1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.16 relative to 1.0.15:
3 * enhancement: add support for fcntl's struct flock to SB-POSIX.
5 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
6 * enhancement: cleaner backtraces for interactive interrupts, as
7 well as other cases where the interesting frames used to be
8 obscured by interrupt handling frames.
9 * enhancement: untracing a whole package using (UNTRACE "FOO") is
10 now supported, and tracing a whole package using (TRACE "FOO") now
11 traces SETF-functions as well.
12 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
13 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
14 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
15 * unreadably printed representation of hash-tables now includes
17 * bug fix: partially fixed #188: type propagation from assignments
18 is now more efficient.
19 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
20 'DOUBLE-FLOAT) are not flushable.
21 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
22 full address of the object, and none of the tag bits.
23 * bug fix: readably printing hash-tables now respects other printer
24 control variables. (reported by Cedric St-Jean)
25 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
26 keyword with MAKE-HASH-TABLE.
27 * bug fix: export SB-POSIX:MKSTEMP.
28 * bug fix: SORT was not interrupt safe.
29 * bug fix: XREF accounts for the last node of each basic-block as
31 * bug fix: MAKE-INSTANCE optimizations interacted badly with
32 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
33 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
34 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
35 create obsolete instances in the case of redefinition or
36 obsoletion of a superclass. (thanks to Andy Hefner)
37 * bug fix: Support for the Alpha architecture has been revived; it had
38 suffered somewhat from lack of maintenance since sbcl-1.0.
39 * improvements to the Windows port:
40 ** The system detects the codepage to use at startup, instead of
41 using the value from the saved core. (thanks to Kei Suzuki)
43 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
44 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
45 (see documentation for details.)
46 * revived support for OpenBSD (contributed by Josh Elsasser)
47 * partially fixed bug #108: ROOM no longer suffers from occasional
48 (AVER (SAP= CURRENT END)) failures.
49 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
50 bodies no longer cause a WARNING to be signalled. (reported by
52 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
53 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
54 the bug in TRUNCATE.) (reported by Michael Weber)
55 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
56 no samples. (reported by Andy Hefner)
57 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
58 no longer appear as (NIL ...) frames in backtraces.
59 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
60 is now actually exported from SB-EXT as documented. (reported by
62 * bug fix: sb-aclrepl now correctly understands how to inspect
63 single-floats on 64-bit platforms where single-floats are not boxed.
64 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
65 yet been finalized. (reported by Levente Meszaros)
66 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
67 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
68 SBCL-specific optimize qualities.
70 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
71 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
72 an executable in the search path, and does so in the child
73 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
74 been removed; it can be found in the CVS history, for somebody who
75 needs that search behavior (see the manual).
76 * minor incompatible change: compiler policy re. weakening type
77 checks has changed: now type checks are weakened only if SAFETY < 2
79 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
80 unparsing of directory pathnames as files. Analogously,
81 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
82 filename to parse into a directory pathname.
83 * enhancement: implicit generic function creation now signals a
84 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
85 users can bind handlers for and muffle around calls to LOAD.
86 * enhancement: RUN-PROGRAM allows unicode arguments and environments
87 to be used (using the default stream external format), and allows
88 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
89 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
90 strings and vectors whose element-type or simplicity is not fully
91 known at compile-time.
92 * optimization: STRING-TRIM and related functions no longer allocate
93 a new string when no trimming needs to be performed. These functions
94 are also faster than before when the input string has been declared
96 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
97 * optimization: READ-LINE is significantly faster for files containing
99 * optimization: non-open coded uses of character comparison operators
100 (e.g. char=) no longer cons when called with more than one parameter
101 on platforms supporting dynamic-extent allocation.
102 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
103 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
104 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
106 * bug fix: some sequence functions elided bounds checking when
108 * bug fix: too liberal weakening of union-type checks when SPEED >
110 * bug fix: more bogus fixnum declarations in ROOM implementation
113 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
114 * new feature: MAKE-HASH-TABLE now experimentally accepts a
115 :SYNCHRONIZED argument, which makes the hash-table safe for
116 concurrent accesses (but not iteration.) See also:
117 SB-EXT:WITH-LOCKED-HASH-TABLE, and
118 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
119 * optimization: CONCATENATE on strings is an order of magnitude faster
120 in code compiled with (> SPEED SPACE).
121 * optimization: SUBSEQ is ~50% faster on lists.
122 * bug fix: bug 417 fixed -- source location reporting is now more robust.
123 * bug fix: SUBSEQ on a list will now correctly signal an error if if
124 END is smaller then START.
125 * bug fix: SB-PROFILE will no longer report extra consing for nested
126 calls to profiled functions.
127 * bug fix: ROOM implementation had bogus fixnum declarations which
128 could cause type-errors when calling ROOM on large images.
129 * bug fix: if file compilation is aborted, the partial fasl is now
130 deleted, and COMPILE-FILE returns NIL as the primary value.
131 * bug fix: number of thread safety issues relating to SBCL's internal
132 hash-table usage have been fixed.
133 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
134 be returned from its body when the values were being returned
135 using unknown-values return convection and the W-P-O was wrapped
136 inside an UNWIND-PROTECT.
137 * bug fix: sb-posix should now compile again under Windows, enabling
140 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
141 * incompatible change: hash-table accessor functions are no longer
142 automatically protected by locks. Concurrent accesses on the same hash-table
143 from multiple threads can give inconsistent results or even corrupt the
144 hash-table completely. Multi-threaded applications should do their own
145 locking at the correct granularity. In the current implementation it is
146 still safe to have multiple readers access the same table, but it's not
147 guaranteed that this property will be maintained in future releases.
148 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
149 and will signal an error at runtime.
150 * enhancement: SB-THREAD package now exports a semaphore interface.
151 * enhancement: CONS can now stack-allocate on x86 and
152 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
154 * enhancement: nested lists can now be stack allocated on
155 platforms providing stack allocation support.
156 * enhancement: dynamic-extent support has been extended to support
157 cases where there are multiple possible sources for the stack
159 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
160 if the mutex is uncontested on Linux.
161 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
162 as specified by the CLHS. (thanks to Tobias Rittweiler)
163 * bug fix: NaN comparison now works on x86-64.
164 * bug fix: CLOSE :ABORT T on a stream with pending output now
166 * bug fix: instances of non-standard metaclasses using standard
167 instance structure protocol sometimes missed the slot type checks
169 * bug fix: known functions can be inlined in high-debug code.
171 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
172 * minor incompatible change: the MSI installer on Windows no longer
173 associates .lisp and .fasl files with the installed SBCL.
174 * minor incompatible change: :UNIX is no longer present in *FEATURES*
175 on Windows. (thanks to Luis Oliviera)
176 * new feature: SB-CLTL2 contrib module now implements
177 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
178 * optimization: scavenging weak pointers is now more efficient,
179 requiring O(1) instead of O(N) per weak pointer to identify
180 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
181 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
182 method bodies using constant slot-names when the first argument is
183 a specializer parameter for the method.
184 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
185 names, when the class of the instance is a direct instance
186 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
187 as slow as the constant slot-name case.)
188 * optimization: member type construction is now O(N) instead
190 * optimization: UNION and NUNION are now O(N+M) for large
191 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
193 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
194 now more readable in environments like Slime which display it.
195 (thanks to Tobias C. Rittweiler)
196 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
197 was able to infer the correct argument type for the object on which
198 the CAS operation was being performed.
199 * bug fix: copy propagation interfered with parallel assignment
200 semantics in local calls. (reported by Paul Khuong)
201 * bug fix: the signed modular fixnum shift compiled to wrong code on
202 x86 and x86-64. (spotted by a slight modification to some of
205 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
206 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
207 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
208 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
210 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
211 and generic function now print the object identity for anonymous
212 instances. (thanks to Attila Lendvai)
213 * enhancement: as an extension to MOP, SBCL now supports SETF of
214 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
215 (thanks to Attila Lendvai)
216 * enhancement: sb-cover has an alternate annotation mode which puts
217 the form annotations at the car of the form, instead of annotating
218 it completely. The benefit of this mode is that it shows better
219 which forms were instrumented by the compiler.
220 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
221 by printer control variables. (reported by Dan Corkill)
222 * bug fix: system leaked memory when delayed output was performed by
223 the OS in smaller chunks then expected. (thanks to David Smith)
224 * bug fix: system leaked memory when file streams were not closed
226 * bug fix: large objects written to slow streams that were modified
227 after the write could end up with the modified state written to
228 the underlying file descriptor.
229 * bug fix: multiple threads operating in parallel on the same stream
230 could cause buffer-overflows.
231 * bug fix: source location information is stored correctly
232 (broken since 1.0.6). This bug would generally show up as the
233 Slime debugger highlighting the wrong form when the sldb-show-source
235 * bug fix: Branch forms are again annotated as branches in the sb-cover
237 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
238 on x86-64. (reported by Gregory Vanuxem)
239 * bug fix: EQUALP could return wrong results for structures with raw
240 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
241 word). (reported by Vjacheslav Fyodorov)
242 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
245 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
246 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
247 atomic compare-and-swap operations on threaded platforms.
248 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
249 allows assigning a global minimum value to optimization qualities
250 (overriding proclamations and declarations).
251 * enhancement: closed over variables can be stack-allocated on x86
253 * performance bug fix: GETHASH and (SETF GETHASH) are once again
255 * optimization: slot definition lookup is now O(1). This speeds up
256 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
257 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
258 * optimization: ASSOC and MEMBER can now be open-coded for all
259 combinations of keyword arguments when second argument is constant
260 and SPEED >= SPACE. In other cases a specialized version is
262 * bug fix: using obsoleted structure instances with TYPEP and
263 generic functions now signals a sensible error.
264 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
265 (reported by Kristoffer Kvello)
266 * bug fix: backtrace construction is now more careful when making
267 lisp-objects from pointers on the stack, to avoid creating bogus
268 objects that can be seen by the GC.
269 * bug fix: defaulting of values in contexts expecting more than 7
270 variables now works on x86-64. (reported by Christopher Laux)
271 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
273 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
274 as the property-list of a symbol.
275 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
276 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
277 situations requested, are once again file-compileable. (reported
280 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
281 * MOP improvement: support for user-defined subclasses of
282 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
283 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
284 the expansion of DEFMETHOD, is responsible for generating a form
285 which creates a list of specializers when evaluated. Additional
286 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
287 debugging and introspective support.
288 * minor incompatible change: the (unsupported) spinlock interface
289 has changed: free spinlock now has the value NIL, and a held spinlock
290 has the owning thread as its value.
291 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
292 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
294 * enhancement: name of a socket-stream is now "a socket" instead of
296 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
297 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
298 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
299 for generic functions. (Thanks to Tobias C. Rittweiler)
300 * optimization: bignum printing speed has been improved by 20-40%
301 (depending on the bignum size.)
302 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
304 * bug fix: the cache used by the CLOS to store precomputed effective
305 methods, slot offsets, and constant return values is now thread and
307 * bug fix: generic function dispatch function updating is now thread
308 and interrupt safe (in the sense that the known issues have been
310 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
311 * bug fix: interning EQL-specializers is now thread and interrupt safe.
312 * bug fix: asdf systems with dependencies to the SB-POSIX or
313 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
314 * bug fix: interrupt safety of applicable method computation has been
317 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
318 * new contrib: sb-cover, an experimental code coverage tool, is included
320 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
321 significantly faster.
322 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
323 produce more efficient code on x86-64 if the compiler can determine
324 that (+ INDEX <constant>) does not require a bounds check and FOO
325 has an element type at least 8 bits wide.
326 * enhancement: a new, experimental synchronous timeout facility is
327 provided. Refer to SB-SYS:WITH-DEADLINE for details.
328 * enhancement: when a symbol name conflict error arises, the
329 conflicting symbols are always printed with a package prefix.
330 (thanks to Kevin Reid)
331 * enhancement: stepping is now once again supported on the SPARC and
332 MIPS platforms. (It is also now more likely to work on CheneyGC
334 * enhancement: sb-sprof can now also track and report accurate call
336 * bug fixes: the treatment of non-standard subclasses of
337 SB-MOP:SPECIALIZER is more correct.
338 * incompatible change: PURIFY no longer copies the data from the
339 dynamic space into the static and read-only spaces on platforms
340 that use the generational garbage collector
341 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
343 * bug fix: GC race condition occasionally resulting in crashes with
344 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
346 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
348 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
349 a STYLE-WARNING to be signalled (regression from 1.0.4.)
350 * bug fix: an asynchronous interrupt could previously leave the
351 system running with GC inhibited.
352 * bug fix: a DECLARE form evaluated at top-level now causes an error
353 rather than silently (or verbosely) returning NIL.
354 * bug fix: trying to dissassemble functions compiled at high DEBUG could
355 result in a error being signalled due to source form lookup errors.
356 (reported by Peter Graves)
358 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
359 * incompatible change: removed writer methods for host-ent-name,
360 host-ent-addresses -- changing the values did not update the DNS
362 * minor incompatible change: changed experimental JOIN-THREAD interface
363 * documentation: the manual now lists reader and writer methods
364 in class slot documentation sections. (thanks to Richard M Kreuter)
365 * documentation: unwinding from asyncronous events has been
366 documented as unsafe.
367 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
368 in multithreaded application code.
369 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
370 platforms. (thanks to James Anderson for the optimization hint)
371 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
373 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
374 variants no longer cons.
375 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
376 their NOT- variants no longer cons.
377 * optimization: EQUAL hash tables no longer use SXHASH for objects
378 of all data types, but instead use an EQL hash for types for which
379 EQUAL is the same as EQL.
380 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
381 are significantly faster.
382 * optimization: new STRING-OUTPUT-STREAM implementation is much
383 faster for large outputs and conses 30% less on average.
384 * enhancement: XREF information is now collected for references made
385 to global variables using SYMBOL-VALUE and a constant argument.
386 * enhancement: SIGINT now causes a specific condition
387 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
388 * bug fix: FILE-POSITION used to signal an error in some cases where
389 ANSI requires it to return NIL.
390 * bug fix: ADJUST-ARRAY is now interrupt-safe.
391 * bug fix: adding and removing fd-handlers is now interrupt-safe.
392 * bug fix: inlined calls to C now ensure 16byte stack alignment on
394 * bug fix: bad type declaration in the CLOS implementation has
395 been fixed. (reported by James Anderson)
396 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
397 (thanks to Lutz Euler)
398 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
399 required by the CLHS. (reported by Samium Gromoff)
400 * bug fix: dead unbound variable references now signal an error.
401 (reported by Marco Monteiro)
402 * bug fix: / with an unused value was being deleted in safe code.
403 (thanks to Marco Monteiro and Kevin Reid)
404 * bug fix: number of characters that can be written onto a single
405 line in a file is unlimited.
406 * bug fix: some GC deadlocks caused by asynchronous interrupts have
407 been fixed by inhibiting interrupts for when GC is disabled.
408 * bug fix: some interrupt safety issues with GETHASH, (SETF
409 GETHASH), CLRHASH and REMHASH have been fixed.
410 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
411 specifier no longer causes infinite recursion.
412 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
413 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
414 * bug fix: an &environment argument with the correct variable information
415 is passed to macros that are expanded during byte compilation.
416 (reported by Samium Gromoff)
417 * bug fix: a logic error was causing the hash function for bignums to
418 have excessive amounts of collisions. (reported by Faré Rideau)
419 * bug fix: modifying the contents of an array could change the return
420 value of SXHASH on that array, which is only allowed for strings
421 and bit vectors. (bug introduced in 0.9.16)
422 * improvement: the x86-64/darwin port now passes all tests (except
423 for the debugger tests) but should still be considered
424 experimental until this is fixed.
425 * improvement: a style-warning is signaled for CASE (etc) clauses with
426 duplicate keys. (thanks to Kevin Reid)
427 * improvement: macos/x86 and macos/x86-64 now use mach exception
428 handlers for dealing with illegal instructions (for trapping and
429 error handling) and memory protection violations (for GC).
431 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
432 * new platform: experimental support for x86-64/darwin (MacOS).
433 * incompatible change: the thread-safe (on most platforms) getaddrinfo
434 and getnameinfo sockets functions are used instead of gethostbyaddr
435 and gethostbyname, on platforms where the newer functions are available.
436 As a result, the ALIASES field of HOST-ENT will always be NIL on these
438 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
439 don't print the startup banner, but behave as if --noinform was passed
440 as a command line argument. (thanks to Kevin Reid)
441 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
442 * optimization: code using alien values with undeclared types is much faster.
443 * optimization: the compiler is now able to open code SEARCH in more cases.
444 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
445 * bug fix: using standardized COMMON-LISP special variables as loop
446 variables no longer signals bogus package lock violations. (reported
448 * bug fix: declaring local loop variables to be of a range-limited type
449 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
450 (reported by Andras Simon)
451 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
452 bugs remain on x86-64.)
453 * bug fix: the #= and ## reader macros now interact reasonably with
454 funcallable instances.
455 * bug fix: type-checks for function arguments were compiled using the
456 compiler policy settings of the wrong lexical environment (in 1.0.2
458 * bug fix: SHADOW accepts characters as string designators, as required
459 by the spec (thanks to Eric Marsden)
460 * bug fix: fixed GC safety issues when foreign functions are called with
461 non-base strings as arguments
462 * bug fix: more consistent error output for fatal-compiler-warnings, like
464 * bug fix: functions with proper names were showing up as NIL in
465 backtraces (reported by Edi Weitz, regression in 1.0.2)
467 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
468 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
469 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
470 produce more efficient code on the x86 if the compiler can determine
471 that (+ INDEX <constant>) does not require a bounds check and FOO
472 has an element type at least 8 bits wide.
473 * bug fix: references to undefined variables are handled the same way
474 in toplevel forms as in the normal compiler
475 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
477 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
478 (thanks to Jon Buller)
479 * improvement: faster compilation times for complex functions
480 * improvement: added readlink support to SB-POSIX (thanks to Richard
483 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
484 * improvement: experimental support for mach exception handling on
485 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
487 * improvement: support for GBK external format.
488 (thanks to Chun Tian (binghe))
489 * improvement: the debugger now displays variables that have been closed
490 over, in code compiled with (DEBUG 2) or higher
491 * improvement: support for executable cores on NetBSD (thanks to
493 * new feature: added a RESTART-FRAME debugger command
494 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
495 be used to provide an implementation for FILE-POSITION on Gray streams
496 (thanks to Eric Marsden)
497 * new feature: add syslog, openlog and closelog support to SB-POSIX
498 (thanks to Richard Kreuter)
499 * optimization: the function call overhead in code compiled with
500 a high DEBUG optimization setting is significantly
501 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
502 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
503 (thanks to Tony Martinez)
504 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
505 evaluated once (reported by Kevin Reid)
506 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
507 works again (thanks to Kilian Sprotte)
508 * bug fix: an error is signaled for tagbodies with duplicate tags
509 (thanks to Stephen Wilson)
510 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
511 * bug fix: Win32 port can now handle foreign code unwinding Lisp
512 stack frames from alien callbacks.
513 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
514 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
515 65k pages (thanks to David Woodhouse)
516 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
518 changes in sbcl-1.0.1 relative to sbcl-1.0:
519 * new platform: FreeBSD/x86-64, including support for threading.
520 * new feature: the compiler stores cross-referencing information
521 about function calls (who-calls), macroexpansion (who-macroexpands)
522 and special variables (who-binds, who-sets, who-references) for code
523 compiled with (< SPACE 3). This information is available through the
524 sb-introspect contrib.
525 * new feature: users may subclass SEQUENCE, and have instances of
526 these classes interoperate with standard Common Lisp functions if
527 a number of methods are defined. (This feature is experimental
528 and the interface subject to change based on feedback from SBCL
529 users and the general community)
530 * improvement: sb-sprof traces call stacks to an arbitrary depth on
531 x86/x86-64, rather than the previous fixed depth of 8
532 * improvement: another pthread back-end of mutex "pthread-futex".
533 * bug fix: non-ascii command-line arguments are processed correctly
534 (thanks to Yaroslav Kavenchuk)
535 * bug fix: non-required arguments were not passed correctly when a method
536 defined using DEFMETHOD was called from a mop-generated method using
537 CALL-NEXT-METHOD (reported by Pascal Costanza)
538 * bug fix: recursion is now permitted in accessors through
539 SLOT-UNBOUND. (reported by Pascal Costanza)
540 * bug fix: an error was signaled at startup if the HOME environment
541 variable was defined, but had an empty value (reported by Peter Van Eynde)
542 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
543 signal a PROGRAM-ERROR, not a BUG.
544 * bug fix: SB-POSIX:READDIR works when built with large file support.
545 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
547 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
548 for better automatic indentation support. (thanks to Matt Pillsbury)
549 * optimization: loading generic functions no longer takes O(n^2) time,
550 proportional to the amount of methods in the generic function
551 (reported by Todd Sabin and Jeremy Brown)
552 * optimization: the FIND and POSITION family of sequence functions
553 are significantly faster on arrays whose element types have been
555 * improvements to the Windows port:
556 ** Intermittent heap corruption problems have been fixed. (thanks
557 to Alastair Bridgewater)
558 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
560 ** Lisp is able to unwind foreign exception frames from alien
561 callbacks. (thanks to Alastair Bridgewater)
563 changes in sbcl-1.0 relative to sbcl-0.9.18:
564 * improvement: experimental support for threading on FreeBSD/x86.
565 (thanks to NIIMI Satoshi)
566 * improvement: runtime option --dynamic-space-size can be used
567 to set the size of the dynamic space reserved on startup.
568 * improvement: floating point modes in effect are now saved in
569 core, and restored on startup.
570 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
571 startup, not time since first call to GET-INTERNAL-REAL-TIME.
572 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
573 threads are not running after *SAVE-HOOKS* have run.
574 * improvement: writes to CLOS instance slots are type-checked in code
575 compiled with (SAFETY 3)
576 * improvement: floating-point exception handling on FreeBSD (thanks to
578 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
579 (thanks to Zach Beane)
580 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
582 * improvement: added support for the Shift-JIS external format.
583 (contributed by NIIMI Satoshi)
584 * improvement: callbacks are supported on Linux/PPC. (thanks to
586 * bug fix: compiler bug triggered by a (non-standard) VALUES
587 declaration in a LET* was fixed. (reported by Kaersten Poeck)
588 * bug fix: file compiler no longer confuses validated and already
589 dumped structurres. (reported by Kaersten Poeck)
590 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
591 fill-pointer signals a type-error as required. (thanks to
593 * bug fix: disassemly of funcallable instances works.
594 * bug fix: single stepping on PPC.
595 * bug fix: fix thread-safety problems in the type system (generally
596 manifesting as nonsensical errors like "STRING is a bad type specifier
597 for sequences" or "The value 1 is not of type FIXNUM")
598 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
599 (reported by Josip Gracin)
600 * bug fix: an error is signaled for attempts to displace arrays with
601 incompatible element types (thanks to Mario Mommer)
602 * bug fix: more correct handling of wide characters in debug info
603 (bug reported by Attila Lendvai and fixed by Juho Snellman)
604 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
605 and don't cause extra consing
606 * optimization: MAP and MAP-INTO are significantly faster on vectors
607 whose elements types have been declared.
608 * Improvements to SB-SPROF:
609 ** Support for allocation profiling
610 ** Reduced profiling overhead, especially for long profiling runs
611 * Improvements to the Windows port:
612 ** floating point exceptions are now reported correctly.
613 ** stack exhaustion detection works partially.
614 ** more accurate GET-INTERNAL-REAL-TIME.
615 ** better RUN-PROGRAM behaviour in case of errors in the child
617 ** PROBE-FILE now simplifies pathnames correctly.
618 ** DIRECTORY now works correctly with :WILD-INFERIORS.
619 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
620 1st 1970 (although time timezone data will be unreliable).
621 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
623 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
624 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
625 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
627 * bug fix: two potential GC deadlocks affecting threaded builds.
628 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
629 returns T (reported by Anton Kazennikov)
630 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
631 with non-variable places
632 * bug fix: remove a race condition in the setting of
633 funcallable-instance functions, this should make threaded CLOS
634 code more stable against memory faults.
635 * bug fix: corruption of specials when unbinding is interrupted by an
636 asynchronous unwind (reported by Hannu Koivisto)
637 * improvement: the debugger will now also display local variables that
638 are only used once, for code compiled with a DEBUG optimization quality
641 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
642 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
643 * incompatible change: External-format support for FFI calls. The
644 SB-ALIEN:C-STRING no longer implies an ASCII
645 external-format. Instead the string is subject to external-format
646 conversion. Additionally return values of type SB-ALIEN:C-STRING
647 are by default of type (SIMPLE-ARRAY CHARACTER), not
648 SIMPLE-BASE-STRING. For an alien type that behaves like the old
649 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
650 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
651 * incompatible change: SB-EXT package no longer contains the
652 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
653 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
655 * incompatible change: the single-stepper is no longer available
656 on Alpha, Mips and Sparc platforms.
657 * minor incompatible change: the direct superclasses of
658 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
659 not (STANDARD-OBJECT FUNCTION). This makes the
660 class-precedence-lists of GENERIC-FUNCTION and
661 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
663 * minor incompatible change: the default stream external format on
664 non-unicode SBCL is no longer detected from the locale, but is always
666 * new feature: Add a version of evaluator that uses an interpreter instead
667 of the compiler. EVAL still uses the compiler by default, to switch it
668 to use the interpreter, set the value of the variable
669 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
670 * minor incompatible change: the single-stepper REPL has been merged
671 with the normal debugger (see the "Stepping" heading of the debugger help
672 for more details). The debugger command STEP will no longer switch
673 to the single-stepper REPL.
674 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
675 for a type now works.
676 * bug fix: loading of default sysinit file works. (thanks to Leonid
678 * bug fix: better detection of circularities in the file-compiler.
679 (reported by Marco Monteiro)
680 * bug fix: the CL pathname functions now work with files that have
681 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
682 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
683 systems with Unix98 pty semantics.
684 * bug fix: ASDF-INSTALL will now work with bsd tar.
685 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
687 * bug fix: timers expiring in dead threads no longer cause a
688 type-error (reported by Paul "Nonny Mouse").
689 * bug fix: thanks to more lightweight single-stepper instrumentation,
690 code compiled with (DEBUG 3) will compile and execute significantly faster,
691 and will have more accurate type-inferencing than before
692 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
693 whose bindings are modified
694 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
695 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
696 ** replace dummy implementations of CL:MACHINE-INSTANCE and
697 CL:SOFTWARE-VERSION with proper version
699 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
700 * feature: implemented the READER-METHOD-CLASS and
701 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
702 as specified by AMOP.
703 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
705 * optimization: faster LOGCOUNT implementation on x86 and x86-64
706 (thanks to Lutz Euler)
707 * optimization: hashing of general arrays and vectors has been
708 improved. (reported by Any Fingerhut)
709 * enhancement: SB-INTROSPECT is now able to find definitions of
710 profiled functions. (thanks to Troels Henriksen)
711 * enhancement: compiler-macro expansion applies now to FUNCALL forms
712 as well, allowing compiler-macros for SETF-functions to expand.
713 * enhancement: step-instrumentation no longer wraps known
714 single-value functions with multiple-value context, allowing
715 better type inference.
716 * fixed bug #337: use of MAKE-METHOD in method combination now works
717 even in the presence of user-defined method classes. (reported by
718 Bruno Haible and Pascal Costanza)
719 * fixed bug #339(c): if there are applicable methods not part of any
720 long-form method-combination group, call INVALID-METHOD-ERROR.
721 (reported by Bruno Haible)
722 * fixed bug #361: the :FUNCTION initarg in the protocol for
723 initialization of methods can now be used to override
724 internally-produced optimized functions. (reported by Bruno
726 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
727 system-provided lambda expression no longer cause warnings about
728 unbound #:|pv-table| symbols.
729 * bug fix: improved the handling of type declarations and the
730 detection of violations for keyword arguments with non-constant
732 * bug fix: potentially erronous calls to PATHNAME and
733 MERGE-PATHNAMES were being flushed in some cases.
734 (reported by Richard Kreuter)
735 * bug fix: compiled calls to TYPEP were mishandling obsolete
736 instances. (reported by James Bielman and Attila Lendvai)
737 * bug fix: format strings with ~> without matching ~< no longer
738 trigger an AVER, but signal an understandable error instead.
739 (reported by Antonio Martinez)
740 * bug fix: specifying an output-file without a pathname-type for
741 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
742 of input-file instead of "fasl". (reported by Robert Dodier)
743 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
744 (reported by James Y Knight).
745 * bug fix: compiler-macros-function did not consider the environment
746 argument for shadowing by local functions.
747 * bug fix: compiler-macros expansion was inhibited by local INLINE
749 * bug fix: inline expansions of known functions were subject to
750 step-instrumentation in high debug policies, leading to problems
752 * bug fix: compiler failed to differentiate between different CONS
754 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
755 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
756 * bug fix: #368: incorrect use of expressed vs. upgraded array
758 * bug fix: #306a: more precise unions of array types.
759 * thread-safety improvements:
760 ** CONDITION-WAIT could return early on Linux, if the thread was
761 interrupted and subsequently continued with SIGCONT.
762 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
764 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
765 * added support for the ucs-2 external format. (contributed by Ivan
767 * minor incompatible change: pretty printing of objects of type
768 (cons symbol) is, in the default pprint-dispatch-table, now
769 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
771 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
772 called later in a class's lifetime, possibly as late as when the
773 first instance of the class is created. Previously,
774 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
775 class became finalizeable.
776 * fixed bug: FILE-POSITION sometimes returned inconsistent results
777 for multibyte external-format streams. (thanks to "vbzoli")
778 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
779 with :ALLOCATION :CLASS inherited from superclasses of the
781 * fixed bug: anonymous classes can now be created using the :NAME
782 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
783 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
784 * fixed bug: core-files saved with :EXECUTABLE T can again be
785 executed when SBCL_HOME isn't set. (reported by James Knight)
786 * fixed bug: toplevel LOCALLY forms with declarations could
787 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
788 * fixed bug: printing from several different threads using different
789 values of *print-case* could cause invalid output, due to
790 some internal special variables of the printer not being bound
791 thread-locally (reported by Max Mikhanosha)
792 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
793 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
794 (reported by Pascal Costanza)
795 * fixed bug: FINALIZE-INHERITANCE is called from
796 REINITIALIZE-INSTANCE on classes when the class has previously
797 been finalized, as required by AMOP.
798 * minor code generation optimizations:
799 ** better register allocation in CLOS dispatching functions
800 ** overflow detection when coercing signed bytes to fixnums on x86-64
801 ** is now implemented with one IMUL instruction instead of three shifts
802 ** more efficient bit-vector access on x86 and x86-64
803 ** more efficient access to raw structure slots on x86 and x86-64
804 * fixed some bugs revealed by Paul Dietz' test suite:
805 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
808 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
809 * feature: thread support on Solaris/x86, and experimental thread support
811 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
813 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
814 default method for SLOT-UNBOUND.
815 * minor incompatible change: prevent the user from specializing the
816 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
817 somewhat counter to the intent of the protocol, I (CSR) think, and
818 additionally it just doesn't work in SBCL as currently
819 implemented, thanks to optimizations (that are always valid for
820 the other three SLOT-VALUEish functions, but not for the setter).
821 * minor incompatible change: the :SB-LDB feature is now enabled by
822 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
823 the low-level debugger.
824 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
825 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
826 down to an SB-SYS:FD-STREAM, allowing direct communication in
827 more cases. Temporary files and pipes are still used as a fallback
829 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
830 on the non-reentrant C functions localtime(3) and gmtime(3).
831 * bug fix: global optimization policy was not visible in LOCALLY and
833 * bug fix: class objects can be used as specializers in methods.
834 (reported by Pascal Costanza)
835 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
836 with a type error. (reported by blitz_ on #lisp)
837 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
838 failed with a type error. (reported by Pascal Bourguignon)
839 * bug fix: merging pathnames against defaults with :DIRECTORY
840 starting with '(:RELATIVE :BACK) should preserve the :BACK.
841 (reported by James Y Knight)
842 * bug fix: saving large (>2GB) cores on x86-64 now works
843 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
844 constant index and a (simple-array (signed-byte 32)) array
845 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
846 error (patch by Robert J. Macomber)
847 * feature: TIME now displays the amount of run-time spent in GC
848 * bug fix: The debugger now does a better job of respecting
849 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
850 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
851 workaround for bug 403.)
852 * bug fix: floating point exception handling now works on FreeBSD
853 (thanks to NIIMI Satoshi)
854 * fixed some bugs revealed by Paul Dietz' test suite:
855 ** MISC.641: LET-conversion were not supposed to work in late
857 * improvements to the Win32/x86 port:
858 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
859 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
861 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
862 * new feature: source path information is generated for macro-expansion
863 errors for use in IDE's like Slime (thanks to Helmut Eller)
864 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
865 * bug fix: compiler does not loop forever on an invalid type in
867 * improvement: compilation of most CLOS applications is significantly
869 * optimization: added a limited bytecode compiler for simple toplevel
870 forms, speeding up compilation and FASL loading
871 * bug fix: the statistical profiler now properly distinguishes anonymous
874 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
875 * minor incompatible change: in sbcl-0.9.11 (but not earlier
876 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
877 into subclasses of the finalized class. Now user calls to
878 FINALIZE-INHERITANCE finalize just the one class, and calls by the
879 system return before any subclasses are finalized.
880 * minor incompatible change: The reader no longer ignores errors
881 regarding non-existent packages in #+ and #- feature tests.
882 * new feature: command line options --no-sysinit, --no-userinit to
883 inhibit loading the corresponding init files
884 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
885 for sending data through UDP sockets (thanks to François-René Rideau)
886 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
887 error is signalled instead (thanks to François-René Rideau)
888 * minor incompatible change: Error signalling behaviour of lexical
889 operations violating package locks has changed slightly. Refer to
890 documentation on package locks for details.
891 * enhancement: EVAL can process IF-expressions without resorting to the
893 * bug fix: Some operations involving SIN, COS, or TAN weren't being
894 constant-folded properly due to missing out-of-line functions.
895 (reported by Mika Pihlajamäki)
896 * bug fix: LISTEN sometimes returned T even in cases where no data was
897 immediately available from the stream
898 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
899 were reversed. (reported by Levente Mészáros)
900 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
901 when given a vector displaced to an adjustable array. (reported
903 * bug fix: derivation of float boundaries from numbers outside the
904 appropriate float range (reported by John Wiseman)
905 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
907 * bug fix: precomputing cache entries for generic functions with
908 some subclasses of specializers as yet invalid does not attempt to
909 fill a cache line with a negative offset. (reported by Levente
911 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
912 allow condition class objects as arguments to DOCUMENTATION and
913 (SETF DOCUMENTATION); only find and set documentation for
914 structure names for the STRUCTURE doc-type. (suggested by Gary
916 * improvements to the Win32/x86 port:
917 ** (user-homedir-pathname) and default initialization file
918 locations now know about the user's "Documents and Settings"
919 directory (thanks to Yaroslav Kavenchuk)
920 ** run-program is implemented (thanks to Mike Thomas)
921 ** sockets support (thanks to Timothy Ritchey)
922 ** better backtrace support (thanks to Alastair Bridgewater)
923 ** sb-grovel supported
924 ** asdf-install and sb-posix work somewhat
925 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
926 * improvements to the Solaris/x86 port:
927 ** works on Solaris 11/Solaris Express
928 ** floating-point exception handling support
929 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
930 * fixed some bugs revealed by Paul Dietz' test suite:
931 ** REMOVE-METHOD returns its generic function argument even when
932 the method is not one of the generic functions' methods.
933 ** objects declared as MEMBER types can be admissible arguments to
935 ** printing characters should simply be printed by the FORMAT ~:C
937 ** compiler failure when compiling functions with hairy constant
938 defaults for optional parameters.
939 ** compiler produces wrong code when MAYBE-INLINE-expanding a
940 function, which is already optimized.
942 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
943 * new platform: experimental support for SBCL x86/Darwin, including
944 MacOS X 10.4.5 on Intel.
945 * new feature: Unicode character names are now known to the system
946 (through CHAR-NAME and NAME-CHAR).
947 * new feature: the filesystem location of SBCL's core file is
948 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
949 * minor incompatible change: the contrib modules SB-POSIX and
950 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
951 this change is to make it easier to distribute
952 location-independent binaries.
953 * bug fix: as implied by AMOP, standardized classes no longer have
954 slots named by external symbols of public packages. (reported by
956 * optimization: calling functions via constant symbols -- as in
957 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
958 via the function object as in (FUNCALL #'FOO).
959 * optimization: on x86 and x86-64, the calling convention for the
960 case of dealing with an unknown number of values has been altered
961 to be friendlier to the prediction heuristics implemented,
962 particularly with reference to CALL and RET pairing. (thanks to
963 Alastair Bridgewater)
964 * optimization: on x86, the code for access to thread-local symbol
965 values and binding/unbinding of thread-local symbols is smaller.
966 (thanks to Alastair Bridgewater)
967 * enhancement: CONSTANTP is now able to determine constantness of
968 more complex forms, including calls to constant-foldable standardized
969 functions and some special forms beyond QUOTE.
970 * fixed bug: occasional GC crashes on Solaris/x86
971 * optimization: x86-64 supports stack allocation of results of simple
972 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
973 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
974 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
975 (thanks to James Knight)
976 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
977 by the standard (thanks to Stephen Compall)
979 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
980 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
981 be used for bundling the runtime and the core file into one
982 executable binary. This feature is not currently supported on all SBCL
983 platforms. (thanks to James Bielman and NIIMI Satoshi)
984 * new feature: a generational or ephemeral garbage collector is now
985 the default on the PowerPC platform (both Linux and Darwin). The
986 old Cheney (stop and copy) collector is a build-time option.
987 (thanks to Cyrus Harmon, after Raymond Toy)
988 * minor incompatible change: the method by which SBCL finds its
989 contributed modules has changed; it no longer relies on symbolic
990 links from an $SBCL_HOME/systems directory, but searches directly
991 in the subdirectories of $SBCL_HOME.
992 * enhancement: the dynamic heap size on the Linux/PPC platform is
993 markedly larger, even using the older Cheney garbage collector.
994 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
995 certain complicated string types. (reported by Gary King)
996 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
997 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
998 * fixed bug: DOCUMENTATION on structures no longer signals an
999 error if no structure type of the right name exists. (reported by
1001 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1002 applicable non-standard primary or :AROUND method on
1003 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1004 no longer cause unbound variable errors. (reported by Kevin Reid)
1005 * fixed bug: metacircle resolution in cases where methods have slots
1006 added before the slots from STANDARD-METHOD. (reported by Jean
1008 * fixed bug: the Power PC assembly code for calling into lisp
1009 sometimes computed its return address wrongly (depending on the
1010 15th bit of the address, fixed at link time). This used to
1011 manifest itself in a segmentation violation while building PCL.
1012 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1013 many others over the years)
1014 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1015 the stack pointer (treated as a signed integer) is negative.
1016 (thanks to Peter van Eynde)
1018 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1019 * new platform: experimental support for the Windows operating
1020 system has been added. (thanks to Alastair Bridgewater)
1021 * fixed several bugs in and robustified the PPC FFI (including
1022 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1023 * bug fix: allow non-simple string symbol names (reported by Paul
1025 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1027 * bug fix: saving a core corrupted callbacks on x86/x86-64
1028 * bug fix: closed a loophole in metacircularity detection and
1029 grounding in the PCL implementation of CLOS.
1030 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1031 (reported by tomppa on #lisp)
1032 * bug fix: clear the direction flag on Lisp -> C transitions, as
1033 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1034 (reported by Andrej Grozin and Hendrik Maryns)
1035 * optimization: major improvements to GC efficiency on GENCGC platforms
1036 * optimization: faster implementation of EQUAL
1037 * optimization: emit more efficient opcodes for some common
1038 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1039 * optimization: several other minor code-generation improvements on x86-64
1040 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1042 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1043 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1044 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1045 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1046 by ANSI as a generic function, and for consistency (SETF
1047 GENERIC-FUNCTION-NAME) is treated likewise.
1048 * fixed bug #233.b: make constraint propagation notice when a variable
1049 value is changed after it is referenced but before it is used
1050 * fixed bug #296: no more arbitrary behaviour from filenames with
1051 odd characters as --load arguments. (reported by Adam Warner)
1052 * fixed bug #390: :CHARACTER-SET pathname components now work as
1053 expected. (reported by Tim Daly Jr)
1054 * fixed bug #391: complicated :TYPE intersections in slot
1055 definitions no longer cause an error in PCL internals.
1056 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1057 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1058 returning the number of octets which would be written to the
1059 file-stream. (thanks to Robert J. Macomber)
1060 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1061 arguments to the internal stream functions. (thanks to David
1063 * bug fix: stack exhaustion is now handled on the x86/SunOS
1064 platform. (thanks to Daisuke Homma)
1065 * bug fix: more accurate ROOM results on GENCGC platforms
1066 * optimization: improved type inference for arithmetic-for
1067 index variables in LOOP
1068 * optimization: faster floating-point SQRT on x86-64
1069 * fixed some bugs revealed by Paul Dietz' test suite:
1070 ** DOCUMENTATION returns NIL instead of "" for method combinations
1071 that don't have a docstring
1073 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1074 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1075 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1076 therefore are not customizeable by user code (as seems to be at
1077 least permitted and maybe required by AMOP). As a consolation,
1078 however, the SBCL implementation of these functions now calls
1079 REINITIALIZE-INSTANCE as specified by AMOP.
1080 * bug fix: slot-definition documentation is propagated as per ANSI
1081 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1082 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1083 argument, as do eslotd initialization methods. (from Pascal
1084 Costanza's "Closer" project)
1085 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1086 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1088 * bug fix: it is now possible to have more than one subclass of
1089 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1090 (reported by Bruno Haible, Pascal Costanza and others)
1091 * bug fix: the dependent update protocol now works for generic
1092 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1093 and Pascal Costanza)
1094 * bug fix: condition-class instances corresponding to
1095 DEFINE-CONDITION forms are now created eagerly. (reported by
1096 Kalle Olavi Niemitalo on comp.lang.lisp)
1097 * bug fix: floating point printing is more accurate in some
1098 circumstances. (thanks to Simon Alexander)
1099 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1100 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1101 * bug fix: callbacks on OS X now preserve stack-alignment.
1102 (thanks to Cyrus Harmon)
1103 * enhancement: the x86-64 disassembler is much better at
1104 disassembling SSE instructions. (thanks to Lutz Euler)
1105 * enhancement: CHECK-TYPE on a variable now allows type-inference
1106 (thanks to Vincent Arkesteijn)
1107 * optimization: improved performance of EUC-JP external format.
1108 (thanks to NIIMI Satoshi)
1109 * optimization: performance improvements to IO on file streams of
1110 :ELEMENT-TYPE CHARACTER
1111 * optimization: much faster memory allocation on x86-64
1112 * optimization: faster garbage collections (latency and throughput)
1114 * optimization: faster bignum subtraction and division on x86 and x86-64
1115 (thanks to Lutz Euler)
1116 * optimization: more accurate type-inference for loops that use a
1117 floating point index variable or a negative step.
1119 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1120 * bug fix: add a workaround to SBCL looping infinitely at startup on
1121 Linux kernels with apparently buggy implementations of personality().
1122 (thanks to Svein Ove Aas)
1123 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1124 (thanks to David Lichteblau)
1125 * bug fix: local bindings shadow global symbol macros for
1126 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1127 * optimization: non-open coded uses of numeric comparison operators
1128 (e.g. >) no longer cons when called with more than one parameter
1129 on platforms supporting dynamic-extent allocation.
1130 * enhancement: saving cores with foreign code loaded is now supported
1131 on MIPS/Linux in addition to the previously supported platforms.
1132 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1133 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1134 * bug fix: division by zero in sb-sprof when no samples were collected
1135 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1136 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1137 (thanks to Alexander Kjeldaas)
1138 * new feature: ignoring the return values of destructive functions
1139 like NREVERSE signals a compile-time style-warning.
1140 (thanks to Kevin Reid)
1141 * new feature: the alignment of alien structure fields can be
1142 explicitly specified. (thanks to Cyrus Harmon)
1143 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1144 and dump core on SIGQUIT
1146 ** incompatible change: threads do not inherit values of specials
1147 from their parents (see manual)
1148 ** bug fix: threads stacks belonging to dead threads are freed by the
1149 next exiting thread, no need to gc to collect thread stacks anymore
1150 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1151 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1152 ** bug fix: dynamic variable and thread start related gc lossage
1153 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1155 * fixed some bugs revealed by Paul Dietz' test suite:
1156 ** SUBTYPEP is slightly more accurate on heinously complicated
1157 CONS types where some of the members have uncertain (in the
1158 NIL, NIL sense) type relationships to each other.
1160 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1161 no longer purified unless :PURIFY T is explicitly specified.
1162 ** Non-purified cores are significantly smaller than before
1164 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1165 * new feature: timers based on Zach Beane's excellent timer package
1166 * added support for the following external formats: koi8-u,
1167 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1168 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1169 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1170 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1171 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1172 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1173 * incompatible change: a threaded SBCL will no longer revert to
1174 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1175 * bug fix: interrupts are disabled until startup is complete; no
1176 more sigsegvs when receiving a signal to soon
1177 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1179 * bug fix: add a workaround for the memory randomization features in
1180 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1181 workaround will only be in effect on systems with the proc filesystem
1183 * bug fix: printing objects of type HASH-TABLE signals a
1184 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1186 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1187 have been running for over 50 days. (reported by Gilbert Baumann)
1188 * bug fix: the logic for getting names of functions gets less
1189 confused when confronded with alternate-metaclass
1190 funcallable-instances. (reported by Cyrus Harmon)
1191 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1192 consistent, even on internal alternate-metaclass objects.
1193 * bug fix: flush closure information collected by physical
1194 environment analysis prepass before the main pass. (bug reported
1196 * bug fix: compiler pack phase does not modify a hashtable, which is
1197 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1198 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1199 may be expanded. (reported by Alexander <malishev> on c.l.l)
1200 * bug fix: no more style-warnings when compiling code using LAST
1201 under high-SPEED low-SPACE settings. (reported by David Wallin
1203 * fixed some incompatibilities between SBCL's MOP and the MOP
1205 ** the METAOBJECT class is now implemented;
1206 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1207 STANDARD-OBJECT, as required;
1208 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1209 now compatible (as required by VALIDATE-SUPERCLASS); there
1210 remains a consistency requirement over the presence or absence
1211 of the FUNCTION class in the superclass list of finalized
1212 classes; see the manual for more details;
1213 ** the :AROUND method for COMPUTE-SLOTS on
1214 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1215 requested slot ordering.
1217 ** bug fix: parent thread now can be gc'ed even with a live
1219 ** bug fix: binding a special with PROGV to no value is not
1220 the same as that symbol not having been bound (thanks to
1222 ** bug fix: binding specials is thread safe (thanks to
1224 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1225 and SB-SPROF do not die with 'no handler for signal XX in
1226 interrupt_handle_now(..)' anymore
1227 ** bug fix: WITH-TIMEOUT works with multiple running threads
1228 * fixed some bugs revealed by Paul Dietz' test suite:
1229 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1230 the :method-class keyword argument.
1232 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1233 * new port: the Solaris operating system on x86 processors is now
1234 mostly supported, though some rough edges in the environment
1235 remain. (thanks to Daisuke Homma)
1236 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1237 space for its heap. (thanks to Thiemo Seufer)
1238 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1239 closures. (thanks to Thiemo Seufer)
1240 * minor incompatible change: eof selects abort in the debugger.
1241 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1242 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1244 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1245 deprecated, and will go away in a future revision of SBCL.
1246 * minor incompatible change: GC-ON and GC-OFF are no longer
1247 implemented with a counter, it does not matter how many times gc
1248 is switched on or off
1249 * bug fix: discriminating functions for generic function classes
1250 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1251 make invalid assumptions about method precedence order. (reported
1253 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1254 platforms. (thanks to Thiemo Seufer)
1255 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1256 * bug fix: correct pathname printing with printer escaping is on.
1257 (thanks to Kevin Reid)
1258 * bug fix: complex VOP definitions in "user-space" no longer trigger
1259 package locks. (reported by Zach Beane)
1260 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1261 for funcallable-instance functions, and is no different from
1263 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1264 characters in the current readtable. (reported by Nicholas Neuss)
1265 * bug fix: SUBTYPEP on various CONS types returns more a more
1266 accurate acknowledgment of its certainty.
1267 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1268 lists in some cases. This partially fixes bug 384.
1269 * flush all standard streams before prompting in the REPL and the
1271 * bug fix: signal handling and triggering gc do not conflict
1272 directly anymore, in particular a high frequency sb-sprof does
1273 not prevent gc from running
1274 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1275 approximation for timezone and DST information between the
1276 universal time epoch and the smallest negative 32-bit time_t.
1277 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1278 year 1899 so long as the encoded time is non-negative.
1279 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1280 the correct answer for non-constant rotate amounts when performing
1281 an inline 32-bit rotation.
1283 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1284 there is only one thread in the session
1285 ** bug fix: memory leak for streams created in one thread and
1286 written to in another
1287 ** bug fix: lockup when compiled with gcc4
1288 ** bug fix: race that allows the gc to be triggered when gc is
1290 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1291 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1293 ** bug fix: finished threads are gc'ed properly
1294 * fixed some bugs revealed by Paul Dietz' test suite:
1295 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1296 has arguments with a different set of applicable methods from
1297 the orignal arguments.
1298 ** The type error thrown by MAP now has a correct expected-type
1300 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1301 name a compiled function.
1302 ** Three MISC tests where a large bignum was improperly coerced to
1303 a float that couldn't represent that bignum during type
1304 derivation were fixed.
1305 ** SUBTYPEP can now handle types involving the negation of a
1306 list-form FUNCTION type.
1307 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1308 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1309 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1311 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1312 * New feature: Experimental support for bivalent streams: streams
1313 opened with :element-type :default now allow character and binary
1314 (unsigned-byte 8) I/O
1315 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1316 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1318 * bug fix: no more highly sporadic "couldn't check whether ~S is
1319 readable" when reading a stream and an interrupt hits in the middle
1320 of a select system call
1321 * compiler better recognizes complex arrays (reported by Eduardo
1323 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1324 for variables on SBCL built with linkage-tables. (reported by Luis
1326 * various error reporting improvements.
1327 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1328 (Thanks to Thiemo Seufer)
1329 * bug fix: nested reader invokations work correctly
1330 * bug fix: it is possible to have simultaneous references to foreign
1331 code and foreign data with the same name.
1333 ** added x86-64 support
1334 ** incompatible change: the threading api now works with thread
1335 objects instead of thread ids
1336 ** bug fix: threads are protected from signals and interruption when
1337 starting up or going down
1338 ** bug fix: a race where an exiting thread could lose its stack to gc
1339 ** bug fix: don't halt on infinite error in threads if possible
1340 ** fixed numerous gc deadlocks introduced in the pthread merge
1341 ** bug fix: fixed thread safety issues in read and print
1342 ** bug fix: debugger doesn't hang on session lock if interrupted at
1343 an inappropriate moment
1344 ** bug fix: run-program is now thread safe(r)
1345 ** bug fix: inner with-recursive-lock no longer releases the mutex
1346 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1347 * fixed some bugs revealed by Paul Dietz' test suite:
1348 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1349 EXPECTED-TYPE slots filled.
1350 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1352 ** MAKE-CONDITION accepts classes as type-designators.
1353 ** COMPILE may never return NIL.
1354 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1355 range before calling Unix time functions
1357 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1358 * numerous signal handling fixes to increase stability
1359 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1360 * minor incompatible change: we now correctly canonize default
1361 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1362 as per the MOP, rather than the historical (INITARG INITFUNCTION
1363 INITFORM). (reported by Bruno Haible)
1364 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1365 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1366 for more information.
1367 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1368 pathname is a directory pathname.
1369 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1370 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1372 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1373 :METACLASS argument in addition to a class name. (reported by
1374 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1375 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1376 (reported by Bruno Haible; patch from Gerd Moellmann)
1377 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1379 * bug fix: more cleanups to the floating point exception handling on
1380 x86-64 (thanks to James Knight)
1381 * bug fix: the compiler does not try to constant fold calls of
1382 COERCE and BIT-* functions when they return freshly constructed
1383 objects. (reported by Nikodemus Siivola and Paul Dietz)
1384 * optimization: improved the MIPS versions of generic (in the
1385 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1386 * optimization: direct conversion of (unsigned-byte 32) to floats on
1387 the PowerPC platform.
1388 * optimization: structure instances with raw slots now use less
1389 memory, and probably show better memory locality. (thanks to
1391 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1392 are treated as requests for stack allocation on the x86-64,
1393 Alpha-32, PPC and SPARC platforms.
1394 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1395 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1396 (thanks to Tony Martinez)
1398 ** gcing a dead thread can no longer lead to lockups
1399 ** threads block signals until they are set up properly
1400 ** errno is no longer shared by threads
1401 ** interrupt-thread restores the eflags register on x86
1402 ** fixed some lockups due to gc/thread interaction
1403 * fixed some bugs revealed by Paul Dietz' test suite:
1404 ** invalid dotted lists no longer raise a read error when
1405 *READ-SUPPRESS* is T
1406 ** signal an error if a symbol that names a declaration is used
1407 as the name of a type, or vice versa
1408 ** allow using the (declare (typespec var*)) abbreviation for
1409 (declare (type typespec var*)) with all type specifiers
1410 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1411 ** always return NIL from PPRINT-POP when OBJECT is NIL
1412 ** don't signal errors when pretty-printing malformed LABELS,
1413 FLET or MACROLET forms
1414 ** declarations in a DOLIST body are also in scope for the
1416 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1418 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1421 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1422 * fixed cross-compiler leakages that prevented building a 32-bit
1423 target with a 64-bit host compiler.
1424 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1425 opened with :IF-EXISTS :OVERWRITE.
1426 * fixed bug 281: error for an invalid qualifier in a short-form method
1427 combination method is not signalled until the faulty method is called.
1428 * bug fix: iteration variable type inferrer failed to deal with open
1429 intervals. (reported by Alan Shields)
1430 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1431 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1432 or more runtime options were provided to the sbcl binary.
1433 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1434 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1435 * optimization: there's now a fast-path for fixnum arguments in the
1436 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1437 * optimization: the code generated on x86-64 is more compact thanks
1438 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1439 * bug fix: floating-point exception handling now partly works on
1440 x86-64. (Thanks to James Knight)
1441 * improvement to the MIPS backend from Thiemo Seufer: C-style
1442 64-bit long long arguments and return values to alien functions
1444 * fixed some bugs revealed by Paul Dietz' test suite:
1445 ** the type-error signalled from WARN has a filled-in DATUM slot.
1446 ** the type-error required when a stream is not associated with
1447 a file has the stream as its datum.
1448 ** type-errors on single-floats on x86-64 no longer have
1449 :INVALID-OBJECT as the datum
1450 ** the type-errors signalled for invalid function names now have
1451 a correct expected type
1452 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1453 for structures defined with a :TYPE. Documentation strings for
1454 typed structures are no longer immediately discarded
1455 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1456 broadcast streams and synonym streams. FILE-LENGTH now also works
1457 on broadcast streams.
1459 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1460 * incompatible change: the --noprogrammer option, deprecated since
1461 version 0.7.5, has been removed. Please use the equivalent
1462 --disable-debugger option instead.
1463 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1465 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1466 inoperational for a while now) has been completely removed.
1467 * null lexical environments are now printed as #<NULL-LEXENV>,
1468 significantly reducing the amount of clutter in typical backtraces.
1469 * documentation on weak pointers, finalization, and after GC hooks
1470 has been added to the manual.
1471 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1472 as other specialized array types, is much faster. SUBSEQ and
1473 COPY-SEQ on such arrays have also been sped up.
1474 * optimization: EQL is now more efficient when at least other argument
1475 is known to be of type (OR FIXNUM (NOT NUMBER)).
1476 * fixed bug: the runtime is now less vulnerable to changes in the
1477 size of the SBCL object on OS X, and virtual memory is reserved for
1478 all spaces that need to be at a fixed address.
1479 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1480 * fixed bug: finalizers and after GC hooks that cause consing are now
1482 * fixed bug: compiler error messages and summaries are now printed to
1483 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1484 * fixed inference of the upper bound of an iteration variable.
1485 (reported by Rajat Datta).
1486 * fixed bug 211e: calling local functions with duplicated constant
1487 keyword argument no longer causes a bogus style warning about an
1489 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1490 ftype declaration to be disregarded. (reported by Dave Roberts)
1491 * fixed bug 373: caused by erronous compilation of references to alien
1492 variables in the runtime on ppc/darwin.
1493 * fixed bug 376: CONJUGATE type deriver.
1494 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1495 (reported by Baughn on #lisp)
1496 * a cross-compiler bug on non-x86 platforms has been identified and
1497 fixed. (thanks to Bruno Haible)
1498 * improvements to the MIPS runtime code for increased stability.
1499 (thanks to Thiemo Seufer)
1500 * increased the maximimum compact environment size to allow
1501 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1502 * partial workaround for bug 135: don't name the function we're
1503 calling for hairy cases of EVAL, so as not to accumulate one environment
1504 entry per call to EVAL. (reported by Kevin Reid)
1505 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1506 * optimization: type testing for non-vector arrays should be faster.
1507 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1509 * bug fix: setting 31st element of a bit vector to zero did not work
1511 * bug fix: redefining a class definition which failed due to a
1512 previous accessor / function clash now works (but see BUGS entry
1513 #380 for more problems in this area). (thanks to Zach Beane)
1514 * the long form of DEFINE-METHOD-COMBINATION disables method group
1515 checking when given a single method group with pattern *. (thanks
1517 * on x86 compiler supports stack allocation of results of simple
1518 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1519 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1520 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1521 * fixed some bugs related to Unicode integration:
1522 ** the restarts for recovering from input and output encoding
1523 errors only appear when there is in fact such an error to
1525 * fixed some bugs revealed by Paul Dietz' test suite:
1526 ** MISC.549 and similar: late transformation of unsafe type
1527 assertions into derived types caused unexpected code
1529 ** SCALE-FLOAT type deriver is less wrong.
1530 ** type derivers for EXP, LOG and similar functions compute result
1531 types for complex arguments better.
1532 ** (MISC.563) CONJUGATE type deriver works for very restricted
1534 ** out-of-line type testers for character strings are available.
1535 ** EQUAL compiler transform understands specialness of objects
1537 ** accessing double-floats stored on the stack now works on x86-64.
1538 ** debugger internals could sometimes create invalid lispobjs,
1539 resulting in GC crashes.
1540 ** MISC.548: type check weakening can convert required type into
1542 ** initialization forms for bindings are not in scope of free special
1545 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1546 * incompatible change: thread support for non-NPTL systems has
1547 been removed - locking is buggy and unreliable. A threaded
1548 SBCL build will now warn at startup and refuse to create
1549 new threads, unless futex support is detected in the kernel
1550 * incompatible change: the top level REPL now has only an ABORT
1551 restart associated with it, not TOPLEVEL and ABORT as it used to.
1552 TOP and TOPLEVEL are now available as debugger commands for
1553 returning to the top level.
1554 * incompatible change: forms evaluated in the REPL now use the
1555 global optimization policy.
1556 * incompatible change: user- and system-initialization files are
1557 no longer processed with LOAD, but by READ and EVAL; hence the
1558 global optimization policy, startup package, readtable, etc,
1560 * The .fasl file format number has been incremented because of
1561 various incompatible changes.
1562 * internal entry point details and argument counts no longer appear
1563 in backtraces unless explicitly requested by setting
1564 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1565 * built-in and standard functions no longer have names like "top
1566 level local call to FOO".
1567 * fixed bug 32: functions defined in non-null lexical environments
1568 now have more legible printed representation
1569 * fixed bug 33: functions defined in non-null lexical environemnts
1570 are now more amenable to inspection by INSPECT.
1571 * workaround for bug 354: XEPs no longer appear in backtraces unless
1572 explicitly requested.
1573 * fixed bug: receiving the signal which results from attempting to
1574 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1575 SIGBUS on other *BSDs) on architectures where the C stack is also
1576 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1577 to debugger confusion or wild instability. (reported by Cheuksan
1579 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1580 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1581 (reported by Lutz Euler)
1582 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1583 compiled code. (reported by Paul Dietz)
1584 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1585 specialized parameter to a method to something that is not TYPEP
1586 the specializer is now possible.
1587 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1588 face of package deletion.
1589 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1590 pathname. (reported by David Tolpin on c.l.l)
1591 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1592 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1593 STRING-STREAM and FILE-STREAM, respectively.
1594 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1595 than previously; in addition, its test suite is now run on build.
1596 (thanks to Luke Gorrie)
1597 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1598 * optimization: added a immediate representation for single-floats
1600 * fixed some bugs revealed by Paul Dietz' test suite:
1601 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1602 ** attempting to create a package with a colliding nickname causes
1603 correctable errors to be signalled.
1604 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1605 ** several x86-64 backend bugs related to sign-extension of immediate
1608 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1609 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1610 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1612 * fixed loading of multiply forward-referenced layouts.
1613 (thanks to Cheuksan Wang)
1614 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1615 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1616 (see COMPILE-FILE documentation for details of :PRINT option)
1617 * fixed bugs 19 and 317: fixed-format floating point printing is
1618 more accurate. This also fixes a bug reported by Adam Warner
1619 related to the ~@F format directive.
1620 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1622 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1623 dispatch table if the from-char is a dispatch macro character.
1624 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1625 by the wordsize no longer ignore the last word. (reported by Lutz
1627 * fixed bug in type checking of dynamic-extent variables. (reported
1629 * optimization: sequence traversal functions use their freedom to
1630 coerce function designators to functions.
1631 * optimization: code with many calls to CLOS methods specialized on
1632 CLOS classes has had redundant type checks removed.
1633 * optimization: added declarations to speed up operations that access
1634 the internal character database (for example STRING-UPCASE)
1635 * optimization: comparison operations between floats and sufficiently small
1636 fixnums no longer create extra rationals
1637 * fixed some bugs related to Unicode integration:
1638 ** portions of multibyte characters at the end of buffers for
1639 character-based file input are correctly transferred to the
1640 start of the buffer at the next read.
1641 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1642 passing it through to OPEN.
1643 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1644 argument given to internal calls to OPEN.
1645 ** the built-in comment readers (introduced by character sequences
1646 ";" and "#|") are more forgiving to encoding errors; they will
1647 STYLE-WARN and then attempt to resync the stream at a character
1648 boundary. (thanks to Teemu Kalvas)
1649 * fixed some bugs in the x86-64 port:
1650 ** Negative short int return values from c-calls are sign-extended
1652 ** The stack is aligned to 16-bytes for c-calls, as required by
1653 the ABI. (thanks to Cheuksan Wang)
1654 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1655 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1656 for enums. (thanks to Vincent Arkesteijn)
1657 ** Multiple small optimizations and bugfixes for floating point
1659 * fixed some bugs revealed by Paul Dietz' test suite:
1660 ** Space, Tab, Linefeed, Return and Page have the invalid
1661 secondary constituent character trait.
1662 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1664 ** WITH-INPUT-FROM-STRING should only update the index place on
1666 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1667 works more reliably.
1668 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1669 with one-dimensional arrays with fill pointers.
1670 ** TYPE-OF failed on a complex with an integer realpart and a
1672 ** compiler failure during type inference for the code of form
1673 (IF (EQL X (THE ...)) ...) (MISC.535).
1675 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1676 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1677 platform. The port supports 61-bit fixnums, large memory spaces
1678 and reloading shared object files.
1679 * enhancement: saving cores with foreign code loaded is now
1680 supported on ppc/Darwin in addition to the previously supported
1682 * enhancement: the statistical profiler now walks deeper into the
1683 call stack for more meaningful call-graphs and accrued time
1684 reports (x86/x86-64 only). It also now reports time spent in
1686 * enhancement: it is now possible to trace most individual methods
1687 of a generic function in addition to tracing the generic function
1689 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1691 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1692 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1693 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1694 produces an error. (thanks to Vincent Arkesteijn)
1695 * bug fix: NAMESTRING on pathnames with :WILD components in their
1696 directories works correctly. (thanks to Artem V. Andreev)
1697 * fixed bug 125: compiler preserves identity of closures. (reported
1699 * bug fixed: functions with &REST arguments sometimes failed with
1700 "Undefined function" when compiled with (DEBUG 3). (reported by
1702 * bug fix: overflow during compiling of setting element of a bit
1703 vector with constant index and value. (reported by Timmy Douglas)
1704 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1705 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1706 * on x86 compiler partially supports stack allocation of dynamic-extent
1708 * GO and RETURN-FROM do not check the extent of their exit points
1709 when compiled with SAFETY 0.
1710 * fixed some bugs related to Unicode integration:
1711 ** encoding and decoding errors are now much more robustly
1712 handled; it should now be possible to recover even from invalid
1713 input or output to the terminal. (thanks to Teemu Kalvas)
1714 ** provided a first cut at implementing STRING-TO-OCTETS and
1715 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1716 ** altered the SB-MD5 contributed module slightly, changing the
1717 interface just enough for it to be supportable for builds where
1718 lisp characters are not eight bits.
1719 * fixed some bugs revealed by Paul Dietz' test suite:
1720 ** the FORMATTER-generated functions for ~V[ conditionals require
1721 the correct number of arguments.
1722 ** READ-FROM-STRING returns the mandated second value when applied
1723 to displaced strings.
1724 ** the #\Rubout and #\Backspace characters are treated as invalid
1725 constituent characters by the tokenizer.
1727 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1728 * new feature: reloading changed shared object files with
1729 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1730 * new feature: references to foreign variables and functions
1731 can now be compiled and loaded before the corresponding shared
1732 object file is loaded, as long as the foreign definitions are
1733 available at runtime.
1734 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1735 supported. (thanks to Dan Debertin)
1736 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1737 just keywords. (thanks to Vincent Arkesteijn)
1738 * fixed bug #331: structure-class instances corresponding to
1739 DEFSTRUCT forms are now created eagerly.
1740 * fixed bug #345: backtraces from calls to undefined functions work
1741 on x86 as well. Related bug #61 is now also partially fixed on x86
1742 (backtraces from throws to unknown catch tags.)
1743 * bug fix: lambda-list parsing is now stricter vrt. order and number
1744 of lambda-list keywords.
1745 * bug fix: as specified by AMOP, an error is signalled if a
1746 class-option appears multiple times in a DEFCLASS form. (reported
1748 * bug fix: RANDOM can be compiled when the compiler derives the type
1749 of its numeric argument as a disjoint set of small integers.
1750 (reported by Paul Dietz)
1751 * bug fix: starting a core saved with shared objects loaded when
1752 those objects are not available no longer causes threaded SBCL to
1753 hang. (reported by Sean Ross)
1754 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1756 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1757 argument, which must always be NIL. (reported by Kalle Niemitalo)
1758 * bug fix: printing 1.0d+23 no longer results in an error.
1759 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1760 * bug fix: structure slot setters preserve evaluation order. (thanks
1762 * fixed some bugs related to Unicode integration:
1763 ** RUN-PROGRAM can allow its child to take input from a Lisp
1764 stream. (reported by Stefan Scholl)
1765 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1766 locales. (reported by Ken Causey)
1767 * fixed some bugs revealed by Paul Dietz' test suite:
1768 ** INCF, DECF and REMF evaluate their place form as specified in
1770 ** FORMATTER expands ~{ ~} iteration directives with V or #
1771 parameters correctly.
1772 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1773 consequent uses no arguments correctly.
1774 ** the system has a partial understanding of the (COMPLEX RATIO)
1776 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1779 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1780 * new feature: a build-time option (controlled by the :SB-UNICODE
1781 keyword feature, enabled by default) for building the system with
1782 support for the entire 21-bit character space defined by the
1784 * new feature: the system now has rudimentary external-format
1785 support; the primary user-visible change at this time is that
1786 characters with the high bit set (such as lower-case-e-acute) will
1787 print correctly to a terminal in a UTF-8 environment.
1788 * minor incompatible change: BASE-CHAR no longer names a class;
1789 however, CHARACTER continues to do so, as required by ANSI.
1790 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1791 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1792 has been moved to the SB-EXT package (temporarily re-exported from
1794 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1795 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1797 * fixed bug #351: better error-handlind and reporting for malformed
1799 * fixed bug #350: bignum-printing is now more memory-efficient,
1800 allowing printing of very large bignums, eg. (expt 2 10000000).
1801 (reported by Bruno Haible)
1802 * fixed bug #302: better primitive-type selection for intersection
1804 * fixed bug #308: non-graphic characters now all have names, as
1805 required. (reported by Bruno Haible)
1806 * bug fix: redefining a class with different superclasses now correctly
1807 removes it from the direct-subclasses of its previous superclasses.
1808 (reported by David Morse)
1809 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1810 new value now works. (reported by Bruno Haible)
1811 * bug fix: correct canonicalization of multiple non-standard slot
1812 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1813 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1814 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1815 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1816 now exists, an signals an error.
1817 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1818 during macroexpansion for non-iteration variables as well. (reported
1819 by Bruno Haible for CMUCL)
1820 * bug fix: Cyclic structures and unprintable objects in compiler
1821 messages no longer cause errors. (reported by Bruno Haible)
1822 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1823 and READ-FROM-STRING all now return a primary value of NIL if
1824 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1825 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1826 (reported by Bruno Haible for CMUCL)
1827 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1828 do not designate a condition. (reported by Bruno Haible for
1830 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1831 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1832 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1833 permitting resolution in favour of any of the conflicting symbols.
1834 (reported by Bruno Haible for CMUCL)
1835 * bug fix: EQUAL compiler optimizations is less aggressive on
1836 strings which can potentially compare true despite having distinct
1837 specialized array element types.
1838 * bug fix: unit enumerations can be defined without dividing by
1839 zero. (thanks to Vincent Arkesteijn)
1840 * FORMAT compile-time argument count checking has been enhanced.
1841 (report from Bruno Haible for CMUCL)
1842 * a partial workaround for the bug 262: the compiler does not try to
1843 inline-expand a local function doing RETURN-FROM from a deleted
1844 BLOCK. (thanks to Peter Denno for the bug report and to David
1845 Wragg for the simple test case)
1846 * fixed some bugs revealed by Paul Dietz' test suite:
1847 ** RENAME-PACKAGE allows all package designators as new package
1849 ** constraint propagation and lambda variable substitution are
1850 more cautious in dealing with partially deleted code.
1851 ** compiler sometimes lost reoptimization passes.
1852 ** CERROR, given a condition as condition designator, passes its
1853 remaining arguments to the continue format control without
1855 ** Case-altering FORMAT directives work correctly on non-ASCII
1857 ** The REAL type specifier handles bounds outside the floating
1858 point ranges without signalling FLOATING-POINT-OVERFLOW.
1859 ** Functions with IR1-transformations can create intercomponent
1860 references to global functions.
1861 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1863 ** FORMAT ~R treats a nil value for its first parameter correctly.
1865 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1866 * enhancement: saving cores with foreign code loaded is now
1867 supported on x86/NetBSD and sparc/Linux in addition to the previously
1868 supported platforms.
1869 * bug fix: on some platforms repeated installations caused multiple
1870 copies of HTML documentation to be installed -- should not happen
1871 any more. (reported by Stefan Scholl)
1872 * bug fix: parsing self-recursive alien record types multiple times
1873 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1874 original patch by Helmut Eller for CMUCL)
1875 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1876 (thanks to Richard Kreuter)
1877 * bug fix: defining classes whose accessors are methods on existing
1878 generic functions in other (locked) packages no longer signals
1879 bogus package lock violations. (reported by François-René Rideau)
1880 * bug fix: special variables as DEFMETHOD parameters no longer have
1881 associated bogus type declarations. (reported by David Wragg and
1883 * bug fix: read-write consistency on streams of element-type
1884 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1885 * bug fix: redefiniton of the only method of a generic function with
1886 no DEFGENERIC no longer emits a full WARNING. In addition,
1887 redefinition of generic functions with no DEFGENERIC to an
1888 incompatible lambda list now signals an error. (thanks to Zach
1890 * bug fix: DEFGENERIC now works even when there's a function of the
1891 same name in an enclosing lexical environment. (thanks to Zach
1893 * fixed compiler failure, caused by instrumenting code during
1894 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1895 * optimization: added loop analysis and improved register allocation
1896 * fixed some bugs revealed by Paul Dietz' test suite:
1897 ** POSITION on displaced vectors with non-zero displacement
1898 returns the right answer.
1899 ** (SIMPLE-STRING) is a valid type specifier for sequence
1901 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1903 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1904 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1906 ** The justification version of the FORMAT ~< directive treats
1907 non-zero minpad parameter correctly.
1909 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1910 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1911 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1912 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1913 the supported interface.
1914 * new feature: Single-stepping of code compiled with DEBUG 2 or
1915 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1916 * new feature: saving cores with foreign code loaded is now
1917 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1918 Timothy Moore's work for CMUCL)
1919 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1920 parameters to * instead of NIL if no initform is supplied.
1921 (reported by Johan Bockgaard)
1922 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1923 correctly when FUNCALL appears as the car of the form. Note:
1924 despite this FUNCALL forms are not currently subject to
1925 compiler-macro expansion. (port of Raymond Toy's fix for the
1926 same from CMUCL, reported by Johan Bockgaard)
1927 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1928 (thanks for Teemu Kalvas)
1929 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1930 case instead of terminating the iteration (thanks for Julian
1931 Squires, Sean Champ and Raymond Toy)
1932 * bug fix: incorrect expansion of defgeneric that caused a style
1933 warning. (thanks for Zach Beane)
1934 * bug fix: slot accessor effective method computation works properly
1935 for classes with multiple non-standard applicable methods on
1936 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1937 * on x86 compiler supports stack allocation of results of LIST and
1938 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1939 CMUCL implementation by Gerd Moellmann)
1940 * fixed some bugs revealed by Paul Dietz' test suite:
1941 ** FORMAT strings with both the ~<~:;~> form of the justification
1942 directive and pretty-printing directives cause an error.
1944 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1945 * incompatible change: the internal functions
1946 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1947 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1948 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1949 instead of the old functions.
1950 * new feature: on platforms where "dladdr" is available foreign
1951 function names now appear in backtraces. (based on Helmut Eller's
1953 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1954 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1956 * documentation: ASDF manual is now distributed with SBCL.
1957 * bug fix: SBCL can now load its contributed modules with REQUIRE
1958 even if the system-provided entries have been removed from
1959 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1961 * bug fix: backtraces involving undefined functions or assembly
1962 routines are more informative. (thanks to Brian Downing)
1963 * bug fix: mutually referent alien structures now work correctly.
1964 (reported by Rick Taube)
1965 * bug fix: structures defined by WITH-ALIEN can be referred to
1966 within other definitions in the same WITH-ALIEN.
1967 * bug fix: division operators (MOD, TRUNCATE and the like) with
1968 constant zero divisors and integer dividends no longer generate
1970 * bug fix: provide default methods for INPUT-STREAM-P and
1971 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1972 * bug fix: improve the detection and resolution of MOP metacycles
1973 (where effective method computation depends on the generic function
1974 having its effective method computed). (reported by Bruno Haible)
1975 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1976 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1977 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1978 * bug fix: pathnames with relative directory components can be
1979 represented relative to default pathnames.
1980 * optimization: in taking the GCD of bignums, reduce the two bignums
1981 to approximately the same size (using Euclid's algorithm) before
1982 applying the more sophisticated binary GCD. (thanks to Juho
1984 * optimization: COUNT on bitvectors now operates word-at-a-time.
1985 * optimization: ASH with a positive, but not necessarily constant,
1986 (leftwards) shift, when in a modular context, is compiled to a
1988 * fixed some bugs revealed by Paul Dietz' test suite:
1989 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1990 if the corresponding argument is NIL.
1991 ** FORMAT directives accepting a minpad parameter treat negative
1993 ** literal commas as character parameters to FORMAT directives are
1995 ** literal spaces directly after ~<Newline> directives within a
1996 format-logical-block (~:< ~@:>) do not induce :FILL-style
1997 conditional newlines.
1998 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1999 as specified: it no longer includes conditional newlines).
2000 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2002 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2003 :PER-LINE-PREFIX argument does not evaluate to a string.
2005 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2006 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2007 the manual for details; for now, package locks can be disabled by
2008 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2009 if no major problems are found then it is likely that they will be
2010 compiled in unconditionally.
2011 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2012 unsupported operators on all platforms. To load a shared library
2013 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2014 object file, link it into a shared library outside of SBCL and
2015 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2017 * bug workaround: ROOM T and the GC were not working together
2018 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2019 reporting facility aren't preserved. That reporting has been
2020 disabled, so now until and unless someone figures out how to make it
2021 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2022 * minor incompatible change: as threatened around sbcl-0.8.0, the
2023 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2024 The symbols are also exported from SB-PCL for backwards
2025 compatibility, but more so than before SB-PCL should be treated as
2026 an implementation-internal package.
2027 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2029 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2030 It is known as of this release not to work on the Alpha, however.
2031 * fixed bug #167: errors signalled due to illegal syntax in method
2032 bodies are now more legible.
2033 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2034 designators and can hence be used with TYPEP.
2035 * fixed bug #333: CHECK-TYPE now ensures that the type error
2036 signalled, if any, has the right object to be accessed by
2037 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2038 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2039 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2041 * fixed bug #334: programmatic addition of slots using specialized
2042 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2043 and :ALLOCATION :CLASS slots.
2044 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2045 float exponents. (rereported by Peter Seibel)
2046 * fixed a bug: #\Space (and other whitespace characters) are no
2047 longer considered to be macro characters in standard syntax by
2048 GET-MACRO-CHARACTER.
2049 * fixed bug: initialization of condition class metaobjects no longer
2050 causes an instance of the condition to be created. (reported by
2052 * fixed bug: it is now possible to have slots such that
2053 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2054 description is neither :INSTANCE nor :CLASS.
2055 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2056 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2057 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2058 non-standard SLOT-DEFINITION-ALLOCATION values.
2059 * fixed bugs: various race conditions handling exiting threads.
2060 CL-PPCRE's thread torture test now passes.
2061 * fixed arguably-a-bug: GC time proportional to number of threads in
2062 system even when most of them are idle
2063 * optimization: improved performance of BIT and SBIT on bit-vectors.
2064 * .fasl file incompatibility: The fasl file version number has
2065 been incremented because of changes associated with package locks.
2067 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2068 * minor incompatible change: the system no longer provides
2069 optimization hints (conditions which are TYPEP
2070 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2071 mode; these hints are still emitted when higher SPEED optimization
2073 * new contrib module: a sampling profiler (profiling by statistical
2074 sampling, rather than by instrumenting functions) is available as
2075 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2076 * the behaviour of the standard function ED is now customizeable by
2077 third parties through a hook variable: see ED's documentation
2078 string for information on the protocol.
2079 * the compiler no longer emits efficiency notes for (FUNCALL X)
2080 when the type of X is uncertain under default optimization
2082 * fixed bug 276: mutating a binding of a specialized parameter to a
2083 method to something that is not TYPEP the specializer is now
2085 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2086 exhibit the required behaviour on the x86 platform. (thanks to
2087 Peter van Eynde, Eric Marsden and Bruno Haible)
2088 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2089 even for difficult arguments. (reported by Peter Graves)
2090 * fixed bug 141a: the backquote printer now descends quoted
2092 * fixed another bug in backquote printing: no more destructive
2093 modification of the form's list structure. (reported by Brian
2095 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2096 move between its address being taken and the call to
2097 interrupt_thread, fixing a crashing race condition.
2098 * the SB-POSIX contrib implementation has been adjusted so that it
2099 no longer exhibits ridiculously poor performance when constructing
2100 instances corresponding to C structs.
2102 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2103 * minor incompatible change: the sb-grovel contrib now treats C
2104 structures as alien (in the sense of SB-ALIEN) objects rather than
2105 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2106 has implications for memory management of client code
2107 (sb-grovel-returned objects must now be manually managed) and for
2108 type safety (alien objects now have full types).
2109 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2110 used to control emission of compiler diagnostics, rather than the
2111 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2112 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2113 quality should be considered deprecated.
2114 * install.sh now installs the user manual as well
2115 * (not quite a new documentable feature, but worth considering in
2116 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2117 beginnings of a semantically meaningful condition hierarchy is
2118 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2120 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2121 designator as the defaults argument.
2122 * fixed bug: Displaced arrays whose displaced-to array has become
2123 too small now cause ARRAY-DIMENSION to signal an error, providing
2124 for safer bounds-checking. (reported by Bruno Haible)
2125 * fixed bug: DEFCLASS slot definitions with identical :READER and
2126 :WRITER names now signal a reasonable error. (reported by Thomas
2128 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2130 * fixed bug: Invalid slot specification errors now print correctly.
2131 (thanks to Zach Beane)
2132 * fixed bug 320: Shared to local slot value transfers in class
2133 redefinitions now happen corrently from superclasses as
2134 well. (reported by Bruno Haible)
2135 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2137 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2138 improper lists correctly. (reported by Bruno Haible)
2139 * fixed bug 313: source-transform for <fun-name> was erroneously
2140 applied to a call of a value of a variable with name <fun-name>.
2141 (reported by Antonio Menezes Leitao)
2142 * fixed bug 307: The obsolete instance protocol ensures that
2143 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2144 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2145 remove dead unknown-values globs from the middle of the stack.
2146 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2147 easier to resume long computations after using *BREAK-ON-SIGNALS*
2148 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2149 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2150 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2151 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2152 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2153 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2155 * fixed a (fairly theoretical) bug in string printing: if
2156 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2157 does not have array-element-type equal to the most general string
2159 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2160 function name. (thanks to Nikodemus Siivola)
2161 * fixed bug: compile-time format string checker failed on
2162 non-closed ~{. (reported by Thomas F Burdick)
2163 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2164 don't warn on legal loop code involving more than one aggregate
2165 boolean. (thanks to Nikodemus Siivola)
2166 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2167 in keyword position to MAKE-PACKAGE was not being reported as
2169 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2170 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2172 * fixed bugs 280 and 312: the checking for multiple definitions in a
2173 file is less likely to become confused by uses of inline
2175 * fixed bug: the #S reader macro performs the keyword coercion
2176 specified for slot names. (reported by Kalle Niemitalo)
2177 * fixed bug: lambda lists may contain symbols whose names start with
2178 & but are not lambda-list-keywords; their occurrence triggers a
2180 * fixed bug 321: define-method-combination argument lambda lists do
2181 not cause invalid code to be generated when &optional or &aux
2182 variables have default values. (reported by Bruno Haible)
2183 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2184 have CLOS classes; this appears also to have fixed reported
2185 instability in stack exhaustion detection.
2186 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2187 MUFFLE-WARNING when no associated restart is present is now
2189 * optimization: rearranged the expansion of various defining macros
2190 so that each expands into only one top-level form in a
2191 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2193 * optimization: used a previously little-used slot in symbols to
2194 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2196 * fixed some bugs revealed by Paul Dietz' test suite:
2197 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2199 ** Fixed an optimization bug related to inheritance of initforms
2200 from local to shared slots.
2201 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2202 ** CAST optimizer forgot to flush argument derived type.
2203 ** print/read consistency on floats is now orders of magnitude
2204 more likely. (thanks also to Bruno Haible for a similar report
2206 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2207 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2208 for objects of type REAL. Make it so.
2209 ** SXHASH is specified (infelicitously) to respect similarity,
2210 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2211 it so. (thanks to Markus Ziegler)
2212 ** on the Alpha, the compiler succeeds in compiling functions
2213 returning a known number of arguments greater than 63.
2214 ** fixed handling of invalid NIL arguments in keyword position
2216 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2218 ** provided a readably-printable representation for RANDOM-STATE
2220 ** ensured that pathnames, where they have a namestring, always
2221 print using #P"..." syntax.
2223 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2224 * Support for the forthcoming 2.0 version of the NetBSD kernel
2225 running on x86 hardware has been added. (thanks to Perry
2226 E. Metzger most immediately, and others for their past work)
2227 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2228 current release version; SBCL's OpenBSD support had been broken
2229 since about the time of OpenBSD's switch to ELF binary format.)
2230 * [placeholder for DX summary]
2231 ** user code with &REST lists declared dynamic-extent, under high
2232 speed or space and low safety and debug optimization policy.
2233 * The manual has been converted to Texinfo format and the debugger
2234 chapter from the cmucl manual has been added.
2235 * A facility has been added to extract documentation strings from
2236 sbcl and store them as Texinfo-formatted snippets for inclusion in
2237 the manual (via Texinfo's @include directive)
2238 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2239 the test case to Dave Roberts)
2240 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2241 pointers. (reported by Sean Ross)
2242 * bug fix: PROFILE output is printed nicely even for large numerical
2243 values. (thanks to Zach Beane)
2244 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2245 greater than 32 handle EOF correctly.
2246 * bug fix: on X86 an immediate argument of the IMUL instruction is
2247 correctly printed in disassembly. (thanks to Lutz Euler)
2248 * bug fix: class slots in redefined classes preserve their old
2249 values. (thanks to Bruno Haible and Nikodemus Siivola)
2250 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2251 succeeds. (reported by Marco Baringer)
2252 * fixed some bugs revealed by Paul Dietz' test suite:
2253 ** READ-SEQUENCE now works on ECHO-STREAMs.
2254 ** RATIONALIZE works more according to its specification. (thanks
2257 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2258 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2259 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2260 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2261 should matter to you only if you rebind the printer control
2262 variables and then find you want different bindings in the
2263 debugger than in the ordinary execution of your program.)
2264 * The runtime build system has been tweaked to support building
2265 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2266 assembler and linker. (thanks to Nikodemus Siivola)
2267 * Unbound, undefined, undeclared variables now trigger full
2268 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2269 more often programmer error than deliberate exploitation of undefined
2271 * optimization: the hash algorithm for strings has changed to one
2272 that is less vulnerable to spurious collisions. (thanks to Juho
2274 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2275 needless bounds checking. (thanks to Juho Snellman)
2276 * optimization: implemented multiplication as a modular
2277 (UNSIGNED-BYTE 32) operation on the PPC backend.
2278 * fixed some bugs revealed by Paul Dietz' test suite:
2279 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2281 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2282 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2283 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2285 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2287 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2289 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2290 their output stream on EOF from read.
2291 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2292 have been read to end-of-file.
2293 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2295 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2296 description of determination of which consecutive characters
2298 ** Printing the "Space" character with escaping on now yields "#\\ ",
2299 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2300 ** Reading floating-point numbers with *READ-BASE* set to a number
2301 less than 10 works correctly.
2302 ** Reading floating-point numbers with *READ-BASE* set to a number
2303 more than 10 works correctly.
2304 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2305 the readtable currently in effect.
2307 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2308 * minor incompatible change: parsing of namestrings on a physical
2309 (Unix) host has changed; numbers after the final #\. in a
2310 namestring are no longer interpreted as a version field. This is
2311 intented to be largely invisible to the user, except that the
2312 meaning of the namestring "*.*.*" has changed: it now refers to a
2313 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2314 should usually be replaced by
2315 (make-pathname :name :wild :type :wild :version :wild)
2316 with the added benefit that this is more likely to be portable.
2317 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2318 option to OPEN now signals an error if the file being opened
2319 exists; this may have an impact on existing code.
2320 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2321 signals received, on the PowerPC platforms (both Linux and
2322 Darwin). (thanks to Pierre Mai for pointing out the location of
2324 * several fixes on OS X: The system now builds and runs cleanly on
2325 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2326 (thanks to Brian Mastenbrook)
2327 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2328 second-resolution: integer multiples of 1/3600 between -24 and 24.
2329 (thanks to Vincent Arkesteijn)
2330 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2331 type when called with 1 argument; PEEK-CHAR checked type of
2332 PEEK-TYPE only after having read first character from a
2333 stream. (reported by Peter Graves)
2334 * bug fix: the garbage collector now has much better locality
2335 behaviour, and in particular no longer treats all memory as being
2336 exhausted when many small objects point to each other in a deeply
2338 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2339 recognized as being TYPEP their class.
2340 * bug fix: the PUSHNEW documentation string has been corrected.
2341 (thanks to Vincent Arkesteijn)
2342 * bug fix: defaulting of the value for the last of an atypically
2343 large number of multiple values being bound was not being
2344 performed correctly on the Alpha or PPC platforms
2345 * optimization: implemented multiplication as a modular
2346 (UNSIGNED-BYTE 32) operation on the x86 backend.
2347 * optimization: SEARCH on simple-base-strings can now be open-coded.
2348 (see also contrib/compiler-extras.lisp for inspiration for
2349 teaching the compiler about the Boyer-Moore algorithm).
2350 * value, returned by MAX (and MIN) called with several EQUALP, but
2351 not EQL, arguments now does not depend on compiler settings.
2352 * fixed some bugs revealed by Paul Dietz' test suite:
2353 ** in stack analysis liveness information is propagated from
2354 non-local entry points.
2355 ** pathwise CAST removing failed when the CAST node did not start
2357 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2358 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2360 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2361 host is already defined.
2362 ** RENAME-FILE works on streams instead of signalling an internal
2364 ** PEEK-CHAR uses the current readtable when determining whether
2365 or not a character is whitespace.
2366 ** MERGE-PATHNAMES handles the case when the pathname does not
2367 specify a name while the default-pathname specifies a version.
2368 ** Pathnames now stand a better chance of respecting print/read
2370 ** Attempting to use standardized file system operators with a
2371 pathname with invalid :DIRECTORY components signals a
2373 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2375 ** DIRECTORY on logical pathnames is more correct.
2376 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2377 signal a TYPE-ERROR if their argument is not a stream.
2378 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2379 designator argument does not designate a stream.
2380 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2381 examining the synonym.
2382 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2384 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2385 element-type larger than ([UN]SIGNED-BYTE 32).
2387 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2388 * When built with the :SB-FUTEX feature, threaded builds now take
2389 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2390 for faster/more reliable mutex and condition variable support.
2391 * Incompatible change (but one you probably shouldn't have been using
2392 anyway): the interface and code for arbitrating between multiple
2393 threads in the same user session has been redesigned.
2394 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2395 error when called without an explicit environment argument.
2396 (thanks to Nikodemus Siivola)
2397 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2398 confused about their position. (thanks to Adam Warner and Gerd
2400 * bug fix: Pretty printing backquoted forms with unquotations in the
2401 argument list position of various code constructs such as LAMBDA
2402 now works correctly. (reported by Paul Dietz)
2403 * bug fix: Pretty printing unquotations no longer loses all
2404 stream position information.
2405 * optimization: performance of string output streams is now less
2406 poor for multiple small sequence writes.
2407 * optimization: performance of CSUBTYPEP in the presence of complex
2408 expressions involving CONS and NOT many times has been improved.
2409 (reported by Paul Dietz)
2410 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2412 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2413 streams for the REPL, for the debugger, and so on. (thanks to
2415 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2416 * fixed some bugs revealed by Paul Dietz' test suite:
2417 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2418 is ignored if it is too small, rather than propagating through
2419 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2420 ** extremely complex negations of CONS types were not being
2421 sufficiently canonicalized, leading to inconsistencies in
2423 ** VALUES tranformer lost derived type.
2425 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2426 * fixed a bootstrapping bug: the build process no longer assumes
2427 that the various BOOLE-related constants have the same value in
2428 host and target lisps. (noted by Paul Dietz' test suite on an
2429 SBCL binary built from CLISP)
2430 * The system can now be dynamically linked on the MIPS platform,
2431 which enables dynamic loading of foreign code from Lisp. (thanks
2432 to Ralf Baechle for discussions on the MIPS ABI)
2433 * The system now records debugging information for its own source
2434 files in a filesystem-position-independent manner, relative to
2435 the "SYS" logical host.
2436 * fixed a compiler bug: MV-LET convertion did not check references
2437 to the "max args" entry point. (reported by Brian Downing)
2438 * tweaked disassembly notes to be less confident about proclaiming
2439 some instruction as an LRA. (thanks to Brian Downing)
2440 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2441 now each have their own history, command character, and other
2442 characteristics. (thanks to David Lichteblau)
2443 * fixed some bugs revealed by Paul Dietz' test suite:
2444 ** compiler failure in compiling LOGAND expressions including a
2446 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2447 shift greater than 32.
2448 ** FLUSH-DEST did not mark blocks for type check regeneration.
2449 ** HANDLER-CASE failed to accept declarations in handler clauses
2450 in some circumstances.
2452 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2453 * New code in contrib/sb-introspect (still probably not entirely
2454 stable yet) provides some support for smart Lisp development
2455 environments like SLIME.
2456 * The conditions signalled for errors occurring when loading .fasl
2457 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2458 in a way which should help ASDF recover gracefully.
2459 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2460 stuff has been cleaned up. If you code contrib/ stuff, this might
2461 affect you, and you can look at contrib/README, contrib/STANDARDS,
2462 and/or the 0.8.4.27 diff to check.
2463 * In full calls the compiler now does not generate checks for declared
2464 argument types for all arguments.
2465 * various threading fixes
2466 ** and some experimental patches which didn't make it into
2467 the main tree for this release, but which are shipped in
2468 contrib/experimental-thread.patch as a possible fix for some
2469 failures (deadlock, spinning...) in GC-intensive multithreaded
2471 * fixed PPC build problem (source code incompatibility of different
2472 library versions): added offsetof() hackery which attempts to divine
2473 where glibc maintainers put uc_mcontext today
2474 * fixed bug 282: compiler does not trust type assertions while passing
2475 arguments to a full call.
2476 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2477 &OPTIONAL VALUES type parameter.
2478 * fix bug 214: algorithm for noting rejected templates is now more
2479 similar to that of template seletion. (also reported by rydis on
2481 * fixed bug 141b: printing backquoted information readably and prettily
2482 inserts a space where necessary.
2483 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2484 (CHARACTER 10) are now reported as errors, rather than propagated
2485 as unknown types. (reported by piso on #lisp)
2486 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2487 with values NIL and :ERROR. (thanks to Milan Zamazal)
2488 * fixed bug 191c: CLOS now does proper keyword argument checking as
2489 described in CLHS 7.6.5 and 7.6.5.1.
2490 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2491 longer raise an error; further, using a list as a for-as-arithmetic
2492 counter now raises a meaningful error.
2493 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2494 now understood by sequence creation functions such as MAKE-SEQUENCE
2496 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2497 a STRING-INPUT-STREAM.
2498 * compiler enhancement: SIGNUM is now better able to derive the type
2500 * type declarations inside WITH-SLOTS are checked. (reported by
2502 * fixed some bugs revealed by Paul Dietz' test suite:
2503 ** incorrect optimization of TRUNCATE for a positive first
2504 argument and negative second.
2505 ** compiler failure in let-convertion during flushing dead code.
2506 ** compiler failure while deriving type of TRUNCATE on an
2507 interval, containing 0.
2508 ** ASH of a negative bignum by a negative bignum count now returns
2510 ** intersection of CONS types now canonicalizes properly, fixing
2511 inconsistencies in SUBTYPEP.
2513 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2514 * incompatible change: The --disable-debugger command line
2515 option now clobbers the debugger at a more fundamental
2516 level, by redefining #'INVOKE-DEBUGGER instead of by
2517 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2518 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2519 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2520 enter the debugger REPL and then suffer recursive errors
2521 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2522 while under the new system, BREAK in --disable-debugger mode
2523 terminates the system just as an unhandled error would.
2524 * fixed compiler performance when processing loops with a step >1;
2525 * bug fix: DOCUMENTATION now retrieves generic function
2526 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2527 support has been systematized, and now supports the methods
2528 specified by ANSI, along with a default method and a method for
2529 slot documentation. (reported by Nathan Froyd)
2530 * bug fix: effective methods associated with a generic function are
2531 no longer cached over a change of that generic function's method
2532 combination. (reported by Andreas Fuchs)
2533 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2534 names a non-existent directory. (thanks to Andreas Fuchs)
2535 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2536 the PPC platform, lead to stack corruption; this has been fixed.
2537 (reported by Rainer Joswig)
2538 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2539 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2540 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2541 arithmetic in general, is now much more reliable on the PPC
2543 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2544 platform now returns the right answer.
2545 * optimization: restored some effective method precomputation in
2546 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2547 precomputation is now tunable.
2548 * optimization: compiler-internal data structure use has been
2549 reviewed, and changes have been made that should improve the
2550 performance of the compiler by about 20%.
2551 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2552 simple-base-strings and simple-bit-vectors is improved.
2553 * optimization: the optimization of 32-bit logical and arithmetic
2554 functions introduced in version 0.8.3 on the x86 has been
2555 implemented on the mips, ppc and sparc platforms; an
2556 implementation of the same facility, but for 64-bit arithmetic,
2557 has been added for the alpha.
2558 * microoptimization: the compiler is better able to make use of the
2559 x86 LEA instruction for multiplication by constants.
2560 * bug fix: in some situations compiler did not report usage of
2561 generic arithmetic in (SPEED 3) policy.
2562 * bug 145b fix: compiler used wrong type specifier while converting
2563 MEMBER-types to numeric.
2564 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2566 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2567 index argument now works properly on the Alpha platform.
2568 * bug fix: floating point exception treatment on the Alpha platform
2570 * bug fix: FILE-POSITION works much better on string input and
2571 output streams. (thanks to Nikodemus Siivola)
2572 * bug fix: many threading/garbage collection symptoms sorted.
2573 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2574 might be pseudo-atomic.
2575 * internal change: Stopping for GC is now done with signals not
2576 ptrace. GC is now done in whichever thread wanted it, instead of
2578 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2580 * bug fix: result form in DO is not contained in the implicit
2582 * incompatible change: ICR structure is changed; the value part of
2583 CONTINUATION is now called LVAR; corresponding functions are
2584 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2585 * added type deriver for ISQRT (thanks to Robert E. Brown).
2586 * bug fix: better support for loading from the command line when an
2587 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2589 * fixed some bugs revealed by Paul Dietz' test suite:
2590 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2591 ** ROUND and FROUND now give the right answer when given very
2592 small float arguments.
2593 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2595 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2596 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2597 ** LCM should return a non-negative integer.
2598 ** PARSE-INTEGER returned the index of a terminator instead of the
2599 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2600 ** PARSE-INTEGER returned an incorrect index being applied to a
2602 ** LCM with two arguments of 0 returns 0 rather than signalling
2604 ** unsigned addition of a 32-bit constant with the high bit set no
2605 longer causes an internal compiler error on the x86.
2606 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2607 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2608 with negative last argument.
2609 ** byte specifiers with zero size and position no longer cause
2610 an error during type derivation.
2611 ** bignum multiplication on the Alpha platform now returns the
2613 * porting: The system now builds on SuSE AMD64, although it still
2614 generates a 32-bit binary.
2615 * .fasl file incompatibility: The fasl file version number has
2616 been incremented (because of the changes to internal compiler
2617 data structures referred to above).
2619 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2620 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2621 more accurately, on the Darwin kernel running on PowerPC hardware.
2622 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2623 * Compiler code deletion notes now signal a condition of type
2624 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2625 an associated MUFFLE-WARNING restart.
2626 * The compiler now performs limited argument count validation of
2627 constant format strings in FORMAT, and where appropriate in ERROR,
2628 CERROR and WARN. (thanks to Gerd Moellmann)
2629 * New ASDF-INSTALL contrib can be used for automatic download and
2630 installation of third-party Lisp code from CCLAN or other sites
2632 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2633 which forces another thread to execute a function supplied by the
2635 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2636 now accept and act upon their :ELEMENT-TYPE keyword argument.
2637 (reported by Martin Atzmueller, Edi Weitz)
2638 * bug fix: FILE-POSITION now accepts position designators up to
2639 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2640 is the greater. (thanks to Patrik Nordebo)
2641 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2642 Kalle Olavi Niemitalo)
2643 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2644 printable. (reported by Eric Marsden)
2645 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2646 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2647 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2648 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2649 (thanks to Stig Sandoe)
2650 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2651 UNREAD-CHAR now correctly includes the unread character in the
2652 target sequence. (thanks to Gerd Moellmann)
2653 * bug fix in threaded builds: the system can now be suspended and
2654 resumed by shell job control with minimal disruption.
2655 * bug fixes in times and timezones >2038AD
2656 * better handling of "where is GNU make?" problem in build scripts
2657 (thanks to Nikodemus Siivola)
2658 * new optimization: inside a named function any reference to a
2659 function with the same name is considered to be a self-reference;
2660 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2661 optimization quality.
2662 * new optimization on x86: logical functions and + now have
2663 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2664 used when the result is truncated to 32 bits.
2665 * VALUES declaration is partially enabled.
2666 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2667 * fixed some bugs revealed by Paul Dietz' test suite:
2668 ** The system now obeys the constraint imposed by
2669 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2670 types form a lattice under type intersection.
2671 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2672 ** ASSOC now ignores NIL elements in an alist.
2673 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2674 and (1+ MOST-POSITIVE-FIXNUM) answers.
2675 ** The addition of a method with invalid qualifiers to a generic
2676 function does not cause an error to be signalled immediately;
2677 a warning is signalled, and the error is generated only on
2678 calling the generic function.
2679 * changed .fasl file version number, in part to add type codes for
2680 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2681 obscure ANSI requirements
2683 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2684 * fixed bug 148: failure to inline-expand a local function left
2685 garbage, confusing the compiler.
2686 * fixed bugs 3cd: structure slot readers perform type check if the
2687 slot can have an invalid value (i.e. it is either not initialized
2688 or can be written with a less specific slot writer).
2689 * bug fix: the compiler now traps array references to elements off
2690 the end of an array; previously, the bounds checking in some
2691 circumstances could go off-by-one.
2692 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2694 * type declarations for array element types now obey the description
2695 on the CLHS page "Declaration TYPE", as per discussions on
2696 sbcl-help around 2003-05-08. This means that a declaration
2697 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2698 declaration, all references to BAR will be asserted or assumed
2699 (with THE, so dependent on compiler policy) to involve objects of
2700 type FOO. Note that no such declaration is implied in
2701 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2702 * declared types of functions from the "Conditions"
2703 chapter. (reported by Paul Dietz)
2704 * bug fix: CERROR accepts a function as its first argument.
2705 * bug fix: NTH an NTHCDR accept a bignum as index
2706 arguments. (reported by Adam Warner)
2707 * optimization: character compare routines now optimize comparing
2708 against a constant character. (reported by Gilbert Baumann)
2709 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2710 argument now works properly on the MIPS platform.
2711 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2712 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2713 declared to have a functional type, does not cause an error
2715 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2716 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2717 * fixed bug: if last continuation of a deleted block has a
2718 destination, this destination should be deleted too. (reported by
2720 * fixed a bug in the bootstrap process: the host compiler's values
2721 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2722 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2723 test case from Patrik Nordebo)
2724 * improved the ability of the disassembler on the PPC platform to
2725 provide helpful disassembly notes.
2726 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2727 the class in more cases than previously.
2728 * bug fix: FILE-POSITION now understands :START and :END for
2729 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2730 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2731 8) data. (thanks to David Lichteblau)
2732 * bug fix: it is possible to add a method to a generic function
2733 without lambda list.
2734 * bug fix: reader failed to signal END-OF-FILE inside an
2735 object representation. (reported by Nikodemus Siivola)
2736 * fixed some bugs revealed by Paul Dietz' test suite:
2737 ** LAST and [N]BUTLAST should accept a bignum.
2738 ** condition slot accessors are methods.
2739 ** (VECTOR NIL) is a subtype of STRING.
2741 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2742 * minor incompatible change: some nonsensical specialized lambda
2743 lists (used in DEFMETHOD) which were previously ignored now signal
2745 * minor incompatible change: the system is now aware of the types of
2746 variables in the COMMON-LISP package, and will signal errors for
2747 most violations of these type constraints (where previously they
2748 were silently accepted).
2749 * minor incompatible change: COMPILE-FILE now uses the freedom
2750 afforded (ANSI 3.2.2.3) to use derived function types for
2751 functions defined in the same file. This also permits the system
2752 to warn on static type mismatches and function
2753 redefinition. (Currently it does not work with high DEBUG level.)
2754 * minor incompatible change: VALUES declaration is disabled.
2755 * When issuing notes, the compiler now signals a condition of type
2756 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2757 restart for use in user handlers. It is expected that the
2758 COMPILER-NOTE condition will eventually become a condition
2759 supertype to a hierarchy of note types, which will then be
2760 handleable in a similar fashion. However, at the moment, no such
2761 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2762 but it's an implementation detail, not a classification for the
2764 * Changes in type checking closed the following bugs:
2765 ** type checking of unused values (192b, 194d, 203);
2766 ** template selection based on unsafe type assertions (192c, 236);
2767 ** type checking in branches (194bc).
2768 * A short form of VALUES type specifier has ANSI meaning (it has
2769 increased the number of situations when SBCL cannot perform type
2771 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2772 keywords or constants is permissible.
2773 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2774 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2775 classes. (thanks to Antonio Martinez)
2776 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2777 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2778 argument to a pretty-print stream. Instead, it leaves any such
2779 operations to DESCRIBE-OBJECT methods.
2780 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2782 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2783 lambda lists. (thanks to David Lichteblau)
2784 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2785 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2786 (thanks to Gerd Moellmann)
2787 * fixed bug 246: increased compilation speed of long
2788 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2790 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2792 * DEFCONSTANT now throws a condition of type
2793 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2794 constant to a non-EQL value; CONTINUE and ABORT restarts
2795 respectively change and preserve the value.
2796 * fixed bug 63: The code walker, part of the implementation of CLOS,
2797 is now better at handling symbol macros.
2798 * bug fix: There is no longer an internal implementation type named
2799 CL:LENGTH. (reported by Raymond Toy)
2800 * bug fix: In macro-like defining macros/special operators the
2801 implicit block does not enclose the lambda list.
2802 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2803 enclosed in parentheses) are not suitable as type specifiers, and
2804 their use properly signals an error now.
2805 * bug fix: An argument count mismatch for a type specifier in code
2806 being compiled no longer causes an unhandled error at compile
2807 time, but signals a compile-time warning.
2808 * fixed simple vector readable printing
2809 * bug fix: DESCRIBE takes more care over whether the class
2810 precedence list slot of a class is bound before accessing it.
2811 (reported by Markus Krummenacker)
2812 * bug fix: FORMATTER can successfully compile pretty-printer format
2813 strings which use variants of the ~* directive inside.
2814 * bug fix: SEARCH now applies its TEST predicate to the elements of
2815 the arguments in the correct order. (thanks to Wolfhard Buss)
2816 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2817 in type inference. (thanks to Robert E. Brown)
2818 * bug fix: Reading in symbols with an explicit package name of ""
2819 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2820 * fixed some bugs revealed by Paul Dietz' test suite:
2821 ** NIL is now allowed as a structure slot name.
2822 ** Arbitrary numbers, not just REALs, are allowed in certain
2823 circumstances in LOOP for-as-arithmetic clauses.
2824 ** Multiple class redefinitions before slot access no longer
2825 causes a type error.
2826 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2827 association between the name and a class.
2828 ** Generic functions with non-standard method-combination and over
2829 five methods all of which return constants no longer return NIL
2830 after the first few invocations. (thanks to Gerd Moellmann)
2831 ** CALL-NEXT-METHOD with no arguments now passes the original
2832 values of the arguments, even in the presence of assignment.
2833 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2835 ** DISASSEMBLE works with closures and funcallable instances.
2836 ** ADD-METHOD now returns the generic function, not the new method.
2837 ** FIND-METHOD signals an error if the lengths of the specializers
2838 is incompatible with the generic function, even if the ERRORP
2840 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2841 which its argument is a member.
2842 ** DEFCLASS only redefines the class named by its class-name
2843 argument if that name is the proper name of the class;
2844 otherwise, it creates a new class.
2845 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2846 of the UNBOUND-SLOT condition to the name of the slot.
2847 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2848 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2849 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2850 SLOT-MAKUNBOUND in the specified fashion.
2852 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2853 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2854 cross-compilation host. As a consequence, we can now bootstrap our
2855 way up to SBCL starting with a bare gcc toolchain and human-readable
2856 source code (first the source to CLISP, then the source to SBCL).
2857 * A contributed module containing a partial implementation of the
2858 simple-streams interface has been included. (thanks to Rudi
2860 * A contributed module implementing the RFC1321 Message Digest
2861 Algorithm, known as MD5, has been included.
2862 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2863 no longer has any effect, as the code controlled by this feature
2864 has been deleted. (As far as we know, no-one has ever built using
2865 this feature, and its semantics were confused in any case).
2866 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2867 work on conditions (as required by the ANSI specification),
2868 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2869 expected behaviour on conditions. Users should note, however,
2870 that such behaviour is not required by the ANSI specification,
2871 and so use of this behaviour may render their code unportable.
2872 * fixed some bugs revealed by Paul Dietz' test suite:
2873 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2875 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2876 ** SXHASH on condition objects no longer returns NIL.
2877 ** :ALLOCATION :CLASS slots are better treated; their values are
2878 updated on class redefinition, and initforms inherited from
2879 superclasses are applied.
2880 ** REMOVE-METHOD returns its generic function argument even when
2881 no method was removed.
2882 ** SHARED-INITIALIZE now initializes the values of the requested
2883 slots, including those with :ALLOCATION :CLASS.
2884 ** ALLOCATE-INSTANCE now works on structure classes defined via
2885 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2887 ** SLOT-EXISTS-P now works on conditions, as well as structures
2889 ** MAKE-LOAD-FORM now has the required methods on
2890 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2891 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2892 keyword, and now implements the SLOT-NAMES argument.
2893 ** methods with &OPTIONAL arguments no longer allow too many
2894 arguments to be passed in the call without error.
2895 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2896 option is consistent with the required arguments of the generic
2897 function lambda list.
2898 * bug fix: REQUIRE accepts a string designator. (Thanks to
2900 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2901 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2902 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2903 * bug fix: The compiler now checks for duplicated variables in macro
2905 * bug fix: SETQ on globals returns the correct value.
2906 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2907 not a COMPILER-ERROR (followed by some other strange error on
2908 choosing the CONTINUE restart).
2909 * bug fix: make.sh and friends are now more consistent in the way that
2910 they look for GNU "make".
2912 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2913 * experimental native threads support (on x86 Linux >=2.4 only).
2914 This is not compiled in by default: you need to add :SB-THREAD to
2915 the target features. See the "Beyond ANSI" chapter of the manual
2917 * fix for longstanding nonANSIism: The old distinction between
2918 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2919 The return value from CL:FIND-CLASS is now a CLOS class, and
2920 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2921 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2922 * An interface to the MetaObject Protocol, as described in Kiczales,
2923 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2924 MIT Press, 1991, is available from the SB-MOP package.
2925 * incompatible change: the SB-PCL package should now be considered
2926 a private implementation detail, and no longer a semi-private MOP
2928 * minor incompatible change: due to rearrangement for threads, the
2929 control stack and binding stack are now allocated at arbitrary
2930 addresses instead of being hardcoded per-port. Users affected by
2931 this probably have to be doing advanced things with shared
2932 libraries, and will know who they are.
2933 * minor incompatible change: Previously, all --eval forms used were
2934 processed with READ before any of them were processed with EVAL.
2935 Now each --eval form is processed with both READ and EVAL before
2936 the next --eval form is processed. (Thus package operations like
2937 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2938 work as the user might reasonably expect.)
2939 * minor incompatible change: *STANDARD-INPUT* is now only an
2940 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2942 * minor incompatible change: Y-OR-N-P is now character-oriented,
2943 not line oriented. Also, YES-OR-NO-P now works without errors.
2944 (thanks to Antonio Martinez)
2945 * sb-aclrepl module improvements: an integrated inspector, added
2946 repl features, and a bug fix to :trace command.
2947 * Known functions, which cannot be open coded by the backend, are
2948 considered to be able to check types of their arguments. (fixing
2949 a bug report by Nathan J. Froyd)
2950 * fixed a bug in computing method discriminating functions: It is
2951 now possible to define methods specialized on classes which have
2952 forward-referenced superclasses. (thanks to Gerd Moellmann)
2953 * fixed evaluation order in optional entries (reported by Gilbert
2955 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2956 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2957 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2958 required-by-AMOP NAME argument, as well as CLASS and
2959 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2960 * fixed bug 20: DEFMETHOD can define methods using names that are
2961 not the proper names of classes to designate class specializers.
2962 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2963 Unix file descriptors, instead of blowing up. (thanks to Antonio
2965 * Garbage collection refactoring: user-visible change is that a
2966 call to the GC function during WITHOUT-GCING will not do garbage
2967 collection until the end of the WITHOUT-GCING. If you were doing
2968 this you were probably losing anyway.
2969 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2970 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2971 TYPEP the latter but not the former.
2972 * The compiler issues a full WARNING for calls to undefined functions
2973 with names from the CL package.
2974 * MAP-INTO for a vector destination is open coded. (reported by
2975 Brian Downing on c.l.l)
2976 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2977 documentation string.
2978 * fixed some bugs revealed by Paul Dietz' test suite:
2979 ** COPY-ALIST now signals an error if its argument is a dotted
2981 ** Condition slots are now accessed more correctly in the presence
2982 of multiple initargs for a given slot.
2983 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2984 exclude restarts of the same name associated with a different
2986 ** DEFCLASS of forward-referenced classes with another
2987 forward-referenced class in the superclasses list no longer
2989 ** Condition slots are now initialized once each, not multiple
2990 times. (thanks to Gerd Moellmann)
2991 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2993 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2994 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2995 arguments contain duplicated elements.
2996 ** RESTART-CASE understands local macros.
2997 ** RESTART-CASE associates exactly its own restarts with a condition.
2998 ** ENDP in safe mode checks its argument to be of type LIST.
2999 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3000 in question is unbound.
3001 ** Optimized MAKE-INSTANCE functions no longer cause internal
3002 assertion failures in the presence of duplicate initargs.
3003 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3004 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3005 propagated even in the ctor optimized implementation of
3007 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3009 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3010 * a better implementation of SXHASH on (simple) bit vectors,
3011 measured both in execution speed and in distribution of results
3012 over the positive fixnums, has been installed. Likewise, a better
3013 implementation of EQUAL for simple bit vectors is now available.
3014 * fixed CEILING optimization for a divisor of form 2^k.
3015 * fixed bug 240 (emitting extra style warnings "using the lexical
3016 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3017 by Antonio Martinez)
3018 * fixed SXHASH, giving different results for NIL depending on type
3019 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3020 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3021 variable. (found by Rolf Wester)
3022 * SBCL does not ignore type declarations for special
3023 variables. (reported by rif on c.l.l 2003-03-05)
3024 * some bug fixes in contrib/sb-aclrepl/
3025 * fixed some bugs revealed by Paul Dietz' test suite:
3026 ** a bug in the CONS type specifier, whereby the CAR and CDR
3027 types got intertwined, has been fixed;
3028 ** the type system is now able to reason about the interaction
3029 between INTEGER and RATIO types more completely;
3030 ** APPEND, [N]REVERSE and NRECONC check that those their
3031 arguments, which must be proper lists, are really so;
3032 ** An array specialized to be unable to hold elements has been
3033 implemented, as required -- yes, really -- by ANSI;
3034 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3035 on malformed property lists;
3037 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3038 * incompatible packaging change: in line with Unix convention,
3039 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3040 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3041 * REQUIRE and PROVIDE are now optionally capable of doing something
3042 useful. See the documentation string for REQUIRE.
3043 * infrastructure for a managed SBCL contrib system: contributed
3044 modules in this release include:
3045 ** the ASDF system definition facility;
3046 ** an interface to the BSD Sockets API;
3047 ** an ACL-like convenience interface to the repl;
3048 (thanks to Kevin Rosenberg)
3049 ** an implementation of ROTATE-BYTE, with efficient implementation
3051 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3052 gives a closer approximation to the right answer than previously.
3053 (thanks to Raymond Toy)
3054 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3055 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3056 optional environment arguments, as required by ANSI.
3057 * fixed bugs in other functions taking environment objects, allowing
3058 calls with an explicit NIL environment argument to be compiled
3060 * fixed bug 228: primary return values from
3061 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3062 COMPILE or FUNCTION.
3063 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3064 :TYPE will no longer signal errors on innocuous objects.
3065 * fixed bug 231b: SETQ is better at respecting type declarations in
3066 the lexical environment.
3067 * fixed a bug in DEFCLASS: classes named by symbols with no or
3068 unprintable packages can now be defined.
3069 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3070 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3071 * fixed compiler failure related to checking types of functions.
3072 (reported by Robert E. Brown)
3073 * the compiler is now much more consistent in its error-checking
3074 treatment of bounding index arguments to sequence functions: in
3075 (SAFETY 3) code, errors will be signalled in almost all cases if
3076 invalid sequence bounding indices are passed to functions defined
3077 by ANSI to operate on sequences.
3078 * fixed a bug in the build procedure: documentation of SBCL-specific
3079 packages is now preserved and available in the final Lisp image.
3080 * lifted FDEFINITION lookup out of loops in the implementation of
3081 many list operations. (thanks to Robert E. Brown)
3082 * fixed a bug in the reader: the #n# reader macro now works for
3083 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3084 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3085 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3086 E. Brown and Tony Martinez respectively)
3087 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3088 SLOT-BOUNDP: in particular, it is now able to optimize them much
3089 better, and is now not vulnerable to having packages renamed.
3090 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3091 always called when a slot is not present in an instance. (thanks
3093 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3094 longer signal an error on unreading a character following EOF on
3095 the previous constituent stream. (thanks to Tony Martinez)
3096 * fixed some bugs revealed by Paul Dietz' test suite:
3097 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3098 not just nonnegative fixnums;
3099 ** the logical bit-array operators such as BIT-AND now accept an
3100 explicit NIL for their "opt-arg" argument (to indicate a
3101 freshly-consed result bit-array);
3102 ** ELT now signals an error on an invalid sequence index in safe
3104 ** the type system is now cleverer about negations of numeric
3105 types, and consequently understands the BIGNUM and RATIO types
3107 ** the type system is now cleverer about the interaction between
3108 INTEGER and RATIO types: while bugs still remain, many more
3109 cases are accurately computed;
3110 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3111 if it is in the last clause;
3112 ** CONSTANTLY now correctly returns a side-effect-free function in
3114 ** DECLARE is no longer treated as a special-operator; in
3115 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3116 * incremented fasl file version number due to the change in the
3117 DEFSTRUCT-SLOT-DESCRIPTION structure.
3119 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3120 * minor incompatible change: code processed by the "interpreter" or
3121 EVAL now has a compilation optimization policy of (DEBUG 2)
3122 (changed from (DEBUG 1)) to improve debuggability of interactive
3123 development, and to allow the use of the debug RETURN command in
3125 * an experimental implementation of the RETURN command for the
3126 debugger has been included. (thanks to Frederik Kuivinen)
3127 * fixed bug 62: constraints were not propagated into a loop.
3128 * fixed bug in embedded calls of SORT (reported and investigated by
3130 * fixed some bugs revealed by Paul Dietz' test suite:
3131 ** printing and reading of arrays with some dimensions having
3132 length 0 (thanks to Gerd Moellmann);
3133 ** BOA constructor with &AUX argument without a default value does
3134 not cause a type error;
3135 ** CONSTANTP now returns true for all self-evaluating objects.
3137 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3138 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3139 accessors that are related by inheritance, as specified in the
3140 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3141 Valtteri Vuorikoski)
3142 * The compiler is now able to inline functions that were defined in
3143 a complex lexical environment (e.g. inside a MACROLET).
3144 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3146 * Support for the upcoming FreeBSD-5.0 release has been included.
3147 (thanks to Dag-Erling Smorgrav)
3148 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3149 effect when it is not in a toplevel context.
3150 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3151 stand a better chance of being correct. (thanks to Gerd
3153 * fixed bug in COERCE, which now signals an error on coercing a
3154 rational to a bounded real type which excludes the expected
3156 * The compiler is now able to derive types more accurately from the
3157 COERCE and COMPILE functions.
3158 * fixed bug 223: functional binding is considered to be constant
3159 only for symbols in the CL package.
3160 * fixed bug 231: SETQ did not check the type of a variable being set
3161 (reported by Robert E. Brown)
3162 * A new optimization for MAKE-INSTANCE has been included, fixing
3163 various bugs (including relating to :ALLOCATION :CLASS slots and
3164 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3166 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3167 ** As required by ANSI, LOOP now disallows anonymous collection
3168 clauses such as COLLECT I in conjunction with aggregate boolean
3169 clauses such as THEREIS (= I 1);
3170 ** LOOP now signals an error when any variable is reused in the
3171 same loop (including the potentially useful construct analogous
3172 to WITH A = 1 WITH A = (1+ A);
3173 ** IT is only a special loop symbol within the first clause of a
3174 conditional loop clause;
3175 ** LOOP with a typed iteration variable over a hashtable now
3176 signals a type error iff it should.
3177 * fixed some other bugs revealed by Paul Dietz' test suite:
3178 ** FILE-STREAM now names the class previously known as FD-STREAM;
3179 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3180 argument) no longer signals an error;
3181 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3182 of the slot symbol, rather than using the current package
3183 ((:CONC-NAME "") continues to intern the slot's name in the
3185 * incremented fasl file version number, because of the incompatible
3186 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3187 of the new implementation of DEFINE-COMPILER-MACRO.
3189 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3190 * Support for building SBCL for MIPS platforms running in
3191 little-endian mode has now been checked in, and basic
3192 functionality on said platforms verified.
3193 * minor incompatible change: PCL now records the pathname of a file
3194 in which methods and the like are defined, rather than its
3196 * minor incompatible change: TRUENAME now considers the truename of
3197 a file naming a directory to be the pathname with :DIRECTORY
3198 component indicating that directory.
3199 * minor incompatible change: a NAMED clause in the extended form of
3200 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3201 reason for the previous behaviour is unclear.
3202 * more systematization and improvement of CLOS and MOP conformance
3203 in PCL (thanks to Gerd Moellman and Pierre Mai):
3204 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3206 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3207 primary methods with no specializers;
3208 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3210 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3211 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3212 FORWARD-REFERENCED-CLASSes; error reporting on
3213 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3215 ** SXHASH on CLOS instances now uses a slot internal to the
3216 instance to return different numbers on distinct instances,
3217 while preserving the same return value through invocations of
3219 ** DEFMETHOD signals errors when methods with longer incongruent
3220 lambda lists are added to generic functions;
3221 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3222 CLASS, as specified in AMOP;
3223 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3224 on the order returned by the primary method for classes of
3225 class STANDARD-CLASS;
3226 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3227 * fixed some bugs shown by Paul Dietz' test suite:
3228 ** DOLIST puts its body in TAGBODY;
3229 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3231 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3232 value producing form;
3233 ** if more variables are given to PROGV than values, extra
3234 variables are bound and made to have no value;
3235 ** NSUBSTITUTE on list arguments gets the right answer with
3237 ** ELT signals an error of type TYPE-ERROR when the index argument
3238 is not a valid sequence index;
3239 ** LOOP signals (at macroexpansion time) an error of type
3240 PROGRAM-ERROR when duplicate variable names are found;
3241 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3242 ** FUNCALL of special-operators now cause an error of type
3244 ** PSETQ now works as required in the presence of side-effecting
3245 symbol-macro places;
3246 ** NCONC accepts any object as its last argument;
3247 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3249 ** loop-for-as-package does not require a package to be explicitely
3251 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3253 * fixed bug 166: compiler preserves "there is a way to go"
3254 invariant when deleting code.
3255 * fixed bug 172: macro lambda lists with required arguments after
3256 &REST arguments now cause an error to be signalled. (thanks to
3258 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3259 bug. (thanks to Matthew Danish)
3260 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3262 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3263 arguments are lexically rebound. (thanks to Gerd Moellmann and
3265 * fixed bug 194: error messages are now more informative when there
3266 is no primary method applicable in a call to a generic
3267 function. (thanks to Gerd Moellmann)
3268 * fixed bug in command line argument checking (thanks to Julian
3270 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3271 (thanks to Matthew Danish)
3272 * incremented fasl file version number, because of the
3273 SXHASH-related changes in the layout of CLOS data structures
3275 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3276 * minor incompatible change: The runtime (the Unix executable named
3277 "sbcl") is now much pickier about the .core files it will load.
3278 Essentially it now requires .core files to descend from the same
3279 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3280 as the runtime does. (The intent is to prevent the crashes which
3281 can occur, and which can even be reported as mysterious failures,
3282 when people patch the sources or change the build parameters
3283 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3284 sbcl and .core files.)
3285 * fixed bug: VALUES-LIST is no longer optimized away.
3286 * fixed bug 142: The FFI conversion of C string values to Lisp
3287 string values no longer conses excessively. (thanks to Nathan
3288 Froyd porting Raymond Toy's fix to CMU CL)
3289 * began to systematize and improve MOP conformance in PCL (thanks to
3290 Nathan Froyd, Gerd Moellman and Pierre Mai):
3291 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3293 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3294 ** FINALIZE-INHERITANCE is now called on class finalization;
3295 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3296 argument precedence order.
3297 * fixed bug 202: The compiler no longer fails on functions whose
3298 derived types contradict their declared type.
3299 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3300 so it can be non-toplevel.
3301 * The fasl file version number has changed (because of the new
3302 implementation of DEFMACRO).
3303 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3304 safe code, that any length requirement by their type-specifier
3305 argument is valid. The exceptions to this are described in bug
3307 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3308 functions now check the types of their inputs as required by ANSI.
3309 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3310 introduced symbol is DECLAREd to be SPECIAL.
3311 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3312 * fixed bug: :COUNT argument to sequence functions may be negative
3313 * fixed bug: body of DO-SYMBOLS may contain declarations
3314 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3315 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3316 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3317 symbol macro only once
3318 * fixed printing of call frame when argument list is unavailable
3319 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3320 * compiler no longer signals WARNING on unknown keyword
3323 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3324 * A beta-quality port to the mips architecture running Linux,
3325 based on the old CMUCL backend, has been made. It has been tested
3326 on a big-endian kernel, and works sufficiently well to be able to
3327 rebuild itself; it has not been tested in little-endian mode.
3328 * fixed an inconsistency between gencgc.c and purify.c which made
3329 dumping/loading .core files unreliable
3330 * fixed bug 120a: The compiler now deals correctly with IFs where
3331 the consequent is the same as the alternative, instead of
3332 misderiving the return type. (thanks to Alexey Dejneka)
3333 * fixed bug 113: Logical pathnames are now dumpable (the logical
3334 host is resolved at load-time, throwing an error if it is not
3336 * fixed bug 174: FORMAT's error message is slightly clearer when a
3337 non-printing character is used in a format directive.
3338 * fixed several bugs in compiler checking of type declarations, i.e.
3339 violations of the Python "declarations are assertions" principle
3340 (thanks to Alexey Dejneka)
3341 * fixed several bugs in PCL's error checking (thanks to Gerd
3343 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3345 * fixed bug in compilation of functions as first class values
3346 (thanks to Antonio Martinez-Shotton)
3347 * The compiler's handling TYPE-ERRORs which it can prove will
3348 inevitably happen at runtime has been cleaned up and corrected
3349 in several ways. (thanks to Alexey Dejneka)
3350 * improved argument type checking for various basic arithmetic
3351 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3352 lavished on them in the past that they can be compiled in many
3353 ways in different special cases
3354 * fixed bug 181: compiler checks validity of user supplied type
3356 * cleaned up code flushing in optimization: Function calls which
3357 should signal errors for safety purposes (e.g. which ANSI says
3358 should signal errors when their arguments are of incorrect type)
3359 are no longer optimized away.
3360 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3361 * incremented fasl file version number, because changes in the
3362 implementation of sequence functions like COERCE caused
3363 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3364 in old inline expansions) to become undefined. (Actually these
3365 changes were later undone, so we might very well be binary
3366 compatible with 0.7.7 after all, but leaving the version number
3367 incremented seemed like the simplest and most conservative
3370 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3371 * An alpha-quality port to the parisc architecture running Linux,
3372 based on the old CMUCL backend, has been made. This, even more so
3373 than the other backends, should be considered still a work in
3374 progress; known problems include that the Linux kernel in 64-bit
3375 mode does not propagate the correct sigcontext structure to
3376 userspace, and consequently SBCL on a parisc64 kernel will not
3378 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3379 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3380 suppressed.) Also now that the compiler is looking at declarations
3381 in the environment, it checks optimization declarations as well,
3382 and suppresses inlining when (> DEBUG SPEED).
3383 * More fixes have been made to treatment of floating point exception
3384 treatment and other Unix signals. In particular, floating point
3385 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3386 * The detection and handling of control stack exhaustion (infinite
3387 or very deeply nested recursion) has changed. Stack exhaustion
3388 detection is now done by write-protecting pages at the OS level
3389 and applies at all optimization settings; when found, a
3390 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3391 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3392 be caught using IGNORE-ERRORS.
3393 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3394 correctly when transforming recursive LABELS functions to LETs.
3395 (thanks to Alexey Dejneka)
3396 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3397 that are names of constants or global variables.
3398 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3399 alien routines with docstrings.
3400 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3401 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3403 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3404 (thanks to Eric Marsden)
3405 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3406 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3407 * Bug fix: The SPARC backend can now compile functions involving
3408 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3409 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3410 to a functions expecting a PATHNAME-DESIGNATOR.
3411 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3412 lambda lists. (thanks to Alexey Dejneka)
3413 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3414 functions so that EXPORTing the name of the class doesn't cause
3415 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3416 to Antonio Martinez for reporting this)
3417 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3418 constant-folded without failing an assertion. (thanks to Einar
3419 Floystad Dorum for reporting this)
3420 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3421 types are now dealt with more correctly by the compiler.
3422 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3423 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3424 behaviour analogous to RENAME-FILE. This puts its behaviour more
3425 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3427 * The fasl file version number has changed again. (because of the
3428 bug fix involving the names of PCL MAKE-INSTANCE functions)
3430 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3431 * bug fix: Floating point exceptions are treated much more
3432 consistently on the x86/Linux and PPC/Linux platforms.
3433 * Array initialization with :INITIAL-ELEMENT is now much faster for
3434 cases when the compiler cannot open code the array creation, but
3435 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3436 array accesses have also seen a speed increase.
3437 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3438 specified. (This was at the root of some bad interactions between
3439 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3441 * bug fix: Internal error arguments for undefined functions are now
3442 computed correctly on the PPC/Linux platform.
3443 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3444 Raymond Toy's patch for CMU CL)
3445 * Support for the Solaris 9 operating environment has been included
3446 (thanks to Daniel Merritt)
3447 * A very ugly but hopefully complete draft of the missing FFI chapter
3448 of the manual has been created by reformatting the corresponding
3449 CMU CL manual chapter into (currently very ugly and incoherent)
3450 DocBook and bringing it up to date for SBCL behavior. Thus, the
3451 manual is now essentially complete, at least by my extreme
3452 once-and-only-once standards, whereby it's acceptable to refer to
3453 the doc strings of SB-EXT functions as the primary documentation.
3454 * The fasl file version number has changed again, due to cleanup of
3455 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3456 is no longer a static symbol.)
3458 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3459 * SBCL now builds with OpenMCL (version 0.12) as the
3460 cross-compilation host; also, more progress has been made toward
3461 bootstrapping under CLISP.
3462 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3464 * bug 158 fixed: The compiler can now deal with integer loop
3465 increments different from 1; fixing this turned out also to fix
3467 * bug 169 fixed: no more bogus warnings about using lexical bindings
3468 despite the presence of perfectly good SPECIAL declarations (thanks
3469 to David Lichteblau)
3470 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3471 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3472 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3474 * bug 180 fixed: Method combination specifications no longer ignore
3475 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3476 * bug fix: Structure type predicate functions now check their argument
3477 count as they should.
3478 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3479 correctly. (thanks to Pierre Mai)
3480 * minor incompatible change: The --noprogrammer option is deprecated
3481 in favor of the new --disable-debugger option, which is very similar.
3482 (The major difference is that it takes effect at a slightly different
3483 time at startup, causing handling of errors in --sysinit and
3484 --userinit files will be affected differently.) The
3485 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3486 been added to allow this functionality to be controlled from ordinary
3487 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3488 maintainers, who might want to run non-interactive scripts to
3489 build SBCL cores which will later be used interactively.)
3490 * minor incompatible change: The LOAD function no longer, when given
3491 a wild pathname to load, loads all files matching that pathname.
3492 Instead, an error of type FILE-ERROR is signalled.
3494 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3495 * bug 147 fixed: The compiler preserves its block link/count
3496 invariants more correctly now so that it doesn't crash. (thanks
3498 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3500 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3501 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3502 no-value-for-this-character as NIL (as specified by ANSI).
3503 * HOST-NAMESTRING on physical pathnames now returns a string that is
3504 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3505 (thanks to Christophe Rhodes)
3506 * The Alpha port handles icache flushing more correctly. (thanks to
3508 * More progress has been made toward bootstrapping under CLISP. (thanks
3509 to Christophe Rhodes)
3510 * The fasl file format has changed again, because dynamic loading
3511 on OpenBSD (which has non-ELF object files) motivated some cleanups
3512 in the way that foreign symbols are transformed and passed around.
3513 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3514 is no longer treated as whitespace by the reader, but instead as
3515 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3516 |A
\7fB|, instead of A as it used to.
3518 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3519 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3520 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3521 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3522 this as of 0.7.1.45, but was left out of the previous news section
3523 (thanks to Dan Barlow)
3524 * SBCL now runs on the Solaris operating system on SPARC architectures
3525 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3526 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3527 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3528 * SPARC backend cleanups, allowing builds of cores optimized for V8
3529 and V9 SPARCS, and also emission of code targeted to a particular
3530 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3532 * SBCL is closer to bootstrapping under CLISP, thanks to various
3533 fixes by Christophe Rhodes.
3534 * The fasl file format has changed again, to allow the compiler's
3535 INFO database to support symbol macros.
3536 * The user manual (in doc/) is formatted into HTML more nicely.
3537 (thanks to coreythomas)
3538 * The system is smarter about SUBTYPEP relationships, especially
3539 those involving NOT types (including types such as ATOM which are
3540 represented internally using NOT types). Thus SUBTYPEP is less
3541 likely to return (VALUES NIL NIL) in general, and in particular
3542 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3544 * The fasl file format has changed again, because the internal
3545 representation of types now includes a new slot to support the new
3546 SUBTYPEP-of-NOT-types logic.
3547 * (not a change in the main branch of SBCL, but a related prototype
3548 which can hopefully be merged into the main branch of SBCL in the
3549 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3550 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3551 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3552 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3553 dispatch macro character combinations. (thanks to Alexey Dejneka)
3554 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3555 that superficially look like logical namestrings correctly.
3556 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3558 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3559 default supplied-p) for &optional and &key arguments. (thanks to
3562 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3563 * incompatible change: The compiler is now less aggressive about
3564 tail call optimization, doing it only when (> SPACE DEBUG) or
3565 (> SPEED DEBUG). (This is an incompatible change because there are
3566 programs which relied on the old CMU-CL-style behavior to optimize
3567 away their unbounded recursion which will now die of stack overflow.)
3568 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3569 for non-GENCGC systems has been increased to 20M (since that
3570 seems much closer to the likely performance optimum for modern
3571 systems than the old 4M value was)
3572 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3573 and *DEBUG-PRINT-LEVEL*
3574 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3575 of CMU CL's support for SPARC, and various endianness and other
3576 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3577 * new syntactic sugar for the Unix command line: --load foo.bar is now
3578 an alternate notation for --eval '(load "foo.bar")'.
3580 ** The system now detects stack overflow and handles it gracefully,
3581 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3582 optimization settings. (This is a good thing in general, and
3583 its introduction in this version should be particularly timely
3584 for anyone whose code fails because of suppression of tail
3586 ** The system now hunts for the C variable "environ" in a more
3587 devious way, to avoid segfaults when the C library version differs
3588 between compile time and run time. (thanks to Christophe Rhodes)
3589 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3590 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3591 ** The compiler no longer issues bogus style warnings for undefined
3592 classes in the same source file as the DEFCLASSes which defined
3593 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3595 ** fixes in CONDITION class precedence list for undefined function
3596 errors (thanks to Alexei Dejneka)
3597 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3598 correctly. (thanks to Dan Barlow)
3599 ** portability fixes aiming at bootstrapping under CLISP (thanks
3600 to Dave McDonald and Christophe Rhodes)
3601 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3602 ** fixes in type translation and and type inference (thanks to
3604 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3605 ** various fixes in the new ports (thanks to Dan Barlow)
3606 * several changes related to debugging:
3607 ** suppression of tail recursion, as noted above
3608 ** stack overflow detection, as noted above
3609 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3610 is now the default. (For some time encapsulation has been more
3611 reliable than the breakpoint-based :ENCAPSULATE NIL
3612 implementation, at least on X86 systems; and I just noticed that
3613 encapsulation also seems closer to the spirit of the ANSI
3616 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3618 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3619 up properly again. (There was a packaging bug in 0.7.0 which
3620 left their definitions in SB-SYS::LOAD-FOREIGN and
3621 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3622 vital for most things which interface to C-level interfaces,
3623 like extensions working with sockets or databases or
3624 Perl-compatible regexes or whatever, and the need to fix
3625 this bug is the main reason that 0.7.1 was released so
3627 ** DEFGENERIC is now choosier about the methods it redefines, so that
3628 reLOADing a previously-LOADed file containing DEFGENERICs does
3629 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3630 cycle now works as it should. (thanks to Alexey Dejneka)
3631 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3632 fixed by Christophe Rhodes. (He actually submitted this patch
3633 months ago, and I delayed until after 0.7.0.)
3634 ** Bug 111 (internal compiler confusion about runtime checks on
3635 FUNCTION types) was fixed by Alexey Dejneka.
3636 * Some internal cleanups (getting rid of variables which aren't
3637 needed now that the byte interpreter is gone) caused the fasl
3638 file format number to change again.
3640 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3641 * major incompatible change: The default fasl file extension, i.e. the
3642 default extension for files produced by COMPILE-FILE, has changed
3643 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3645 ** There are many changes in the implementation of the compiler.
3646 SBCL is now essentially a compiler-only implementation of ANSI
3647 Common Lisp. EVAL still "interprets" a few special cases, but
3648 almost all the interesting cases are handled by creating
3649 a LAMBDA expression, calling COMPILE on it, then calling
3650 FUNCALL on the result.
3651 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3652 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3653 Since the code is newer, there might still be some new bugs
3654 (though not as many as before Martin Atzmueller's fixes:-). But
3655 the new code is substantially simpler and clearer, and hopefully
3656 any remaining bugs will be simpler, less fundamental, and more
3657 fixable then the bugs in the old code.
3658 ** The revised compiler is still a little unsteady on its feet.
3660 *** The debugging information it produces (particularly the names
3661 of FUNCTION objects) is sometimes much less useful than what
3662 the old compiler produced.
3663 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3664 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3665 within a MACROLET) has been temporarily weakened.
3666 ** There are new compiler optimizations for various functions:
3667 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3668 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3669 *** the math functions TRUNCATE, FLOOR, and CEILING
3670 *** the function-of-all-trades COERCE
3671 Mostly these should be transparent, but there's one
3672 potentially-annoying problem (bug 117): when the compiler
3673 inline-expands a function and does type analysis on the result,
3674 it can create control paths which have type mismatches, and
3675 when it can't prove that those control paths aren't taken,
3676 it will issue WARNINGs about the type mismatches. This is
3677 a particular problem in practice for the new sequence functions.
3678 It's not clear how this should be fixed, and for now, a
3679 workaround is given in the entry for 117 in the BUGS file.
3680 ** (Because of the interaction between the two previous items --
3681 occasional inlining problems and new inline expansions -- some
3682 of the new sequence function optimizations won't really kick in
3683 completely until debugging information, and then inlining, are
3684 straightened out in some future version.)
3685 * minor incompatible changes:
3686 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3687 DIRECTORY no longer implicitly promotes NIL slots of its
3688 pathname argument to :WILD. In particular, when you ask for the
3689 contents of a directory (which you used to be able to do without
3690 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3691 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3692 ** changes in behavior that ANSI explicitly defines to be
3693 implementation dependent:
3694 *** The new compiler-only implementation still conforms with ANSI,
3695 but acts a little different than before. Besides the obvious
3696 changes in performance tradeoffs (that the cost per form passed
3697 to EVAL has gone up, and the cost per form executed by EVAL
3698 has gone down), the behavior of the system changes a little
3699 because there are no longer any interpreted function objects.
3700 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3701 e.g. doing COMPILE on the output of interactive DEFUN is
3703 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3705 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3706 has changed from (:CL) to NIL.
3707 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3708 e.g. #\Newline and #\Tab, don't have names specified in the
3709 ANSI Common Lisp standard, is now based on their ASCII symbolic
3710 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3711 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3712 are no longer used for output.
3713 ** changes in internal implementation constants:
3714 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3715 4 million. (If your application spends a lot of time GCing and
3716 you have a lot of RAM, you might want to experiment with
3717 increasing it even more.)
3718 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3719 However, almost all old code should still continue to work without
3720 immediate update, as SB-C-CALL is now a (deprecated) nickname
3722 ** Old operator names in the style DEF-FOO are now deprecated in
3723 favor of new corresponding names DEFINE-FOO, for consistency with
3724 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3725 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3726 internal symbols, but a few supported extensions like
3727 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3728 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3729 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3730 etc. as you get deeper into recursive calls to the debugger
3731 command loop, instead of the old "5]", "5]]", "5]]]"
3732 sequence. (I was motivated to do this when squabbles between
3733 ILISP and SBCL left me very deeply nested in the debugger. In the
3734 short term, this change will probably provoke more ILISP/SBCL
3735 squabbles, but hopefully it will be an improvement in the long run.)
3736 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3737 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3738 function, since declaiming return types involving aliens is
3739 (1) annoyingly messy to do by hand and (2) vital to efficient
3740 compilation of code which calls such functions.
3741 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3742 longer reexported by the SB-EXT package. They're solely useful
3743 for alien code, so it seems more logical that you should get
3744 them from the SB-ALIEN package, not in SB-EXT.
3745 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3746 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3747 features. Instead, the code that they used to control is always
3748 built into the system.
3749 * many other bug fixes
3750 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3751 advantage of the new EVAL-WHEN stuff and to clean them up in
3752 general, and they are now more ANSI-compliant in a number of
3753 ways. Martin Atzmueller is responsible for a lot of this.
3754 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3756 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3757 *** correct ERROR type for various file operations
3758 *** some fixes for Lisp streams
3759 *** DEFMETHOD syntax checking
3760 *** changing old weird representation of debug information as
3761 strings (which, among their other deficiencies, don't transform
3762 correctly when you rename packages, and don't change their
3763 print representation when you change things like *PACKAGE*
3764 and *PRINT-LENGTH*) to symbols and lists of symbols
3765 He also made several improvements and fixed several bugs in DESCRIBE.
3766 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3768 *** misbehavior of WRITE-STRING/WRITE-LINE
3769 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3770 and several other LOOP problems as well
3771 *** DIRECTORY when similar filenames are present
3772 *** DEFGENERIC with :METHOD options
3773 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3774 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3775 *** argument ordering in FIND with :TEST option
3776 *** mishandled package designator argument in APROPOS-LIST
3777 *** various problems in the backquote readmacro
3778 *** a bug in APROPOS
3779 *** probably some others that I'm not describing very well here,
3780 since the CVS log documents them by reference to sbcl-devel
3781 messages, and the SourceForge archives aren't working well.:-(
3782 ** Dan Barlow improved the Alpha port (and is making progress on the
3783 PPC port, for those of you who think different).
3784 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3785 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3786 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3787 continues to work on the SPARC port (for those of you in a position
3788 to look down upon our little PC-compatible boxes from a great height).
3789 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3790 and uses that copy, rather than the current dynamic value, when
3791 it's trying to decide whether to truncate output. Thus e.g.
3792 (let ((*print-lines* 50))
3793 (pprint-logical-block (stream nil)
3795 (let ((*print-lines* 8))
3796 (print (aref possiblybigthings i) stream)))))
3797 should now truncate the logical block only at 50 lines, instead of
3798 often truncating it at 8 lines, as it did before.
3799 * The doc/cmucl/ directory, containing old CMU CL documentation
3800 from the time of the fork, is no longer part of the base system.
3801 SourceForge has shut down its anonymous FTP service, and with it
3802 my original plan for distributing the old CMU CL documentation
3803 there. For now, if you need these files you can download an old
3804 SBCL source release and extract them from it.
3805 * The fasl file version number changed again, for dozens of reasons,
3806 some of which are apparent above.
3808 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3809 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3810 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3811 patch, so that DISASSEMBLE output is much nicer.
3812 * The code in the SB-PROFILE package now seems reasonably stable.
3813 I still haven't decided what the final interface should look like
3814 (I'd like PROFILE to interact cleanly with TRACE, since both
3815 facilities use function encapsulation) but if you have a need
3816 for profiling now, you can probably use it successfully with
3817 the current CMU-CL-style interface.
3818 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3819 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3820 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3821 link now returns the dangling link itself, and for similar
3822 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3823 link itself. (In these cases the old code signalled an error and
3824 looped endlessly, respectively.) Thus, DIRECTORY now works even
3825 in the presence of dangling and cyclic symbolic links.
3826 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3827 is now a supported extension again, since the consensus on
3828 sbcl-devel was that it can be useful for ordinary development
3829 work, not just for debugging SBCL itself.
3830 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3831 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3832 that currently-defined functions might be redefined later with
3833 different return types.
3834 * Hash tables can be printed readably, as inspired by CMU CL code
3835 of Eric Marsden and SBCL code of Martin Atzmueller.
3836 * better error handling in CLOS method combination, thanks to
3837 Martin Atzmueller porting Pierre Mai's CMU CL patches
3838 * more overflow fixes for >16Mbyte I/O buffers
3839 * A bug in READ has been fixed, so that now a single Ctrl-D
3840 character suffices to cause end-of-file on character streams.
3841 In particular, now you only need one Ctrl-D at the command
3842 line (not two) to exit SBCL.
3843 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3845 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3846 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3847 does the right thing.
3848 * The implementation of some type tests, especially for CONDITION
3849 types, is now tidier and maybe faster, due to CMU CL code
3850 originally by Douglas Crosher, ported by Martin Atzmueller.
3851 * Some math functions have been fixed, and there are new
3852 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3853 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3854 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3855 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3856 likely to be built into sbcl-0.7.0.)
3857 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3858 CL, ported to SBCL by Martin Atzmueller)
3859 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3860 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3861 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3862 * There's a new slam.sh hack to shorten the edit/compile/debug
3863 cycle for low-level changes to SBCL itself, and a new
3864 :SB-AFTER-XC-CORE target feature to control the generation of
3865 the after-xc.core file needed by slam.sh.
3866 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3867 COMPILE-FILE is no longer supported, so that now every function
3868 gets an entry point, so that block compilation looks a little
3869 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3870 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3871 returns the number of bytes consed since the system started,
3872 rather than the number consed since the first time the function
3873 was called. (The new definition parallels ANSI functions like
3874 CL:GET-INTERNAL-RUN-TIME.)
3875 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3876 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3877 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3878 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3879 * Fasl file version numbers are now independent of the target CPU,
3880 since historically most system changes which required version
3881 number changes have affected all CPUs equally. Similarly,
3882 the byte fasl file version is now equal to the ordinary
3885 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3886 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3887 is no longer recognized. I apologize for this, because it was
3888 listed in SB-EXT as a supported extension, but I found that
3889 its existing behavior was poorly specified, as well as incorrectly
3890 specified, and it looked like too much of a mess to straighten it
3891 out. I have enough on my hands trying to get ANSI stuff to work..
3892 * many patches ported from CMU CL by Martin Atzmueller, with
3893 half a dozen bug fixes in pretty-printing and the debugger, and
3894 half a dozen others elsewhere
3895 * fixed bug 13: Floating point infinities are now supported again.
3896 They might still be a little bit flaky, but thanks to bug reports
3897 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3898 as flaky as they were.
3899 * The --noprogrammer command line option is now supported. (Its
3900 behavior is slightly different in detail from what the old man
3901 page claimed it would do, but it's still appropriate under the
3902 same circumstances that the man page talks about.)
3903 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3904 are now supported, and enabled by default. Thus, the compiler can
3905 handle many floating point and complex operations much less
3906 inefficiently. (Thus e.g. you can implement a complex FFT
3908 * The compiler now detects type mismatches between DECLAIM FTYPE
3909 and DEFUN better, and implements CHECK-TYPE more correctly, and
3910 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3911 thanks to patches from Martin Atzmueller.
3912 * various fixes to make the cross-compiler more portable to
3913 ANSI-conforming-but-different cross-compilation hosts (notably
3914 Lispworks for Windows, following bug reports from Arthur Lemmens)
3915 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3916 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3917 fixed thanks to Pierre Mai's CMU CL patches.
3918 * fixes to keep the system from overflowing internal counters when
3919 it tries to use i/o buffers larger than 16M bytes
3920 * fixed bug 45a: Various internal functions required to support
3921 complex special functions have been merged from CMU CL sources.
3922 (When I was first setting up SBCL, I misunderstood a compile-time
3923 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3924 * improved support for type intersection and union, fixing bug 12
3925 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3926 more obscure bugs as well
3927 * some steps toward byte-compiling non-performance-critical
3928 parts of the system, courtesy of patches from Martin Atzmueller
3929 * Christophe Rhodes has made some debian packages of sbcl at
3930 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3931 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3932 stable, but are nonetheless usable. When he's ready, I'd be happy
3933 to add them to the SourceForge "File Releases" section. (And if
3934 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3935 * new fasl file format version number (because of changes in
3936 internal representation of (OR ..) types to accommodate the new
3937 support for (AND ..) types, among other things)
3939 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3940 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3942 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3944 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3945 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3946 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3947 * DESCRIBE now works on structure objects again.
3948 * Most function call argument type mismatches are now handled as
3949 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3950 know whether the function will be redefined before the call is
3951 executed. (The compiler could flag local calls with full WARNINGs,
3952 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3953 it doesn't keep track of enough information to know whether calls
3954 are local in this sense.)
3955 * Compiler output is now more verbose, with messages truncated
3956 later than before. (There should be some supported way for users
3957 to override the default verbosity, but I haven't decided how to
3958 provide it yet, so this behavior is still controlled by the internal
3959 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3960 src/compiler/ir1util.lisp.)
3961 * Fasl file format version numbers have increased again, because
3962 support for the Gray streams extension changes the layout of the
3963 system's STREAM objects.
3964 * The Gray subclassable streams extension now works, thanks to a
3965 patch from Martin Atzmueller.
3966 * The full LOAD-FOREIGN extension (not just the primitive
3967 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3968 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3969 but like most other programs, it defaults to copying the Unix
3970 environment from the original process instead of starting the
3971 new process in an empty environment.
3972 * Extensions which manipulate the Unix environment now support
3973 an :ENVIRONMENT keyword option which doesn't smash case or
3974 do other bad things. The CMU-CL-style :ENV option is retained
3975 for porting convenience.
3976 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3977 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3979 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3981 * A patch from Martin Atzmueller seems to have solved the SIGINT
3982 problem, and as far as we know, signal-handling now works cleanly.
3983 (If you find any new bugs, please report them!)
3984 * The system no longer defaults Lisp source file names to types
3985 ".l", ".cl", or ".lsp", but only to ".lisp".
3986 * The compiler no longer uses special default file extensions for
3987 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3988 seems to expect a single default extension for all compiled code,
3989 and there's no compelling reason to try to stretch the standard
3990 to allow two different extensions.) Instead, byte-compiled files
3991 default to the same extension as native-compiled files.
3992 * Fasl file format version numbers have increased again, because
3993 a rearrangement of internal implementation packages made some
3994 dumped symbols in old fasl files unreadable in new cores.
3995 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3996 many fewer weird special cases.
3997 * Bug #17 (differing COMPILE-FILE behavior between logical and
3998 physical pathnames) has been fixed, and some related misbehavior too,
3999 thanks to a patch from Martin Atzmueller.
4000 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4001 by Tim Moore, ported to SBCL by Martin Atzmueller.
4002 * Martin Atzmueller fixed several filesystem-related problems,
4003 including bug #36, in part by porting CMU CL patches, which were
4004 written in part by Paul Werkowski.
4005 * More compiler warnings in src/runtime/ are gone, thanks to
4006 more patches from Martin Atzmueller.
4007 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4010 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4012 * DESCRIBE now works on CONDITION objects.
4013 * The debugger now handles errors which arise when trying to print
4014 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4016 * The build system now uses an additional file, customize-target-features.lisp,
4017 to allow local modifications to the target *FEATURES* list. (The point of
4018 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4019 debugging features enabled, without having to worry about propagating it
4020 into everyone's system when I do a "cvs update".) When no
4021 customize-target-features.lisp file exists, the target *FEATURES* list
4022 should be constructed the same way as before.
4023 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4024 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4025 warn about behavior which is undefined under the ANSI spec. Note: This
4026 is specified by ANSI, but it's not very popular with programmers.
4027 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4028 macro in the SBCL sources for an example of a workaround which you
4029 might use to make portable ANSI-standard code which does what you want.
4030 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4031 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4032 when it's used as a non-toplevel form. (This is required in order
4033 to implement the DEFCONSTANT-EQX macro.)
4034 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4035 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4036 funny things, due to bugs in the implementation of EVAL-WHEN
4037 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4038 non-toplevel DEFCONSTANTs are uncommon.)
4039 * The core file version number and fasl file version number have been
4040 incremented, because the old noncompliant DEFCONSTANT behavior involved
4041 calling functions which no longer exist, and because I also took the
4042 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4043 * fixed bug 1 (error handling before read-eval-print loop starts), and
4044 redid debugger restarts and related debugger commands somewhat while
4046 ** The QUIT debugger command is gone, since it did something
4047 rather different than the SB-EXT:QUIT command, and since it never
4048 worked properly outside the main toplevel read/eval/print loop.
4049 Invoking the new TOPLEVEL restart provides the same functionality.
4050 ** The GO debugger command is also gone, since you can just invoke
4051 the CONTINUE restart directly instead.
4052 ** The TOP debugger command is also gone, since it's redundant with the
4053 FRAME 0 command, and since it interfered with abbreviations for the
4055 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4057 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4058 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4059 some porting work by Martin Atzmueller)
4060 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4061 with a boatload of patches which he ported to SBCL. Now that those
4063 ** The system tries to make sure that its low-priority messages
4064 are prefixed by semicolons, to help people who like to use
4065 syntax highlighting in their ILISP buffer. (This patch
4066 was originally due to Raymond Toy.)
4067 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4068 patches originally written by Raymond Toy.
4069 ** The compiler understands coercion between single-value and
4070 multiple-VALUES type expressions better, getting rid of some very
4071 weird behavior, thanks to patches originally by Robert MacLachlan
4072 and Douglas Crosher.
4073 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4074 lambda lists, thanks to a patch originally by Pierre Mai.
4075 ** The system no longer bogusly warns about "abbreviated type
4077 ** The compiler gets less confused by inlining and RETURN-FROM,
4078 thanks to some patches originally by Tim Moore.
4079 ** The system no longer hangs when dumping circular lists to fasl
4080 files, thanks to a patch originally from Douglas Crosher.
4081 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4082 undefined function error.
4083 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4084 in structure slot initforms) for now, documented workaround instead:-|
4085 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4086 * fixed bug 5: added stubs for various Gray stream functions called
4087 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4088 installed, at least appropriate type errors are generated
4089 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4090 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4091 consistently in DEFMETHOD forms.
4092 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4093 it doesn't seem to affect SBCL after all
4094 * The C runtime system now builds with better optimization and many
4095 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4097 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4099 * The system is now under CVS at SourceForge (instead of the
4100 CVS repository on my home machine).
4101 * The new signal handling code has been tweaked to treat register
4102 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4103 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4104 downstream, e.g. in debug-int.lisp, has implicit dependencies
4105 on the unsignedness of integer representation of machine words,
4106 and that caused the system to bomb out with infinite regress
4107 when trying to recover from type errors involving signed values,
4108 e.g. (BUTLAST '(1 2 3) -1).)
4109 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4110 (This was one of the bugs Peter Van Eynde reported back in July.)
4111 * The system now uses code inspired by Colin Walters' O(N)
4112 implementation of MAP (from the cmucl-imp@cons.org mailing
4113 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4114 inline the MAP operation, and there is more than one
4115 sequence argument to the MAP call (so that it can't just
4116 do ETYPECASE once and for all based on the type of the
4117 single sequence argument). (The old non-inline implementation
4118 of the general M-argument sequence-of-length-N case required
4119 O(M*N*N) time when any of the sequence arguments were LISTs.)
4120 * The QUIT :UNIX-CODE keyword argument has been renamed to
4121 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4122 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4124 * Raymond Wiker's patches to port dynamic loading from Linux to
4125 FreeBSD have been added.
4126 * The BUGS file is now more nearly up to date, thanks in large part
4127 to Martin Atzmueller's review of it.
4128 * The debugger now flushes standard output streams before it begins
4129 its output ("debugger invoked" and so forth).
4130 * The core version number and fasl file version number have both
4131 been incremented, because of incompatible changes in the layout
4133 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4134 used to not be called for a saved Lisp image.)
4135 * Martin Atzmueller's version of a patch to fix a compiler crash,
4136 as posted on sbcl-devel 13 September 2000, has been installed.
4137 * Instead of installing Martin Atzmueller's patch for the
4138 compiler transform for SUBSEQ, I deleted the compiler transform,
4139 and transforms for some similar consing operations.
4140 * A bug in signal handling which kept TRACE from working on OpenBSD
4142 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4143 away by constant folding
4144 * The system now defines its address space constants in one place
4145 (in the Lisp sources), and propagates them automatically elsewhere
4146 (through GENESIS and the sbcl.h file). Therefore, patching the
4147 address map is less unnecessarily tedious and error-prone. The
4148 Lisp names of address space constants have also been systematized.
4149 * CVS tags like dollar-Header-dollar have been removed from
4150 the sources, because they have never saved me trouble and
4151 they've been source of trouble working with patches and other
4152 diff-related operations.
4153 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4154 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4156 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4158 * The system has been ported to OpenBSD.
4159 * The system now compiles with a simple "sh make.sh" on the systems
4160 that it's supported on. I.e., now you no longer need to tweak
4161 text in the target-features.lisp-expr and symlinks in src/runtime/
4162 by hand, the make.sh takes care of it for you.
4163 * The system is no longer so grossly inefficient when compiling code
4164 involving vectors implemented as general (not simple) vectors (VECTOR T),
4165 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4166 which dares to use the various sequence functions on non-simple
4167 vectors, takes less of a performance hit.
4168 * There is now a primitive type predicate VECTOR-T-P
4169 to test for the (VECTOR T) type, so that e.g.
4170 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4171 can now be compiled with some semblance of efficiency. (The old code
4172 turned the type declaration into a full call to %TYPEP at runtime!)
4173 * AREF on (VECTOR T) is still not fast, since it's still compiled
4174 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4175 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4176 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4177 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4178 without making full calls to SUBTYPEP at runtime.
4179 (Some analogous efficiency issues for non-simple vectors specialized to
4180 element types other than T, or for non-simple multidimensional arrays,
4181 have not been addressed. They could almost certainly be handled the
4182 same way if anyone is motivated to do so.)
4183 * The changes in array handling break binary compatibility, so
4184 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4185 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4186 instead of (VALUES T T).
4187 * By following the instructions that Dan Barlow posted to sbcl-devel
4188 on 2 July 2000, I was able to enable primitive dynamic object
4189 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4190 functionality is not implemented (since it calls ld to resolve
4191 library references automatically, requiring RUN-PROGRAM for its
4192 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4193 try to resolve library references) is now supported.
4194 * The system now flushes the standard output streams when it terminates,
4195 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4196 them at several other probably-convenient times, e.g. in each pass of
4197 the toplevel read-eval-print loop, and after evaluating a form given
4198 as an "--eval" command-line option. (These changes were motivated by a
4199 discussion of stream flushing issues on cmucl-imp in August 2000.)
4200 * The source transform for TYPEP of array types no longer assumes
4201 that an array whose element type is a not-yet-defined type
4202 is implemented as an array of T, but instead punts, so that the
4203 type will be interpreted at runtime.
4204 * There is now some support for cross-compiling in make.sh: each of
4205 the phases of make.sh has its own script. (This should be transparent
4206 to people doing ordinary, non-cross-compile builds.)
4207 * Since my laptop doesn't have hundreds of megabytes of memory like
4208 my desktop machine, I became more motivated to do some items on
4209 my to-do list in order to reduce the size of the system a little:
4210 ** Arrange for various needed-only-at-cold-init things to be
4211 uninterned after cold init. To support this, those things have
4212 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4213 symbols with such names are now uninterned after cold init).
4214 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4215 fasl files for cold load.
4216 ** Remove the old compiler structure pooling code (which used to
4217 be conditional on the target feature :SB-ALLOC) completely.
4218 ** Redo the representation of some data in cold init to be more compact.
4219 (I also looked into supporting byte compiled code at bootstrap time,
4220 which would probably reduce the size of the system a lot, but that
4221 looked too complicated, so I punted for now.)
4222 * The maximum signal nesting depth in the src/runtime/ support code has
4223 been reduced from 4096 to 256. (I don't know any reason for the very
4224 large old value. If the new smaller value turns out to break something,
4225 I'll probably just bump it back up.)
4226 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4228 * Many, many bugs reported by Peter Van Eynde have been added to
4229 the BUGS list; some have even been fixed.
4230 * While enabling dynamic object file loading, I tried to make the
4231 code easier to understand, renaming various functions and variables
4232 with less ambiguous names, and changing some function calling
4233 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4234 * While trying to figure out how to do the OpenBSD port, I tried to
4235 clean up some of the code in src/runtime/. In particular, I dropped
4236 support for non-POSIX signal handling, added various comments,
4237 tweaked the code to reduce the number of compilation warnings, and
4238 renamed some files to increase consistency.
4239 * To support the new automatic configuration functionality in make.sh,
4240 the source file target-features.lisp-expr has been replaced with the
4241 source file base-target-features.lisp-expr and the machine-generated
4242 file local-target-features.lisp-expr.
4243 * fixed a stupid quoting error in make.sh so that using CMU CL
4244 "lisp -batch" as cross-compilation host works again
4246 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4248 * DESCRIBE no longer tries to call itself recursively to describe
4249 bound/fbound values, so that it no longer fails on symbols which are
4250 bound to themselves (like keywords, T, and NIL).
4251 * DESCRIBE now works on generic functions.
4252 * The printer now prints less-screwed-up representations of closures
4253 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4254 * A private symbol is used instead of the :EMPTY keyword previously
4255 used to mark empty slots in hash tables. Thus
4256 (DEFVAR *HT* (MAKE-HASH-TABLE))
4257 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4258 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4259 now does what ANSI says that it should. (You can still get
4260 similar noncompliant behavior if bang on the hash table
4261 implementation with all the symbols you get back from
4262 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4263 This breaks binary compatibility, since tests for equality to
4264 :EMPTY are wired into things like the macroexpansion of
4265 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4267 * There's now a minimal placeholder implementation for CL:STEP,
4268 as required by ANSI.
4269 * An obscure bug in the interaction of the normal compiler, the byte
4270 compiler, inlining, and structure predicates has been patched
4271 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4272 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4274 * Missing ordinary arguments in a macro call are now detected even
4275 when the macro lambda list contains &KEY or &REST.
4276 * The debugger no longer complains about encountering the top of the
4277 stack when you type "FRAME 0" to explicitly instruct it to go to
4278 the top of the stack. And it now prints the frame you request even
4279 if it's the current frame (instead of saying "You are here.").
4280 * As specified by ANSI, the system now always prints keywords
4281 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4283 * The default initial SIZE of HASH-TABLEs is now smaller.
4284 * Type information from CLOS class dispatch is now propagated
4285 into DEFMETHOD bodies, so that e.g.
4286 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4288 is now basically equivalent to
4289 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4290 (DECLARE (TYPE SINGLE-FLOAT X))
4292 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4293 operation, without having to do run-time type dispatch.
4294 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4295 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4296 * Since I don't seem to be making big file reorganizations very often
4297 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4298 is overflowing my ability to conveniently back them up), I've finally
4299 checked the system into CVS. (The CVS repository is on my home system,
4300 not at SourceForge -- putting it on SourceForge might come later.)
4301 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4302 high-level GC-NOTIFY-FOO functions send their output. (There's
4303 still very little control of where low-level verbose GC functions
4304 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4305 less than it used to -- the GC-NOTIFY-FOO functions are now under
4306 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4307 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4308 in only one place in the source code, and propagates it automatically
4309 everywhere that it's needed. Thus e.g. when I bump the version from
4310 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4311 * The C source files now include boilerplate legalese and documentation
4312 at the head of each file (just as the Lisp source files already did).
4313 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4314 to his page will be replaced with a link to his new CLiki service.
4316 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4318 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4319 * The build process now looks for GNU make under the default name "gmake",
4320 instead of "make" as it used to. If GNU make is not available as "gmake"
4321 on your system, you can change this default behavior by setting the
4322 GNUMAKE environment variable.
4323 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4324 can build without error under CMU CL.
4326 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4328 * There is now a partial SBCL user manual (with some new text and some
4329 text cribbed from the CMU CL manual).
4330 * The beginnings of a profiler have been added (starting with the
4331 CMU CL profiler and simplifying and cleaning up). Eventually the
4332 main interface should be through the TRACE macro, but for now,
4333 it's still accessed through vaguely CMU-CL-style functions and macros
4334 exported from the package SB-PROFILE.
4335 * Some problems left over from porting CMU CL to the new
4336 cross-compilation bootstrap process have been cleaned up:
4337 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4338 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4339 part when getting rid of the compiler *BACKEND* stuff.)
4340 ** The value of *NULL-TYPE* was screwed up, because it was
4341 being initialized before the type system knew the final
4342 definition of the 'NULL type. This screwed up several key
4343 optimizations in the compiler, causing inefficiency in all sorts
4344 of places. (I found it because I wanted to understand why
4345 GET-INTERNAL-RUN-TIME was consing.)
4346 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4347 PROCLAIM FTYPE information. Unfortunately this broke binary
4348 compatibility again, since now the forms output by DEFGENERIC
4349 to refer to functions which didn't exist in 0.6.3.
4350 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4351 can use the new (as of 0.6.3) transform for SOME into MAP into
4353 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4354 values to (INTEGER 0 1000000), so that the time code will no longer
4355 occasionally get blown up by Linux returning 1000000 microseconds
4356 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4357 its output conform to the ANSI spec. (Alas, this makes its output
4358 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4359 * A full call to MAP NIL with a single sequence argument no longer conses.
4360 * fixes to problems pointed out by Martin Atzmueller:
4361 * The manual page no longer talks about multiprocessing as though
4362 it were currently supported.
4363 * The ILISP support patches have been removed from the distribution,
4364 because as of version 5.10.1, ILISP now supports SBCL without us
4365 having to maintain patches.
4366 * added a modified version of Raymond Toy's recent CMU CL patch for
4367 EQUALP comparison of HASH-TABLE
4369 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4371 * The system still can't cross-compile itself with
4372 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4373 entails), but at least it can compile more of itself that way
4374 than it used to be able to, and various buglets which were uncovered
4375 by trying to cross-compile itself that way have now been fixed.
4376 * This release breaks binary compatibility again. This time
4377 at least I've incremented the FASL file format version to 2, so that the
4378 problem can be detected reliably instead of just causing weird errors.
4379 * various new style warnings:
4380 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4381 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4382 ** using the lexical binding of a variable named in the *FOO* style
4383 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4384 as specified by ANSI.
4385 * *RANDOM-STATE* is no longer automatically initialized from
4386 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4387 default behavior of the system is to repeat its behavior every time
4388 it's run. If you'd like to change this behavior, you can always
4389 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4390 old convention there was no comparably easy way to get the system to
4391 repeat its behavior every time it was run.
4392 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4393 FTYPE declarations has been removed, in favor of their ANSI
4394 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4395 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4396 the types of their sequence arguments can be determined at compile time.
4397 This is done through a new open code expansion for MAP which eliminates
4398 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4399 argument types can be determined at compile time.
4400 * The optimizer now transforms COERCE into an identity operation when it
4401 can prove that the coerced object is already of the correct type. (This
4402 can be a win for machine generated code, including the output of other
4403 optimization transforms, such as the MAP transform above.)
4404 * Credit information has been moved from source file headers into CREDITS.
4405 * Source file headers have been made more standard.
4406 * The CASE macro now compiles without complaining even when it has
4409 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4411 * (Note that the way that the PCL macroexpansions were rewritten
4412 to accommodate the change in DEFGENERIC below breaks binary
4413 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4414 not run under sbcl-0.6.2. Once we get out of alpha releases,
4415 i.e. hit release 1.0.0, we'll probably try to maintain binary
4416 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4417 and sbcl-1.4.4. Until then, however, it might be fairly common
4418 for maintenance releases to break binary compatibility.)
4419 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4421 * The reporting of unhandled signals has been changed to print some
4422 explanatory text as well as the report form. (Previously only
4423 the report form was printed.)
4424 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4425 it defines, so that the compiler no longer issues undefined function
4426 warnings for compiled-but-not-yet-loaded generic functions.
4427 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4428 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4429 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4430 section of the ANSI spec are supported.
4431 * The "" nickname for the "KEYWORD" package has been removed.
4432 The reader still handles symbol tokens which begin with a package marker
4433 as keywords, but it doesn't expose its mechanism for doing so in the
4434 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4435 * The system now issues STYLE-WARNINGs for contradictory TYPE
4436 proclamations. (Warnings for contradictory FTYPE proclamations would
4437 be nice too, but those can't be done usefully unless the type system
4438 is made smarter about FUNCTION types.)
4439 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4440 systematized, so that "*target-*.lisp is supposed to exist only on the
4441 target and imply that there's a related file which exists on the
4442 host, and *host-*.lisp is supposed to exist only on the host and imply
4443 that there's a related file which exists on the target. This involves a
4444 lot of renaming. Hopefully the acute confusion caused by the renaming
4445 will be justified by the reduction in chronic confusion..
4446 ** runtime-type.lisp -> early-target-type.lisp
4447 ** target-type.lisp -> late-target-type.lisp
4448 ** early-host-format.lisp -> early-format.lisp
4449 ** late-host-format.lisp -> late-format.lisp
4450 ** host-error.lisp -> misc-error.lisp
4451 ** early-error.lisp -> early-target-error.lisp
4452 ** late-error.lisp -> late-target-error.lisp
4453 ** host-defboot.lisp -> early-defboot.lisp
4454 ** code/misc.lisp -> code/target-misc.lisp
4455 ** code/host-misc.lisp -> code/misc.lisp
4456 ** code/numbers.lisp -> code/target-numbers.lisp
4457 ** code/early-numbers.lisp -> numbers.lisp
4458 ** early-host-type.lisp -> early-type.lisp
4459 ** late-host-type.lisp -> late-type.lisp
4460 ** host-typep.lisp -> typep.lisp
4461 ** load.lisp -> target-load.lisp
4462 ** host-load.lisp -> load.lisp
4463 ** host-disassem.lisp -> disassem.lisp
4464 ** host-insts.lisp -> insts.lisp
4465 ** byte-comp.lisp -> target-byte-comp.lisp
4466 ** host-byte-comp.lisp -> byte-comp.lisp
4467 ** host-signal.lisp -> signal.lisp
4468 ** host-defstruct.lisp -> defstruct.lisp
4469 ** late-target-type.lisp -> deftypes-for-target.lisp
4470 Furthermore, several other previously target-only files foo.lisp (e.g.
4471 hash-table.lisp and random.lisp) have been split into a target-and-host
4472 foo.lisp file and a target-only target-foo.lisp file, with their key type
4473 definitions in the target-and-host part, so that the cross-compiler will
4474 know more about target types.
4475 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4476 gone away. In their place are various *BACKEND-FOO* variables
4477 corresponding to the slots of the old structure.
4478 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4479 into the target SBCL has been fixed.
4480 * The chill.lisp system for loading cold code into a running SBCL
4482 * Support for the CMU CL "scavenger hook" extension has been removed.
4483 (It was undocumented and unused in the CMU CL sources that SBCL was
4484 derived from, and stale in sbcl-0.6.1.)
4485 * Various errors in the cross-compiler type system were detected
4486 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4487 (enabling various consistency checks). Many of them were fixed,
4488 but some hard problems remain, so the compiler is back to
4489 running without *TYPE-SYSTEM-INITIALIZED* for now.
4490 * As part of the cross-compiler type system cleanup, I implemented
4491 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4492 * I have started adding UNCROSS calls throughout the type system
4493 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4494 on cross-compiler input files will be able to go away, and various
4496 * CONSTANTP now returns true for quoted forms (as explicitly required
4499 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4501 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4502 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4503 (MAKE-LIST -1). In the longer term, it ought to become true
4504 that declarations are assertions even at SAFETY 1. For now, it's not
4505 quite true even at SAFETY 3, but it's at least more nearly true..
4506 (Note that this change seems to increases the size of the system by
4507 O(5%) and to decrease the speed of the compiler by 20% or more.)
4508 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4509 for the problem of printing dozens of lines of distracting information
4510 about low-level system machinery as part of the top stack frame
4511 on entry to the debugger when an undefined function was called.
4512 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4513 is not reset to COMMON-LISP-USER.
4514 * Compilation of stuff related to dyncount.lisp has been made conditional
4515 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4516 smaller. The various dyncount-related symbols have been moved into
4517 a new "SB-DYNCOUNT" package.
4518 * tty-inspect.lisp has been renamed to inspect.lisp.
4519 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4520 feature has gone away. (When we eventually port to other flavors of
4521 libc and/or Unix, we'll try to make the differences between flavors
4522 invisible at the user level.)
4523 * Various other *FEATURES* tags, and/or their associated conditionals,
4524 have been removed if obsolescent, or given better documentation, or
4525 sometimes given more-mnemonic names.
4527 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4529 * tidied up "make.sh" script
4530 * tidied up system directory structure
4531 * better "clean.sh" behavior
4532 * added doc/FOR-CMUCL-DEVELOPERS
4533 * many many small tweaks to output format, e.g. removing possibly-confusing
4534 trailing #\. character in DESCRIBE-INSTANCE
4535 * (EQUALP #\A 'A) no longer signals an error.
4536 * new hashing code, including EQUALP hashing
4537 * tidied up Lisp initialization and toplevel
4538 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4539 * command line argument processing
4540 * added POSIX-GETENV function to deal with Unix-ish environment variables
4541 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4542 terminating SBCL on EOF
4543 * non-verbose GC by default
4544 * There is no more "sbcl" shell script; the sbcl file is now the C
4545 runtime executable (just like CMU CL).
4546 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4548 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4549 symbol and package information as Lisp native symbol and package objects
4550 instead of strings naming symbols and strings naming packages. This way,
4551 whenever packages are renamed (as in warm init), debug information is
4552 transformed along with everything else.
4553 * tweaked the optimization policy declarations which control the building
4554 of SBCL itself. Now, among other things, the system no longer saves
4555 source location debugging information. (This helps two problems at once
4556 by reducing SBCL size and by keeping SBCL from trying to look for its
4557 sources -- which may not exist -- when reporting errors.)
4558 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4559 debugging and testing purposes
4560 * cleaned up printing, making the printer call PRINT-OBJECT for
4561 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4562 methods, giving nearly-ANSI behavior
4563 * converted almost all special variables to use *FOO* naming convention
4564 * deleted PARSE-TIME functionality, since it can be done portably
4565 * moved some files out of cold init into warm init
4566 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4568 * regularized formatting of source files
4569 * added an install.sh script
4570 * fixed ridiculous memory usage of cross-compiler by making
4571 compiler/alloc.lisp not try to do pooling unless it can hook
4572 itself into the GC of the cross-compilation host. Now the system
4573 builds nicely on my old laptop.
4574 * added :SB-ALLOC in target-features.lisp-expr
4575 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4576 was not implemented)
4577 * re-did condition handling and note reporting in the compiler. Notes
4578 are no longer handled by signalling conditions. Style warnings
4579 and warnings are handled more correctly and reported in such a way
4580 that it's easy to find one or the other in your output (so that you
4581 can e.g. figure out which of many problems caused COMPILE-FILE to
4583 * changed the severity of several compiler warnings from full WARNING
4584 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4585 compiler note reporting so that it doesn't use the condition system
4586 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4587 COMPILE-FILE command)
4588 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4589 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4590 no longer supported, since their implementation was deeply intertwingled
4591 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4592 * removed bogus "support" for compiler macros named (SETF FOO), and
4593 removed the compiler macro for SETF INFO (but only after making a fool
4594 of myself on the cmucl-imp mailing list by posting a bogus patch for
4595 DEFINE-COMPILER-MACRO..)
4596 * Compiled files containing forms which have side effects on the Lisp
4597 reader (such as DEFPACKAGE forms) are now handled more correctly.
4598 (Compiler queuing of top level lambdas has been suppressed by setting
4599 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4600 * deleted various currently-unused source files, e.g. gengc.lisp. They
4601 may be added back at some point e.g. when porting to other architectures,
4602 but until they are it's distracting to distribute them and to try to
4604 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4605 there were so many of them they're just distractions, and UNCROSS is
4606 known to be able to handle the current sources
4607 * moved PROFILE functionality into TRACE, so that it will be clear
4608 how the wrapping and unwrapping of functions when you profile them
4609 interacts with the wrapping and unwrapping of functions when you
4610 trace them. (Actually, the functionality isn't there yet, but at least
4611 the interface specification is there. Hopefully, the functionality will
4612 arrive with some future maintenance release.)
4613 * removed host-oops.lisp
4614 * changed signature of QUIT function to allow UNIX-CODE argument
4615 * fixed READ-SEQUENCE bug
4616 * tweaked verbose GC output so that it looks more like the progress
4617 output that ANSI specifies for functions like LOAD
4618 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4619 * added <http://sbcl.sourceforge.com> to the banner information printed by