ((*print-lines* 50))
3925 (pprint-logical-block (stream nil)
3927 (let ((*print-lines* 8))
3928 (print (aref possiblybigthings i) stream)))))
3929 should now truncate the logical block only at 50 lines, instead of
3930 often truncating it at 8 lines, as it did before.
3931 * The doc/cmucl/ directory, containing old CMU CL documentation
3932 from the time of the fork, is no longer part of the base system.
3933 SourceForge has shut down its anonymous FTP service, and with it
3934 my original plan for distributing the old CMU CL documentation
3935 there. For now, if you need these files you can download an old
3936 SBCL source release and extract them from it.
3937 * The fasl file version number changed again, for dozens of reasons,
3938 some of which are apparent above.
3940 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3941 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3942 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3943 patch, so that DISASSEMBLE output is much nicer.
3944 * The code in the SB-PROFILE package now seems reasonably stable.
3945 I still haven't decided what the final interface should look like
3946 (I'd like PROFILE to interact cleanly with TRACE, since both
3947 facilities use function encapsulation) but if you have a need
3948 for profiling now, you can probably use it successfully with
3949 the current CMU-CL-style interface.
3950 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3951 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3952 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3953 link now returns the dangling link itself, and for similar
3954 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3955 link itself. (In these cases the old code signalled an error and
3956 looped endlessly, respectively.) Thus, DIRECTORY now works even
3957 in the presence of dangling and cyclic symbolic links.
3958 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3959 is now a supported extension again, since the consensus on
3960 sbcl-devel was that it can be useful for ordinary development
3961 work, not just for debugging SBCL itself.
3962 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3963 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3964 that currently-defined functions might be redefined later with
3965 different return types.
3966 * Hash tables can be printed readably, as inspired by CMU CL code
3967 of Eric Marsden and SBCL code of Martin Atzmueller.
3968 * better error handling in CLOS method combination, thanks to
3969 Martin Atzmueller porting Pierre Mai's CMU CL patches
3970 * more overflow fixes for >16Mbyte I/O buffers
3971 * A bug in READ has been fixed, so that now a single Ctrl-D
3972 character suffices to cause end-of-file on character streams.
3973 In particular, now you only need one Ctrl-D at the command
3974 line (not two) to exit SBCL.
3975 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3977 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3978 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3979 does the right thing.
3980 * The implementation of some type tests, especially for CONDITION
3981 types, is now tidier and maybe faster, due to CMU CL code
3982 originally by Douglas Crosher, ported by Martin Atzmueller.
3983 * Some math functions have been fixed, and there are new
3984 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3985 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3986 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3987 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3988 likely to be built into sbcl-0.7.0.)
3989 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3990 CL, ported to SBCL by Martin Atzmueller)
3991 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3992 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3993 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3994 * There's a new slam.sh hack to shorten the edit/compile/debug
3995 cycle for low-level changes to SBCL itself, and a new
3996 :SB-AFTER-XC-CORE target feature to control the generation of
3997 the after-xc.core file needed by slam.sh.
3998 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3999 COMPILE-FILE is no longer supported, so that now every function
4000 gets an entry point, so that block compilation looks a little
4001 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4002 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4003 returns the number of bytes consed since the system started,
4004 rather than the number consed since the first time the function
4005 was called. (The new definition parallels ANSI functions like
4006 CL:GET-INTERNAL-RUN-TIME.)
4007 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4008 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4009 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4010 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4011 * Fasl file version numbers are now independent of the target CPU,
4012 since historically most system changes which required version
4013 number changes have affected all CPUs equally. Similarly,
4014 the byte fasl file version is now equal to the ordinary
4017 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4018 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4019 is no longer recognized. I apologize for this, because it was
4020 listed in SB-EXT as a supported extension, but I found that
4021 its existing behavior was poorly specified, as well as incorrectly
4022 specified, and it looked like too much of a mess to straighten it
4023 out. I have enough on my hands trying to get ANSI stuff to work..
4024 * many patches ported from CMU CL by Martin Atzmueller, with
4025 half a dozen bug fixes in pretty-printing and the debugger, and
4026 half a dozen others elsewhere
4027 * fixed bug 13: Floating point infinities are now supported again.
4028 They might still be a little bit flaky, but thanks to bug reports
4029 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4030 as flaky as they were.
4031 * The --noprogrammer command line option is now supported. (Its
4032 behavior is slightly different in detail from what the old man
4033 page claimed it would do, but it's still appropriate under the
4034 same circumstances that the man page talks about.)
4035 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4036 are now supported, and enabled by default. Thus, the compiler can
4037 handle many floating point and complex operations much less
4038 inefficiently. (Thus e.g. you can implement a complex FFT
4040 * The compiler now detects type mismatches between DECLAIM FTYPE
4041 and DEFUN better, and implements CHECK-TYPE more correctly, and
4042 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4043 thanks to patches from Martin Atzmueller.
4044 * various fixes to make the cross-compiler more portable to
4045 ANSI-conforming-but-different cross-compilation hosts (notably
4046 Lispworks for Windows, following bug reports from Arthur Lemmens)
4047 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4048 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4049 fixed thanks to Pierre Mai's CMU CL patches.
4050 * fixes to keep the system from overflowing internal counters when
4051 it tries to use i/o buffers larger than 16M bytes
4052 * fixed bug 45a: Various internal functions required to support
4053 complex special functions have been merged from CMU CL sources.
4054 (When I was first setting up SBCL, I misunderstood a compile-time
4055 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4056 * improved support for type intersection and union, fixing bug 12
4057 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4058 more obscure bugs as well
4059 * some steps toward byte-compiling non-performance-critical
4060 parts of the system, courtesy of patches from Martin Atzmueller
4061 * Christophe Rhodes has made some debian packages of sbcl at
4062 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4063 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4064 stable, but are nonetheless usable. When he's ready, I'd be happy
4065 to add them to the SourceForge "File Releases" section. (And if
4066 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4067 * new fasl file format version number (because of changes in
4068 internal representation of (OR ..) types to accommodate the new
4069 support for (AND ..) types, among other things)
4071 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4072 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4074 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4076 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4077 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4078 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4079 * DESCRIBE now works on structure objects again.
4080 * Most function call argument type mismatches are now handled as
4081 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4082 know whether the function will be redefined before the call is
4083 executed. (The compiler could flag local calls with full WARNINGs,
4084 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4085 it doesn't keep track of enough information to know whether calls
4086 are local in this sense.)
4087 * Compiler output is now more verbose, with messages truncated
4088 later than before. (There should be some supported way for users
4089 to override the default verbosity, but I haven't decided how to
4090 provide it yet, so this behavior is still controlled by the internal
4091 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4092 src/compiler/ir1util.lisp.)
4093 * Fasl file format version numbers have increased again, because
4094 support for the Gray streams extension changes the layout of the
4095 system's STREAM objects.
4096 * The Gray subclassable streams extension now works, thanks to a
4097 patch from Martin Atzmueller.
4098 * The full LOAD-FOREIGN extension (not just the primitive
4099 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4100 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4101 but like most other programs, it defaults to copying the Unix
4102 environment from the original process instead of starting the
4103 new process in an empty environment.
4104 * Extensions which manipulate the Unix environment now support
4105 an :ENVIRONMENT keyword option which doesn't smash case or
4106 do other bad things. The CMU-CL-style :ENV option is retained
4107 for porting convenience.
4108 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4109 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4111 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4113 * A patch from Martin Atzmueller seems to have solved the SIGINT
4114 problem, and as far as we know, signal-handling now works cleanly.
4115 (If you find any new bugs, please report them!)
4116 * The system no longer defaults Lisp source file names to types
4117 ".l", ".cl", or ".lsp", but only to ".lisp".
4118 * The compiler no longer uses special default file extensions for
4119 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4120 seems to expect a single default extension for all compiled code,
4121 and there's no compelling reason to try to stretch the standard
4122 to allow two different extensions.) Instead, byte-compiled files
4123 default to the same extension as native-compiled files.
4124 * Fasl file format version numbers have increased again, because
4125 a rearrangement of internal implementation packages made some
4126 dumped symbols in old fasl files unreadable in new cores.
4127 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4128 many fewer weird special cases.
4129 * Bug #17 (differing COMPILE-FILE behavior between logical and
4130 physical pathnames) has been fixed, and some related misbehavior too,
4131 thanks to a patch from Martin Atzmueller.
4132 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4133 by Tim Moore, ported to SBCL by Martin Atzmueller.
4134 * Martin Atzmueller fixed several filesystem-related problems,
4135 including bug #36, in part by porting CMU CL patches, which were
4136 written in part by Paul Werkowski.
4137 * More compiler warnings in src/runtime/ are gone, thanks to
4138 more patches from Martin Atzmueller.
4139 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4142 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4144 * DESCRIBE now works on CONDITION objects.
4145 * The debugger now handles errors which arise when trying to print
4146 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4148 * The build system now uses an additional file, customize-target-features.lisp,
4149 to allow local modifications to the target *FEATURES* list. (The point of
4150 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4151 debugging features enabled, without having to worry about propagating it
4152 into everyone's system when I do a "cvs update".) When no
4153 customize-target-features.lisp file exists, the target *FEATURES* list
4154 should be constructed the same way as before.
4155 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4156 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4157 warn about behavior which is undefined under the ANSI spec. Note: This
4158 is specified by ANSI, but it's not very popular with programmers.
4159 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4160 macro in the SBCL sources for an example of a workaround which you
4161 might use to make portable ANSI-standard code which does what you want.
4162 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4163 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4164 when it's used as a non-toplevel form. (This is required in order
4165 to implement the DEFCONSTANT-EQX macro.)
4166 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4167 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4168 funny things, due to bugs in the implementation of EVAL-WHEN
4169 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4170 non-toplevel DEFCONSTANTs are uncommon.)
4171 * The core file version number and fasl file version number have been
4172 incremented, because the old noncompliant DEFCONSTANT behavior involved
4173 calling functions which no longer exist, and because I also took the
4174 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4175 * fixed bug 1 (error handling before read-eval-print loop starts), and
4176 redid debugger restarts and related debugger commands somewhat while
4178 ** The QUIT debugger command is gone, since it did something
4179 rather different than the SB-EXT:QUIT command, and since it never
4180 worked properly outside the main toplevel read/eval/print loop.
4181 Invoking the new TOPLEVEL restart provides the same functionality.
4182 ** The GO debugger command is also gone, since you can just invoke
4183 the CONTINUE restart directly instead.
4184 ** The TOP debugger command is also gone, since it's redundant with the
4185 FRAME 0 command, and since it interfered with abbreviations for the
4187 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4189 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4190 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4191 some porting work by Martin Atzmueller)
4192 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4193 with a boatload of patches which he ported to SBCL. Now that those
4195 ** The system tries to make sure that its low-priority messages
4196 are prefixed by semicolons, to help people who like to use
4197 syntax highlighting in their ILISP buffer. (This patch
4198 was originally due to Raymond Toy.)
4199 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4200 patches originally written by Raymond Toy.
4201 ** The compiler understands coercion between single-value and
4202 multiple-VALUES type expressions better, getting rid of some very
4203 weird behavior, thanks to patches originally by Robert MacLachlan
4204 and Douglas Crosher.
4205 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4206 lambda lists, thanks to a patch originally by Pierre Mai.
4207 ** The system no longer bogusly warns about "abbreviated type
4209 ** The compiler gets less confused by inlining and RETURN-FROM,
4210 thanks to some patches originally by Tim Moore.
4211 ** The system no longer hangs when dumping circular lists to fasl
4212 files, thanks to a patch originally from Douglas Crosher.
4213 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4214 undefined function error.
4215 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4216 in structure slot initforms) for now, documented workaround instead:-|
4217 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4218 * fixed bug 5: added stubs for various Gray stream functions called
4219 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4220 installed, at least appropriate type errors are generated
4221 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4222 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4223 consistently in DEFMETHOD forms.
4224 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4225 it doesn't seem to affect SBCL after all
4226 * The C runtime system now builds with better optimization and many
4227 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4229 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4231 * The system is now under CVS at SourceForge (instead of the
4232 CVS repository on my home machine).
4233 * The new signal handling code has been tweaked to treat register
4234 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4235 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4236 downstream, e.g. in debug-int.lisp, has implicit dependencies
4237 on the unsignedness of integer representation of machine words,
4238 and that caused the system to bomb out with infinite regress
4239 when trying to recover from type errors involving signed values,
4240 e.g. (BUTLAST '(1 2 3) -1).)
4241 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4242 (This was one of the bugs Peter Van Eynde reported back in July.)
4243 * The system now uses code inspired by Colin Walters' O(N)
4244 implementation of MAP (from the cmucl-imp@cons.org mailing
4245 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4246 inline the MAP operation, and there is more than one
4247 sequence argument to the MAP call (so that it can't just
4248 do ETYPECASE once and for all based on the type of the
4249 single sequence argument). (The old non-inline implementation
4250 of the general M-argument sequence-of-length-N case required
4251 O(M*N*N) time when any of the sequence arguments were LISTs.)
4252 * The QUIT :UNIX-CODE keyword argument has been renamed to
4253 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4254 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4256 * Raymond Wiker's patches to port dynamic loading from Linux to
4257 FreeBSD have been added.
4258 * The BUGS file is now more nearly up to date, thanks in large part
4259 to Martin Atzmueller's review of it.
4260 * The debugger now flushes standard output streams before it begins
4261 its output ("debugger invoked" and so forth).
4262 * The core version number and fasl file version number have both
4263 been incremented, because of incompatible changes in the layout
4265 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4266 used to not be called for a saved Lisp image.)
4267 * Martin Atzmueller's version of a patch to fix a compiler crash,
4268 as posted on sbcl-devel 13 September 2000, has been installed.
4269 * Instead of installing Martin Atzmueller's patch for the
4270 compiler transform for SUBSEQ, I deleted the compiler transform,
4271 and transforms for some similar consing operations.
4272 * A bug in signal handling which kept TRACE from working on OpenBSD
4274 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4275 away by constant folding
4276 * The system now defines its address space constants in one place
4277 (in the Lisp sources), and propagates them automatically elsewhere
4278 (through GENESIS and the sbcl.h file). Therefore, patching the
4279 address map is less unnecessarily tedious and error-prone. The
4280 Lisp names of address space constants have also been systematized.
4281 * CVS tags like dollar-Header-dollar have been removed from
4282 the sources, because they have never saved me trouble and
4283 they've been source of trouble working with patches and other
4284 diff-related operations.
4285 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4286 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4288 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4290 * The system has been ported to OpenBSD.
4291 * The system now compiles with a simple "sh make.sh" on the systems
4292 that it's supported on. I.e., now you no longer need to tweak
4293 text in the target-features.lisp-expr and symlinks in src/runtime/
4294 by hand, the make.sh takes care of it for you.
4295 * The system is no longer so grossly inefficient when compiling code
4296 involving vectors implemented as general (not simple) vectors (VECTOR T),
4297 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4298 which dares to use the various sequence functions on non-simple
4299 vectors, takes less of a performance hit.
4300 * There is now a primitive type predicate VECTOR-T-P
4301 to test for the (VECTOR T) type, so that e.g.
4302 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4303 can now be compiled with some semblance of efficiency. (The old code
4304 turned the type declaration into a full call to %TYPEP at runtime!)
4305 * AREF on (VECTOR T) is still not fast, since it's still compiled
4306 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4307 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4308 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4309 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4310 without making full calls to SUBTYPEP at runtime.
4311 (Some analogous efficiency issues for non-simple vectors specialized to
4312 element types other than T, or for non-simple multidimensional arrays,
4313 have not been addressed. They could almost certainly be handled the
4314 same way if anyone is motivated to do so.)
4315 * The changes in array handling break binary compatibility, so
4316 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4317 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4318 instead of (VALUES T T).
4319 * By following the instructions that Dan Barlow posted to sbcl-devel
4320 on 2 July 2000, I was able to enable primitive dynamic object
4321 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4322 functionality is not implemented (since it calls ld to resolve
4323 library references automatically, requiring RUN-PROGRAM for its
4324 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4325 try to resolve library references) is now supported.
4326 * The system now flushes the standard output streams when it terminates,
4327 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4328 them at several other probably-convenient times, e.g. in each pass of
4329 the toplevel read-eval-print loop, and after evaluating a form given
4330 as an "--eval" command-line option. (These changes were motivated by a
4331 discussion of stream flushing issues on cmucl-imp in August 2000.)
4332 * The source transform for TYPEP of array types no longer assumes
4333 that an array whose element type is a not-yet-defined type
4334 is implemented as an array of T, but instead punts, so that the
4335 type will be interpreted at runtime.
4336 * There is now some support for cross-compiling in make.sh: each of
4337 the phases of make.sh has its own script. (This should be transparent
4338 to people doing ordinary, non-cross-compile builds.)
4339 * Since my laptop doesn't have hundreds of megabytes of memory like
4340 my desktop machine, I became more motivated to do some items on
4341 my to-do list in order to reduce the size of the system a little:
4342 ** Arrange for various needed-only-at-cold-init things to be
4343 uninterned after cold init. To support this, those things have
4344 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4345 symbols with such names are now uninterned after cold init).
4346 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4347 fasl files for cold load.
4348 ** Remove the old compiler structure pooling code (which used to
4349 be conditional on the target feature :SB-ALLOC) completely.
4350 ** Redo the representation of some data in cold init to be more compact.
4351 (I also looked into supporting byte compiled code at bootstrap time,
4352 which would probably reduce the size of the system a lot, but that
4353 looked too complicated, so I punted for now.)
4354 * The maximum signal nesting depth in the src/runtime/ support code has
4355 been reduced from 4096 to 256. (I don't know any reason for the very
4356 large old value. If the new smaller value turns out to break something,
4357 I'll probably just bump it back up.)
4358 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4360 * Many, many bugs reported by Peter Van Eynde have been added to
4361 the BUGS list; some have even been fixed.
4362 * While enabling dynamic object file loading, I tried to make the
4363 code easier to understand, renaming various functions and variables
4364 with less ambiguous names, and changing some function calling
4365 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4366 * While trying to figure out how to do the OpenBSD port, I tried to
4367 clean up some of the code in src/runtime/. In particular, I dropped
4368 support for non-POSIX signal handling, added various comments,
4369 tweaked the code to reduce the number of compilation warnings, and
4370 renamed some files to increase consistency.
4371 * To support the new automatic configuration functionality in make.sh,
4372 the source file target-features.lisp-expr has been replaced with the
4373 source file base-target-features.lisp-expr and the machine-generated
4374 file local-target-features.lisp-expr.
4375 * fixed a stupid quoting error in make.sh so that using CMU CL
4376 "lisp -batch" as cross-compilation host works again
4378 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4380 * DESCRIBE no longer tries to call itself recursively to describe
4381 bound/fbound values, so that it no longer fails on symbols which are
4382 bound to themselves (like keywords, T, and NIL).
4383 * DESCRIBE now works on generic functions.
4384 * The printer now prints less-screwed-up representations of closures
4385 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4386 * A private symbol is used instead of the :EMPTY keyword previously
4387 used to mark empty slots in hash tables. Thus
4388 (DEFVAR *HT* (MAKE-HASH-TABLE))
4389 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4390 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4391 now does what ANSI says that it should. (You can still get
4392 similar noncompliant behavior if bang on the hash table
4393 implementation with all the symbols you get back from
4394 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4395 This breaks binary compatibility, since tests for equality to
4396 :EMPTY are wired into things like the macroexpansion of
4397 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4399 * There's now a minimal placeholder implementation for CL:STEP,
4400 as required by ANSI.
4401 * An obscure bug in the interaction of the normal compiler, the byte
4402 compiler, inlining, and structure predicates has been patched
4403 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4404 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4406 * Missing ordinary arguments in a macro call are now detected even
4407 when the macro lambda list contains &KEY or &REST.
4408 * The debugger no longer complains about encountering the top of the
4409 stack when you type "FRAME 0" to explicitly instruct it to go to
4410 the top of the stack. And it now prints the frame you request even
4411 if it's the current frame (instead of saying "You are here.").
4412 * As specified by ANSI, the system now always prints keywords
4413 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4415 * The default initial SIZE of HASH-TABLEs is now smaller.
4416 * Type information from CLOS class dispatch is now propagated
4417 into DEFMETHOD bodies, so that e.g.
4418 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4420 is now basically equivalent to
4421 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4422 (DECLARE (TYPE SINGLE-FLOAT X))
4424 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4425 operation, without having to do run-time type dispatch.
4426 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4427 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4428 * Since I don't seem to be making big file reorganizations very often
4429 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4430 is overflowing my ability to conveniently back them up), I've finally
4431 checked the system into CVS. (The CVS repository is on my home system,
4432 not at SourceForge -- putting it on SourceForge might come later.)
4433 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4434 high-level GC-NOTIFY-FOO functions send their output. (There's
4435 still very little control of where low-level verbose GC functions
4436 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4437 less than it used to -- the GC-NOTIFY-FOO functions are now under
4438 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4439 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4440 in only one place in the source code, and propagates it automatically
4441 everywhere that it's needed. Thus e.g. when I bump the version from
4442 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4443 * The C source files now include boilerplate legalese and documentation
4444 at the head of each file (just as the Lisp source files already did).
4445 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4446 to his page will be replaced with a link to his new CLiki service.
4448 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4450 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4451 * The build process now looks for GNU make under the default name "gmake",
4452 instead of "make" as it used to. If GNU make is not available as "gmake"
4453 on your system, you can change this default behavior by setting the
4454 GNUMAKE environment variable.
4455 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4456 can build without error under CMU CL.
4458 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4460 * There is now a partial SBCL user manual (with some new text and some
4461 text cribbed from the CMU CL manual).
4462 * The beginnings of a profiler have been added (starting with the
4463 CMU CL profiler and simplifying and cleaning up). Eventually the
4464 main interface should be through the TRACE macro, but for now,
4465 it's still accessed through vaguely CMU-CL-style functions and macros
4466 exported from the package SB-PROFILE.
4467 * Some problems left over from porting CMU CL to the new
4468 cross-compilation bootstrap process have been cleaned up:
4469 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4470 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4471 part when getting rid of the compiler *BACKEND* stuff.)
4472 ** The value of *NULL-TYPE* was screwed up, because it was
4473 being initialized before the type system knew the final
4474 definition of the 'NULL type. This screwed up several key
4475 optimizations in the compiler, causing inefficiency in all sorts
4476 of places. (I found it because I wanted to understand why
4477 GET-INTERNAL-RUN-TIME was consing.)
4478 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4479 PROCLAIM FTYPE information. Unfortunately this broke binary
4480 compatibility again, since now the forms output by DEFGENERIC
4481 to refer to functions which didn't exist in 0.6.3.
4482 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4483 can use the new (as of 0.6.3) transform for SOME into MAP into
4485 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4486 values to (INTEGER 0 1000000), so that the time code will no longer
4487 occasionally get blown up by Linux returning 1000000 microseconds
4488 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4489 its output conform to the ANSI spec. (Alas, this makes its output
4490 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4491 * A full call to MAP NIL with a single sequence argument no longer conses.
4492 * fixes to problems pointed out by Martin Atzmueller:
4493 * The manual page no longer talks about multiprocessing as though
4494 it were currently supported.
4495 * The ILISP support patches have been removed from the distribution,
4496 because as of version 5.10.1, ILISP now supports SBCL without us
4497 having to maintain patches.
4498 * added a modified version of Raymond Toy's recent CMU CL patch for
4499 EQUALP comparison of HASH-TABLE
4501 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4503 * The system still can't cross-compile itself with
4504 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4505 entails), but at least it can compile more of itself that way
4506 than it used to be able to, and various buglets which were uncovered
4507 by trying to cross-compile itself that way have now been fixed.
4508 * This release breaks binary compatibility again. This time
4509 at least I've incremented the FASL file format version to 2, so that the
4510 problem can be detected reliably instead of just causing weird errors.
4511 * various new style warnings:
4512 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4513 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4514 ** using the lexical binding of a variable named in the *FOO* style
4515 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4516 as specified by ANSI.
4517 * *RANDOM-STATE* is no longer automatically initialized from
4518 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4519 default behavior of the system is to repeat its behavior every time
4520 it's run. If you'd like to change this behavior, you can always
4521 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4522 old convention there was no comparably easy way to get the system to
4523 repeat its behavior every time it was run.
4524 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4525 FTYPE declarations has been removed, in favor of their ANSI
4526 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4527 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4528 the types of their sequence arguments can be determined at compile time.
4529 This is done through a new open code expansion for MAP which eliminates
4530 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4531 argument types can be determined at compile time.
4532 * The optimizer now transforms COERCE into an identity operation when it
4533 can prove that the coerced object is already of the correct type. (This
4534 can be a win for machine generated code, including the output of other
4535 optimization transforms, such as the MAP transform above.)
4536 * Credit information has been moved from source file headers into CREDITS.
4537 * Source file headers have been made more standard.
4538 * The CASE macro now compiles without complaining even when it has
4541 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4543 * (Note that the way that the PCL macroexpansions were rewritten
4544 to accommodate the change in DEFGENERIC below breaks binary
4545 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4546 not run under sbcl-0.6.2. Once we get out of alpha releases,
4547 i.e. hit release 1.0.0, we'll probably try to maintain binary
4548 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4549 and sbcl-1.4.4. Until then, however, it might be fairly common
4550 for maintenance releases to break binary compatibility.)
4551 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4553 * The reporting of unhandled signals has been changed to print some
4554 explanatory text as well as the report form. (Previously only
4555 the report form was printed.)
4556 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4557 it defines, so that the compiler no longer issues undefined function
4558 warnings for compiled-but-not-yet-loaded generic functions.
4559 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4560 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4561 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4562 section of the ANSI spec are supported.
4563 * The "" nickname for the "KEYWORD" package has been removed.
4564 The reader still handles symbol tokens which begin with a package marker
4565 as keywords, but it doesn't expose its mechanism for doing so in the
4566 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4567 * The system now issues STYLE-WARNINGs for contradictory TYPE
4568 proclamations. (Warnings for contradictory FTYPE proclamations would
4569 be nice too, but those can't be done usefully unless the type system
4570 is made smarter about FUNCTION types.)
4571 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4572 systematized, so that "*target-*.lisp is supposed to exist only on the
4573 target and imply that there's a related file which exists on the
4574 host, and *host-*.lisp is supposed to exist only on the host and imply
4575 that there's a related file which exists on the target. This involves a
4576 lot of renaming. Hopefully the acute confusion caused by the renaming
4577 will be justified by the reduction in chronic confusion..
4578 ** runtime-type.lisp -> early-target-type.lisp
4579 ** target-type.lisp -> late-target-type.lisp
4580 ** early-host-format.lisp -> early-format.lisp
4581 ** late-host-format.lisp -> late-format.lisp
4582 ** host-error.lisp -> misc-error.lisp
4583 ** early-error.lisp -> early-target-error.lisp
4584 ** late-error.lisp -> late-target-error.lisp
4585 ** host-defboot.lisp -> early-defboot.lisp
4586 ** code/misc.lisp -> code/target-misc.lisp
4587 ** code/host-misc.lisp -> code/misc.lisp
4588 ** code/numbers.lisp -> code/target-numbers.lisp
4589 ** code/early-numbers.lisp -> numbers.lisp
4590 ** early-host-type.lisp -> early-type.lisp
4591 ** late-host-type.lisp -> late-type.lisp
4592 ** host-typep.lisp -> typep.lisp
4593 ** load.lisp -> target-load.lisp
4594 ** host-load.lisp -> load.lisp
4595 ** host-disassem.lisp -> disassem.lisp
4596 ** host-insts.lisp -> insts.lisp
4597 ** byte-comp.lisp -> target-byte-comp.lisp
4598 ** host-byte-comp.lisp -> byte-comp.lisp
4599 ** host-signal.lisp -> signal.lisp
4600 ** host-defstruct.lisp -> defstruct.lisp
4601 ** late-target-type.lisp -> deftypes-for-target.lisp
4602 Furthermore, several other previously target-only files foo.lisp (e.g.
4603 hash-table.lisp and random.lisp) have been split into a target-and-host
4604 foo.lisp file and a target-only target-foo.lisp file, with their key type
4605 definitions in the target-and-host part, so that the cross-compiler will
4606 know more about target types.
4607 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4608 gone away. In their place are various *BACKEND-FOO* variables
4609 corresponding to the slots of the old structure.
4610 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4611 into the target SBCL has been fixed.
4612 * The chill.lisp system for loading cold code into a running SBCL
4614 * Support for the CMU CL "scavenger hook" extension has been removed.
4615 (It was undocumented and unused in the CMU CL sources that SBCL was
4616 derived from, and stale in sbcl-0.6.1.)
4617 * Various errors in the cross-compiler type system were detected
4618 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4619 (enabling various consistency checks). Many of them were fixed,
4620 but some hard problems remain, so the compiler is back to
4621 running without *TYPE-SYSTEM-INITIALIZED* for now.
4622 * As part of the cross-compiler type system cleanup, I implemented
4623 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4624 * I have started adding UNCROSS calls throughout the type system
4625 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4626 on cross-compiler input files will be able to go away, and various
4628 * CONSTANTP now returns true for quoted forms (as explicitly required
4631 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4633 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4634 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4635 (MAKE-LIST -1). In the longer term, it ought to become true
4636 that declarations are assertions even at SAFETY 1. For now, it's not
4637 quite true even at SAFETY 3, but it's at least more nearly true..
4638 (Note that this change seems to increases the size of the system by
4639 O(5%) and to decrease the speed of the compiler by 20% or more.)
4640 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4641 for the problem of printing dozens of lines of distracting information
4642 about low-level system machinery as part of the top stack frame
4643 on entry to the debugger when an undefined function was called.
4644 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4645 is not reset to COMMON-LISP-USER.
4646 * Compilation of stuff related to dyncount.lisp has been made conditional
4647 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4648 smaller. The various dyncount-related symbols have been moved into
4649 a new "SB-DYNCOUNT" package.
4650 * tty-inspect.lisp has been renamed to inspect.lisp.
4651 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4652 feature has gone away. (When we eventually port to other flavors of
4653 libc and/or Unix, we'll try to make the differences between flavors
4654 invisible at the user level.)
4655 * Various other *FEATURES* tags, and/or their associated conditionals,
4656 have been removed if obsolescent, or given better documentation, or
4657 sometimes given more-mnemonic names.
4659 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4661 * tidied up "make.sh" script
4662 * tidied up system directory structure
4663 * better "clean.sh" behavior
4664 * added doc/FOR-CMUCL-DEVELOPERS
4665 * many many small tweaks to output format, e.g. removing possibly-confusing
4666 trailing #\. character in DESCRIBE-INSTANCE
4667 * (EQUALP #\A 'A) no longer signals an error.
4668 * new hashing code, including EQUALP hashing
4669 * tidied up Lisp initialization and toplevel
4670 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4671 * command line argument processing
4672 * added POSIX-GETENV function to deal with Unix-ish environment variables
4673 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4674 terminating SBCL on EOF
4675 * non-verbose GC by default
4676 * There is no more "sbcl" shell script; the sbcl file is now the C
4677 runtime executable (just like CMU CL).
4678 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4680 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4681 symbol and package information as Lisp native symbol and package objects
4682 instead of strings naming symbols and strings naming packages. This way,
4683 whenever packages are renamed (as in warm init), debug information is
4684 transformed along with everything else.
4685 * tweaked the optimization policy declarations which control the building
4686 of SBCL itself. Now, among other things, the system no longer saves
4687 source location debugging information. (This helps two problems at once
4688 by reducing SBCL size and by keeping SBCL from trying to look for its
4689 sources -- which may not exist -- when reporting errors.)
4690 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4691 debugging and testing purposes
4692 * cleaned up printing, making the printer call PRINT-OBJECT for
4693 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4694 methods, giving nearly-ANSI behavior
4695 * converted almost all special variables to use *FOO* naming convention
4696 * deleted PARSE-TIME functionality, since it can be done portably
4697 * moved some files out of cold init into warm init
4698 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4700 * regularized formatting of source files
4701 * added an install.sh script
4702 * fixed ridiculous memory usage of cross-compiler by making
4703 compiler/alloc.lisp not try to do pooling unless it can hook
4704 itself into the GC of the cross-compilation host. Now the system
4705 builds nicely on my old laptop.
4706 * added :SB-ALLOC in target-features.lisp-expr
4707 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4708 was not implemented)
4709 * re-did condition handling and note reporting in the compiler. Notes
4710 are no longer handled by signalling conditions. Style warnings
4711 and warnings are handled more correctly and reported in such a way
4712 that it's easy to find one or the other in your output (so that you
4713 can e.g. figure out which of many problems caused COMPILE-FILE to
4715 * changed the severity of several compiler warnings from full WARNING
4716 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4717 compiler note reporting so that it doesn't use the condition system
4718 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4719 COMPILE-FILE command)
4720 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4721 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4722 no longer supported, since their implementation was deeply intertwingled
4723 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4724 * removed bogus "support" for compiler macros named (SETF FOO), and
4725 removed the compiler macro for SETF INFO (but only after making a fool
4726 of myself on the cmucl-imp mailing list by posting a bogus patch for
4727 DEFINE-COMPILER-MACRO..)
4728 * Compiled files containing forms which have side effects on the Lisp
4729 reader (such as DEFPACKAGE forms) are now handled more correctly.
4730 (Compiler queuing of top level lambdas has been suppressed by setting
4731 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4732 * deleted various currently-unused source files, e.g. gengc.lisp. They
4733 may be added back at some point e.g. when porting to other architectures,
4734 but until they are it's distracting to distribute them and to try to
4736 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4737 there were so many of them they're just distractions, and UNCROSS is
4738 known to be able to handle the current sources
4739 * moved PROFILE functionality into TRACE, so that it will be clear
4740 how the wrapping and unwrapping of functions when you profile them
4741 interacts with the wrapping and unwrapping of functions when you
4742 trace them. (Actually, the functionality isn't there yet, but at least
4743 the interface specification is there. Hopefully, the functionality will
4744 arrive with some future maintenance release.)
4745 * removed host-oops.lisp
4746 * changed signature of QUIT function to allow UNIX-CODE argument
4747 * fixed READ-SEQUENCE bug
4748 * tweaked verbose GC output so that it looks more like the progress
4749 output that ANSI specifies for functions like LOAD
4750 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4751 * added <http://sbcl.sourceforge.com> to the banner information printed by