1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
3 * enhancement: cleaner backtraces for interactive interrupts, as
4 well as other cases where the interesting frames used to be
5 obscured by interrupt handling frames.
6 * enhancement: untracing a whole package using (UNTRACE "FOO") is
7 now supported, and tracing a whole package using (TRACE "FOO") now
8 traces SETF-functions as well.
9 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
10 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
11 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
12 * unreadably printed representation of hash-tables now includes
14 * bug fix: partially fixed #188: type propagation from assignments
15 is now more efficient.
16 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
17 'DOUBLE-FLOAT) are not flushable.
18 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
19 full address of the object, and none of the tag bits.
20 * bug fix: readably printing hash-tables now respects other printer
21 control variables. (reported by Cedric St-Jean)
22 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
23 keyword with MAKE-HASH-TABLE.
24 * bug fix: export SB-POSIX:MKSTEMP.
25 * bug fix: SORT was not interrupt safe.
26 * bug fix: XREF accounts for the last node of each basic-block as
28 * bug fix: MAKE-INSTANCE optimizations interacted badly with
29 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
30 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
31 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
32 create obsolete instances in the case of redefinition or
33 obsoletion of a superclass. (thanks to Andy Hefner)
34 * bug fix: Support for the Alpha architecture has been revived; it had
35 suffered somewhat from lack of maintenance since sbcl-1.0.
36 * improvements to the Windows port:
37 ** The system detects the codepage to use at startup, instead of
38 using the value from the saved core. (thanks to Kei Suzuki)
40 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
41 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
42 (see documentation for details.)
43 * revived support for OpenBSD (contributed by Josh Elsasser)
44 * partially fixed bug #108: ROOM no longer suffers from occasional
45 (AVER (SAP= CURRENT END)) failures.
46 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
47 bodies no longer cause a WARNING to be signalled. (reported by
49 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
50 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
51 the bug in TRUNCATE.) (reported by Michael Weber)
52 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
53 no samples. (reported by Andy Hefner)
54 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
55 no longer appear as (NIL ...) frames in backtraces.
56 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
57 is now actually exported from SB-EXT as documented. (reported by
59 * bug fix: sb-aclrepl now correctly understands how to inspect
60 single-floats on 64-bit platforms where single-floats are not boxed.
61 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
62 yet been finalized. (reported by Levente Meszaros)
63 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
64 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
65 SBCL-specific optimize qualities.
67 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
68 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
69 an executable in the search path, and does so in the child
70 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
71 been removed; it can be found in the CVS history, for somebody who
72 needs that search behavior (see the manual).
73 * minor incompatible change: compiler policy re. weakening type
74 checks has changed: now type checks are weakened only if SAFETY < 2
76 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
77 unparsing of directory pathnames as files. Analogously,
78 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
79 filename to parse into a directory pathname.
80 * enhancement: implicit generic function creation now signals a
81 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
82 users can bind handlers for and muffle around calls to LOAD.
83 * enhancement: RUN-PROGRAM allows unicode arguments and environments
84 to be used (using the default stream external format), and allows
85 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
86 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
87 strings and vectors whose element-type or simplicity is not fully
88 known at compile-time.
89 * optimization: STRING-TRIM and related functions no longer allocate
90 a new string when no trimming needs to be performed. These functions
91 are also faster than before when the input string has been declared
93 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
94 * optimization: READ-LINE is significantly faster for files containing
96 * optimization: non-open coded uses of character comparison operators
97 (e.g. char=) no longer cons when called with more than one parameter
98 on platforms supporting dynamic-extent allocation.
99 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
100 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
101 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
103 * bug fix: some sequence functions elided bounds checking when
105 * bug fix: too liberal weakening of union-type checks when SPEED >
107 * bug fix: more bogus fixnum declarations in ROOM implementation
110 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
111 * new feature: MAKE-HASH-TABLE now experimentally accepts a
112 :SYNCHRONIZED argument, which makes the hash-table safe for
113 concurrent accesses (but not iteration.) See also:
114 SB-EXT:WITH-LOCKED-HASH-TABLE, and
115 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
116 * optimization: CONCATENATE on strings is an order of magnitude faster
117 in code compiled with (> SPEED SPACE).
118 * optimization: SUBSEQ is ~50% faster on lists.
119 * bug fix: bug 417 fixed -- source location reporting is now more robust.
120 * bug fix: SUBSEQ on a list will now correctly signal an error if if
121 END is smaller then START.
122 * bug fix: SB-PROFILE will no longer report extra consing for nested
123 calls to profiled functions.
124 * bug fix: ROOM implementation had bogus fixnum declarations which
125 could cause type-errors when calling ROOM on large images.
126 * bug fix: if file compilation is aborted, the partial fasl is now
127 deleted, and COMPILE-FILE returns NIL as the primary value.
128 * bug fix: number of thread safety issues relating to SBCL's internal
129 hash-table usage have been fixed.
130 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
131 be returned from its body when the values were being returned
132 using unknown-values return convection and the W-P-O was wrapped
133 inside an UNWIND-PROTECT.
134 * bug fix: sb-posix should now compile again under Windows, enabling
137 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
138 * incompatible change: hash-table accessor functions are no longer
139 automatically protected by locks. Concurrent accesses on the same hash-table
140 from multiple threads can give inconsistent results or even corrupt the
141 hash-table completely. Multi-threaded applications should do their own
142 locking at the correct granularity. In the current implementation it is
143 still safe to have multiple readers access the same table, but it's not
144 guaranteed that this property will be maintained in future releases.
145 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
146 and will signal an error at runtime.
147 * enhancement: SB-THREAD package now exports a semaphore interface.
148 * enhancement: CONS can now stack-allocate on x86 and
149 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
151 * enhancement: nested lists can now be stack allocated on
152 platforms providing stack allocation support.
153 * enhancement: dynamic-extent support has been extended to support
154 cases where there are multiple possible sources for the stack
156 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
157 if the mutex is uncontested on Linux.
158 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
159 as specified by the CLHS. (thanks to Tobias Rittweiler)
160 * bug fix: NaN comparison now works on x86-64.
161 * bug fix: CLOSE :ABORT T on a stream with pending output now
163 * bug fix: instances of non-standard metaclasses using standard
164 instance structure protocol sometimes missed the slot type checks
166 * bug fix: known functions can be inlined in high-debug code.
168 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
169 * minor incompatible change: the MSI installer on Windows no longer
170 associates .lisp and .fasl files with the installed SBCL.
171 * minor incompatible change: :UNIX is no longer present in *FEATURES*
172 on Windows. (thanks to Luis Oliviera)
173 * new feature: SB-CLTL2 contrib module now implements
174 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
175 * optimization: scavenging weak pointers is now more efficient,
176 requiring O(1) instead of O(N) per weak pointer to identify
177 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
178 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
179 method bodies using constant slot-names when the first argument is
180 a specializer parameter for the method.
181 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
182 names, when the class of the instance is a direct instance
183 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
184 as slow as the constant slot-name case.)
185 * optimization: member type construction is now O(N) instead
187 * optimization: UNION and NUNION are now O(N+M) for large
188 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
190 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
191 now more readable in environments like Slime which display it.
192 (thanks to Tobias C. Rittweiler)
193 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
194 was able to infer the correct argument type for the object on which
195 the CAS operation was being performed.
196 * bug fix: copy propagation interfered with parallel assignment
197 semantics in local calls. (reported by Paul Khuong)
198 * bug fix: the signed modular fixnum shift compiled to wrong code on
199 x86 and x86-64. (spotted by a slight modification to some of
202 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
203 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
204 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
205 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
207 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
208 and generic function now print the object identity for anonymous
209 instances. (thanks to Attila Lendvai)
210 * enhancement: as an extension to MOP, SBCL now supports SETF of
211 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
212 (thanks to Attila Lendvai)
213 * enhancement: sb-cover has an alternate annotation mode which puts
214 the form annotations at the car of the form, instead of annotating
215 it completely. The benefit of this mode is that it shows better
216 which forms were instrumented by the compiler.
217 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
218 by printer control variables. (reported by Dan Corkill)
219 * bug fix: system leaked memory when delayed output was performed by
220 the OS in smaller chunks then expected. (thanks to David Smith)
221 * bug fix: system leaked memory when file streams were not closed
223 * bug fix: large objects written to slow streams that were modified
224 after the write could end up with the modified state written to
225 the underlying file descriptor.
226 * bug fix: multiple threads operating in parallel on the same stream
227 could cause buffer-overflows.
228 * bug fix: source location information is stored correctly
229 (broken since 1.0.6). This bug would generally show up as the
230 Slime debugger highlighting the wrong form when the sldb-show-source
232 * bug fix: Branch forms are again annotated as branches in the sb-cover
234 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
235 on x86-64. (reported by Gregory Vanuxem)
236 * bug fix: EQUALP could return wrong results for structures with raw
237 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
238 word). (reported by Vjacheslav Fyodorov)
239 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
242 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
243 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
244 atomic compare-and-swap operations on threaded platforms.
245 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
246 allows assigning a global minimum value to optimization qualities
247 (overriding proclamations and declarations).
248 * enhancement: closed over variables can be stack-allocated on x86
250 * performance bug fix: GETHASH and (SETF GETHASH) are once again
252 * optimization: slot definition lookup is now O(1). This speeds up
253 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
254 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
255 * optimization: ASSOC and MEMBER can now be open-coded for all
256 combinations of keyword arguments when second argument is constant
257 and SPEED >= SPACE. In other cases a specialized version is
259 * bug fix: using obsoleted structure instances with TYPEP and
260 generic functions now signals a sensible error.
261 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
262 (reported by Kristoffer Kvello)
263 * bug fix: backtrace construction is now more careful when making
264 lisp-objects from pointers on the stack, to avoid creating bogus
265 objects that can be seen by the GC.
266 * bug fix: defaulting of values in contexts expecting more than 7
267 variables now works on x86-64. (reported by Christopher Laux)
268 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
270 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
271 as the property-list of a symbol.
272 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
273 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
274 situations requested, are once again file-compileable. (reported
277 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
278 * MOP improvement: support for user-defined subclasses of
279 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
280 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
281 the expansion of DEFMETHOD, is responsible for generating a form
282 which creates a list of specializers when evaluated. Additional
283 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
284 debugging and introspective support.
285 * minor incompatible change: the (unsupported) spinlock interface
286 has changed: free spinlock now has the value NIL, and a held spinlock
287 has the owning thread as its value.
288 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
289 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
291 * enhancement: name of a socket-stream is now "a socket" instead of
293 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
294 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
295 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
296 for generic functions. (Thanks to Tobias C. Rittweiler)
297 * optimization: bignum printing speed has been improved by 20-40%
298 (depending on the bignum size.)
299 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
301 * bug fix: the cache used by the CLOS to store precomputed effective
302 methods, slot offsets, and constant return values is now thread and
304 * bug fix: generic function dispatch function updating is now thread
305 and interrupt safe (in the sense that the known issues have been
307 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
308 * bug fix: interning EQL-specializers is now thread and interrupt safe.
309 * bug fix: asdf systems with dependencies to the SB-POSIX or
310 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
311 * bug fix: interrupt safety of applicable method computation has been
314 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
315 * new contrib: sb-cover, an experimental code coverage tool, is included
317 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
318 significantly faster.
319 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
320 produce more efficient code on x86-64 if the compiler can determine
321 that (+ INDEX <constant>) does not require a bounds check and FOO
322 has an element type at least 8 bits wide.
323 * enhancement: a new, experimental synchronous timeout facility is
324 provided. Refer to SB-SYS:WITH-DEADLINE for details.
325 * enhancement: when a symbol name conflict error arises, the
326 conflicting symbols are always printed with a package prefix.
327 (thanks to Kevin Reid)
328 * enhancement: stepping is now once again supported on the SPARC and
329 MIPS platforms. (It is also now more likely to work on CheneyGC
331 * enhancement: sb-sprof can now also track and report accurate call
333 * bug fixes: the treatment of non-standard subclasses of
334 SB-MOP:SPECIALIZER is more correct.
335 * incompatible change: PURIFY no longer copies the data from the
336 dynamic space into the static and read-only spaces on platforms
337 that use the generational garbage collector
338 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
340 * bug fix: GC race condition occasionally resulting in crashes with
341 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
343 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
345 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
346 a STYLE-WARNING to be signalled (regression from 1.0.4.)
347 * bug fix: an asynchronous interrupt could previously leave the
348 system running with GC inhibited.
349 * bug fix: a DECLARE form evaluated at top-level now causes an error
350 rather than silently (or verbosely) returning NIL.
351 * bug fix: trying to dissassemble functions compiled at high DEBUG could
352 result in a error being signalled due to source form lookup errors.
353 (reported by Peter Graves)
355 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
356 * incompatible change: removed writer methods for host-ent-name,
357 host-ent-addresses -- changing the values did not update the DNS
359 * minor incompatible change: changed experimental JOIN-THREAD interface
360 * documentation: the manual now lists reader and writer methods
361 in class slot documentation sections. (thanks to Richard M Kreuter)
362 * documentation: unwinding from asyncronous events has been
363 documented as unsafe.
364 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
365 in multithreaded application code.
366 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
367 platforms. (thanks to James Anderson for the optimization hint)
368 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
370 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
371 variants no longer cons.
372 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
373 their NOT- variants no longer cons.
374 * optimization: EQUAL hash tables no longer use SXHASH for objects
375 of all data types, but instead use an EQL hash for types for which
376 EQUAL is the same as EQL.
377 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
378 are significantly faster.
379 * optimization: new STRING-OUTPUT-STREAM implementation is much
380 faster for large outputs and conses 30% less on average.
381 * enhancement: XREF information is now collected for references made
382 to global variables using SYMBOL-VALUE and a constant argument.
383 * enhancement: SIGINT now causes a specific condition
384 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
385 * bug fix: FILE-POSITION used to signal an error in some cases where
386 ANSI requires it to return NIL.
387 * bug fix: ADJUST-ARRAY is now interrupt-safe.
388 * bug fix: adding and removing fd-handlers is now interrupt-safe.
389 * bug fix: inlined calls to C now ensure 16byte stack alignment on
391 * bug fix: bad type declaration in the CLOS implementation has
392 been fixed. (reported by James Anderson)
393 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
394 (thanks to Lutz Euler)
395 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
396 required by the CLHS. (reported by Samium Gromoff)
397 * bug fix: dead unbound variable references now signal an error.
398 (reported by Marco Monteiro)
399 * bug fix: / with an unused value was being deleted in safe code.
400 (thanks to Marco Monteiro and Kevin Reid)
401 * bug fix: number of characters that can be written onto a single
402 line in a file is unlimited.
403 * bug fix: some GC deadlocks caused by asynchronous interrupts have
404 been fixed by inhibiting interrupts for when GC is disabled.
405 * bug fix: some interrupt safety issues with GETHASH, (SETF
406 GETHASH), CLRHASH and REMHASH have been fixed.
407 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
408 specifier no longer causes infinite recursion.
409 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
410 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
411 * bug fix: an &environment argument with the correct variable information
412 is passed to macros that are expanded during byte compilation.
413 (reported by Samium Gromoff)
414 * bug fix: a logic error was causing the hash function for bignums to
415 have excessive amounts of collisions. (reported by Faré Rideau)
416 * bug fix: modifying the contents of an array could change the return
417 value of SXHASH on that array, which is only allowed for strings
418 and bit vectors. (bug introduced in 0.9.16)
419 * improvement: the x86-64/darwin port now passes all tests (except
420 for the debugger tests) but should still be considered
421 experimental until this is fixed.
422 * improvement: a style-warning is signaled for CASE (etc) clauses with
423 duplicate keys. (thanks to Kevin Reid)
424 * improvement: macos/x86 and macos/x86-64 now use mach exception
425 handlers for dealing with illegal instructions (for trapping and
426 error handling) and memory protection violations (for GC).
428 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
429 * new platform: experimental support for x86-64/darwin (MacOS).
430 * incompatible change: the thread-safe (on most platforms) getaddrinfo
431 and getnameinfo sockets functions are used instead of gethostbyaddr
432 and gethostbyname, on platforms where the newer functions are available.
433 As a result, the ALIASES field of HOST-ENT will always be NIL on these
435 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
436 don't print the startup banner, but behave as if --noinform was passed
437 as a command line argument. (thanks to Kevin Reid)
438 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
439 * optimization: code using alien values with undeclared types is much faster.
440 * optimization: the compiler is now able to open code SEARCH in more cases.
441 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
442 * bug fix: using standardized COMMON-LISP special variables as loop
443 variables no longer signals bogus package lock violations. (reported
445 * bug fix: declaring local loop variables to be of a range-limited type
446 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
447 (reported by Andras Simon)
448 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
449 bugs remain on x86-64.)
450 * bug fix: the #= and ## reader macros now interact reasonably with
451 funcallable instances.
452 * bug fix: type-checks for function arguments were compiled using the
453 compiler policy settings of the wrong lexical environment (in 1.0.2
455 * bug fix: SHADOW accepts characters as string designators, as required
456 by the spec (thanks to Eric Marsden)
457 * bug fix: fixed GC safety issues when foreign functions are called with
458 non-base strings as arguments
459 * bug fix: more consistent error output for fatal-compiler-warnings, like
461 * bug fix: functions with proper names were showing up as NIL in
462 backtraces (reported by Edi Weitz, regression in 1.0.2)
464 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
465 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
466 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
467 produce more efficient code on the x86 if the compiler can determine
468 that (+ INDEX <constant>) does not require a bounds check and FOO
469 has an element type at least 8 bits wide.
470 * bug fix: references to undefined variables are handled the same way
471 in toplevel forms as in the normal compiler
472 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
474 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
475 (thanks to Jon Buller)
476 * improvement: faster compilation times for complex functions
477 * improvement: added readlink support to SB-POSIX (thanks to Richard
480 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
481 * improvement: experimental support for mach exception handling on
482 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
484 * improvement: support for GBK external format.
485 (thanks to Chun Tian (binghe))
486 * improvement: the debugger now displays variables that have been closed
487 over, in code compiled with (DEBUG 2) or higher
488 * improvement: support for executable cores on NetBSD (thanks to
490 * new feature: added a RESTART-FRAME debugger command
491 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
492 be used to provide an implementation for FILE-POSITION on Gray streams
493 (thanks to Eric Marsden)
494 * new feature: add syslog, openlog and closelog support to SB-POSIX
495 (thanks to Richard Kreuter)
496 * optimization: the function call overhead in code compiled with
497 a high DEBUG optimization setting is significantly
498 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
499 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
500 (thanks to Tony Martinez)
501 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
502 evaluated once (reported by Kevin Reid)
503 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
504 works again (thanks to Kilian Sprotte)
505 * bug fix: an error is signaled for tagbodies with duplicate tags
506 (thanks to Stephen Wilson)
507 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
508 * bug fix: Win32 port can now handle foreign code unwinding Lisp
509 stack frames from alien callbacks.
510 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
511 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
512 65k pages (thanks to David Woodhouse)
513 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
515 changes in sbcl-1.0.1 relative to sbcl-1.0:
516 * new platform: FreeBSD/x86-64, including support for threading.
517 * new feature: the compiler stores cross-referencing information
518 about function calls (who-calls), macroexpansion (who-macroexpands)
519 and special variables (who-binds, who-sets, who-references) for code
520 compiled with (< SPACE 3). This information is available through the
521 sb-introspect contrib.
522 * new feature: users may subclass SEQUENCE, and have instances of
523 these classes interoperate with standard Common Lisp functions if
524 a number of methods are defined. (This feature is experimental
525 and the interface subject to change based on feedback from SBCL
526 users and the general community)
527 * improvement: sb-sprof traces call stacks to an arbitrary depth on
528 x86/x86-64, rather than the previous fixed depth of 8
529 * improvement: another pthread back-end of mutex "pthread-futex".
530 * bug fix: non-ascii command-line arguments are processed correctly
531 (thanks to Yaroslav Kavenchuk)
532 * bug fix: non-required arguments were not passed correctly when a method
533 defined using DEFMETHOD was called from a mop-generated method using
534 CALL-NEXT-METHOD (reported by Pascal Costanza)
535 * bug fix: recursion is now permitted in accessors through
536 SLOT-UNBOUND. (reported by Pascal Costanza)
537 * bug fix: an error was signaled at startup if the HOME environment
538 variable was defined, but had an empty value (reported by Peter Van Eynde)
539 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
540 signal a PROGRAM-ERROR, not a BUG.
541 * bug fix: SB-POSIX:READDIR works when built with large file support.
542 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
544 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
545 for better automatic indentation support. (thanks to Matt Pillsbury)
546 * optimization: loading generic functions no longer takes O(n^2) time,
547 proportional to the amount of methods in the generic function
548 (reported by Todd Sabin and Jeremy Brown)
549 * optimization: the FIND and POSITION family of sequence functions
550 are significantly faster on arrays whose element types have been
552 * improvements to the Windows port:
553 ** Intermittent heap corruption problems have been fixed. (thanks
554 to Alastair Bridgewater)
555 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
557 ** Lisp is able to unwind foreign exception frames from alien
558 callbacks. (thanks to Alastair Bridgewater)
560 changes in sbcl-1.0 relative to sbcl-0.9.18:
561 * improvement: experimental support for threading on FreeBSD/x86.
562 (thanks to NIIMI Satoshi)
563 * improvement: runtime option --dynamic-space-size can be used
564 to set the size of the dynamic space reserved on startup.
565 * improvement: floating point modes in effect are now saved in
566 core, and restored on startup.
567 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
568 startup, not time since first call to GET-INTERNAL-REAL-TIME.
569 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
570 threads are not running after *SAVE-HOOKS* have run.
571 * improvement: writes to CLOS instance slots are type-checked in code
572 compiled with (SAFETY 3)
573 * improvement: floating-point exception handling on FreeBSD (thanks to
575 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
576 (thanks to Zach Beane)
577 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
579 * improvement: added support for the Shift-JIS external format.
580 (contributed by NIIMI Satoshi)
581 * improvement: callbacks are supported on Linux/PPC. (thanks to
583 * bug fix: compiler bug triggered by a (non-standard) VALUES
584 declaration in a LET* was fixed. (reported by Kaersten Poeck)
585 * bug fix: file compiler no longer confuses validated and already
586 dumped structurres. (reported by Kaersten Poeck)
587 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
588 fill-pointer signals a type-error as required. (thanks to
590 * bug fix: disassemly of funcallable instances works.
591 * bug fix: single stepping on PPC.
592 * bug fix: fix thread-safety problems in the type system (generally
593 manifesting as nonsensical errors like "STRING is a bad type specifier
594 for sequences" or "The value 1 is not of type FIXNUM")
595 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
596 (reported by Josip Gracin)
597 * bug fix: an error is signaled for attempts to displace arrays with
598 incompatible element types (thanks to Mario Mommer)
599 * bug fix: more correct handling of wide characters in debug info
600 (bug reported by Attila Lendvai and fixed by Juho Snellman)
601 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
602 and don't cause extra consing
603 * optimization: MAP and MAP-INTO are significantly faster on vectors
604 whose elements types have been declared.
605 * Improvements to SB-SPROF:
606 ** Support for allocation profiling
607 ** Reduced profiling overhead, especially for long profiling runs
608 * Improvements to the Windows port:
609 ** floating point exceptions are now reported correctly.
610 ** stack exhaustion detection works partially.
611 ** more accurate GET-INTERNAL-REAL-TIME.
612 ** better RUN-PROGRAM behaviour in case of errors in the child
614 ** PROBE-FILE now simplifies pathnames correctly.
615 ** DIRECTORY now works correctly with :WILD-INFERIORS.
616 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
617 1st 1970 (although time timezone data will be unreliable).
618 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
620 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
621 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
622 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
624 * bug fix: two potential GC deadlocks affecting threaded builds.
625 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
626 returns T (reported by Anton Kazennikov)
627 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
628 with non-variable places
629 * bug fix: remove a race condition in the setting of
630 funcallable-instance functions, this should make threaded CLOS
631 code more stable against memory faults.
632 * bug fix: corruption of specials when unbinding is interrupted by an
633 asynchronous unwind (reported by Hannu Koivisto)
634 * improvement: the debugger will now also display local variables that
635 are only used once, for code compiled with a DEBUG optimization quality
638 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
639 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
640 * incompatible change: External-format support for FFI calls. The
641 SB-ALIEN:C-STRING no longer implies an ASCII
642 external-format. Instead the string is subject to external-format
643 conversion. Additionally return values of type SB-ALIEN:C-STRING
644 are by default of type (SIMPLE-ARRAY CHARACTER), not
645 SIMPLE-BASE-STRING. For an alien type that behaves like the old
646 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
647 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
648 * incompatible change: SB-EXT package no longer contains the
649 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
650 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
652 * incompatible change: the single-stepper is no longer available
653 on Alpha, Mips and Sparc platforms.
654 * minor incompatible change: the direct superclasses of
655 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
656 not (STANDARD-OBJECT FUNCTION). This makes the
657 class-precedence-lists of GENERIC-FUNCTION and
658 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
660 * minor incompatible change: the default stream external format on
661 non-unicode SBCL is no longer detected from the locale, but is always
663 * new feature: Add a version of evaluator that uses an interpreter instead
664 of the compiler. EVAL still uses the compiler by default, to switch it
665 to use the interpreter, set the value of the variable
666 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
667 * minor incompatible change: the single-stepper REPL has been merged
668 with the normal debugger (see the "Stepping" heading of the debugger help
669 for more details). The debugger command STEP will no longer switch
670 to the single-stepper REPL.
671 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
672 for a type now works.
673 * bug fix: loading of default sysinit file works. (thanks to Leonid
675 * bug fix: better detection of circularities in the file-compiler.
676 (reported by Marco Monteiro)
677 * bug fix: the CL pathname functions now work with files that have
678 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
679 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
680 systems with Unix98 pty semantics.
681 * bug fix: ASDF-INSTALL will now work with bsd tar.
682 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
684 * bug fix: timers expiring in dead threads no longer cause a
685 type-error (reported by Paul "Nonny Mouse").
686 * bug fix: thanks to more lightweight single-stepper instrumentation,
687 code compiled with (DEBUG 3) will compile and execute significantly faster,
688 and will have more accurate type-inferencing than before
689 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
690 whose bindings are modified
691 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
692 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
693 ** replace dummy implementations of CL:MACHINE-INSTANCE and
694 CL:SOFTWARE-VERSION with proper version
696 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
697 * feature: implemented the READER-METHOD-CLASS and
698 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
699 as specified by AMOP.
700 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
702 * optimization: faster LOGCOUNT implementation on x86 and x86-64
703 (thanks to Lutz Euler)
704 * optimization: hashing of general arrays and vectors has been
705 improved. (reported by Any Fingerhut)
706 * enhancement: SB-INTROSPECT is now able to find definitions of
707 profiled functions. (thanks to Troels Henriksen)
708 * enhancement: compiler-macro expansion applies now to FUNCALL forms
709 as well, allowing compiler-macros for SETF-functions to expand.
710 * enhancement: step-instrumentation no longer wraps known
711 single-value functions with multiple-value context, allowing
712 better type inference.
713 * fixed bug #337: use of MAKE-METHOD in method combination now works
714 even in the presence of user-defined method classes. (reported by
715 Bruno Haible and Pascal Costanza)
716 * fixed bug #339(c): if there are applicable methods not part of any
717 long-form method-combination group, call INVALID-METHOD-ERROR.
718 (reported by Bruno Haible)
719 * fixed bug #361: the :FUNCTION initarg in the protocol for
720 initialization of methods can now be used to override
721 internally-produced optimized functions. (reported by Bruno
723 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
724 system-provided lambda expression no longer cause warnings about
725 unbound #:|pv-table| symbols.
726 * bug fix: improved the handling of type declarations and the
727 detection of violations for keyword arguments with non-constant
729 * bug fix: potentially erronous calls to PATHNAME and
730 MERGE-PATHNAMES were being flushed in some cases.
731 (reported by Richard Kreuter)
732 * bug fix: compiled calls to TYPEP were mishandling obsolete
733 instances. (reported by James Bielman and Attila Lendvai)
734 * bug fix: format strings with ~> without matching ~< no longer
735 trigger an AVER, but signal an understandable error instead.
736 (reported by Antonio Martinez)
737 * bug fix: specifying an output-file without a pathname-type for
738 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
739 of input-file instead of "fasl". (reported by Robert Dodier)
740 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
741 (reported by James Y Knight).
742 * bug fix: compiler-macros-function did not consider the environment
743 argument for shadowing by local functions.
744 * bug fix: compiler-macros expansion was inhibited by local INLINE
746 * bug fix: inline expansions of known functions were subject to
747 step-instrumentation in high debug policies, leading to problems
749 * bug fix: compiler failed to differentiate between different CONS
751 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
752 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
753 * bug fix: #368: incorrect use of expressed vs. upgraded array
755 * bug fix: #306a: more precise unions of array types.
756 * thread-safety improvements:
757 ** CONDITION-WAIT could return early on Linux, if the thread was
758 interrupted and subsequently continued with SIGCONT.
759 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
761 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
762 * added support for the ucs-2 external format. (contributed by Ivan
764 * minor incompatible change: pretty printing of objects of type
765 (cons symbol) is, in the default pprint-dispatch-table, now
766 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
768 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
769 called later in a class's lifetime, possibly as late as when the
770 first instance of the class is created. Previously,
771 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
772 class became finalizeable.
773 * fixed bug: FILE-POSITION sometimes returned inconsistent results
774 for multibyte external-format streams. (thanks to "vbzoli")
775 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
776 with :ALLOCATION :CLASS inherited from superclasses of the
778 * fixed bug: anonymous classes can now be created using the :NAME
779 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
780 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
781 * fixed bug: core-files saved with :EXECUTABLE T can again be
782 executed when SBCL_HOME isn't set. (reported by James Knight)
783 * fixed bug: toplevel LOCALLY forms with declarations could
784 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
785 * fixed bug: printing from several different threads using different
786 values of *print-case* could cause invalid output, due to
787 some internal special variables of the printer not being bound
788 thread-locally (reported by Max Mikhanosha)
789 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
790 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
791 (reported by Pascal Costanza)
792 * fixed bug: FINALIZE-INHERITANCE is called from
793 REINITIALIZE-INSTANCE on classes when the class has previously
794 been finalized, as required by AMOP.
795 * minor code generation optimizations:
796 ** better register allocation in CLOS dispatching functions
797 ** overflow detection when coercing signed bytes to fixnums on x86-64
798 ** is now implemented with one IMUL instruction instead of three shifts
799 ** more efficient bit-vector access on x86 and x86-64
800 ** more efficient access to raw structure slots on x86 and x86-64
801 * fixed some bugs revealed by Paul Dietz' test suite:
802 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
805 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
806 * feature: thread support on Solaris/x86, and experimental thread support
808 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
810 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
811 default method for SLOT-UNBOUND.
812 * minor incompatible change: prevent the user from specializing the
813 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
814 somewhat counter to the intent of the protocol, I (CSR) think, and
815 additionally it just doesn't work in SBCL as currently
816 implemented, thanks to optimizations (that are always valid for
817 the other three SLOT-VALUEish functions, but not for the setter).
818 * minor incompatible change: the :SB-LDB feature is now enabled by
819 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
820 the low-level debugger.
821 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
822 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
823 down to an SB-SYS:FD-STREAM, allowing direct communication in
824 more cases. Temporary files and pipes are still used as a fallback
826 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
827 on the non-reentrant C functions localtime(3) and gmtime(3).
828 * bug fix: global optimization policy was not visible in LOCALLY and
830 * bug fix: class objects can be used as specializers in methods.
831 (reported by Pascal Costanza)
832 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
833 with a type error. (reported by blitz_ on #lisp)
834 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
835 failed with a type error. (reported by Pascal Bourguignon)
836 * bug fix: merging pathnames against defaults with :DIRECTORY
837 starting with '(:RELATIVE :BACK) should preserve the :BACK.
838 (reported by James Y Knight)
839 * bug fix: saving large (>2GB) cores on x86-64 now works
840 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
841 constant index and a (simple-array (signed-byte 32)) array
842 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
843 error (patch by Robert J. Macomber)
844 * feature: TIME now displays the amount of run-time spent in GC
845 * bug fix: The debugger now does a better job of respecting
846 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
847 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
848 workaround for bug 403.)
849 * bug fix: floating point exception handling now works on FreeBSD
850 (thanks to NIIMI Satoshi)
851 * fixed some bugs revealed by Paul Dietz' test suite:
852 ** MISC.641: LET-conversion were not supposed to work in late
854 * improvements to the Win32/x86 port:
855 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
856 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
858 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
859 * new feature: source path information is generated for macro-expansion
860 errors for use in IDE's like Slime (thanks to Helmut Eller)
861 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
862 * bug fix: compiler does not loop forever on an invalid type in
864 * improvement: compilation of most CLOS applications is significantly
866 * optimization: added a limited bytecode compiler for simple toplevel
867 forms, speeding up compilation and FASL loading
868 * bug fix: the statistical profiler now properly distinguishes anonymous
871 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
872 * minor incompatible change: in sbcl-0.9.11 (but not earlier
873 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
874 into subclasses of the finalized class. Now user calls to
875 FINALIZE-INHERITANCE finalize just the one class, and calls by the
876 system return before any subclasses are finalized.
877 * minor incompatible change: The reader no longer ignores errors
878 regarding non-existent packages in #+ and #- feature tests.
879 * new feature: command line options --no-sysinit, --no-userinit to
880 inhibit loading the corresponding init files
881 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
882 for sending data through UDP sockets (thanks to François-René Rideau)
883 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
884 error is signalled instead (thanks to François-René Rideau)
885 * minor incompatible change: Error signalling behaviour of lexical
886 operations violating package locks has changed slightly. Refer to
887 documentation on package locks for details.
888 * enhancement: EVAL can process IF-expressions without resorting to the
890 * bug fix: Some operations involving SIN, COS, or TAN weren't being
891 constant-folded properly due to missing out-of-line functions.
892 (reported by Mika Pihlajamäki)
893 * bug fix: LISTEN sometimes returned T even in cases where no data was
894 immediately available from the stream
895 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
896 were reversed. (reported by Levente Mészáros)
897 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
898 when given a vector displaced to an adjustable array. (reported
900 * bug fix: derivation of float boundaries from numbers outside the
901 appropriate float range (reported by John Wiseman)
902 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
904 * bug fix: precomputing cache entries for generic functions with
905 some subclasses of specializers as yet invalid does not attempt to
906 fill a cache line with a negative offset. (reported by Levente
908 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
909 allow condition class objects as arguments to DOCUMENTATION and
910 (SETF DOCUMENTATION); only find and set documentation for
911 structure names for the STRUCTURE doc-type. (suggested by Gary
913 * improvements to the Win32/x86 port:
914 ** (user-homedir-pathname) and default initialization file
915 locations now know about the user's "Documents and Settings"
916 directory (thanks to Yaroslav Kavenchuk)
917 ** run-program is implemented (thanks to Mike Thomas)
918 ** sockets support (thanks to Timothy Ritchey)
919 ** better backtrace support (thanks to Alastair Bridgewater)
920 ** sb-grovel supported
921 ** asdf-install and sb-posix work somewhat
922 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
923 * improvements to the Solaris/x86 port:
924 ** works on Solaris 11/Solaris Express
925 ** floating-point exception handling support
926 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
927 * fixed some bugs revealed by Paul Dietz' test suite:
928 ** REMOVE-METHOD returns its generic function argument even when
929 the method is not one of the generic functions' methods.
930 ** objects declared as MEMBER types can be admissible arguments to
932 ** printing characters should simply be printed by the FORMAT ~:C
934 ** compiler failure when compiling functions with hairy constant
935 defaults for optional parameters.
936 ** compiler produces wrong code when MAYBE-INLINE-expanding a
937 function, which is already optimized.
939 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
940 * new platform: experimental support for SBCL x86/Darwin, including
941 MacOS X 10.4.5 on Intel.
942 * new feature: Unicode character names are now known to the system
943 (through CHAR-NAME and NAME-CHAR).
944 * new feature: the filesystem location of SBCL's core file is
945 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
946 * minor incompatible change: the contrib modules SB-POSIX and
947 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
948 this change is to make it easier to distribute
949 location-independent binaries.
950 * bug fix: as implied by AMOP, standardized classes no longer have
951 slots named by external symbols of public packages. (reported by
953 * optimization: calling functions via constant symbols -- as in
954 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
955 via the function object as in (FUNCALL #'FOO).
956 * optimization: on x86 and x86-64, the calling convention for the
957 case of dealing with an unknown number of values has been altered
958 to be friendlier to the prediction heuristics implemented,
959 particularly with reference to CALL and RET pairing. (thanks to
960 Alastair Bridgewater)
961 * optimization: on x86, the code for access to thread-local symbol
962 values and binding/unbinding of thread-local symbols is smaller.
963 (thanks to Alastair Bridgewater)
964 * enhancement: CONSTANTP is now able to determine constantness of
965 more complex forms, including calls to constant-foldable standardized
966 functions and some special forms beyond QUOTE.
967 * fixed bug: occasional GC crashes on Solaris/x86
968 * optimization: x86-64 supports stack allocation of results of simple
969 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
970 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
971 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
972 (thanks to James Knight)
973 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
974 by the standard (thanks to Stephen Compall)
976 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
977 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
978 be used for bundling the runtime and the core file into one
979 executable binary. This feature is not currently supported on all SBCL
980 platforms. (thanks to James Bielman and NIIMI Satoshi)
981 * new feature: a generational or ephemeral garbage collector is now
982 the default on the PowerPC platform (both Linux and Darwin). The
983 old Cheney (stop and copy) collector is a build-time option.
984 (thanks to Cyrus Harmon, after Raymond Toy)
985 * minor incompatible change: the method by which SBCL finds its
986 contributed modules has changed; it no longer relies on symbolic
987 links from an $SBCL_HOME/systems directory, but searches directly
988 in the subdirectories of $SBCL_HOME.
989 * enhancement: the dynamic heap size on the Linux/PPC platform is
990 markedly larger, even using the older Cheney garbage collector.
991 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
992 certain complicated string types. (reported by Gary King)
993 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
994 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
995 * fixed bug: DOCUMENTATION on structures no longer signals an
996 error if no structure type of the right name exists. (reported by
998 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
999 applicable non-standard primary or :AROUND method on
1000 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1001 no longer cause unbound variable errors. (reported by Kevin Reid)
1002 * fixed bug: metacircle resolution in cases where methods have slots
1003 added before the slots from STANDARD-METHOD. (reported by Jean
1005 * fixed bug: the Power PC assembly code for calling into lisp
1006 sometimes computed its return address wrongly (depending on the
1007 15th bit of the address, fixed at link time). This used to
1008 manifest itself in a segmentation violation while building PCL.
1009 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1010 many others over the years)
1011 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1012 the stack pointer (treated as a signed integer) is negative.
1013 (thanks to Peter van Eynde)
1015 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1016 * new platform: experimental support for the Windows operating
1017 system has been added. (thanks to Alastair Bridgewater)
1018 * fixed several bugs in and robustified the PPC FFI (including
1019 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1020 * bug fix: allow non-simple string symbol names (reported by Paul
1022 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1024 * bug fix: saving a core corrupted callbacks on x86/x86-64
1025 * bug fix: closed a loophole in metacircularity detection and
1026 grounding in the PCL implementation of CLOS.
1027 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1028 (reported by tomppa on #lisp)
1029 * bug fix: clear the direction flag on Lisp -> C transitions, as
1030 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1031 (reported by Andrej Grozin and Hendrik Maryns)
1032 * optimization: major improvements to GC efficiency on GENCGC platforms
1033 * optimization: faster implementation of EQUAL
1034 * optimization: emit more efficient opcodes for some common
1035 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1036 * optimization: several other minor code-generation improvements on x86-64
1037 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1039 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1040 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1041 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1042 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1043 by ANSI as a generic function, and for consistency (SETF
1044 GENERIC-FUNCTION-NAME) is treated likewise.
1045 * fixed bug #233.b: make constraint propagation notice when a variable
1046 value is changed after it is referenced but before it is used
1047 * fixed bug #296: no more arbitrary behaviour from filenames with
1048 odd characters as --load arguments. (reported by Adam Warner)
1049 * fixed bug #390: :CHARACTER-SET pathname components now work as
1050 expected. (reported by Tim Daly Jr)
1051 * fixed bug #391: complicated :TYPE intersections in slot
1052 definitions no longer cause an error in PCL internals.
1053 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1054 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1055 returning the number of octets which would be written to the
1056 file-stream. (thanks to Robert J. Macomber)
1057 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1058 arguments to the internal stream functions. (thanks to David
1060 * bug fix: stack exhaustion is now handled on the x86/SunOS
1061 platform. (thanks to Daisuke Homma)
1062 * bug fix: more accurate ROOM results on GENCGC platforms
1063 * optimization: improved type inference for arithmetic-for
1064 index variables in LOOP
1065 * optimization: faster floating-point SQRT on x86-64
1066 * fixed some bugs revealed by Paul Dietz' test suite:
1067 ** DOCUMENTATION returns NIL instead of "" for method combinations
1068 that don't have a docstring
1070 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1071 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1072 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1073 therefore are not customizeable by user code (as seems to be at
1074 least permitted and maybe required by AMOP). As a consolation,
1075 however, the SBCL implementation of these functions now calls
1076 REINITIALIZE-INSTANCE as specified by AMOP.
1077 * bug fix: slot-definition documentation is propagated as per ANSI
1078 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1079 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1080 argument, as do eslotd initialization methods. (from Pascal
1081 Costanza's "Closer" project)
1082 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1083 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1085 * bug fix: it is now possible to have more than one subclass of
1086 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1087 (reported by Bruno Haible, Pascal Costanza and others)
1088 * bug fix: the dependent update protocol now works for generic
1089 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1090 and Pascal Costanza)
1091 * bug fix: condition-class instances corresponding to
1092 DEFINE-CONDITION forms are now created eagerly. (reported by
1093 Kalle Olavi Niemitalo on comp.lang.lisp)
1094 * bug fix: floating point printing is more accurate in some
1095 circumstances. (thanks to Simon Alexander)
1096 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1097 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1098 * bug fix: callbacks on OS X now preserve stack-alignment.
1099 (thanks to Cyrus Harmon)
1100 * enhancement: the x86-64 disassembler is much better at
1101 disassembling SSE instructions. (thanks to Lutz Euler)
1102 * enhancement: CHECK-TYPE on a variable now allows type-inference
1103 (thanks to Vincent Arkesteijn)
1104 * optimization: improved performance of EUC-JP external format.
1105 (thanks to NIIMI Satoshi)
1106 * optimization: performance improvements to IO on file streams of
1107 :ELEMENT-TYPE CHARACTER
1108 * optimization: much faster memory allocation on x86-64
1109 * optimization: faster garbage collections (latency and throughput)
1111 * optimization: faster bignum subtraction and division on x86 and x86-64
1112 (thanks to Lutz Euler)
1113 * optimization: more accurate type-inference for loops that use a
1114 floating point index variable or a negative step.
1116 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1117 * bug fix: add a workaround to SBCL looping infinitely at startup on
1118 Linux kernels with apparently buggy implementations of personality().
1119 (thanks to Svein Ove Aas)
1120 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1121 (thanks to David Lichteblau)
1122 * bug fix: local bindings shadow global symbol macros for
1123 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1124 * optimization: non-open coded uses of numeric comparison operators
1125 (e.g. >) no longer cons when called with more than one parameter
1126 on platforms supporting dynamic-extent allocation.
1127 * enhancement: saving cores with foreign code loaded is now supported
1128 on MIPS/Linux in addition to the previously supported platforms.
1129 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1130 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1131 * bug fix: division by zero in sb-sprof when no samples were collected
1132 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1133 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1134 (thanks to Alexander Kjeldaas)
1135 * new feature: ignoring the return values of destructive functions
1136 like NREVERSE signals a compile-time style-warning.
1137 (thanks to Kevin Reid)
1138 * new feature: the alignment of alien structure fields can be
1139 explicitly specified. (thanks to Cyrus Harmon)
1140 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1141 and dump core on SIGQUIT
1143 ** incompatible change: threads do not inherit values of specials
1144 from their parents (see manual)
1145 ** bug fix: threads stacks belonging to dead threads are freed by the
1146 next exiting thread, no need to gc to collect thread stacks anymore
1147 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1148 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1149 ** bug fix: dynamic variable and thread start related gc lossage
1150 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1152 * fixed some bugs revealed by Paul Dietz' test suite:
1153 ** SUBTYPEP is slightly more accurate on heinously complicated
1154 CONS types where some of the members have uncertain (in the
1155 NIL, NIL sense) type relationships to each other.
1157 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1158 no longer purified unless :PURIFY T is explicitly specified.
1159 ** Non-purified cores are significantly smaller than before
1161 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1162 * new feature: timers based on Zach Beane's excellent timer package
1163 * added support for the following external formats: koi8-u,
1164 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1165 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1166 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1167 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1168 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1169 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1170 * incompatible change: a threaded SBCL will no longer revert to
1171 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1172 * bug fix: interrupts are disabled until startup is complete; no
1173 more sigsegvs when receiving a signal to soon
1174 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1176 * bug fix: add a workaround for the memory randomization features in
1177 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1178 workaround will only be in effect on systems with the proc filesystem
1180 * bug fix: printing objects of type HASH-TABLE signals a
1181 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1183 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1184 have been running for over 50 days. (reported by Gilbert Baumann)
1185 * bug fix: the logic for getting names of functions gets less
1186 confused when confronded with alternate-metaclass
1187 funcallable-instances. (reported by Cyrus Harmon)
1188 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1189 consistent, even on internal alternate-metaclass objects.
1190 * bug fix: flush closure information collected by physical
1191 environment analysis prepass before the main pass. (bug reported
1193 * bug fix: compiler pack phase does not modify a hashtable, which is
1194 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1195 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1196 may be expanded. (reported by Alexander <malishev> on c.l.l)
1197 * bug fix: no more style-warnings when compiling code using LAST
1198 under high-SPEED low-SPACE settings. (reported by David Wallin
1200 * fixed some incompatibilities between SBCL's MOP and the MOP
1202 ** the METAOBJECT class is now implemented;
1203 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1204 STANDARD-OBJECT, as required;
1205 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1206 now compatible (as required by VALIDATE-SUPERCLASS); there
1207 remains a consistency requirement over the presence or absence
1208 of the FUNCTION class in the superclass list of finalized
1209 classes; see the manual for more details;
1210 ** the :AROUND method for COMPUTE-SLOTS on
1211 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1212 requested slot ordering.
1214 ** bug fix: parent thread now can be gc'ed even with a live
1216 ** bug fix: binding a special with PROGV to no value is not
1217 the same as that symbol not having been bound (thanks to
1219 ** bug fix: binding specials is thread safe (thanks to
1221 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1222 and SB-SPROF do not die with 'no handler for signal XX in
1223 interrupt_handle_now(..)' anymore
1224 ** bug fix: WITH-TIMEOUT works with multiple running threads
1225 * fixed some bugs revealed by Paul Dietz' test suite:
1226 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1227 the :method-class keyword argument.
1229 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1230 * new port: the Solaris operating system on x86 processors is now
1231 mostly supported, though some rough edges in the environment
1232 remain. (thanks to Daisuke Homma)
1233 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1234 space for its heap. (thanks to Thiemo Seufer)
1235 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1236 closures. (thanks to Thiemo Seufer)
1237 * minor incompatible change: eof selects abort in the debugger.
1238 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1239 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1241 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1242 deprecated, and will go away in a future revision of SBCL.
1243 * minor incompatible change: GC-ON and GC-OFF are no longer
1244 implemented with a counter, it does not matter how many times gc
1245 is switched on or off
1246 * bug fix: discriminating functions for generic function classes
1247 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1248 make invalid assumptions about method precedence order. (reported
1250 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1251 platforms. (thanks to Thiemo Seufer)
1252 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1253 * bug fix: correct pathname printing with printer escaping is on.
1254 (thanks to Kevin Reid)
1255 * bug fix: complex VOP definitions in "user-space" no longer trigger
1256 package locks. (reported by Zach Beane)
1257 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1258 for funcallable-instance functions, and is no different from
1260 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1261 characters in the current readtable. (reported by Nicholas Neuss)
1262 * bug fix: SUBTYPEP on various CONS types returns more a more
1263 accurate acknowledgment of its certainty.
1264 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1265 lists in some cases. This partially fixes bug 384.
1266 * flush all standard streams before prompting in the REPL and the
1268 * bug fix: signal handling and triggering gc do not conflict
1269 directly anymore, in particular a high frequency sb-sprof does
1270 not prevent gc from running
1271 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1272 approximation for timezone and DST information between the
1273 universal time epoch and the smallest negative 32-bit time_t.
1274 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1275 year 1899 so long as the encoded time is non-negative.
1276 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1277 the correct answer for non-constant rotate amounts when performing
1278 an inline 32-bit rotation.
1280 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1281 there is only one thread in the session
1282 ** bug fix: memory leak for streams created in one thread and
1283 written to in another
1284 ** bug fix: lockup when compiled with gcc4
1285 ** bug fix: race that allows the gc to be triggered when gc is
1287 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1288 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1290 ** bug fix: finished threads are gc'ed properly
1291 * fixed some bugs revealed by Paul Dietz' test suite:
1292 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1293 has arguments with a different set of applicable methods from
1294 the orignal arguments.
1295 ** The type error thrown by MAP now has a correct expected-type
1297 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1298 name a compiled function.
1299 ** Three MISC tests where a large bignum was improperly coerced to
1300 a float that couldn't represent that bignum during type
1301 derivation were fixed.
1302 ** SUBTYPEP can now handle types involving the negation of a
1303 list-form FUNCTION type.
1304 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1305 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1306 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1308 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1309 * New feature: Experimental support for bivalent streams: streams
1310 opened with :element-type :default now allow character and binary
1311 (unsigned-byte 8) I/O
1312 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1313 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1315 * bug fix: no more highly sporadic "couldn't check whether ~S is
1316 readable" when reading a stream and an interrupt hits in the middle
1317 of a select system call
1318 * compiler better recognizes complex arrays (reported by Eduardo
1320 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1321 for variables on SBCL built with linkage-tables. (reported by Luis
1323 * various error reporting improvements.
1324 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1325 (Thanks to Thiemo Seufer)
1326 * bug fix: nested reader invokations work correctly
1327 * bug fix: it is possible to have simultaneous references to foreign
1328 code and foreign data with the same name.
1330 ** added x86-64 support
1331 ** incompatible change: the threading api now works with thread
1332 objects instead of thread ids
1333 ** bug fix: threads are protected from signals and interruption when
1334 starting up or going down
1335 ** bug fix: a race where an exiting thread could lose its stack to gc
1336 ** bug fix: don't halt on infinite error in threads if possible
1337 ** fixed numerous gc deadlocks introduced in the pthread merge
1338 ** bug fix: fixed thread safety issues in read and print
1339 ** bug fix: debugger doesn't hang on session lock if interrupted at
1340 an inappropriate moment
1341 ** bug fix: run-program is now thread safe(r)
1342 ** bug fix: inner with-recursive-lock no longer releases the mutex
1343 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1344 * fixed some bugs revealed by Paul Dietz' test suite:
1345 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1346 EXPECTED-TYPE slots filled.
1347 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1349 ** MAKE-CONDITION accepts classes as type-designators.
1350 ** COMPILE may never return NIL.
1351 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1352 range before calling Unix time functions
1354 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1355 * numerous signal handling fixes to increase stability
1356 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1357 * minor incompatible change: we now correctly canonize default
1358 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1359 as per the MOP, rather than the historical (INITARG INITFUNCTION
1360 INITFORM). (reported by Bruno Haible)
1361 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1362 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1363 for more information.
1364 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1365 pathname is a directory pathname.
1366 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1367 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1369 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1370 :METACLASS argument in addition to a class name. (reported by
1371 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1372 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1373 (reported by Bruno Haible; patch from Gerd Moellmann)
1374 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1376 * bug fix: more cleanups to the floating point exception handling on
1377 x86-64 (thanks to James Knight)
1378 * bug fix: the compiler does not try to constant fold calls of
1379 COERCE and BIT-* functions when they return freshly constructed
1380 objects. (reported by Nikodemus Siivola and Paul Dietz)
1381 * optimization: improved the MIPS versions of generic (in the
1382 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1383 * optimization: direct conversion of (unsigned-byte 32) to floats on
1384 the PowerPC platform.
1385 * optimization: structure instances with raw slots now use less
1386 memory, and probably show better memory locality. (thanks to
1388 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1389 are treated as requests for stack allocation on the x86-64,
1390 Alpha-32, PPC and SPARC platforms.
1391 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1392 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1393 (thanks to Tony Martinez)
1395 ** gcing a dead thread can no longer lead to lockups
1396 ** threads block signals until they are set up properly
1397 ** errno is no longer shared by threads
1398 ** interrupt-thread restores the eflags register on x86
1399 ** fixed some lockups due to gc/thread interaction
1400 * fixed some bugs revealed by Paul Dietz' test suite:
1401 ** invalid dotted lists no longer raise a read error when
1402 *READ-SUPPRESS* is T
1403 ** signal an error if a symbol that names a declaration is used
1404 as the name of a type, or vice versa
1405 ** allow using the (declare (typespec var*)) abbreviation for
1406 (declare (type typespec var*)) with all type specifiers
1407 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1408 ** always return NIL from PPRINT-POP when OBJECT is NIL
1409 ** don't signal errors when pretty-printing malformed LABELS,
1410 FLET or MACROLET forms
1411 ** declarations in a DOLIST body are also in scope for the
1413 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1415 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1418 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1419 * fixed cross-compiler leakages that prevented building a 32-bit
1420 target with a 64-bit host compiler.
1421 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1422 opened with :IF-EXISTS :OVERWRITE.
1423 * fixed bug 281: error for an invalid qualifier in a short-form method
1424 combination method is not signalled until the faulty method is called.
1425 * bug fix: iteration variable type inferrer failed to deal with open
1426 intervals. (reported by Alan Shields)
1427 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1428 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1429 or more runtime options were provided to the sbcl binary.
1430 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1431 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1432 * optimization: there's now a fast-path for fixnum arguments in the
1433 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1434 * optimization: the code generated on x86-64 is more compact thanks
1435 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1436 * bug fix: floating-point exception handling now partly works on
1437 x86-64. (Thanks to James Knight)
1438 * improvement to the MIPS backend from Thiemo Seufer: C-style
1439 64-bit long long arguments and return values to alien functions
1441 * fixed some bugs revealed by Paul Dietz' test suite:
1442 ** the type-error signalled from WARN has a filled-in DATUM slot.
1443 ** the type-error required when a stream is not associated with
1444 a file has the stream as its datum.
1445 ** type-errors on single-floats on x86-64 no longer have
1446 :INVALID-OBJECT as the datum
1447 ** the type-errors signalled for invalid function names now have
1448 a correct expected type
1449 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1450 for structures defined with a :TYPE. Documentation strings for
1451 typed structures are no longer immediately discarded
1452 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1453 broadcast streams and synonym streams. FILE-LENGTH now also works
1454 on broadcast streams.
1456 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1457 * incompatible change: the --noprogrammer option, deprecated since
1458 version 0.7.5, has been removed. Please use the equivalent
1459 --disable-debugger option instead.
1460 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1462 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1463 inoperational for a while now) has been completely removed.
1464 * null lexical environments are now printed as #<NULL-LEXENV>,
1465 significantly reducing the amount of clutter in typical backtraces.
1466 * documentation on weak pointers, finalization, and after GC hooks
1467 has been added to the manual.
1468 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1469 as other specialized array types, is much faster. SUBSEQ and
1470 COPY-SEQ on such arrays have also been sped up.
1471 * optimization: EQL is now more efficient when at least other argument
1472 is known to be of type (OR FIXNUM (NOT NUMBER)).
1473 * fixed bug: the runtime is now less vulnerable to changes in the
1474 size of the SBCL object on OS X, and virtual memory is reserved for
1475 all spaces that need to be at a fixed address.
1476 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1477 * fixed bug: finalizers and after GC hooks that cause consing are now
1479 * fixed bug: compiler error messages and summaries are now printed to
1480 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1481 * fixed inference of the upper bound of an iteration variable.
1482 (reported by Rajat Datta).
1483 * fixed bug 211e: calling local functions with duplicated constant
1484 keyword argument no longer causes a bogus style warning about an
1486 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1487 ftype declaration to be disregarded. (reported by Dave Roberts)
1488 * fixed bug 373: caused by erronous compilation of references to alien
1489 variables in the runtime on ppc/darwin.
1490 * fixed bug 376: CONJUGATE type deriver.
1491 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1492 (reported by Baughn on #lisp)
1493 * a cross-compiler bug on non-x86 platforms has been identified and
1494 fixed. (thanks to Bruno Haible)
1495 * improvements to the MIPS runtime code for increased stability.
1496 (thanks to Thiemo Seufer)
1497 * increased the maximimum compact environment size to allow
1498 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1499 * partial workaround for bug 135: don't name the function we're
1500 calling for hairy cases of EVAL, so as not to accumulate one environment
1501 entry per call to EVAL. (reported by Kevin Reid)
1502 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1503 * optimization: type testing for non-vector arrays should be faster.
1504 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1506 * bug fix: setting 31st element of a bit vector to zero did not work
1508 * bug fix: redefining a class definition which failed due to a
1509 previous accessor / function clash now works (but see BUGS entry
1510 #380 for more problems in this area). (thanks to Zach Beane)
1511 * the long form of DEFINE-METHOD-COMBINATION disables method group
1512 checking when given a single method group with pattern *. (thanks
1514 * on x86 compiler supports stack allocation of results of simple
1515 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1516 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1517 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1518 * fixed some bugs related to Unicode integration:
1519 ** the restarts for recovering from input and output encoding
1520 errors only appear when there is in fact such an error to
1522 * fixed some bugs revealed by Paul Dietz' test suite:
1523 ** MISC.549 and similar: late transformation of unsafe type
1524 assertions into derived types caused unexpected code
1526 ** SCALE-FLOAT type deriver is less wrong.
1527 ** type derivers for EXP, LOG and similar functions compute result
1528 types for complex arguments better.
1529 ** (MISC.563) CONJUGATE type deriver works for very restricted
1531 ** out-of-line type testers for character strings are available.
1532 ** EQUAL compiler transform understands specialness of objects
1534 ** accessing double-floats stored on the stack now works on x86-64.
1535 ** debugger internals could sometimes create invalid lispobjs,
1536 resulting in GC crashes.
1537 ** MISC.548: type check weakening can convert required type into
1539 ** initialization forms for bindings are not in scope of free special
1542 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1543 * incompatible change: thread support for non-NPTL systems has
1544 been removed - locking is buggy and unreliable. A threaded
1545 SBCL build will now warn at startup and refuse to create
1546 new threads, unless futex support is detected in the kernel
1547 * incompatible change: the top level REPL now has only an ABORT
1548 restart associated with it, not TOPLEVEL and ABORT as it used to.
1549 TOP and TOPLEVEL are now available as debugger commands for
1550 returning to the top level.
1551 * incompatible change: forms evaluated in the REPL now use the
1552 global optimization policy.
1553 * incompatible change: user- and system-initialization files are
1554 no longer processed with LOAD, but by READ and EVAL; hence the
1555 global optimization policy, startup package, readtable, etc,
1557 * The .fasl file format number has been incremented because of
1558 various incompatible changes.
1559 * internal entry point details and argument counts no longer appear
1560 in backtraces unless explicitly requested by setting
1561 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1562 * built-in and standard functions no longer have names like "top
1563 level local call to FOO".
1564 * fixed bug 32: functions defined in non-null lexical environments
1565 now have more legible printed representation
1566 * fixed bug 33: functions defined in non-null lexical environemnts
1567 are now more amenable to inspection by INSPECT.
1568 * workaround for bug 354: XEPs no longer appear in backtraces unless
1569 explicitly requested.
1570 * fixed bug: receiving the signal which results from attempting to
1571 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1572 SIGBUS on other *BSDs) on architectures where the C stack is also
1573 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1574 to debugger confusion or wild instability. (reported by Cheuksan
1576 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1577 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1578 (reported by Lutz Euler)
1579 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1580 compiled code. (reported by Paul Dietz)
1581 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1582 specialized parameter to a method to something that is not TYPEP
1583 the specializer is now possible.
1584 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1585 face of package deletion.
1586 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1587 pathname. (reported by David Tolpin on c.l.l)
1588 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1589 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1590 STRING-STREAM and FILE-STREAM, respectively.
1591 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1592 than previously; in addition, its test suite is now run on build.
1593 (thanks to Luke Gorrie)
1594 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1595 * optimization: added a immediate representation for single-floats
1597 * fixed some bugs revealed by Paul Dietz' test suite:
1598 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1599 ** attempting to create a package with a colliding nickname causes
1600 correctable errors to be signalled.
1601 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1602 ** several x86-64 backend bugs related to sign-extension of immediate
1605 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1606 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1607 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1609 * fixed loading of multiply forward-referenced layouts.
1610 (thanks to Cheuksan Wang)
1611 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1612 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1613 (see COMPILE-FILE documentation for details of :PRINT option)
1614 * fixed bugs 19 and 317: fixed-format floating point printing is
1615 more accurate. This also fixes a bug reported by Adam Warner
1616 related to the ~@F format directive.
1617 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1619 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1620 dispatch table if the from-char is a dispatch macro character.
1621 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1622 by the wordsize no longer ignore the last word. (reported by Lutz
1624 * fixed bug in type checking of dynamic-extent variables. (reported
1626 * optimization: sequence traversal functions use their freedom to
1627 coerce function designators to functions.
1628 * optimization: code with many calls to CLOS methods specialized on
1629 CLOS classes has had redundant type checks removed.
1630 * optimization: added declarations to speed up operations that access
1631 the internal character database (for example STRING-UPCASE)
1632 * optimization: comparison operations between floats and sufficiently small
1633 fixnums no longer create extra rationals
1634 * fixed some bugs related to Unicode integration:
1635 ** portions of multibyte characters at the end of buffers for
1636 character-based file input are correctly transferred to the
1637 start of the buffer at the next read.
1638 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1639 passing it through to OPEN.
1640 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1641 argument given to internal calls to OPEN.
1642 ** the built-in comment readers (introduced by character sequences
1643 ";" and "#|") are more forgiving to encoding errors; they will
1644 STYLE-WARN and then attempt to resync the stream at a character
1645 boundary. (thanks to Teemu Kalvas)
1646 * fixed some bugs in the x86-64 port:
1647 ** Negative short int return values from c-calls are sign-extended
1649 ** The stack is aligned to 16-bytes for c-calls, as required by
1650 the ABI. (thanks to Cheuksan Wang)
1651 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1652 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1653 for enums. (thanks to Vincent Arkesteijn)
1654 ** Multiple small optimizations and bugfixes for floating point
1656 * fixed some bugs revealed by Paul Dietz' test suite:
1657 ** Space, Tab, Linefeed, Return and Page have the invalid
1658 secondary constituent character trait.
1659 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1661 ** WITH-INPUT-FROM-STRING should only update the index place on
1663 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1664 works more reliably.
1665 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1666 with one-dimensional arrays with fill pointers.
1667 ** TYPE-OF failed on a complex with an integer realpart and a
1669 ** compiler failure during type inference for the code of form
1670 (IF (EQL X (THE ...)) ...) (MISC.535).
1672 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1673 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1674 platform. The port supports 61-bit fixnums, large memory spaces
1675 and reloading shared object files.
1676 * enhancement: saving cores with foreign code loaded is now
1677 supported on ppc/Darwin in addition to the previously supported
1679 * enhancement: the statistical profiler now walks deeper into the
1680 call stack for more meaningful call-graphs and accrued time
1681 reports (x86/x86-64 only). It also now reports time spent in
1683 * enhancement: it is now possible to trace most individual methods
1684 of a generic function in addition to tracing the generic function
1686 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1688 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1689 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1690 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1691 produces an error. (thanks to Vincent Arkesteijn)
1692 * bug fix: NAMESTRING on pathnames with :WILD components in their
1693 directories works correctly. (thanks to Artem V. Andreev)
1694 * fixed bug 125: compiler preserves identity of closures. (reported
1696 * bug fixed: functions with &REST arguments sometimes failed with
1697 "Undefined function" when compiled with (DEBUG 3). (reported by
1699 * bug fix: overflow during compiling of setting element of a bit
1700 vector with constant index and value. (reported by Timmy Douglas)
1701 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1702 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1703 * on x86 compiler partially supports stack allocation of dynamic-extent
1705 * GO and RETURN-FROM do not check the extent of their exit points
1706 when compiled with SAFETY 0.
1707 * fixed some bugs related to Unicode integration:
1708 ** encoding and decoding errors are now much more robustly
1709 handled; it should now be possible to recover even from invalid
1710 input or output to the terminal. (thanks to Teemu Kalvas)
1711 ** provided a first cut at implementing STRING-TO-OCTETS and
1712 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1713 ** altered the SB-MD5 contributed module slightly, changing the
1714 interface just enough for it to be supportable for builds where
1715 lisp characters are not eight bits.
1716 * fixed some bugs revealed by Paul Dietz' test suite:
1717 ** the FORMATTER-generated functions for ~V[ conditionals require
1718 the correct number of arguments.
1719 ** READ-FROM-STRING returns the mandated second value when applied
1720 to displaced strings.
1721 ** the #\Rubout and #\Backspace characters are treated as invalid
1722 constituent characters by the tokenizer.
1724 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1725 * new feature: reloading changed shared object files with
1726 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1727 * new feature: references to foreign variables and functions
1728 can now be compiled and loaded before the corresponding shared
1729 object file is loaded, as long as the foreign definitions are
1730 available at runtime.
1731 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1732 supported. (thanks to Dan Debertin)
1733 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1734 just keywords. (thanks to Vincent Arkesteijn)
1735 * fixed bug #331: structure-class instances corresponding to
1736 DEFSTRUCT forms are now created eagerly.
1737 * fixed bug #345: backtraces from calls to undefined functions work
1738 on x86 as well. Related bug #61 is now also partially fixed on x86
1739 (backtraces from throws to unknown catch tags.)
1740 * bug fix: lambda-list parsing is now stricter vrt. order and number
1741 of lambda-list keywords.
1742 * bug fix: as specified by AMOP, an error is signalled if a
1743 class-option appears multiple times in a DEFCLASS form. (reported
1745 * bug fix: RANDOM can be compiled when the compiler derives the type
1746 of its numeric argument as a disjoint set of small integers.
1747 (reported by Paul Dietz)
1748 * bug fix: starting a core saved with shared objects loaded when
1749 those objects are not available no longer causes threaded SBCL to
1750 hang. (reported by Sean Ross)
1751 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1753 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1754 argument, which must always be NIL. (reported by Kalle Niemitalo)
1755 * bug fix: printing 1.0d+23 no longer results in an error.
1756 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1757 * bug fix: structure slot setters preserve evaluation order. (thanks
1759 * fixed some bugs related to Unicode integration:
1760 ** RUN-PROGRAM can allow its child to take input from a Lisp
1761 stream. (reported by Stefan Scholl)
1762 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1763 locales. (reported by Ken Causey)
1764 * fixed some bugs revealed by Paul Dietz' test suite:
1765 ** INCF, DECF and REMF evaluate their place form as specified in
1767 ** FORMATTER expands ~{ ~} iteration directives with V or #
1768 parameters correctly.
1769 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1770 consequent uses no arguments correctly.
1771 ** the system has a partial understanding of the (COMPLEX RATIO)
1773 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1776 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1777 * new feature: a build-time option (controlled by the :SB-UNICODE
1778 keyword feature, enabled by default) for building the system with
1779 support for the entire 21-bit character space defined by the
1781 * new feature: the system now has rudimentary external-format
1782 support; the primary user-visible change at this time is that
1783 characters with the high bit set (such as lower-case-e-acute) will
1784 print correctly to a terminal in a UTF-8 environment.
1785 * minor incompatible change: BASE-CHAR no longer names a class;
1786 however, CHARACTER continues to do so, as required by ANSI.
1787 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1788 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1789 has been moved to the SB-EXT package (temporarily re-exported from
1791 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1792 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1794 * fixed bug #351: better error-handlind and reporting for malformed
1796 * fixed bug #350: bignum-printing is now more memory-efficient,
1797 allowing printing of very large bignums, eg. (expt 2 10000000).
1798 (reported by Bruno Haible)
1799 * fixed bug #302: better primitive-type selection for intersection
1801 * fixed bug #308: non-graphic characters now all have names, as
1802 required. (reported by Bruno Haible)
1803 * bug fix: redefining a class with different superclasses now correctly
1804 removes it from the direct-subclasses of its previous superclasses.
1805 (reported by David Morse)
1806 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1807 new value now works. (reported by Bruno Haible)
1808 * bug fix: correct canonicalization of multiple non-standard slot
1809 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1810 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1811 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1812 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1813 now exists, an signals an error.
1814 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1815 during macroexpansion for non-iteration variables as well. (reported
1816 by Bruno Haible for CMUCL)
1817 * bug fix: Cyclic structures and unprintable objects in compiler
1818 messages no longer cause errors. (reported by Bruno Haible)
1819 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1820 and READ-FROM-STRING all now return a primary value of NIL if
1821 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1822 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1823 (reported by Bruno Haible for CMUCL)
1824 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1825 do not designate a condition. (reported by Bruno Haible for
1827 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1828 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1829 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1830 permitting resolution in favour of any of the conflicting symbols.
1831 (reported by Bruno Haible for CMUCL)
1832 * bug fix: EQUAL compiler optimizations is less aggressive on
1833 strings which can potentially compare true despite having distinct
1834 specialized array element types.
1835 * bug fix: unit enumerations can be defined without dividing by
1836 zero. (thanks to Vincent Arkesteijn)
1837 * FORMAT compile-time argument count checking has been enhanced.
1838 (report from Bruno Haible for CMUCL)
1839 * a partial workaround for the bug 262: the compiler does not try to
1840 inline-expand a local function doing RETURN-FROM from a deleted
1841 BLOCK. (thanks to Peter Denno for the bug report and to David
1842 Wragg for the simple test case)
1843 * fixed some bugs revealed by Paul Dietz' test suite:
1844 ** RENAME-PACKAGE allows all package designators as new package
1846 ** constraint propagation and lambda variable substitution are
1847 more cautious in dealing with partially deleted code.
1848 ** compiler sometimes lost reoptimization passes.
1849 ** CERROR, given a condition as condition designator, passes its
1850 remaining arguments to the continue format control without
1852 ** Case-altering FORMAT directives work correctly on non-ASCII
1854 ** The REAL type specifier handles bounds outside the floating
1855 point ranges without signalling FLOATING-POINT-OVERFLOW.
1856 ** Functions with IR1-transformations can create intercomponent
1857 references to global functions.
1858 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1860 ** FORMAT ~R treats a nil value for its first parameter correctly.
1862 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1863 * enhancement: saving cores with foreign code loaded is now
1864 supported on x86/NetBSD and sparc/Linux in addition to the previously
1865 supported platforms.
1866 * bug fix: on some platforms repeated installations caused multiple
1867 copies of HTML documentation to be installed -- should not happen
1868 any more. (reported by Stefan Scholl)
1869 * bug fix: parsing self-recursive alien record types multiple times
1870 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1871 original patch by Helmut Eller for CMUCL)
1872 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1873 (thanks to Richard Kreuter)
1874 * bug fix: defining classes whose accessors are methods on existing
1875 generic functions in other (locked) packages no longer signals
1876 bogus package lock violations. (reported by François-René Rideau)
1877 * bug fix: special variables as DEFMETHOD parameters no longer have
1878 associated bogus type declarations. (reported by David Wragg and
1880 * bug fix: read-write consistency on streams of element-type
1881 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1882 * bug fix: redefiniton of the only method of a generic function with
1883 no DEFGENERIC no longer emits a full WARNING. In addition,
1884 redefinition of generic functions with no DEFGENERIC to an
1885 incompatible lambda list now signals an error. (thanks to Zach
1887 * bug fix: DEFGENERIC now works even when there's a function of the
1888 same name in an enclosing lexical environment. (thanks to Zach
1890 * fixed compiler failure, caused by instrumenting code during
1891 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1892 * optimization: added loop analysis and improved register allocation
1893 * fixed some bugs revealed by Paul Dietz' test suite:
1894 ** POSITION on displaced vectors with non-zero displacement
1895 returns the right answer.
1896 ** (SIMPLE-STRING) is a valid type specifier for sequence
1898 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1900 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1901 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1903 ** The justification version of the FORMAT ~< directive treats
1904 non-zero minpad parameter correctly.
1906 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1907 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1908 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1909 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1910 the supported interface.
1911 * new feature: Single-stepping of code compiled with DEBUG 2 or
1912 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1913 * new feature: saving cores with foreign code loaded is now
1914 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1915 Timothy Moore's work for CMUCL)
1916 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1917 parameters to * instead of NIL if no initform is supplied.
1918 (reported by Johan Bockgaard)
1919 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1920 correctly when FUNCALL appears as the car of the form. Note:
1921 despite this FUNCALL forms are not currently subject to
1922 compiler-macro expansion. (port of Raymond Toy's fix for the
1923 same from CMUCL, reported by Johan Bockgaard)
1924 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1925 (thanks for Teemu Kalvas)
1926 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1927 case instead of terminating the iteration (thanks for Julian
1928 Squires, Sean Champ and Raymond Toy)
1929 * bug fix: incorrect expansion of defgeneric that caused a style
1930 warning. (thanks for Zach Beane)
1931 * bug fix: slot accessor effective method computation works properly
1932 for classes with multiple non-standard applicable methods on
1933 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1934 * on x86 compiler supports stack allocation of results of LIST and
1935 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1936 CMUCL implementation by Gerd Moellmann)
1937 * fixed some bugs revealed by Paul Dietz' test suite:
1938 ** FORMAT strings with both the ~<~:;~> form of the justification
1939 directive and pretty-printing directives cause an error.
1941 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1942 * incompatible change: the internal functions
1943 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1944 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1945 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1946 instead of the old functions.
1947 * new feature: on platforms where "dladdr" is available foreign
1948 function names now appear in backtraces. (based on Helmut Eller's
1950 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1951 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1953 * documentation: ASDF manual is now distributed with SBCL.
1954 * bug fix: SBCL can now load its contributed modules with REQUIRE
1955 even if the system-provided entries have been removed from
1956 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1958 * bug fix: backtraces involving undefined functions or assembly
1959 routines are more informative. (thanks to Brian Downing)
1960 * bug fix: mutually referent alien structures now work correctly.
1961 (reported by Rick Taube)
1962 * bug fix: structures defined by WITH-ALIEN can be referred to
1963 within other definitions in the same WITH-ALIEN.
1964 * bug fix: division operators (MOD, TRUNCATE and the like) with
1965 constant zero divisors and integer dividends no longer generate
1967 * bug fix: provide default methods for INPUT-STREAM-P and
1968 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1969 * bug fix: improve the detection and resolution of MOP metacycles
1970 (where effective method computation depends on the generic function
1971 having its effective method computed). (reported by Bruno Haible)
1972 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1973 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1974 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1975 * bug fix: pathnames with relative directory components can be
1976 represented relative to default pathnames.
1977 * optimization: in taking the GCD of bignums, reduce the two bignums
1978 to approximately the same size (using Euclid's algorithm) before
1979 applying the more sophisticated binary GCD. (thanks to Juho
1981 * optimization: COUNT on bitvectors now operates word-at-a-time.
1982 * optimization: ASH with a positive, but not necessarily constant,
1983 (leftwards) shift, when in a modular context, is compiled to a
1985 * fixed some bugs revealed by Paul Dietz' test suite:
1986 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1987 if the corresponding argument is NIL.
1988 ** FORMAT directives accepting a minpad parameter treat negative
1990 ** literal commas as character parameters to FORMAT directives are
1992 ** literal spaces directly after ~<Newline> directives within a
1993 format-logical-block (~:< ~@:>) do not induce :FILL-style
1994 conditional newlines.
1995 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1996 as specified: it no longer includes conditional newlines).
1997 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1999 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2000 :PER-LINE-PREFIX argument does not evaluate to a string.
2002 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2003 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2004 the manual for details; for now, package locks can be disabled by
2005 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2006 if no major problems are found then it is likely that they will be
2007 compiled in unconditionally.
2008 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2009 unsupported operators on all platforms. To load a shared library
2010 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2011 object file, link it into a shared library outside of SBCL and
2012 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2014 * bug workaround: ROOM T and the GC were not working together
2015 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2016 reporting facility aren't preserved. That reporting has been
2017 disabled, so now until and unless someone figures out how to make it
2018 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2019 * minor incompatible change: as threatened around sbcl-0.8.0, the
2020 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2021 The symbols are also exported from SB-PCL for backwards
2022 compatibility, but more so than before SB-PCL should be treated as
2023 an implementation-internal package.
2024 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2026 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2027 It is known as of this release not to work on the Alpha, however.
2028 * fixed bug #167: errors signalled due to illegal syntax in method
2029 bodies are now more legible.
2030 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2031 designators and can hence be used with TYPEP.
2032 * fixed bug #333: CHECK-TYPE now ensures that the type error
2033 signalled, if any, has the right object to be accessed by
2034 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2035 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2036 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2038 * fixed bug #334: programmatic addition of slots using specialized
2039 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2040 and :ALLOCATION :CLASS slots.
2041 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2042 float exponents. (rereported by Peter Seibel)
2043 * fixed a bug: #\Space (and other whitespace characters) are no
2044 longer considered to be macro characters in standard syntax by
2045 GET-MACRO-CHARACTER.
2046 * fixed bug: initialization of condition class metaobjects no longer
2047 causes an instance of the condition to be created. (reported by
2049 * fixed bug: it is now possible to have slots such that
2050 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2051 description is neither :INSTANCE nor :CLASS.
2052 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2053 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2054 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2055 non-standard SLOT-DEFINITION-ALLOCATION values.
2056 * fixed bugs: various race conditions handling exiting threads.
2057 CL-PPCRE's thread torture test now passes.
2058 * fixed arguably-a-bug: GC time proportional to number of threads in
2059 system even when most of them are idle
2060 * optimization: improved performance of BIT and SBIT on bit-vectors.
2061 * .fasl file incompatibility: The fasl file version number has
2062 been incremented because of changes associated with package locks.
2064 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2065 * minor incompatible change: the system no longer provides
2066 optimization hints (conditions which are TYPEP
2067 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2068 mode; these hints are still emitted when higher SPEED optimization
2070 * new contrib module: a sampling profiler (profiling by statistical
2071 sampling, rather than by instrumenting functions) is available as
2072 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2073 * the behaviour of the standard function ED is now customizeable by
2074 third parties through a hook variable: see ED's documentation
2075 string for information on the protocol.
2076 * the compiler no longer emits efficiency notes for (FUNCALL X)
2077 when the type of X is uncertain under default optimization
2079 * fixed bug 276: mutating a binding of a specialized parameter to a
2080 method to something that is not TYPEP the specializer is now
2082 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2083 exhibit the required behaviour on the x86 platform. (thanks to
2084 Peter van Eynde, Eric Marsden and Bruno Haible)
2085 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2086 even for difficult arguments. (reported by Peter Graves)
2087 * fixed bug 141a: the backquote printer now descends quoted
2089 * fixed another bug in backquote printing: no more destructive
2090 modification of the form's list structure. (reported by Brian
2092 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2093 move between its address being taken and the call to
2094 interrupt_thread, fixing a crashing race condition.
2095 * the SB-POSIX contrib implementation has been adjusted so that it
2096 no longer exhibits ridiculously poor performance when constructing
2097 instances corresponding to C structs.
2099 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2100 * minor incompatible change: the sb-grovel contrib now treats C
2101 structures as alien (in the sense of SB-ALIEN) objects rather than
2102 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2103 has implications for memory management of client code
2104 (sb-grovel-returned objects must now be manually managed) and for
2105 type safety (alien objects now have full types).
2106 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2107 used to control emission of compiler diagnostics, rather than the
2108 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2109 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2110 quality should be considered deprecated.
2111 * install.sh now installs the user manual as well
2112 * (not quite a new documentable feature, but worth considering in
2113 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2114 beginnings of a semantically meaningful condition hierarchy is
2115 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2117 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2118 designator as the defaults argument.
2119 * fixed bug: Displaced arrays whose displaced-to array has become
2120 too small now cause ARRAY-DIMENSION to signal an error, providing
2121 for safer bounds-checking. (reported by Bruno Haible)
2122 * fixed bug: DEFCLASS slot definitions with identical :READER and
2123 :WRITER names now signal a reasonable error. (reported by Thomas
2125 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2127 * fixed bug: Invalid slot specification errors now print correctly.
2128 (thanks to Zach Beane)
2129 * fixed bug 320: Shared to local slot value transfers in class
2130 redefinitions now happen corrently from superclasses as
2131 well. (reported by Bruno Haible)
2132 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2134 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2135 improper lists correctly. (reported by Bruno Haible)
2136 * fixed bug 313: source-transform for <fun-name> was erroneously
2137 applied to a call of a value of a variable with name <fun-name>.
2138 (reported by Antonio Menezes Leitao)
2139 * fixed bug 307: The obsolete instance protocol ensures that
2140 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2141 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2142 remove dead unknown-values globs from the middle of the stack.
2143 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2144 easier to resume long computations after using *BREAK-ON-SIGNALS*
2145 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2146 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2147 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2148 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2149 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2150 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2152 * fixed a (fairly theoretical) bug in string printing: if
2153 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2154 does not have array-element-type equal to the most general string
2156 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2157 function name. (thanks to Nikodemus Siivola)
2158 * fixed bug: compile-time format string checker failed on
2159 non-closed ~{. (reported by Thomas F Burdick)
2160 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2161 don't warn on legal loop code involving more than one aggregate
2162 boolean. (thanks to Nikodemus Siivola)
2163 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2164 in keyword position to MAKE-PACKAGE was not being reported as
2166 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2167 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2169 * fixed bugs 280 and 312: the checking for multiple definitions in a
2170 file is less likely to become confused by uses of inline
2172 * fixed bug: the #S reader macro performs the keyword coercion
2173 specified for slot names. (reported by Kalle Niemitalo)
2174 * fixed bug: lambda lists may contain symbols whose names start with
2175 & but are not lambda-list-keywords; their occurrence triggers a
2177 * fixed bug 321: define-method-combination argument lambda lists do
2178 not cause invalid code to be generated when &optional or &aux
2179 variables have default values. (reported by Bruno Haible)
2180 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2181 have CLOS classes; this appears also to have fixed reported
2182 instability in stack exhaustion detection.
2183 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2184 MUFFLE-WARNING when no associated restart is present is now
2186 * optimization: rearranged the expansion of various defining macros
2187 so that each expands into only one top-level form in a
2188 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2190 * optimization: used a previously little-used slot in symbols to
2191 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2193 * fixed some bugs revealed by Paul Dietz' test suite:
2194 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2196 ** Fixed an optimization bug related to inheritance of initforms
2197 from local to shared slots.
2198 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2199 ** CAST optimizer forgot to flush argument derived type.
2200 ** print/read consistency on floats is now orders of magnitude
2201 more likely. (thanks also to Bruno Haible for a similar report
2203 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2204 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2205 for objects of type REAL. Make it so.
2206 ** SXHASH is specified (infelicitously) to respect similarity,
2207 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2208 it so. (thanks to Markus Ziegler)
2209 ** on the Alpha, the compiler succeeds in compiling functions
2210 returning a known number of arguments greater than 63.
2211 ** fixed handling of invalid NIL arguments in keyword position
2213 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2215 ** provided a readably-printable representation for RANDOM-STATE
2217 ** ensured that pathnames, where they have a namestring, always
2218 print using #P"..." syntax.
2220 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2221 * Support for the forthcoming 2.0 version of the NetBSD kernel
2222 running on x86 hardware has been added. (thanks to Perry
2223 E. Metzger most immediately, and others for their past work)
2224 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2225 current release version; SBCL's OpenBSD support had been broken
2226 since about the time of OpenBSD's switch to ELF binary format.)
2227 * [placeholder for DX summary]
2228 ** user code with &REST lists declared dynamic-extent, under high
2229 speed or space and low safety and debug optimization policy.
2230 * The manual has been converted to Texinfo format and the debugger
2231 chapter from the cmucl manual has been added.
2232 * A facility has been added to extract documentation strings from
2233 sbcl and store them as Texinfo-formatted snippets for inclusion in
2234 the manual (via Texinfo's @include directive)
2235 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2236 the test case to Dave Roberts)
2237 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2238 pointers. (reported by Sean Ross)
2239 * bug fix: PROFILE output is printed nicely even for large numerical
2240 values. (thanks to Zach Beane)
2241 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2242 greater than 32 handle EOF correctly.
2243 * bug fix: on X86 an immediate argument of the IMUL instruction is
2244 correctly printed in disassembly. (thanks to Lutz Euler)
2245 * bug fix: class slots in redefined classes preserve their old
2246 values. (thanks to Bruno Haible and Nikodemus Siivola)
2247 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2248 succeeds. (reported by Marco Baringer)
2249 * fixed some bugs revealed by Paul Dietz' test suite:
2250 ** READ-SEQUENCE now works on ECHO-STREAMs.
2251 ** RATIONALIZE works more according to its specification. (thanks
2254 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2255 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2256 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2257 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2258 should matter to you only if you rebind the printer control
2259 variables and then find you want different bindings in the
2260 debugger than in the ordinary execution of your program.)
2261 * The runtime build system has been tweaked to support building
2262 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2263 assembler and linker. (thanks to Nikodemus Siivola)
2264 * Unbound, undefined, undeclared variables now trigger full
2265 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2266 more often programmer error than deliberate exploitation of undefined
2268 * optimization: the hash algorithm for strings has changed to one
2269 that is less vulnerable to spurious collisions. (thanks to Juho
2271 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2272 needless bounds checking. (thanks to Juho Snellman)
2273 * optimization: implemented multiplication as a modular
2274 (UNSIGNED-BYTE 32) operation on the PPC backend.
2275 * fixed some bugs revealed by Paul Dietz' test suite:
2276 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2278 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2279 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2280 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2282 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2284 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2286 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2287 their output stream on EOF from read.
2288 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2289 have been read to end-of-file.
2290 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2292 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2293 description of determination of which consecutive characters
2295 ** Printing the "Space" character with escaping on now yields "#\\ ",
2296 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2297 ** Reading floating-point numbers with *READ-BASE* set to a number
2298 less than 10 works correctly.
2299 ** Reading floating-point numbers with *READ-BASE* set to a number
2300 more than 10 works correctly.
2301 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2302 the readtable currently in effect.
2304 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2305 * minor incompatible change: parsing of namestrings on a physical
2306 (Unix) host has changed; numbers after the final #\. in a
2307 namestring are no longer interpreted as a version field. This is
2308 intented to be largely invisible to the user, except that the
2309 meaning of the namestring "*.*.*" has changed: it now refers to a
2310 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2311 should usually be replaced by
2312 (make-pathname :name :wild :type :wild :version :wild)
2313 with the added benefit that this is more likely to be portable.
2314 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2315 option to OPEN now signals an error if the file being opened
2316 exists; this may have an impact on existing code.
2317 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2318 signals received, on the PowerPC platforms (both Linux and
2319 Darwin). (thanks to Pierre Mai for pointing out the location of
2321 * several fixes on OS X: The system now builds and runs cleanly on
2322 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2323 (thanks to Brian Mastenbrook)
2324 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2325 second-resolution: integer multiples of 1/3600 between -24 and 24.
2326 (thanks to Vincent Arkesteijn)
2327 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2328 type when called with 1 argument; PEEK-CHAR checked type of
2329 PEEK-TYPE only after having read first character from a
2330 stream. (reported by Peter Graves)
2331 * bug fix: the garbage collector now has much better locality
2332 behaviour, and in particular no longer treats all memory as being
2333 exhausted when many small objects point to each other in a deeply
2335 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2336 recognized as being TYPEP their class.
2337 * bug fix: the PUSHNEW documentation string has been corrected.
2338 (thanks to Vincent Arkesteijn)
2339 * bug fix: defaulting of the value for the last of an atypically
2340 large number of multiple values being bound was not being
2341 performed correctly on the Alpha or PPC platforms
2342 * optimization: implemented multiplication as a modular
2343 (UNSIGNED-BYTE 32) operation on the x86 backend.
2344 * optimization: SEARCH on simple-base-strings can now be open-coded.
2345 (see also contrib/compiler-extras.lisp for inspiration for
2346 teaching the compiler about the Boyer-Moore algorithm).
2347 * value, returned by MAX (and MIN) called with several EQUALP, but
2348 not EQL, arguments now does not depend on compiler settings.
2349 * fixed some bugs revealed by Paul Dietz' test suite:
2350 ** in stack analysis liveness information is propagated from
2351 non-local entry points.
2352 ** pathwise CAST removing failed when the CAST node did not start
2354 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2355 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2357 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2358 host is already defined.
2359 ** RENAME-FILE works on streams instead of signalling an internal
2361 ** PEEK-CHAR uses the current readtable when determining whether
2362 or not a character is whitespace.
2363 ** MERGE-PATHNAMES handles the case when the pathname does not
2364 specify a name while the default-pathname specifies a version.
2365 ** Pathnames now stand a better chance of respecting print/read
2367 ** Attempting to use standardized file system operators with a
2368 pathname with invalid :DIRECTORY components signals a
2370 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2372 ** DIRECTORY on logical pathnames is more correct.
2373 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2374 signal a TYPE-ERROR if their argument is not a stream.
2375 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2376 designator argument does not designate a stream.
2377 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2378 examining the synonym.
2379 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2381 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2382 element-type larger than ([UN]SIGNED-BYTE 32).
2384 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2385 * When built with the :SB-FUTEX feature, threaded builds now take
2386 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2387 for faster/more reliable mutex and condition variable support.
2388 * Incompatible change (but one you probably shouldn't have been using
2389 anyway): the interface and code for arbitrating between multiple
2390 threads in the same user session has been redesigned.
2391 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2392 error when called without an explicit environment argument.
2393 (thanks to Nikodemus Siivola)
2394 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2395 confused about their position. (thanks to Adam Warner and Gerd
2397 * bug fix: Pretty printing backquoted forms with unquotations in the
2398 argument list position of various code constructs such as LAMBDA
2399 now works correctly. (reported by Paul Dietz)
2400 * bug fix: Pretty printing unquotations no longer loses all
2401 stream position information.
2402 * optimization: performance of string output streams is now less
2403 poor for multiple small sequence writes.
2404 * optimization: performance of CSUBTYPEP in the presence of complex
2405 expressions involving CONS and NOT many times has been improved.
2406 (reported by Paul Dietz)
2407 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2409 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2410 streams for the REPL, for the debugger, and so on. (thanks to
2412 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2413 * fixed some bugs revealed by Paul Dietz' test suite:
2414 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2415 is ignored if it is too small, rather than propagating through
2416 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2417 ** extremely complex negations of CONS types were not being
2418 sufficiently canonicalized, leading to inconsistencies in
2420 ** VALUES tranformer lost derived type.
2422 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2423 * fixed a bootstrapping bug: the build process no longer assumes
2424 that the various BOOLE-related constants have the same value in
2425 host and target lisps. (noted by Paul Dietz' test suite on an
2426 SBCL binary built from CLISP)
2427 * The system can now be dynamically linked on the MIPS platform,
2428 which enables dynamic loading of foreign code from Lisp. (thanks
2429 to Ralf Baechle for discussions on the MIPS ABI)
2430 * The system now records debugging information for its own source
2431 files in a filesystem-position-independent manner, relative to
2432 the "SYS" logical host.
2433 * fixed a compiler bug: MV-LET convertion did not check references
2434 to the "max args" entry point. (reported by Brian Downing)
2435 * tweaked disassembly notes to be less confident about proclaiming
2436 some instruction as an LRA. (thanks to Brian Downing)
2437 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2438 now each have their own history, command character, and other
2439 characteristics. (thanks to David Lichteblau)
2440 * fixed some bugs revealed by Paul Dietz' test suite:
2441 ** compiler failure in compiling LOGAND expressions including a
2443 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2444 shift greater than 32.
2445 ** FLUSH-DEST did not mark blocks for type check regeneration.
2446 ** HANDLER-CASE failed to accept declarations in handler clauses
2447 in some circumstances.
2449 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2450 * New code in contrib/sb-introspect (still probably not entirely
2451 stable yet) provides some support for smart Lisp development
2452 environments like SLIME.
2453 * The conditions signalled for errors occurring when loading .fasl
2454 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2455 in a way which should help ASDF recover gracefully.
2456 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2457 stuff has been cleaned up. If you code contrib/ stuff, this might
2458 affect you, and you can look at contrib/README, contrib/STANDARDS,
2459 and/or the 0.8.4.27 diff to check.
2460 * In full calls the compiler now does not generate checks for declared
2461 argument types for all arguments.
2462 * various threading fixes
2463 ** and some experimental patches which didn't make it into
2464 the main tree for this release, but which are shipped in
2465 contrib/experimental-thread.patch as a possible fix for some
2466 failures (deadlock, spinning...) in GC-intensive multithreaded
2468 * fixed PPC build problem (source code incompatibility of different
2469 library versions): added offsetof() hackery which attempts to divine
2470 where glibc maintainers put uc_mcontext today
2471 * fixed bug 282: compiler does not trust type assertions while passing
2472 arguments to a full call.
2473 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2474 &OPTIONAL VALUES type parameter.
2475 * fix bug 214: algorithm for noting rejected templates is now more
2476 similar to that of template seletion. (also reported by rydis on
2478 * fixed bug 141b: printing backquoted information readably and prettily
2479 inserts a space where necessary.
2480 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2481 (CHARACTER 10) are now reported as errors, rather than propagated
2482 as unknown types. (reported by piso on #lisp)
2483 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2484 with values NIL and :ERROR. (thanks to Milan Zamazal)
2485 * fixed bug 191c: CLOS now does proper keyword argument checking as
2486 described in CLHS 7.6.5 and 7.6.5.1.
2487 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2488 longer raise an error; further, using a list as a for-as-arithmetic
2489 counter now raises a meaningful error.
2490 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2491 now understood by sequence creation functions such as MAKE-SEQUENCE
2493 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2494 a STRING-INPUT-STREAM.
2495 * compiler enhancement: SIGNUM is now better able to derive the type
2497 * type declarations inside WITH-SLOTS are checked. (reported by
2499 * fixed some bugs revealed by Paul Dietz' test suite:
2500 ** incorrect optimization of TRUNCATE for a positive first
2501 argument and negative second.
2502 ** compiler failure in let-convertion during flushing dead code.
2503 ** compiler failure while deriving type of TRUNCATE on an
2504 interval, containing 0.
2505 ** ASH of a negative bignum by a negative bignum count now returns
2507 ** intersection of CONS types now canonicalizes properly, fixing
2508 inconsistencies in SUBTYPEP.
2510 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2511 * incompatible change: The --disable-debugger command line
2512 option now clobbers the debugger at a more fundamental
2513 level, by redefining #'INVOKE-DEBUGGER instead of by
2514 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2515 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2516 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2517 enter the debugger REPL and then suffer recursive errors
2518 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2519 while under the new system, BREAK in --disable-debugger mode
2520 terminates the system just as an unhandled error would.
2521 * fixed compiler performance when processing loops with a step >1;
2522 * bug fix: DOCUMENTATION now retrieves generic function
2523 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2524 support has been systematized, and now supports the methods
2525 specified by ANSI, along with a default method and a method for
2526 slot documentation. (reported by Nathan Froyd)
2527 * bug fix: effective methods associated with a generic function are
2528 no longer cached over a change of that generic function's method
2529 combination. (reported by Andreas Fuchs)
2530 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2531 names a non-existent directory. (thanks to Andreas Fuchs)
2532 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2533 the PPC platform, lead to stack corruption; this has been fixed.
2534 (reported by Rainer Joswig)
2535 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2536 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2537 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2538 arithmetic in general, is now much more reliable on the PPC
2540 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2541 platform now returns the right answer.
2542 * optimization: restored some effective method precomputation in
2543 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2544 precomputation is now tunable.
2545 * optimization: compiler-internal data structure use has been
2546 reviewed, and changes have been made that should improve the
2547 performance of the compiler by about 20%.
2548 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2549 simple-base-strings and simple-bit-vectors is improved.
2550 * optimization: the optimization of 32-bit logical and arithmetic
2551 functions introduced in version 0.8.3 on the x86 has been
2552 implemented on the mips, ppc and sparc platforms; an
2553 implementation of the same facility, but for 64-bit arithmetic,
2554 has been added for the alpha.
2555 * microoptimization: the compiler is better able to make use of the
2556 x86 LEA instruction for multiplication by constants.
2557 * bug fix: in some situations compiler did not report usage of
2558 generic arithmetic in (SPEED 3) policy.
2559 * bug 145b fix: compiler used wrong type specifier while converting
2560 MEMBER-types to numeric.
2561 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2563 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2564 index argument now works properly on the Alpha platform.
2565 * bug fix: floating point exception treatment on the Alpha platform
2567 * bug fix: FILE-POSITION works much better on string input and
2568 output streams. (thanks to Nikodemus Siivola)
2569 * bug fix: many threading/garbage collection symptoms sorted.
2570 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2571 might be pseudo-atomic.
2572 * internal change: Stopping for GC is now done with signals not
2573 ptrace. GC is now done in whichever thread wanted it, instead of
2575 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2577 * bug fix: result form in DO is not contained in the implicit
2579 * incompatible change: ICR structure is changed; the value part of
2580 CONTINUATION is now called LVAR; corresponding functions are
2581 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2582 * added type deriver for ISQRT (thanks to Robert E. Brown).
2583 * bug fix: better support for loading from the command line when an
2584 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2586 * fixed some bugs revealed by Paul Dietz' test suite:
2587 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2588 ** ROUND and FROUND now give the right answer when given very
2589 small float arguments.
2590 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2592 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2593 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2594 ** LCM should return a non-negative integer.
2595 ** PARSE-INTEGER returned the index of a terminator instead of the
2596 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2597 ** PARSE-INTEGER returned an incorrect index being applied to a
2599 ** LCM with two arguments of 0 returns 0 rather than signalling
2601 ** unsigned addition of a 32-bit constant with the high bit set no
2602 longer causes an internal compiler error on the x86.
2603 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2604 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2605 with negative last argument.
2606 ** byte specifiers with zero size and position no longer cause
2607 an error during type derivation.
2608 ** bignum multiplication on the Alpha platform now returns the
2610 * porting: The system now builds on SuSE AMD64, although it still
2611 generates a 32-bit binary.
2612 * .fasl file incompatibility: The fasl file version number has
2613 been incremented (because of the changes to internal compiler
2614 data structures referred to above).
2616 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2617 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2618 more accurately, on the Darwin kernel running on PowerPC hardware.
2619 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2620 * Compiler code deletion notes now signal a condition of type
2621 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2622 an associated MUFFLE-WARNING restart.
2623 * The compiler now performs limited argument count validation of
2624 constant format strings in FORMAT, and where appropriate in ERROR,
2625 CERROR and WARN. (thanks to Gerd Moellmann)
2626 * New ASDF-INSTALL contrib can be used for automatic download and
2627 installation of third-party Lisp code from CCLAN or other sites
2629 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2630 which forces another thread to execute a function supplied by the
2632 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2633 now accept and act upon their :ELEMENT-TYPE keyword argument.
2634 (reported by Martin Atzmueller, Edi Weitz)
2635 * bug fix: FILE-POSITION now accepts position designators up to
2636 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2637 is the greater. (thanks to Patrik Nordebo)
2638 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2639 Kalle Olavi Niemitalo)
2640 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2641 printable. (reported by Eric Marsden)
2642 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2643 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2644 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2645 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2646 (thanks to Stig Sandoe)
2647 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2648 UNREAD-CHAR now correctly includes the unread character in the
2649 target sequence. (thanks to Gerd Moellmann)
2650 * bug fix in threaded builds: the system can now be suspended and
2651 resumed by shell job control with minimal disruption.
2652 * bug fixes in times and timezones >2038AD
2653 * better handling of "where is GNU make?" problem in build scripts
2654 (thanks to Nikodemus Siivola)
2655 * new optimization: inside a named function any reference to a
2656 function with the same name is considered to be a self-reference;
2657 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2658 optimization quality.
2659 * new optimization on x86: logical functions and + now have
2660 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2661 used when the result is truncated to 32 bits.
2662 * VALUES declaration is partially enabled.
2663 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2664 * fixed some bugs revealed by Paul Dietz' test suite:
2665 ** The system now obeys the constraint imposed by
2666 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2667 types form a lattice under type intersection.
2668 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2669 ** ASSOC now ignores NIL elements in an alist.
2670 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2671 and (1+ MOST-POSITIVE-FIXNUM) answers.
2672 ** The addition of a method with invalid qualifiers to a generic
2673 function does not cause an error to be signalled immediately;
2674 a warning is signalled, and the error is generated only on
2675 calling the generic function.
2676 * changed .fasl file version number, in part to add type codes for
2677 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2678 obscure ANSI requirements
2680 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2681 * fixed bug 148: failure to inline-expand a local function left
2682 garbage, confusing the compiler.
2683 * fixed bugs 3cd: structure slot readers perform type check if the
2684 slot can have an invalid value (i.e. it is either not initialized
2685 or can be written with a less specific slot writer).
2686 * bug fix: the compiler now traps array references to elements off
2687 the end of an array; previously, the bounds checking in some
2688 circumstances could go off-by-one.
2689 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2691 * type declarations for array element types now obey the description
2692 on the CLHS page "Declaration TYPE", as per discussions on
2693 sbcl-help around 2003-05-08. This means that a declaration
2694 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2695 declaration, all references to BAR will be asserted or assumed
2696 (with THE, so dependent on compiler policy) to involve objects of
2697 type FOO. Note that no such declaration is implied in
2698 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2699 * declared types of functions from the "Conditions"
2700 chapter. (reported by Paul Dietz)
2701 * bug fix: CERROR accepts a function as its first argument.
2702 * bug fix: NTH an NTHCDR accept a bignum as index
2703 arguments. (reported by Adam Warner)
2704 * optimization: character compare routines now optimize comparing
2705 against a constant character. (reported by Gilbert Baumann)
2706 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2707 argument now works properly on the MIPS platform.
2708 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2709 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2710 declared to have a functional type, does not cause an error
2712 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2713 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2714 * fixed bug: if last continuation of a deleted block has a
2715 destination, this destination should be deleted too. (reported by
2717 * fixed a bug in the bootstrap process: the host compiler's values
2718 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2719 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2720 test case from Patrik Nordebo)
2721 * improved the ability of the disassembler on the PPC platform to
2722 provide helpful disassembly notes.
2723 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2724 the class in more cases than previously.
2725 * bug fix: FILE-POSITION now understands :START and :END for
2726 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2727 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2728 8) data. (thanks to David Lichteblau)
2729 * bug fix: it is possible to add a method to a generic function
2730 without lambda list.
2731 * bug fix: reader failed to signal END-OF-FILE inside an
2732 object representation. (reported by Nikodemus Siivola)
2733 * fixed some bugs revealed by Paul Dietz' test suite:
2734 ** LAST and [N]BUTLAST should accept a bignum.
2735 ** condition slot accessors are methods.
2736 ** (VECTOR NIL) is a subtype of STRING.
2738 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2739 * minor incompatible change: some nonsensical specialized lambda
2740 lists (used in DEFMETHOD) which were previously ignored now signal
2742 * minor incompatible change: the system is now aware of the types of
2743 variables in the COMMON-LISP package, and will signal errors for
2744 most violations of these type constraints (where previously they
2745 were silently accepted).
2746 * minor incompatible change: COMPILE-FILE now uses the freedom
2747 afforded (ANSI 3.2.2.3) to use derived function types for
2748 functions defined in the same file. This also permits the system
2749 to warn on static type mismatches and function
2750 redefinition. (Currently it does not work with high DEBUG level.)
2751 * minor incompatible change: VALUES declaration is disabled.
2752 * When issuing notes, the compiler now signals a condition of type
2753 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2754 restart for use in user handlers. It is expected that the
2755 COMPILER-NOTE condition will eventually become a condition
2756 supertype to a hierarchy of note types, which will then be
2757 handleable in a similar fashion. However, at the moment, no such
2758 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2759 but it's an implementation detail, not a classification for the
2761 * Changes in type checking closed the following bugs:
2762 ** type checking of unused values (192b, 194d, 203);
2763 ** template selection based on unsafe type assertions (192c, 236);
2764 ** type checking in branches (194bc).
2765 * A short form of VALUES type specifier has ANSI meaning (it has
2766 increased the number of situations when SBCL cannot perform type
2768 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2769 keywords or constants is permissible.
2770 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2771 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2772 classes. (thanks to Antonio Martinez)
2773 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2774 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2775 argument to a pretty-print stream. Instead, it leaves any such
2776 operations to DESCRIBE-OBJECT methods.
2777 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2779 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2780 lambda lists. (thanks to David Lichteblau)
2781 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2782 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2783 (thanks to Gerd Moellmann)
2784 * fixed bug 246: increased compilation speed of long
2785 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2787 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2789 * DEFCONSTANT now throws a condition of type
2790 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2791 constant to a non-EQL value; CONTINUE and ABORT restarts
2792 respectively change and preserve the value.
2793 * fixed bug 63: The code walker, part of the implementation of CLOS,
2794 is now better at handling symbol macros.
2795 * bug fix: There is no longer an internal implementation type named
2796 CL:LENGTH. (reported by Raymond Toy)
2797 * bug fix: In macro-like defining macros/special operators the
2798 implicit block does not enclose the lambda list.
2799 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2800 enclosed in parentheses) are not suitable as type specifiers, and
2801 their use properly signals an error now.
2802 * bug fix: An argument count mismatch for a type specifier in code
2803 being compiled no longer causes an unhandled error at compile
2804 time, but signals a compile-time warning.
2805 * fixed simple vector readable printing
2806 * bug fix: DESCRIBE takes more care over whether the class
2807 precedence list slot of a class is bound before accessing it.
2808 (reported by Markus Krummenacker)
2809 * bug fix: FORMATTER can successfully compile pretty-printer format
2810 strings which use variants of the ~* directive inside.
2811 * bug fix: SEARCH now applies its TEST predicate to the elements of
2812 the arguments in the correct order. (thanks to Wolfhard Buss)
2813 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2814 in type inference. (thanks to Robert E. Brown)
2815 * bug fix: Reading in symbols with an explicit package name of ""
2816 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2817 * fixed some bugs revealed by Paul Dietz' test suite:
2818 ** NIL is now allowed as a structure slot name.
2819 ** Arbitrary numbers, not just REALs, are allowed in certain
2820 circumstances in LOOP for-as-arithmetic clauses.
2821 ** Multiple class redefinitions before slot access no longer
2822 causes a type error.
2823 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2824 association between the name and a class.
2825 ** Generic functions with non-standard method-combination and over
2826 five methods all of which return constants no longer return NIL
2827 after the first few invocations. (thanks to Gerd Moellmann)
2828 ** CALL-NEXT-METHOD with no arguments now passes the original
2829 values of the arguments, even in the presence of assignment.
2830 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2832 ** DISASSEMBLE works with closures and funcallable instances.
2833 ** ADD-METHOD now returns the generic function, not the new method.
2834 ** FIND-METHOD signals an error if the lengths of the specializers
2835 is incompatible with the generic function, even if the ERRORP
2837 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2838 which its argument is a member.
2839 ** DEFCLASS only redefines the class named by its class-name
2840 argument if that name is the proper name of the class;
2841 otherwise, it creates a new class.
2842 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2843 of the UNBOUND-SLOT condition to the name of the slot.
2844 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2845 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2846 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2847 SLOT-MAKUNBOUND in the specified fashion.
2849 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2850 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2851 cross-compilation host. As a consequence, we can now bootstrap our
2852 way up to SBCL starting with a bare gcc toolchain and human-readable
2853 source code (first the source to CLISP, then the source to SBCL).
2854 * A contributed module containing a partial implementation of the
2855 simple-streams interface has been included. (thanks to Rudi
2857 * A contributed module implementing the RFC1321 Message Digest
2858 Algorithm, known as MD5, has been included.
2859 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2860 no longer has any effect, as the code controlled by this feature
2861 has been deleted. (As far as we know, no-one has ever built using
2862 this feature, and its semantics were confused in any case).
2863 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2864 work on conditions (as required by the ANSI specification),
2865 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2866 expected behaviour on conditions. Users should note, however,
2867 that such behaviour is not required by the ANSI specification,
2868 and so use of this behaviour may render their code unportable.
2869 * fixed some bugs revealed by Paul Dietz' test suite:
2870 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2872 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2873 ** SXHASH on condition objects no longer returns NIL.
2874 ** :ALLOCATION :CLASS slots are better treated; their values are
2875 updated on class redefinition, and initforms inherited from
2876 superclasses are applied.
2877 ** REMOVE-METHOD returns its generic function argument even when
2878 no method was removed.
2879 ** SHARED-INITIALIZE now initializes the values of the requested
2880 slots, including those with :ALLOCATION :CLASS.
2881 ** ALLOCATE-INSTANCE now works on structure classes defined via
2882 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2884 ** SLOT-EXISTS-P now works on conditions, as well as structures
2886 ** MAKE-LOAD-FORM now has the required methods on
2887 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2888 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2889 keyword, and now implements the SLOT-NAMES argument.
2890 ** methods with &OPTIONAL arguments no longer allow too many
2891 arguments to be passed in the call without error.
2892 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2893 option is consistent with the required arguments of the generic
2894 function lambda list.
2895 * bug fix: REQUIRE accepts a string designator. (Thanks to
2897 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2898 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2899 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2900 * bug fix: The compiler now checks for duplicated variables in macro
2902 * bug fix: SETQ on globals returns the correct value.
2903 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2904 not a COMPILER-ERROR (followed by some other strange error on
2905 choosing the CONTINUE restart).
2906 * bug fix: make.sh and friends are now more consistent in the way that
2907 they look for GNU "make".
2909 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2910 * experimental native threads support (on x86 Linux >=2.4 only).
2911 This is not compiled in by default: you need to add :SB-THREAD to
2912 the target features. See the "Beyond ANSI" chapter of the manual
2914 * fix for longstanding nonANSIism: The old distinction between
2915 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2916 The return value from CL:FIND-CLASS is now a CLOS class, and
2917 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2918 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2919 * An interface to the MetaObject Protocol, as described in Kiczales,
2920 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2921 MIT Press, 1991, is available from the SB-MOP package.
2922 * incompatible change: the SB-PCL package should now be considered
2923 a private implementation detail, and no longer a semi-private MOP
2925 * minor incompatible change: due to rearrangement for threads, the
2926 control stack and binding stack are now allocated at arbitrary
2927 addresses instead of being hardcoded per-port. Users affected by
2928 this probably have to be doing advanced things with shared
2929 libraries, and will know who they are.
2930 * minor incompatible change: Previously, all --eval forms used were
2931 processed with READ before any of them were processed with EVAL.
2932 Now each --eval form is processed with both READ and EVAL before
2933 the next --eval form is processed. (Thus package operations like
2934 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2935 work as the user might reasonably expect.)
2936 * minor incompatible change: *STANDARD-INPUT* is now only an
2937 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2939 * minor incompatible change: Y-OR-N-P is now character-oriented,
2940 not line oriented. Also, YES-OR-NO-P now works without errors.
2941 (thanks to Antonio Martinez)
2942 * sb-aclrepl module improvements: an integrated inspector, added
2943 repl features, and a bug fix to :trace command.
2944 * Known functions, which cannot be open coded by the backend, are
2945 considered to be able to check types of their arguments. (fixing
2946 a bug report by Nathan J. Froyd)
2947 * fixed a bug in computing method discriminating functions: It is
2948 now possible to define methods specialized on classes which have
2949 forward-referenced superclasses. (thanks to Gerd Moellmann)
2950 * fixed evaluation order in optional entries (reported by Gilbert
2952 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2953 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2954 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2955 required-by-AMOP NAME argument, as well as CLASS and
2956 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2957 * fixed bug 20: DEFMETHOD can define methods using names that are
2958 not the proper names of classes to designate class specializers.
2959 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2960 Unix file descriptors, instead of blowing up. (thanks to Antonio
2962 * Garbage collection refactoring: user-visible change is that a
2963 call to the GC function during WITHOUT-GCING will not do garbage
2964 collection until the end of the WITHOUT-GCING. If you were doing
2965 this you were probably losing anyway.
2966 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2967 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2968 TYPEP the latter but not the former.
2969 * The compiler issues a full WARNING for calls to undefined functions
2970 with names from the CL package.
2971 * MAP-INTO for a vector destination is open coded. (reported by
2972 Brian Downing on c.l.l)
2973 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2974 documentation string.
2975 * fixed some bugs revealed by Paul Dietz' test suite:
2976 ** COPY-ALIST now signals an error if its argument is a dotted
2978 ** Condition slots are now accessed more correctly in the presence
2979 of multiple initargs for a given slot.
2980 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2981 exclude restarts of the same name associated with a different
2983 ** DEFCLASS of forward-referenced classes with another
2984 forward-referenced class in the superclasses list no longer
2986 ** Condition slots are now initialized once each, not multiple
2987 times. (thanks to Gerd Moellmann)
2988 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2990 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2991 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2992 arguments contain duplicated elements.
2993 ** RESTART-CASE understands local macros.
2994 ** RESTART-CASE associates exactly its own restarts with a condition.
2995 ** ENDP in safe mode checks its argument to be of type LIST.
2996 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2997 in question is unbound.
2998 ** Optimized MAKE-INSTANCE functions no longer cause internal
2999 assertion failures in the presence of duplicate initargs.
3000 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3001 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3002 propagated even in the ctor optimized implementation of
3004 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3006 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3007 * a better implementation of SXHASH on (simple) bit vectors,
3008 measured both in execution speed and in distribution of results
3009 over the positive fixnums, has been installed. Likewise, a better
3010 implementation of EQUAL for simple bit vectors is now available.
3011 * fixed CEILING optimization for a divisor of form 2^k.
3012 * fixed bug 240 (emitting extra style warnings "using the lexical
3013 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3014 by Antonio Martinez)
3015 * fixed SXHASH, giving different results for NIL depending on type
3016 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3017 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3018 variable. (found by Rolf Wester)
3019 * SBCL does not ignore type declarations for special
3020 variables. (reported by rif on c.l.l 2003-03-05)
3021 * some bug fixes in contrib/sb-aclrepl/
3022 * fixed some bugs revealed by Paul Dietz' test suite:
3023 ** a bug in the CONS type specifier, whereby the CAR and CDR
3024 types got intertwined, has been fixed;
3025 ** the type system is now able to reason about the interaction
3026 between INTEGER and RATIO types more completely;
3027 ** APPEND, [N]REVERSE and NRECONC check that those their
3028 arguments, which must be proper lists, are really so;
3029 ** An array specialized to be unable to hold elements has been
3030 implemented, as required -- yes, really -- by ANSI;
3031 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3032 on malformed property lists;
3034 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3035 * incompatible packaging change: in line with Unix convention,
3036 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3037 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3038 * REQUIRE and PROVIDE are now optionally capable of doing something
3039 useful. See the documentation string for REQUIRE.
3040 * infrastructure for a managed SBCL contrib system: contributed
3041 modules in this release include:
3042 ** the ASDF system definition facility;
3043 ** an interface to the BSD Sockets API;
3044 ** an ACL-like convenience interface to the repl;
3045 (thanks to Kevin Rosenberg)
3046 ** an implementation of ROTATE-BYTE, with efficient implementation
3048 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3049 gives a closer approximation to the right answer than previously.
3050 (thanks to Raymond Toy)
3051 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3052 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3053 optional environment arguments, as required by ANSI.
3054 * fixed bugs in other functions taking environment objects, allowing
3055 calls with an explicit NIL environment argument to be compiled
3057 * fixed bug 228: primary return values from
3058 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3059 COMPILE or FUNCTION.
3060 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3061 :TYPE will no longer signal errors on innocuous objects.
3062 * fixed bug 231b: SETQ is better at respecting type declarations in
3063 the lexical environment.
3064 * fixed a bug in DEFCLASS: classes named by symbols with no or
3065 unprintable packages can now be defined.
3066 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3067 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3068 * fixed compiler failure related to checking types of functions.
3069 (reported by Robert E. Brown)
3070 * the compiler is now much more consistent in its error-checking
3071 treatment of bounding index arguments to sequence functions: in
3072 (SAFETY 3) code, errors will be signalled in almost all cases if
3073 invalid sequence bounding indices are passed to functions defined
3074 by ANSI to operate on sequences.
3075 * fixed a bug in the build procedure: documentation of SBCL-specific
3076 packages is now preserved and available in the final Lisp image.
3077 * lifted FDEFINITION lookup out of loops in the implementation of
3078 many list operations. (thanks to Robert E. Brown)
3079 * fixed a bug in the reader: the #n# reader macro now works for
3080 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3081 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3082 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3083 E. Brown and Tony Martinez respectively)
3084 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3085 SLOT-BOUNDP: in particular, it is now able to optimize them much
3086 better, and is now not vulnerable to having packages renamed.
3087 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3088 always called when a slot is not present in an instance. (thanks
3090 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3091 longer signal an error on unreading a character following EOF on
3092 the previous constituent stream. (thanks to Tony Martinez)
3093 * fixed some bugs revealed by Paul Dietz' test suite:
3094 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3095 not just nonnegative fixnums;
3096 ** the logical bit-array operators such as BIT-AND now accept an
3097 explicit NIL for their "opt-arg" argument (to indicate a
3098 freshly-consed result bit-array);
3099 ** ELT now signals an error on an invalid sequence index in safe
3101 ** the type system is now cleverer about negations of numeric
3102 types, and consequently understands the BIGNUM and RATIO types
3104 ** the type system is now cleverer about the interaction between
3105 INTEGER and RATIO types: while bugs still remain, many more
3106 cases are accurately computed;
3107 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3108 if it is in the last clause;
3109 ** CONSTANTLY now correctly returns a side-effect-free function in
3111 ** DECLARE is no longer treated as a special-operator; in
3112 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3113 * incremented fasl file version number due to the change in the
3114 DEFSTRUCT-SLOT-DESCRIPTION structure.
3116 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3117 * minor incompatible change: code processed by the "interpreter" or
3118 EVAL now has a compilation optimization policy of (DEBUG 2)
3119 (changed from (DEBUG 1)) to improve debuggability of interactive
3120 development, and to allow the use of the debug RETURN command in
3122 * an experimental implementation of the RETURN command for the
3123 debugger has been included. (thanks to Frederik Kuivinen)
3124 * fixed bug 62: constraints were not propagated into a loop.
3125 * fixed bug in embedded calls of SORT (reported and investigated by
3127 * fixed some bugs revealed by Paul Dietz' test suite:
3128 ** printing and reading of arrays with some dimensions having
3129 length 0 (thanks to Gerd Moellmann);
3130 ** BOA constructor with &AUX argument without a default value does
3131 not cause a type error;
3132 ** CONSTANTP now returns true for all self-evaluating objects.
3134 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3135 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3136 accessors that are related by inheritance, as specified in the
3137 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3138 Valtteri Vuorikoski)
3139 * The compiler is now able to inline functions that were defined in
3140 a complex lexical environment (e.g. inside a MACROLET).
3141 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3143 * Support for the upcoming FreeBSD-5.0 release has been included.
3144 (thanks to Dag-Erling Smorgrav)
3145 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3146 effect when it is not in a toplevel context.
3147 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3148 stand a better chance of being correct. (thanks to Gerd
3150 * fixed bug in COERCE, which now signals an error on coercing a
3151 rational to a bounded real type which excludes the expected
3153 * The compiler is now able to derive types more accurately from the
3154 COERCE and COMPILE functions.
3155 * fixed bug 223: functional binding is considered to be constant
3156 only for symbols in the CL package.
3157 * fixed bug 231: SETQ did not check the type of a variable being set
3158 (reported by Robert E. Brown)
3159 * A new optimization for MAKE-INSTANCE has been included, fixing
3160 various bugs (including relating to :ALLOCATION :CLASS slots and
3161 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3163 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3164 ** As required by ANSI, LOOP now disallows anonymous collection
3165 clauses such as COLLECT I in conjunction with aggregate boolean
3166 clauses such as THEREIS (= I 1);
3167 ** LOOP now signals an error when any variable is reused in the
3168 same loop (including the potentially useful construct analogous
3169 to WITH A = 1 WITH A = (1+ A);
3170 ** IT is only a special loop symbol within the first clause of a
3171 conditional loop clause;
3172 ** LOOP with a typed iteration variable over a hashtable now
3173 signals a type error iff it should.
3174 * fixed some other bugs revealed by Paul Dietz' test suite:
3175 ** FILE-STREAM now names the class previously known as FD-STREAM;
3176 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3177 argument) no longer signals an error;
3178 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3179 of the slot symbol, rather than using the current package
3180 ((:CONC-NAME "") continues to intern the slot's name in the
3182 * incremented fasl file version number, because of the incompatible
3183 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3184 of the new implementation of DEFINE-COMPILER-MACRO.
3186 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3187 * Support for building SBCL for MIPS platforms running in
3188 little-endian mode has now been checked in, and basic
3189 functionality on said platforms verified.
3190 * minor incompatible change: PCL now records the pathname of a file
3191 in which methods and the like are defined, rather than its
3193 * minor incompatible change: TRUENAME now considers the truename of
3194 a file naming a directory to be the pathname with :DIRECTORY
3195 component indicating that directory.
3196 * minor incompatible change: a NAMED clause in the extended form of
3197 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3198 reason for the previous behaviour is unclear.
3199 * more systematization and improvement of CLOS and MOP conformance
3200 in PCL (thanks to Gerd Moellman and Pierre Mai):
3201 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3203 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3204 primary methods with no specializers;
3205 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3207 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3208 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3209 FORWARD-REFERENCED-CLASSes; error reporting on
3210 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3212 ** SXHASH on CLOS instances now uses a slot internal to the
3213 instance to return different numbers on distinct instances,
3214 while preserving the same return value through invocations of
3216 ** DEFMETHOD signals errors when methods with longer incongruent
3217 lambda lists are added to generic functions;
3218 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3219 CLASS, as specified in AMOP;
3220 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3221 on the order returned by the primary method for classes of
3222 class STANDARD-CLASS;
3223 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3224 * fixed some bugs shown by Paul Dietz' test suite:
3225 ** DOLIST puts its body in TAGBODY;
3226 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3228 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3229 value producing form;
3230 ** if more variables are given to PROGV than values, extra
3231 variables are bound and made to have no value;
3232 ** NSUBSTITUTE on list arguments gets the right answer with
3234 ** ELT signals an error of type TYPE-ERROR when the index argument
3235 is not a valid sequence index;
3236 ** LOOP signals (at macroexpansion time) an error of type
3237 PROGRAM-ERROR when duplicate variable names are found;
3238 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3239 ** FUNCALL of special-operators now cause an error of type
3241 ** PSETQ now works as required in the presence of side-effecting
3242 symbol-macro places;
3243 ** NCONC accepts any object as its last argument;
3244 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3246 ** loop-for-as-package does not require a package to be explicitely
3248 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3250 * fixed bug 166: compiler preserves "there is a way to go"
3251 invariant when deleting code.
3252 * fixed bug 172: macro lambda lists with required arguments after
3253 &REST arguments now cause an error to be signalled. (thanks to
3255 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3256 bug. (thanks to Matthew Danish)
3257 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3259 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3260 arguments are lexically rebound. (thanks to Gerd Moellmann and
3262 * fixed bug 194: error messages are now more informative when there
3263 is no primary method applicable in a call to a generic
3264 function. (thanks to Gerd Moellmann)
3265 * fixed bug in command line argument checking (thanks to Julian
3267 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3268 (thanks to Matthew Danish)
3269 * incremented fasl file version number, because of the
3270 SXHASH-related changes in the layout of CLOS data structures
3272 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3273 * minor incompatible change: The runtime (the Unix executable named
3274 "sbcl") is now much pickier about the .core files it will load.
3275 Essentially it now requires .core files to descend from the same
3276 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3277 as the runtime does. (The intent is to prevent the crashes which
3278 can occur, and which can even be reported as mysterious failures,
3279 when people patch the sources or change the build parameters
3280 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3281 sbcl and .core files.)
3282 * fixed bug: VALUES-LIST is no longer optimized away.
3283 * fixed bug 142: The FFI conversion of C string values to Lisp
3284 string values no longer conses excessively. (thanks to Nathan
3285 Froyd porting Raymond Toy's fix to CMU CL)
3286 * began to systematize and improve MOP conformance in PCL (thanks to
3287 Nathan Froyd, Gerd Moellman and Pierre Mai):
3288 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3290 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3291 ** FINALIZE-INHERITANCE is now called on class finalization;
3292 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3293 argument precedence order.
3294 * fixed bug 202: The compiler no longer fails on functions whose
3295 derived types contradict their declared type.
3296 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3297 so it can be non-toplevel.
3298 * The fasl file version number has changed (because of the new
3299 implementation of DEFMACRO).
3300 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3301 safe code, that any length requirement by their type-specifier
3302 argument is valid. The exceptions to this are described in bug
3304 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3305 functions now check the types of their inputs as required by ANSI.
3306 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3307 introduced symbol is DECLAREd to be SPECIAL.
3308 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3309 * fixed bug: :COUNT argument to sequence functions may be negative
3310 * fixed bug: body of DO-SYMBOLS may contain declarations
3311 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3312 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3313 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3314 symbol macro only once
3315 * fixed printing of call frame when argument list is unavailable
3316 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3317 * compiler no longer signals WARNING on unknown keyword
3320 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3321 * A beta-quality port to the mips architecture running Linux,
3322 based on the old CMUCL backend, has been made. It has been tested
3323 on a big-endian kernel, and works sufficiently well to be able to
3324 rebuild itself; it has not been tested in little-endian mode.
3325 * fixed an inconsistency between gencgc.c and purify.c which made
3326 dumping/loading .core files unreliable
3327 * fixed bug 120a: The compiler now deals correctly with IFs where
3328 the consequent is the same as the alternative, instead of
3329 misderiving the return type. (thanks to Alexey Dejneka)
3330 * fixed bug 113: Logical pathnames are now dumpable (the logical
3331 host is resolved at load-time, throwing an error if it is not
3333 * fixed bug 174: FORMAT's error message is slightly clearer when a
3334 non-printing character is used in a format directive.
3335 * fixed several bugs in compiler checking of type declarations, i.e.
3336 violations of the Python "declarations are assertions" principle
3337 (thanks to Alexey Dejneka)
3338 * fixed several bugs in PCL's error checking (thanks to Gerd
3340 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3342 * fixed bug in compilation of functions as first class values
3343 (thanks to Antonio Martinez-Shotton)
3344 * The compiler's handling TYPE-ERRORs which it can prove will
3345 inevitably happen at runtime has been cleaned up and corrected
3346 in several ways. (thanks to Alexey Dejneka)
3347 * improved argument type checking for various basic arithmetic
3348 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3349 lavished on them in the past that they can be compiled in many
3350 ways in different special cases
3351 * fixed bug 181: compiler checks validity of user supplied type
3353 * cleaned up code flushing in optimization: Function calls which
3354 should signal errors for safety purposes (e.g. which ANSI says
3355 should signal errors when their arguments are of incorrect type)
3356 are no longer optimized away.
3357 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3358 * incremented fasl file version number, because changes in the
3359 implementation of sequence functions like COERCE caused
3360 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3361 in old inline expansions) to become undefined. (Actually these
3362 changes were later undone, so we might very well be binary
3363 compatible with 0.7.7 after all, but leaving the version number
3364 incremented seemed like the simplest and most conservative
3367 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3368 * An alpha-quality port to the parisc architecture running Linux,
3369 based on the old CMUCL backend, has been made. This, even more so
3370 than the other backends, should be considered still a work in
3371 progress; known problems include that the Linux kernel in 64-bit
3372 mode does not propagate the correct sigcontext structure to
3373 userspace, and consequently SBCL on a parisc64 kernel will not
3375 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3376 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3377 suppressed.) Also now that the compiler is looking at declarations
3378 in the environment, it checks optimization declarations as well,
3379 and suppresses inlining when (> DEBUG SPEED).
3380 * More fixes have been made to treatment of floating point exception
3381 treatment and other Unix signals. In particular, floating point
3382 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3383 * The detection and handling of control stack exhaustion (infinite
3384 or very deeply nested recursion) has changed. Stack exhaustion
3385 detection is now done by write-protecting pages at the OS level
3386 and applies at all optimization settings; when found, a
3387 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3388 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3389 be caught using IGNORE-ERRORS.
3390 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3391 correctly when transforming recursive LABELS functions to LETs.
3392 (thanks to Alexey Dejneka)
3393 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3394 that are names of constants or global variables.
3395 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3396 alien routines with docstrings.
3397 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3398 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3400 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3401 (thanks to Eric Marsden)
3402 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3403 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3404 * Bug fix: The SPARC backend can now compile functions involving
3405 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3406 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3407 to a functions expecting a PATHNAME-DESIGNATOR.
3408 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3409 lambda lists. (thanks to Alexey Dejneka)
3410 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3411 functions so that EXPORTing the name of the class doesn't cause
3412 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3413 to Antonio Martinez for reporting this)
3414 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3415 constant-folded without failing an assertion. (thanks to Einar
3416 Floystad Dorum for reporting this)
3417 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3418 types are now dealt with more correctly by the compiler.
3419 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3420 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3421 behaviour analogous to RENAME-FILE. This puts its behaviour more
3422 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3424 * The fasl file version number has changed again. (because of the
3425 bug fix involving the names of PCL MAKE-INSTANCE functions)
3427 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3428 * bug fix: Floating point exceptions are treated much more
3429 consistently on the x86/Linux and PPC/Linux platforms.
3430 * Array initialization with :INITIAL-ELEMENT is now much faster for
3431 cases when the compiler cannot open code the array creation, but
3432 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3433 array accesses have also seen a speed increase.
3434 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3435 specified. (This was at the root of some bad interactions between
3436 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3438 * bug fix: Internal error arguments for undefined functions are now
3439 computed correctly on the PPC/Linux platform.
3440 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3441 Raymond Toy's patch for CMU CL)
3442 * Support for the Solaris 9 operating environment has been included
3443 (thanks to Daniel Merritt)
3444 * A very ugly but hopefully complete draft of the missing FFI chapter
3445 of the manual has been created by reformatting the corresponding
3446 CMU CL manual chapter into (currently very ugly and incoherent)
3447 DocBook and bringing it up to date for SBCL behavior. Thus, the
3448 manual is now essentially complete, at least by my extreme
3449 once-and-only-once standards, whereby it's acceptable to refer to
3450 the doc strings of SB-EXT functions as the primary documentation.
3451 * The fasl file version number has changed again, due to cleanup of
3452 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3453 is no longer a static symbol.)
3455 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3456 * SBCL now builds with OpenMCL (version 0.12) as the
3457 cross-compilation host; also, more progress has been made toward
3458 bootstrapping under CLISP.
3459 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3461 * bug 158 fixed: The compiler can now deal with integer loop
3462 increments different from 1; fixing this turned out also to fix
3464 * bug 169 fixed: no more bogus warnings about using lexical bindings
3465 despite the presence of perfectly good SPECIAL declarations (thanks
3466 to David Lichteblau)
3467 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3468 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3469 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3471 * bug 180 fixed: Method combination specifications no longer ignore
3472 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3473 * bug fix: Structure type predicate functions now check their argument
3474 count as they should.
3475 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3476 correctly. (thanks to Pierre Mai)
3477 * minor incompatible change: The --noprogrammer option is deprecated
3478 in favor of the new --disable-debugger option, which is very similar.
3479 (The major difference is that it takes effect at a slightly different
3480 time at startup, causing handling of errors in --sysinit and
3481 --userinit files will be affected differently.) The
3482 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3483 been added to allow this functionality to be controlled from ordinary
3484 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3485 maintainers, who might want to run non-interactive scripts to
3486 build SBCL cores which will later be used interactively.)
3487 * minor incompatible change: The LOAD function no longer, when given
3488 a wild pathname to load, loads all files matching that pathname.
3489 Instead, an error of type FILE-ERROR is signalled.
3491 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3492 * bug 147 fixed: The compiler preserves its block link/count
3493 invariants more correctly now so that it doesn't crash. (thanks
3495 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3497 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3498 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3499 no-value-for-this-character as NIL (as specified by ANSI).
3500 * HOST-NAMESTRING on physical pathnames now returns a string that is
3501 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3502 (thanks to Christophe Rhodes)
3503 * The Alpha port handles icache flushing more correctly. (thanks to
3505 * More progress has been made toward bootstrapping under CLISP. (thanks
3506 to Christophe Rhodes)
3507 * The fasl file format has changed again, because dynamic loading
3508 on OpenBSD (which has non-ELF object files) motivated some cleanups
3509 in the way that foreign symbols are transformed and passed around.
3510 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3511 is no longer treated as whitespace by the reader, but instead as
3512 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3513 |A
\7fB|, instead of A as it used to.
3515 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3516 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3517 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3518 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3519 this as of 0.7.1.45, but was left out of the previous news section
3520 (thanks to Dan Barlow)
3521 * SBCL now runs on the Solaris operating system on SPARC architectures
3522 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3523 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3524 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3525 * SPARC backend cleanups, allowing builds of cores optimized for V8
3526 and V9 SPARCS, and also emission of code targeted to a particular
3527 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3529 * SBCL is closer to bootstrapping under CLISP, thanks to various
3530 fixes by Christophe Rhodes.
3531 * The fasl file format has changed again, to allow the compiler's
3532 INFO database to support symbol macros.
3533 * The user manual (in doc/) is formatted into HTML more nicely.
3534 (thanks to coreythomas)
3535 * The system is smarter about SUBTYPEP relationships, especially
3536 those involving NOT types (including types such as ATOM which are
3537 represented internally using NOT types). Thus SUBTYPEP is less
3538 likely to return (VALUES NIL NIL) in general, and in particular
3539 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3541 * The fasl file format has changed again, because the internal
3542 representation of types now includes a new slot to support the new
3543 SUBTYPEP-of-NOT-types logic.
3544 * (not a change in the main branch of SBCL, but a related prototype
3545 which can hopefully be merged into the main branch of SBCL in the
3546 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3547 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3548 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3549 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3550 dispatch macro character combinations. (thanks to Alexey Dejneka)
3551 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3552 that superficially look like logical namestrings correctly.
3553 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3555 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3556 default supplied-p) for &optional and &key arguments. (thanks to
3559 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3560 * incompatible change: The compiler is now less aggressive about
3561 tail call optimization, doing it only when (> SPACE DEBUG) or
3562 (> SPEED DEBUG). (This is an incompatible change because there are
3563 programs which relied on the old CMU-CL-style behavior to optimize
3564 away their unbounded recursion which will now die of stack overflow.)
3565 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3566 for non-GENCGC systems has been increased to 20M (since that
3567 seems much closer to the likely performance optimum for modern
3568 systems than the old 4M value was)
3569 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3570 and *DEBUG-PRINT-LEVEL*
3571 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3572 of CMU CL's support for SPARC, and various endianness and other
3573 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3574 * new syntactic sugar for the Unix command line: --load foo.bar is now
3575 an alternate notation for --eval '(load "foo.bar")'.
3577 ** The system now detects stack overflow and handles it gracefully,
3578 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3579 optimization settings. (This is a good thing in general, and
3580 its introduction in this version should be particularly timely
3581 for anyone whose code fails because of suppression of tail
3583 ** The system now hunts for the C variable "environ" in a more
3584 devious way, to avoid segfaults when the C library version differs
3585 between compile time and run time. (thanks to Christophe Rhodes)
3586 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3587 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3588 ** The compiler no longer issues bogus style warnings for undefined
3589 classes in the same source file as the DEFCLASSes which defined
3590 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3592 ** fixes in CONDITION class precedence list for undefined function
3593 errors (thanks to Alexei Dejneka)
3594 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3595 correctly. (thanks to Dan Barlow)
3596 ** portability fixes aiming at bootstrapping under CLISP (thanks
3597 to Dave McDonald and Christophe Rhodes)
3598 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3599 ** fixes in type translation and and type inference (thanks to
3601 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3602 ** various fixes in the new ports (thanks to Dan Barlow)
3603 * several changes related to debugging:
3604 ** suppression of tail recursion, as noted above
3605 ** stack overflow detection, as noted above
3606 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3607 is now the default. (For some time encapsulation has been more
3608 reliable than the breakpoint-based :ENCAPSULATE NIL
3609 implementation, at least on X86 systems; and I just noticed that
3610 encapsulation also seems closer to the spirit of the ANSI
3613 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3615 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3616 up properly again. (There was a packaging bug in 0.7.0 which
3617 left their definitions in SB-SYS::LOAD-FOREIGN and
3618 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3619 vital for most things which interface to C-level interfaces,
3620 like extensions working with sockets or databases or
3621 Perl-compatible regexes or whatever, and the need to fix
3622 this bug is the main reason that 0.7.1 was released so
3624 ** DEFGENERIC is now choosier about the methods it redefines, so that
3625 reLOADing a previously-LOADed file containing DEFGENERICs does
3626 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3627 cycle now works as it should. (thanks to Alexey Dejneka)
3628 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3629 fixed by Christophe Rhodes. (He actually submitted this patch
3630 months ago, and I delayed until after 0.7.0.)
3631 ** Bug 111 (internal compiler confusion about runtime checks on
3632 FUNCTION types) was fixed by Alexey Dejneka.
3633 * Some internal cleanups (getting rid of variables which aren't
3634 needed now that the byte interpreter is gone) caused the fasl
3635 file format number to change again.
3637 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3638 * major incompatible change: The default fasl file extension, i.e. the
3639 default extension for files produced by COMPILE-FILE, has changed
3640 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3642 ** There are many changes in the implementation of the compiler.
3643 SBCL is now essentially a compiler-only implementation of ANSI
3644 Common Lisp. EVAL still "interprets" a few special cases, but
3645 almost all the interesting cases are handled by creating
3646 a LAMBDA expression, calling COMPILE on it, then calling
3647 FUNCALL on the result.
3648 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3649 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3650 Since the code is newer, there might still be some new bugs
3651 (though not as many as before Martin Atzmueller's fixes:-). But
3652 the new code is substantially simpler and clearer, and hopefully
3653 any remaining bugs will be simpler, less fundamental, and more
3654 fixable then the bugs in the old code.
3655 ** The revised compiler is still a little unsteady on its feet.
3657 *** The debugging information it produces (particularly the names
3658 of FUNCTION objects) is sometimes much less useful than what
3659 the old compiler produced.
3660 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3661 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3662 within a MACROLET) has been temporarily weakened.
3663 ** There are new compiler optimizations for various functions:
3664 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3665 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3666 *** the math functions TRUNCATE, FLOOR, and CEILING
3667 *** the function-of-all-trades COERCE
3668 Mostly these should be transparent, but there's one
3669 potentially-annoying problem (bug 117): when the compiler
3670 inline-expands a function and does type analysis on the result,
3671 it can create control paths which have type mismatches, and
3672 when it can't prove that those control paths aren't taken,
3673 it will issue WARNINGs about the type mismatches. This is
3674 a particular problem in practice for the new sequence functions.
3675 It's not clear how this should be fixed, and for now, a
3676 workaround is given in the entry for 117 in the BUGS file.
3677 ** (Because of the interaction between the two previous items --
3678 occasional inlining problems and new inline expansions -- some
3679 of the new sequence function optimizations won't really kick in
3680 completely until debugging information, and then inlining, are
3681 straightened out in some future version.)
3682 * minor incompatible changes:
3683 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3684 DIRECTORY no longer implicitly promotes NIL slots of its
3685 pathname argument to :WILD. In particular, when you ask for the
3686 contents of a directory (which you used to be able to do without
3687 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3688 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3689 ** changes in behavior that ANSI explicitly defines to be
3690 implementation dependent:
3691 *** The new compiler-only implementation still conforms with ANSI,
3692 but acts a little different than before. Besides the obvious
3693 changes in performance tradeoffs (that the cost per form passed
3694 to EVAL has gone up, and the cost per form executed by EVAL
3695 has gone down), the behavior of the system changes a little
3696 because there are no longer any interpreted function objects.
3697 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3698 e.g. doing COMPILE on the output of interactive DEFUN is
3700 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3702 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3703 has changed from (:CL) to NIL.
3704 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3705 e.g. #\Newline and #\Tab, don't have names specified in the
3706 ANSI Common Lisp standard, is now based on their ASCII symbolic
3707 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3708 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3709 are no longer used for output.
3710 ** changes in internal implementation constants:
3711 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3712 4 million. (If your application spends a lot of time GCing and
3713 you have a lot of RAM, you might want to experiment with
3714 increasing it even more.)
3715 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3716 However, almost all old code should still continue to work without
3717 immediate update, as SB-C-CALL is now a (deprecated) nickname
3719 ** Old operator names in the style DEF-FOO are now deprecated in
3720 favor of new corresponding names DEFINE-FOO, for consistency with
3721 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3722 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3723 internal symbols, but a few supported extensions like
3724 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3725 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3726 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3727 etc. as you get deeper into recursive calls to the debugger
3728 command loop, instead of the old "5]", "5]]", "5]]]"
3729 sequence. (I was motivated to do this when squabbles between
3730 ILISP and SBCL left me very deeply nested in the debugger. In the
3731 short term, this change will probably provoke more ILISP/SBCL
3732 squabbles, but hopefully it will be an improvement in the long run.)
3733 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3734 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3735 function, since declaiming return types involving aliens is
3736 (1) annoyingly messy to do by hand and (2) vital to efficient
3737 compilation of code which calls such functions.
3738 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3739 longer reexported by the SB-EXT package. They're solely useful
3740 for alien code, so it seems more logical that you should get
3741 them from the SB-ALIEN package, not in SB-EXT.
3742 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3743 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3744 features. Instead, the code that they used to control is always
3745 built into the system.
3746 * many other bug fixes
3747 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3748 advantage of the new EVAL-WHEN stuff and to clean them up in
3749 general, and they are now more ANSI-compliant in a number of
3750 ways. Martin Atzmueller is responsible for a lot of this.
3751 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3753 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3754 *** correct ERROR type for various file operations
3755 *** some fixes for Lisp streams
3756 *** DEFMETHOD syntax checking
3757 *** changing old weird representation of debug information as
3758 strings (which, among their other deficiencies, don't transform
3759 correctly when you rename packages, and don't change their
3760 print representation when you change things like *PACKAGE*
3761 and *PRINT-LENGTH*) to symbols and lists of symbols
3762 He also made several improvements and fixed several bugs in DESCRIBE.
3763 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3765 *** misbehavior of WRITE-STRING/WRITE-LINE
3766 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3767 and several other LOOP problems as well
3768 *** DIRECTORY when similar filenames are present
3769 *** DEFGENERIC with :METHOD options
3770 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3771 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3772 *** argument ordering in FIND with :TEST option
3773 *** mishandled package designator argument in APROPOS-LIST
3774 *** various problems in the backquote readmacro
3775 *** a bug in APROPOS
3776 *** probably some others that I'm not describing very well here,
3777 since the CVS log documents them by reference to sbcl-devel
3778 messages, and the SourceForge archives aren't working well.:-(
3779 ** Dan Barlow improved the Alpha port (and is making progress on the
3780 PPC port, for those of you who think different).
3781 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3782 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3783 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3784 continues to work on the SPARC port (for those of you in a position
3785 to look down upon our little PC-compatible boxes from a great height).
3786 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3787 and uses that copy, rather than the current dynamic value, when
3788 it's trying to decide whether to truncate output. Thus e.g.
3789 (let ((*print-lines* 50))
3790 (pprint-logical-block (stream nil)
3792 (let ((*print-lines* 8))
3793 (print (aref possiblybigthings i) stream)))))
3794 should now truncate the logical block only at 50 lines, instead of
3795 often truncating it at 8 lines, as it did before.
3796 * The doc/cmucl/ directory, containing old CMU CL documentation
3797 from the time of the fork, is no longer part of the base system.
3798 SourceForge has shut down its anonymous FTP service, and with it
3799 my original plan for distributing the old CMU CL documentation
3800 there. For now, if you need these files you can download an old
3801 SBCL source release and extract them from it.
3802 * The fasl file version number changed again, for dozens of reasons,
3803 some of which are apparent above.
3805 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3806 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3807 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3808 patch, so that DISASSEMBLE output is much nicer.
3809 * The code in the SB-PROFILE package now seems reasonably stable.
3810 I still haven't decided what the final interface should look like
3811 (I'd like PROFILE to interact cleanly with TRACE, since both
3812 facilities use function encapsulation) but if you have a need
3813 for profiling now, you can probably use it successfully with
3814 the current CMU-CL-style interface.
3815 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3816 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3817 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3818 link now returns the dangling link itself, and for similar
3819 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3820 link itself. (In these cases the old code signalled an error and
3821 looped endlessly, respectively.) Thus, DIRECTORY now works even
3822 in the presence of dangling and cyclic symbolic links.
3823 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3824 is now a supported extension again, since the consensus on
3825 sbcl-devel was that it can be useful for ordinary development
3826 work, not just for debugging SBCL itself.
3827 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3828 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3829 that currently-defined functions might be redefined later with
3830 different return types.
3831 * Hash tables can be printed readably, as inspired by CMU CL code
3832 of Eric Marsden and SBCL code of Martin Atzmueller.
3833 * better error handling in CLOS method combination, thanks to
3834 Martin Atzmueller porting Pierre Mai's CMU CL patches
3835 * more overflow fixes for >16Mbyte I/O buffers
3836 * A bug in READ has been fixed, so that now a single Ctrl-D
3837 character suffices to cause end-of-file on character streams.
3838 In particular, now you only need one Ctrl-D at the command
3839 line (not two) to exit SBCL.
3840 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3842 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3843 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3844 does the right thing.
3845 * The implementation of some type tests, especially for CONDITION
3846 types, is now tidier and maybe faster, due to CMU CL code
3847 originally by Douglas Crosher, ported by Martin Atzmueller.
3848 * Some math functions have been fixed, and there are new
3849 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3850 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3851 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3852 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3853 likely to be built into sbcl-0.7.0.)
3854 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3855 CL, ported to SBCL by Martin Atzmueller)
3856 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3857 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3858 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3859 * There's a new slam.sh hack to shorten the edit/compile/debug
3860 cycle for low-level changes to SBCL itself, and a new
3861 :SB-AFTER-XC-CORE target feature to control the generation of
3862 the after-xc.core file needed by slam.sh.
3863 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3864 COMPILE-FILE is no longer supported, so that now every function
3865 gets an entry point, so that block compilation looks a little
3866 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3867 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3868 returns the number of bytes consed since the system started,
3869 rather than the number consed since the first time the function
3870 was called. (The new definition parallels ANSI functions like
3871 CL:GET-INTERNAL-RUN-TIME.)
3872 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3873 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3874 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3875 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3876 * Fasl file version numbers are now independent of the target CPU,
3877 since historically most system changes which required version
3878 number changes have affected all CPUs equally. Similarly,
3879 the byte fasl file version is now equal to the ordinary
3882 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3883 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3884 is no longer recognized. I apologize for this, because it was
3885 listed in SB-EXT as a supported extension, but I found that
3886 its existing behavior was poorly specified, as well as incorrectly
3887 specified, and it looked like too much of a mess to straighten it
3888 out. I have enough on my hands trying to get ANSI stuff to work..
3889 * many patches ported from CMU CL by Martin Atzmueller, with
3890 half a dozen bug fixes in pretty-printing and the debugger, and
3891 half a dozen others elsewhere
3892 * fixed bug 13: Floating point infinities are now supported again.
3893 They might still be a little bit flaky, but thanks to bug reports
3894 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3895 as flaky as they were.
3896 * The --noprogrammer command line option is now supported. (Its
3897 behavior is slightly different in detail from what the old man
3898 page claimed it would do, but it's still appropriate under the
3899 same circumstances that the man page talks about.)
3900 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3901 are now supported, and enabled by default. Thus, the compiler can
3902 handle many floating point and complex operations much less
3903 inefficiently. (Thus e.g. you can implement a complex FFT
3905 * The compiler now detects type mismatches between DECLAIM FTYPE
3906 and DEFUN better, and implements CHECK-TYPE more correctly, and
3907 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3908 thanks to patches from Martin Atzmueller.
3909 * various fixes to make the cross-compiler more portable to
3910 ANSI-conforming-but-different cross-compilation hosts (notably
3911 Lispworks for Windows, following bug reports from Arthur Lemmens)
3912 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3913 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3914 fixed thanks to Pierre Mai's CMU CL patches.
3915 * fixes to keep the system from overflowing internal counters when
3916 it tries to use i/o buffers larger than 16M bytes
3917 * fixed bug 45a: Various internal functions required to support
3918 complex special functions have been merged from CMU CL sources.
3919 (When I was first setting up SBCL, I misunderstood a compile-time
3920 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3921 * improved support for type intersection and union, fixing bug 12
3922 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3923 more obscure bugs as well
3924 * some steps toward byte-compiling non-performance-critical
3925 parts of the system, courtesy of patches from Martin Atzmueller
3926 * Christophe Rhodes has made some debian packages of sbcl at
3927 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3928 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3929 stable, but are nonetheless usable. When he's ready, I'd be happy
3930 to add them to the SourceForge "File Releases" section. (And if
3931 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3932 * new fasl file format version number (because of changes in
3933 internal representation of (OR ..) types to accommodate the new
3934 support for (AND ..) types, among other things)
3936 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3937 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3939 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3941 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3942 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3943 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3944 * DESCRIBE now works on structure objects again.
3945 * Most function call argument type mismatches are now handled as
3946 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3947 know whether the function will be redefined before the call is
3948 executed. (The compiler could flag local calls with full WARNINGs,
3949 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3950 it doesn't keep track of enough information to know whether calls
3951 are local in this sense.)
3952 * Compiler output is now more verbose, with messages truncated
3953 later than before. (There should be some supported way for users
3954 to override the default verbosity, but I haven't decided how to
3955 provide it yet, so this behavior is still controlled by the internal
3956 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3957 src/compiler/ir1util.lisp.)
3958 * Fasl file format version numbers have increased again, because
3959 support for the Gray streams extension changes the layout of the
3960 system's STREAM objects.
3961 * The Gray subclassable streams extension now works, thanks to a
3962 patch from Martin Atzmueller.
3963 * The full LOAD-FOREIGN extension (not just the primitive
3964 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3965 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3966 but like most other programs, it defaults to copying the Unix
3967 environment from the original process instead of starting the
3968 new process in an empty environment.
3969 * Extensions which manipulate the Unix environment now support
3970 an :ENVIRONMENT keyword option which doesn't smash case or
3971 do other bad things. The CMU-CL-style :ENV option is retained
3972 for porting convenience.
3973 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3974 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3976 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3978 * A patch from Martin Atzmueller seems to have solved the SIGINT
3979 problem, and as far as we know, signal-handling now works cleanly.
3980 (If you find any new bugs, please report them!)
3981 * The system no longer defaults Lisp source file names to types
3982 ".l", ".cl", or ".lsp", but only to ".lisp".
3983 * The compiler no longer uses special default file extensions for
3984 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3985 seems to expect a single default extension for all compiled code,
3986 and there's no compelling reason to try to stretch the standard
3987 to allow two different extensions.) Instead, byte-compiled files
3988 default to the same extension as native-compiled files.
3989 * Fasl file format version numbers have increased again, because
3990 a rearrangement of internal implementation packages made some
3991 dumped symbols in old fasl files unreadable in new cores.
3992 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3993 many fewer weird special cases.
3994 * Bug #17 (differing COMPILE-FILE behavior between logical and
3995 physical pathnames) has been fixed, and some related misbehavior too,
3996 thanks to a patch from Martin Atzmueller.
3997 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3998 by Tim Moore, ported to SBCL by Martin Atzmueller.
3999 * Martin Atzmueller fixed several filesystem-related problems,
4000 including bug #36, in part by porting CMU CL patches, which were
4001 written in part by Paul Werkowski.
4002 * More compiler warnings in src/runtime/ are gone, thanks to
4003 more patches from Martin Atzmueller.
4004 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4007 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4009 * DESCRIBE now works on CONDITION objects.
4010 * The debugger now handles errors which arise when trying to print
4011 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4013 * The build system now uses an additional file, customize-target-features.lisp,
4014 to allow local modifications to the target *FEATURES* list. (The point of
4015 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4016 debugging features enabled, without having to worry about propagating it
4017 into everyone's system when I do a "cvs update".) When no
4018 customize-target-features.lisp file exists, the target *FEATURES* list
4019 should be constructed the same way as before.
4020 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4021 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4022 warn about behavior which is undefined under the ANSI spec. Note: This
4023 is specified by ANSI, but it's not very popular with programmers.
4024 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4025 macro in the SBCL sources for an example of a workaround which you
4026 might use to make portable ANSI-standard code which does what you want.
4027 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4028 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4029 when it's used as a non-toplevel form. (This is required in order
4030 to implement the DEFCONSTANT-EQX macro.)
4031 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4032 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4033 funny things, due to bugs in the implementation of EVAL-WHEN
4034 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4035 non-toplevel DEFCONSTANTs are uncommon.)
4036 * The core file version number and fasl file version number have been
4037 incremented, because the old noncompliant DEFCONSTANT behavior involved
4038 calling functions which no longer exist, and because I also took the
4039 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4040 * fixed bug 1 (error handling before read-eval-print loop starts), and
4041 redid debugger restarts and related debugger commands somewhat while
4043 ** The QUIT debugger command is gone, since it did something
4044 rather different than the SB-EXT:QUIT command, and since it never
4045 worked properly outside the main toplevel read/eval/print loop.
4046 Invoking the new TOPLEVEL restart provides the same functionality.
4047 ** The GO debugger command is also gone, since you can just invoke
4048 the CONTINUE restart directly instead.
4049 ** The TOP debugger command is also gone, since it's redundant with the
4050 FRAME 0 command, and since it interfered with abbreviations for the
4052 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4054 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4055 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4056 some porting work by Martin Atzmueller)
4057 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4058 with a boatload of patches which he ported to SBCL. Now that those
4060 ** The system tries to make sure that its low-priority messages
4061 are prefixed by semicolons, to help people who like to use
4062 syntax highlighting in their ILISP buffer. (This patch
4063 was originally due to Raymond Toy.)
4064 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4065 patches originally written by Raymond Toy.
4066 ** The compiler understands coercion between single-value and
4067 multiple-VALUES type expressions better, getting rid of some very
4068 weird behavior, thanks to patches originally by Robert MacLachlan
4069 and Douglas Crosher.
4070 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4071 lambda lists, thanks to a patch originally by Pierre Mai.
4072 ** The system no longer bogusly warns about "abbreviated type
4074 ** The compiler gets less confused by inlining and RETURN-FROM,
4075 thanks to some patches originally by Tim Moore.
4076 ** The system no longer hangs when dumping circular lists to fasl
4077 files, thanks to a patch originally from Douglas Crosher.
4078 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4079 undefined function error.
4080 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4081 in structure slot initforms) for now, documented workaround instead:-|
4082 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4083 * fixed bug 5: added stubs for various Gray stream functions called
4084 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4085 installed, at least appropriate type errors are generated
4086 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4087 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4088 consistently in DEFMETHOD forms.
4089 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4090 it doesn't seem to affect SBCL after all
4091 * The C runtime system now builds with better optimization and many
4092 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4094 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4096 * The system is now under CVS at SourceForge (instead of the
4097 CVS repository on my home machine).
4098 * The new signal handling code has been tweaked to treat register
4099 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4100 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4101 downstream, e.g. in debug-int.lisp, has implicit dependencies
4102 on the unsignedness of integer representation of machine words,
4103 and that caused the system to bomb out with infinite regress
4104 when trying to recover from type errors involving signed values,
4105 e.g. (BUTLAST '(1 2 3) -1).)
4106 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4107 (This was one of the bugs Peter Van Eynde reported back in July.)
4108 * The system now uses code inspired by Colin Walters' O(N)
4109 implementation of MAP (from the cmucl-imp@cons.org mailing
4110 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4111 inline the MAP operation, and there is more than one
4112 sequence argument to the MAP call (so that it can't just
4113 do ETYPECASE once and for all based on the type of the
4114 single sequence argument). (The old non-inline implementation
4115 of the general M-argument sequence-of-length-N case required
4116 O(M*N*N) time when any of the sequence arguments were LISTs.)
4117 * The QUIT :UNIX-CODE keyword argument has been renamed to
4118 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4119 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4121 * Raymond Wiker's patches to port dynamic loading from Linux to
4122 FreeBSD have been added.
4123 * The BUGS file is now more nearly up to date, thanks in large part
4124 to Martin Atzmueller's review of it.
4125 * The debugger now flushes standard output streams before it begins
4126 its output ("debugger invoked" and so forth).
4127 * The core version number and fasl file version number have both
4128 been incremented, because of incompatible changes in the layout
4130 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4131 used to not be called for a saved Lisp image.)
4132 * Martin Atzmueller's version of a patch to fix a compiler crash,
4133 as posted on sbcl-devel 13 September 2000, has been installed.
4134 * Instead of installing Martin Atzmueller's patch for the
4135 compiler transform for SUBSEQ, I deleted the compiler transform,
4136 and transforms for some similar consing operations.
4137 * A bug in signal handling which kept TRACE from working on OpenBSD
4139 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4140 away by constant folding
4141 * The system now defines its address space constants in one place
4142 (in the Lisp sources), and propagates them automatically elsewhere
4143 (through GENESIS and the sbcl.h file). Therefore, patching the
4144 address map is less unnecessarily tedious and error-prone. The
4145 Lisp names of address space constants have also been systematized.
4146 * CVS tags like dollar-Header-dollar have been removed from
4147 the sources, because they have never saved me trouble and
4148 they've been source of trouble working with patches and other
4149 diff-related operations.
4150 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4151 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4153 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4155 * The system has been ported to OpenBSD.
4156 * The system now compiles with a simple "sh make.sh" on the systems
4157 that it's supported on. I.e., now you no longer need to tweak
4158 text in the target-features.lisp-expr and symlinks in src/runtime/
4159 by hand, the make.sh takes care of it for you.
4160 * The system is no longer so grossly inefficient when compiling code
4161 involving vectors implemented as general (not simple) vectors (VECTOR T),
4162 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4163 which dares to use the various sequence functions on non-simple
4164 vectors, takes less of a performance hit.
4165 * There is now a primitive type predicate VECTOR-T-P
4166 to test for the (VECTOR T) type, so that e.g.
4167 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4168 can now be compiled with some semblance of efficiency. (The old code
4169 turned the type declaration into a full call to %TYPEP at runtime!)
4170 * AREF on (VECTOR T) is still not fast, since it's still compiled
4171 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4172 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4173 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4174 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4175 without making full calls to SUBTYPEP at runtime.
4176 (Some analogous efficiency issues for non-simple vectors specialized to
4177 element types other than T, or for non-simple multidimensional arrays,
4178 have not been addressed. They could almost certainly be handled the
4179 same way if anyone is motivated to do so.)
4180 * The changes in array handling break binary compatibility, so
4181 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4182 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4183 instead of (VALUES T T).
4184 * By following the instructions that Dan Barlow posted to sbcl-devel
4185 on 2 July 2000, I was able to enable primitive dynamic object
4186 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4187 functionality is not implemented (since it calls ld to resolve
4188 library references automatically, requiring RUN-PROGRAM for its
4189 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4190 try to resolve library references) is now supported.
4191 * The system now flushes the standard output streams when it terminates,
4192 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4193 them at several other probably-convenient times, e.g. in each pass of
4194 the toplevel read-eval-print loop, and after evaluating a form given
4195 as an "--eval" command-line option. (These changes were motivated by a
4196 discussion of stream flushing issues on cmucl-imp in August 2000.)
4197 * The source transform for TYPEP of array types no longer assumes
4198 that an array whose element type is a not-yet-defined type
4199 is implemented as an array of T, but instead punts, so that the
4200 type will be interpreted at runtime.
4201 * There is now some support for cross-compiling in make.sh: each of
4202 the phases of make.sh has its own script. (This should be transparent
4203 to people doing ordinary, non-cross-compile builds.)
4204 * Since my laptop doesn't have hundreds of megabytes of memory like
4205 my desktop machine, I became more motivated to do some items on
4206 my to-do list in order to reduce the size of the system a little:
4207 ** Arrange for various needed-only-at-cold-init things to be
4208 uninterned after cold init. To support this, those things have
4209 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4210 symbols with such names are now uninterned after cold init).
4211 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4212 fasl files for cold load.
4213 ** Remove the old compiler structure pooling code (which used to
4214 be conditional on the target feature :SB-ALLOC) completely.
4215 ** Redo the representation of some data in cold init to be more compact.
4216 (I also looked into supporting byte compiled code at bootstrap time,
4217 which would probably reduce the size of the system a lot, but that
4218 looked too complicated, so I punted for now.)
4219 * The maximum signal nesting depth in the src/runtime/ support code has
4220 been reduced from 4096 to 256. (I don't know any reason for the very
4221 large old value. If the new smaller value turns out to break something,
4222 I'll probably just bump it back up.)
4223 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4225 * Many, many bugs reported by Peter Van Eynde have been added to
4226 the BUGS list; some have even been fixed.
4227 * While enabling dynamic object file loading, I tried to make the
4228 code easier to understand, renaming various functions and variables
4229 with less ambiguous names, and changing some function calling
4230 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4231 * While trying to figure out how to do the OpenBSD port, I tried to
4232 clean up some of the code in src/runtime/. In particular, I dropped
4233 support for non-POSIX signal handling, added various comments,
4234 tweaked the code to reduce the number of compilation warnings, and
4235 renamed some files to increase consistency.
4236 * To support the new automatic configuration functionality in make.sh,
4237 the source file target-features.lisp-expr has been replaced with the
4238 source file base-target-features.lisp-expr and the machine-generated
4239 file local-target-features.lisp-expr.
4240 * fixed a stupid quoting error in make.sh so that using CMU CL
4241 "lisp -batch" as cross-compilation host works again
4243 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4245 * DESCRIBE no longer tries to call itself recursively to describe
4246 bound/fbound values, so that it no longer fails on symbols which are
4247 bound to themselves (like keywords, T, and NIL).
4248 * DESCRIBE now works on generic functions.
4249 * The printer now prints less-screwed-up representations of closures
4250 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4251 * A private symbol is used instead of the :EMPTY keyword previously
4252 used to mark empty slots in hash tables. Thus
4253 (DEFVAR *HT* (MAKE-HASH-TABLE))
4254 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4255 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4256 now does what ANSI says that it should. (You can still get
4257 similar noncompliant behavior if bang on the hash table
4258 implementation with all the symbols you get back from
4259 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4260 This breaks binary compatibility, since tests for equality to
4261 :EMPTY are wired into things like the macroexpansion of
4262 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4264 * There's now a minimal placeholder implementation for CL:STEP,
4265 as required by ANSI.
4266 * An obscure bug in the interaction of the normal compiler, the byte
4267 compiler, inlining, and structure predicates has been patched
4268 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4269 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4271 * Missing ordinary arguments in a macro call are now detected even
4272 when the macro lambda list contains &KEY or &REST.
4273 * The debugger no longer complains about encountering the top of the
4274 stack when you type "FRAME 0" to explicitly instruct it to go to
4275 the top of the stack. And it now prints the frame you request even
4276 if it's the current frame (instead of saying "You are here.").
4277 * As specified by ANSI, the system now always prints keywords
4278 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4280 * The default initial SIZE of HASH-TABLEs is now smaller.
4281 * Type information from CLOS class dispatch is now propagated
4282 into DEFMETHOD bodies, so that e.g.
4283 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4285 is now basically equivalent to
4286 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4287 (DECLARE (TYPE SINGLE-FLOAT X))
4289 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4290 operation, without having to do run-time type dispatch.
4291 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4292 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4293 * Since I don't seem to be making big file reorganizations very often
4294 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4295 is overflowing my ability to conveniently back them up), I've finally
4296 checked the system into CVS. (The CVS repository is on my home system,
4297 not at SourceForge -- putting it on SourceForge might come later.)
4298 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4299 high-level GC-NOTIFY-FOO functions send their output. (There's
4300 still very little control of where low-level verbose GC functions
4301 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4302 less than it used to -- the GC-NOTIFY-FOO functions are now under
4303 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4304 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4305 in only one place in the source code, and propagates it automatically
4306 everywhere that it's needed. Thus e.g. when I bump the version from
4307 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4308 * The C source files now include boilerplate legalese and documentation
4309 at the head of each file (just as the Lisp source files already did).
4310 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4311 to his page will be replaced with a link to his new CLiki service.
4313 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4315 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4316 * The build process now looks for GNU make under the default name "gmake",
4317 instead of "make" as it used to. If GNU make is not available as "gmake"
4318 on your system, you can change this default behavior by setting the
4319 GNUMAKE environment variable.
4320 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4321 can build without error under CMU CL.
4323 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4325 * There is now a partial SBCL user manual (with some new text and some
4326 text cribbed from the CMU CL manual).
4327 * The beginnings of a profiler have been added (starting with the
4328 CMU CL profiler and simplifying and cleaning up). Eventually the
4329 main interface should be through the TRACE macro, but for now,
4330 it's still accessed through vaguely CMU-CL-style functions and macros
4331 exported from the package SB-PROFILE.
4332 * Some problems left over from porting CMU CL to the new
4333 cross-compilation bootstrap process have been cleaned up:
4334 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4335 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4336 part when getting rid of the compiler *BACKEND* stuff.)
4337 ** The value of *NULL-TYPE* was screwed up, because it was
4338 being initialized before the type system knew the final
4339 definition of the 'NULL type. This screwed up several key
4340 optimizations in the compiler, causing inefficiency in all sorts
4341 of places. (I found it because I wanted to understand why
4342 GET-INTERNAL-RUN-TIME was consing.)
4343 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4344 PROCLAIM FTYPE information. Unfortunately this broke binary
4345 compatibility again, since now the forms output by DEFGENERIC
4346 to refer to functions which didn't exist in 0.6.3.
4347 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4348 can use the new (as of 0.6.3) transform for SOME into MAP into
4350 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4351 values to (INTEGER 0 1000000), so that the time code will no longer
4352 occasionally get blown up by Linux returning 1000000 microseconds
4353 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4354 its output conform to the ANSI spec. (Alas, this makes its output
4355 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4356 * A full call to MAP NIL with a single sequence argument no longer conses.
4357 * fixes to problems pointed out by Martin Atzmueller:
4358 * The manual page no longer talks about multiprocessing as though
4359 it were currently supported.
4360 * The ILISP support patches have been removed from the distribution,
4361 because as of version 5.10.1, ILISP now supports SBCL without us
4362 having to maintain patches.
4363 * added a modified version of Raymond Toy's recent CMU CL patch for
4364 EQUALP comparison of HASH-TABLE
4366 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4368 * The system still can't cross-compile itself with
4369 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4370 entails), but at least it can compile more of itself that way
4371 than it used to be able to, and various buglets which were uncovered
4372 by trying to cross-compile itself that way have now been fixed.
4373 * This release breaks binary compatibility again. This time
4374 at least I've incremented the FASL file format version to 2, so that the
4375 problem can be detected reliably instead of just causing weird errors.
4376 * various new style warnings:
4377 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4378 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4379 ** using the lexical binding of a variable named in the *FOO* style
4380 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4381 as specified by ANSI.
4382 * *RANDOM-STATE* is no longer automatically initialized from
4383 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4384 default behavior of the system is to repeat its behavior every time
4385 it's run. If you'd like to change this behavior, you can always
4386 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4387 old convention there was no comparably easy way to get the system to
4388 repeat its behavior every time it was run.
4389 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4390 FTYPE declarations has been removed, in favor of their ANSI
4391 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4392 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4393 the types of their sequence arguments can be determined at compile time.
4394 This is done through a new open code expansion for MAP which eliminates
4395 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4396 argument types can be determined at compile time.
4397 * The optimizer now transforms COERCE into an identity operation when it
4398 can prove that the coerced object is already of the correct type. (This
4399 can be a win for machine generated code, including the output of other
4400 optimization transforms, such as the MAP transform above.)
4401 * Credit information has been moved from source file headers into CREDITS.
4402 * Source file headers have been made more standard.
4403 * The CASE macro now compiles without complaining even when it has
4406 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4408 * (Note that the way that the PCL macroexpansions were rewritten
4409 to accommodate the change in DEFGENERIC below breaks binary
4410 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4411 not run under sbcl-0.6.2. Once we get out of alpha releases,
4412 i.e. hit release 1.0.0, we'll probably try to maintain binary
4413 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4414 and sbcl-1.4.4. Until then, however, it might be fairly common
4415 for maintenance releases to break binary compatibility.)
4416 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4418 * The reporting of unhandled signals has been changed to print some
4419 explanatory text as well as the report form. (Previously only
4420 the report form was printed.)
4421 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4422 it defines, so that the compiler no longer issues undefined function
4423 warnings for compiled-but-not-yet-loaded generic functions.
4424 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4425 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4426 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4427 section of the ANSI spec are supported.
4428 * The "" nickname for the "KEYWORD" package has been removed.
4429 The reader still handles symbol tokens which begin with a package marker
4430 as keywords, but it doesn't expose its mechanism for doing so in the
4431 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4432 * The system now issues STYLE-WARNINGs for contradictory TYPE
4433 proclamations. (Warnings for contradictory FTYPE proclamations would
4434 be nice too, but those can't be done usefully unless the type system
4435 is made smarter about FUNCTION types.)
4436 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4437 systematized, so that "*target-*.lisp is supposed to exist only on the
4438 target and imply that there's a related file which exists on the
4439 host, and *host-*.lisp is supposed to exist only on the host and imply
4440 that there's a related file which exists on the target. This involves a
4441 lot of renaming. Hopefully the acute confusion caused by the renaming
4442 will be justified by the reduction in chronic confusion..
4443 ** runtime-type.lisp -> early-target-type.lisp
4444 ** target-type.lisp -> late-target-type.lisp
4445 ** early-host-format.lisp -> early-format.lisp
4446 ** late-host-format.lisp -> late-format.lisp
4447 ** host-error.lisp -> misc-error.lisp
4448 ** early-error.lisp -> early-target-error.lisp
4449 ** late-error.lisp -> late-target-error.lisp
4450 ** host-defboot.lisp -> early-defboot.lisp
4451 ** code/misc.lisp -> code/target-misc.lisp
4452 ** code/host-misc.lisp -> code/misc.lisp
4453 ** code/numbers.lisp -> code/target-numbers.lisp
4454 ** code/early-numbers.lisp -> numbers.lisp
4455 ** early-host-type.lisp -> early-type.lisp
4456 ** late-host-type.lisp -> late-type.lisp
4457 ** host-typep.lisp -> typep.lisp
4458 ** load.lisp -> target-load.lisp
4459 ** host-load.lisp -> load.lisp
4460 ** host-disassem.lisp -> disassem.lisp
4461 ** host-insts.lisp -> insts.lisp
4462 ** byte-comp.lisp -> target-byte-comp.lisp
4463 ** host-byte-comp.lisp -> byte-comp.lisp
4464 ** host-signal.lisp -> signal.lisp
4465 ** host-defstruct.lisp -> defstruct.lisp
4466 ** late-target-type.lisp -> deftypes-for-target.lisp
4467 Furthermore, several other previously target-only files foo.lisp (e.g.
4468 hash-table.lisp and random.lisp) have been split into a target-and-host
4469 foo.lisp file and a target-only target-foo.lisp file, with their key type
4470 definitions in the target-and-host part, so that the cross-compiler will
4471 know more about target types.
4472 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4473 gone away. In their place are various *BACKEND-FOO* variables
4474 corresponding to the slots of the old structure.
4475 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4476 into the target SBCL has been fixed.
4477 * The chill.lisp system for loading cold code into a running SBCL
4479 * Support for the CMU CL "scavenger hook" extension has been removed.
4480 (It was undocumented and unused in the CMU CL sources that SBCL was
4481 derived from, and stale in sbcl-0.6.1.)
4482 * Various errors in the cross-compiler type system were detected
4483 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4484 (enabling various consistency checks). Many of them were fixed,
4485 but some hard problems remain, so the compiler is back to
4486 running without *TYPE-SYSTEM-INITIALIZED* for now.
4487 * As part of the cross-compiler type system cleanup, I implemented
4488 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4489 * I have started adding UNCROSS calls throughout the type system
4490 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4491 on cross-compiler input files will be able to go away, and various
4493 * CONSTANTP now returns true for quoted forms (as explicitly required
4496 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4498 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4499 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4500 (MAKE-LIST -1). In the longer term, it ought to become true
4501 that declarations are assertions even at SAFETY 1. For now, it's not
4502 quite true even at SAFETY 3, but it's at least more nearly true..
4503 (Note that this change seems to increases the size of the system by
4504 O(5%) and to decrease the speed of the compiler by 20% or more.)
4505 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4506 for the problem of printing dozens of lines of distracting information
4507 about low-level system machinery as part of the top stack frame
4508 on entry to the debugger when an undefined function was called.
4509 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4510 is not reset to COMMON-LISP-USER.
4511 * Compilation of stuff related to dyncount.lisp has been made conditional
4512 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4513 smaller. The various dyncount-related symbols have been moved into
4514 a new "SB-DYNCOUNT" package.
4515 * tty-inspect.lisp has been renamed to inspect.lisp.
4516 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4517 feature has gone away. (When we eventually port to other flavors of
4518 libc and/or Unix, we'll try to make the differences between flavors
4519 invisible at the user level.)
4520 * Various other *FEATURES* tags, and/or their associated conditionals,
4521 have been removed if obsolescent, or given better documentation, or
4522 sometimes given more-mnemonic names.
4524 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4526 * tidied up "make.sh" script
4527 * tidied up system directory structure
4528 * better "clean.sh" behavior
4529 * added doc/FOR-CMUCL-DEVELOPERS
4530 * many many small tweaks to output format, e.g. removing possibly-confusing
4531 trailing #\. character in DESCRIBE-INSTANCE
4532 * (EQUALP #\A 'A) no longer signals an error.
4533 * new hashing code, including EQUALP hashing
4534 * tidied up Lisp initialization and toplevel
4535 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4536 * command line argument processing
4537 * added POSIX-GETENV function to deal with Unix-ish environment variables
4538 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4539 terminating SBCL on EOF
4540 * non-verbose GC by default
4541 * There is no more "sbcl" shell script; the sbcl file is now the C
4542 runtime executable (just like CMU CL).
4543 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4545 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4546 symbol and package information as Lisp native symbol and package objects
4547 instead of strings naming symbols and strings naming packages. This way,
4548 whenever packages are renamed (as in warm init), debug information is
4549 transformed along with everything else.
4550 * tweaked the optimization policy declarations which control the building
4551 of SBCL itself. Now, among other things, the system no longer saves
4552 source location debugging information. (This helps two problems at once
4553 by reducing SBCL size and by keeping SBCL from trying to look for its
4554 sources -- which may not exist -- when reporting errors.)
4555 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4556 debugging and testing purposes
4557 * cleaned up printing, making the printer call PRINT-OBJECT for
4558 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4559 methods, giving nearly-ANSI behavior
4560 * converted almost all special variables to use *FOO* naming convention
4561 * deleted PARSE-TIME functionality, since it can be done portably
4562 * moved some files out of cold init into warm init
4563 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4565 * regularized formatting of source files
4566 * added an install.sh script
4567 * fixed ridiculous memory usage of cross-compiler by making
4568 compiler/alloc.lisp not try to do pooling unless it can hook
4569 itself into the GC of the cross-compilation host. Now the system
4570 builds nicely on my old laptop.
4571 * added :SB-ALLOC in target-features.lisp-expr
4572 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4573 was not implemented)
4574 * re-did condition handling and note reporting in the compiler. Notes
4575 are no longer handled by signalling conditions. Style warnings
4576 and warnings are handled more correctly and reported in such a way
4577 that it's easy to find one or the other in your output (so that you
4578 can e.g. figure out which of many problems caused COMPILE-FILE to
4580 * changed the severity of several compiler warnings from full WARNING
4581 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4582 compiler note reporting so that it doesn't use the condition system
4583 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4584 COMPILE-FILE command)
4585 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4586 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4587 no longer supported, since their implementation was deeply intertwingled
4588 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4589 * removed bogus "support" for compiler macros named (SETF FOO), and
4590 removed the compiler macro for SETF INFO (but only after making a fool
4591 of myself on the cmucl-imp mailing list by posting a bogus patch for
4592 DEFINE-COMPILER-MACRO..)
4593 * Compiled files containing forms which have side effects on the Lisp
4594 reader (such as DEFPACKAGE forms) are now handled more correctly.
4595 (Compiler queuing of top level lambdas has been suppressed by setting
4596 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4597 * deleted various currently-unused source files, e.g. gengc.lisp. They
4598 may be added back at some point e.g. when porting to other architectures,
4599 but until they are it's distracting to distribute them and to try to
4601 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4602 there were so many of them they're just distractions, and UNCROSS is
4603 known to be able to handle the current sources
4604 * moved PROFILE functionality into TRACE, so that it will be clear
4605 how the wrapping and unwrapping of functions when you profile them
4606 interacts with the wrapping and unwrapping of functions when you
4607 trace them. (Actually, the functionality isn't there yet, but at least
4608 the interface specification is there. Hopefully, the functionality will
4609 arrive with some future maintenance release.)
4610 * removed host-oops.lisp
4611 * changed signature of QUIT function to allow UNIX-CODE argument
4612 * fixed READ-SEQUENCE bug
4613 * tweaked verbose GC output so that it looks more like the progress
4614 output that ANSI specifies for functions like LOAD
4615 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4616 * added <http://sbcl.sourceforge.com> to the banner information printed by