1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.16 relative to 1.0.15:
3 * minor incompatible change: change PROBE-FILE back to returning
4 NIL whenever we can't get a truename, as was the case before 1.0.14.
5 * enhancement: add support for fcntl's struct flock to SB-POSIX.
7 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
8 * enhancement: cleaner backtraces for interactive interrupts, as
9 well as other cases where the interesting frames used to be
10 obscured by interrupt handling frames.
11 * enhancement: untracing a whole package using (UNTRACE "FOO") is
12 now supported, and tracing a whole package using (TRACE "FOO") now
13 traces SETF-functions as well.
14 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
15 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
16 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
17 * unreadably printed representation of hash-tables now includes
19 * bug fix: partially fixed #188: type propagation from assignments
20 is now more efficient.
21 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
22 'DOUBLE-FLOAT) are not flushable.
23 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
24 full address of the object, and none of the tag bits.
25 * bug fix: readably printing hash-tables now respects other printer
26 control variables. (reported by Cedric St-Jean)
27 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
28 keyword with MAKE-HASH-TABLE.
29 * bug fix: export SB-POSIX:MKSTEMP.
30 * bug fix: SORT was not interrupt safe.
31 * bug fix: XREF accounts for the last node of each basic-block as
33 * bug fix: MAKE-INSTANCE optimizations interacted badly with
34 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
35 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
36 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
37 create obsolete instances in the case of redefinition or
38 obsoletion of a superclass. (thanks to Andy Hefner)
39 * bug fix: Support for the Alpha architecture has been revived; it had
40 suffered somewhat from lack of maintenance since sbcl-1.0.
41 * improvements to the Windows port:
42 ** The system detects the codepage to use at startup, instead of
43 using the value from the saved core. (thanks to Kei Suzuki)
45 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
46 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
47 (see documentation for details.)
48 * revived support for OpenBSD (contributed by Josh Elsasser)
49 * partially fixed bug #108: ROOM no longer suffers from occasional
50 (AVER (SAP= CURRENT END)) failures.
51 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
52 bodies no longer cause a WARNING to be signalled. (reported by
54 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
55 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
56 the bug in TRUNCATE.) (reported by Michael Weber)
57 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
58 no samples. (reported by Andy Hefner)
59 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
60 no longer appear as (NIL ...) frames in backtraces.
61 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
62 is now actually exported from SB-EXT as documented. (reported by
64 * bug fix: sb-aclrepl now correctly understands how to inspect
65 single-floats on 64-bit platforms where single-floats are not boxed.
66 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
67 yet been finalized. (reported by Levente Meszaros)
68 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
69 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
70 SBCL-specific optimize qualities.
72 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
73 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
74 an executable in the search path, and does so in the child
75 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
76 been removed; it can be found in the CVS history, for somebody who
77 needs that search behavior (see the manual).
78 * minor incompatible change: compiler policy re. weakening type
79 checks has changed: now type checks are weakened only if SAFETY < 2
81 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
82 unparsing of directory pathnames as files. Analogously,
83 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
84 filename to parse into a directory pathname.
85 * enhancement: implicit generic function creation now signals a
86 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
87 users can bind handlers for and muffle around calls to LOAD.
88 * enhancement: RUN-PROGRAM allows unicode arguments and environments
89 to be used (using the default stream external format), and allows
90 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
91 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
92 strings and vectors whose element-type or simplicity is not fully
93 known at compile-time.
94 * optimization: STRING-TRIM and related functions no longer allocate
95 a new string when no trimming needs to be performed. These functions
96 are also faster than before when the input string has been declared
98 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
99 * optimization: READ-LINE is significantly faster for files containing
101 * optimization: non-open coded uses of character comparison operators
102 (e.g. char=) no longer cons when called with more than one parameter
103 on platforms supporting dynamic-extent allocation.
104 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
105 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
106 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
108 * bug fix: some sequence functions elided bounds checking when
110 * bug fix: too liberal weakening of union-type checks when SPEED >
112 * bug fix: more bogus fixnum declarations in ROOM implementation
115 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
116 * new feature: MAKE-HASH-TABLE now experimentally accepts a
117 :SYNCHRONIZED argument, which makes the hash-table safe for
118 concurrent accesses (but not iteration.) See also:
119 SB-EXT:WITH-LOCKED-HASH-TABLE, and
120 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
121 * optimization: CONCATENATE on strings is an order of magnitude faster
122 in code compiled with (> SPEED SPACE).
123 * optimization: SUBSEQ is ~50% faster on lists.
124 * bug fix: bug 417 fixed -- source location reporting is now more robust.
125 * bug fix: SUBSEQ on a list will now correctly signal an error if if
126 END is smaller then START.
127 * bug fix: SB-PROFILE will no longer report extra consing for nested
128 calls to profiled functions.
129 * bug fix: ROOM implementation had bogus fixnum declarations which
130 could cause type-errors when calling ROOM on large images.
131 * bug fix: if file compilation is aborted, the partial fasl is now
132 deleted, and COMPILE-FILE returns NIL as the primary value.
133 * bug fix: number of thread safety issues relating to SBCL's internal
134 hash-table usage have been fixed.
135 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
136 be returned from its body when the values were being returned
137 using unknown-values return convection and the W-P-O was wrapped
138 inside an UNWIND-PROTECT.
139 * bug fix: sb-posix should now compile again under Windows, enabling
142 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
143 * incompatible change: hash-table accessor functions are no longer
144 automatically protected by locks. Concurrent accesses on the same hash-table
145 from multiple threads can give inconsistent results or even corrupt the
146 hash-table completely. Multi-threaded applications should do their own
147 locking at the correct granularity. In the current implementation it is
148 still safe to have multiple readers access the same table, but it's not
149 guaranteed that this property will be maintained in future releases.
150 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
151 and will signal an error at runtime.
152 * enhancement: SB-THREAD package now exports a semaphore interface.
153 * enhancement: CONS can now stack-allocate on x86 and
154 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
156 * enhancement: nested lists can now be stack allocated on
157 platforms providing stack allocation support.
158 * enhancement: dynamic-extent support has been extended to support
159 cases where there are multiple possible sources for the stack
161 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
162 if the mutex is uncontested on Linux.
163 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
164 as specified by the CLHS. (thanks to Tobias Rittweiler)
165 * bug fix: NaN comparison now works on x86-64.
166 * bug fix: CLOSE :ABORT T on a stream with pending output now
168 * bug fix: instances of non-standard metaclasses using standard
169 instance structure protocol sometimes missed the slot type checks
171 * bug fix: known functions can be inlined in high-debug code.
173 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
174 * minor incompatible change: the MSI installer on Windows no longer
175 associates .lisp and .fasl files with the installed SBCL.
176 * minor incompatible change: :UNIX is no longer present in *FEATURES*
177 on Windows. (thanks to Luis Oliviera)
178 * new feature: SB-CLTL2 contrib module now implements
179 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
180 * optimization: scavenging weak pointers is now more efficient,
181 requiring O(1) instead of O(N) per weak pointer to identify
182 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
183 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
184 method bodies using constant slot-names when the first argument is
185 a specializer parameter for the method.
186 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
187 names, when the class of the instance is a direct instance
188 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
189 as slow as the constant slot-name case.)
190 * optimization: member type construction is now O(N) instead
192 * optimization: UNION and NUNION are now O(N+M) for large
193 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
195 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
196 now more readable in environments like Slime which display it.
197 (thanks to Tobias C. Rittweiler)
198 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
199 was able to infer the correct argument type for the object on which
200 the CAS operation was being performed.
201 * bug fix: copy propagation interfered with parallel assignment
202 semantics in local calls. (reported by Paul Khuong)
203 * bug fix: the signed modular fixnum shift compiled to wrong code on
204 x86 and x86-64. (spotted by a slight modification to some of
207 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
208 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
209 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
210 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
212 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
213 and generic function now print the object identity for anonymous
214 instances. (thanks to Attila Lendvai)
215 * enhancement: as an extension to MOP, SBCL now supports SETF of
216 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
217 (thanks to Attila Lendvai)
218 * enhancement: sb-cover has an alternate annotation mode which puts
219 the form annotations at the car of the form, instead of annotating
220 it completely. The benefit of this mode is that it shows better
221 which forms were instrumented by the compiler.
222 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
223 by printer control variables. (reported by Dan Corkill)
224 * bug fix: system leaked memory when delayed output was performed by
225 the OS in smaller chunks then expected. (thanks to David Smith)
226 * bug fix: system leaked memory when file streams were not closed
228 * bug fix: large objects written to slow streams that were modified
229 after the write could end up with the modified state written to
230 the underlying file descriptor.
231 * bug fix: multiple threads operating in parallel on the same stream
232 could cause buffer-overflows.
233 * bug fix: source location information is stored correctly
234 (broken since 1.0.6). This bug would generally show up as the
235 Slime debugger highlighting the wrong form when the sldb-show-source
237 * bug fix: Branch forms are again annotated as branches in the sb-cover
239 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
240 on x86-64. (reported by Gregory Vanuxem)
241 * bug fix: EQUALP could return wrong results for structures with raw
242 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
243 word). (reported by Vjacheslav Fyodorov)
244 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
247 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
248 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
249 atomic compare-and-swap operations on threaded platforms.
250 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
251 allows assigning a global minimum value to optimization qualities
252 (overriding proclamations and declarations).
253 * enhancement: closed over variables can be stack-allocated on x86
255 * performance bug fix: GETHASH and (SETF GETHASH) are once again
257 * optimization: slot definition lookup is now O(1). This speeds up
258 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
259 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
260 * optimization: ASSOC and MEMBER can now be open-coded for all
261 combinations of keyword arguments when second argument is constant
262 and SPEED >= SPACE. In other cases a specialized version is
264 * bug fix: using obsoleted structure instances with TYPEP and
265 generic functions now signals a sensible error.
266 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
267 (reported by Kristoffer Kvello)
268 * bug fix: backtrace construction is now more careful when making
269 lisp-objects from pointers on the stack, to avoid creating bogus
270 objects that can be seen by the GC.
271 * bug fix: defaulting of values in contexts expecting more than 7
272 variables now works on x86-64. (reported by Christopher Laux)
273 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
275 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
276 as the property-list of a symbol.
277 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
278 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
279 situations requested, are once again file-compileable. (reported
282 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
283 * MOP improvement: support for user-defined subclasses of
284 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
285 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
286 the expansion of DEFMETHOD, is responsible for generating a form
287 which creates a list of specializers when evaluated. Additional
288 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
289 debugging and introspective support.
290 * minor incompatible change: the (unsupported) spinlock interface
291 has changed: free spinlock now has the value NIL, and a held spinlock
292 has the owning thread as its value.
293 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
294 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
296 * enhancement: name of a socket-stream is now "a socket" instead of
298 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
299 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
300 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
301 for generic functions. (Thanks to Tobias C. Rittweiler)
302 * optimization: bignum printing speed has been improved by 20-40%
303 (depending on the bignum size.)
304 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
306 * bug fix: the cache used by the CLOS to store precomputed effective
307 methods, slot offsets, and constant return values is now thread and
309 * bug fix: generic function dispatch function updating is now thread
310 and interrupt safe (in the sense that the known issues have been
312 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
313 * bug fix: interning EQL-specializers is now thread and interrupt safe.
314 * bug fix: asdf systems with dependencies to the SB-POSIX or
315 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
316 * bug fix: interrupt safety of applicable method computation has been
319 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
320 * new contrib: sb-cover, an experimental code coverage tool, is included
322 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
323 significantly faster.
324 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
325 produce more efficient code on x86-64 if the compiler can determine
326 that (+ INDEX <constant>) does not require a bounds check and FOO
327 has an element type at least 8 bits wide.
328 * enhancement: a new, experimental synchronous timeout facility is
329 provided. Refer to SB-SYS:WITH-DEADLINE for details.
330 * enhancement: when a symbol name conflict error arises, the
331 conflicting symbols are always printed with a package prefix.
332 (thanks to Kevin Reid)
333 * enhancement: stepping is now once again supported on the SPARC and
334 MIPS platforms. (It is also now more likely to work on CheneyGC
336 * enhancement: sb-sprof can now also track and report accurate call
338 * bug fixes: the treatment of non-standard subclasses of
339 SB-MOP:SPECIALIZER is more correct.
340 * incompatible change: PURIFY no longer copies the data from the
341 dynamic space into the static and read-only spaces on platforms
342 that use the generational garbage collector
343 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
345 * bug fix: GC race condition occasionally resulting in crashes with
346 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
348 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
350 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
351 a STYLE-WARNING to be signalled (regression from 1.0.4.)
352 * bug fix: an asynchronous interrupt could previously leave the
353 system running with GC inhibited.
354 * bug fix: a DECLARE form evaluated at top-level now causes an error
355 rather than silently (or verbosely) returning NIL.
356 * bug fix: trying to dissassemble functions compiled at high DEBUG could
357 result in a error being signalled due to source form lookup errors.
358 (reported by Peter Graves)
360 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
361 * incompatible change: removed writer methods for host-ent-name,
362 host-ent-addresses -- changing the values did not update the DNS
364 * minor incompatible change: changed experimental JOIN-THREAD interface
365 * documentation: the manual now lists reader and writer methods
366 in class slot documentation sections. (thanks to Richard M Kreuter)
367 * documentation: unwinding from asyncronous events has been
368 documented as unsafe.
369 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
370 in multithreaded application code.
371 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
372 platforms. (thanks to James Anderson for the optimization hint)
373 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
375 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
376 variants no longer cons.
377 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
378 their NOT- variants no longer cons.
379 * optimization: EQUAL hash tables no longer use SXHASH for objects
380 of all data types, but instead use an EQL hash for types for which
381 EQUAL is the same as EQL.
382 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
383 are significantly faster.
384 * optimization: new STRING-OUTPUT-STREAM implementation is much
385 faster for large outputs and conses 30% less on average.
386 * enhancement: XREF information is now collected for references made
387 to global variables using SYMBOL-VALUE and a constant argument.
388 * enhancement: SIGINT now causes a specific condition
389 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
390 * bug fix: FILE-POSITION used to signal an error in some cases where
391 ANSI requires it to return NIL.
392 * bug fix: ADJUST-ARRAY is now interrupt-safe.
393 * bug fix: adding and removing fd-handlers is now interrupt-safe.
394 * bug fix: inlined calls to C now ensure 16byte stack alignment on
396 * bug fix: bad type declaration in the CLOS implementation has
397 been fixed. (reported by James Anderson)
398 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
399 (thanks to Lutz Euler)
400 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
401 required by the CLHS. (reported by Samium Gromoff)
402 * bug fix: dead unbound variable references now signal an error.
403 (reported by Marco Monteiro)
404 * bug fix: / with an unused value was being deleted in safe code.
405 (thanks to Marco Monteiro and Kevin Reid)
406 * bug fix: number of characters that can be written onto a single
407 line in a file is unlimited.
408 * bug fix: some GC deadlocks caused by asynchronous interrupts have
409 been fixed by inhibiting interrupts for when GC is disabled.
410 * bug fix: some interrupt safety issues with GETHASH, (SETF
411 GETHASH), CLRHASH and REMHASH have been fixed.
412 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
413 specifier no longer causes infinite recursion.
414 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
415 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
416 * bug fix: an &environment argument with the correct variable information
417 is passed to macros that are expanded during byte compilation.
418 (reported by Samium Gromoff)
419 * bug fix: a logic error was causing the hash function for bignums to
420 have excessive amounts of collisions. (reported by Faré Rideau)
421 * bug fix: modifying the contents of an array could change the return
422 value of SXHASH on that array, which is only allowed for strings
423 and bit vectors. (bug introduced in 0.9.16)
424 * improvement: the x86-64/darwin port now passes all tests (except
425 for the debugger tests) but should still be considered
426 experimental until this is fixed.
427 * improvement: a style-warning is signaled for CASE (etc) clauses with
428 duplicate keys. (thanks to Kevin Reid)
429 * improvement: macos/x86 and macos/x86-64 now use mach exception
430 handlers for dealing with illegal instructions (for trapping and
431 error handling) and memory protection violations (for GC).
433 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
434 * new platform: experimental support for x86-64/darwin (MacOS).
435 * incompatible change: the thread-safe (on most platforms) getaddrinfo
436 and getnameinfo sockets functions are used instead of gethostbyaddr
437 and gethostbyname, on platforms where the newer functions are available.
438 As a result, the ALIASES field of HOST-ENT will always be NIL on these
440 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
441 don't print the startup banner, but behave as if --noinform was passed
442 as a command line argument. (thanks to Kevin Reid)
443 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
444 * optimization: code using alien values with undeclared types is much faster.
445 * optimization: the compiler is now able to open code SEARCH in more cases.
446 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
447 * bug fix: using standardized COMMON-LISP special variables as loop
448 variables no longer signals bogus package lock violations. (reported
450 * bug fix: declaring local loop variables to be of a range-limited type
451 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
452 (reported by Andras Simon)
453 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
454 bugs remain on x86-64.)
455 * bug fix: the #= and ## reader macros now interact reasonably with
456 funcallable instances.
457 * bug fix: type-checks for function arguments were compiled using the
458 compiler policy settings of the wrong lexical environment (in 1.0.2
460 * bug fix: SHADOW accepts characters as string designators, as required
461 by the spec (thanks to Eric Marsden)
462 * bug fix: fixed GC safety issues when foreign functions are called with
463 non-base strings as arguments
464 * bug fix: more consistent error output for fatal-compiler-warnings, like
466 * bug fix: functions with proper names were showing up as NIL in
467 backtraces (reported by Edi Weitz, regression in 1.0.2)
469 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
470 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
471 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
472 produce more efficient code on the x86 if the compiler can determine
473 that (+ INDEX <constant>) does not require a bounds check and FOO
474 has an element type at least 8 bits wide.
475 * bug fix: references to undefined variables are handled the same way
476 in toplevel forms as in the normal compiler
477 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
479 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
480 (thanks to Jon Buller)
481 * improvement: faster compilation times for complex functions
482 * improvement: added readlink support to SB-POSIX (thanks to Richard
485 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
486 * improvement: experimental support for mach exception handling on
487 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
489 * improvement: support for GBK external format.
490 (thanks to Chun Tian (binghe))
491 * improvement: the debugger now displays variables that have been closed
492 over, in code compiled with (DEBUG 2) or higher
493 * improvement: support for executable cores on NetBSD (thanks to
495 * new feature: added a RESTART-FRAME debugger command
496 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
497 be used to provide an implementation for FILE-POSITION on Gray streams
498 (thanks to Eric Marsden)
499 * new feature: add syslog, openlog and closelog support to SB-POSIX
500 (thanks to Richard Kreuter)
501 * optimization: the function call overhead in code compiled with
502 a high DEBUG optimization setting is significantly
503 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
504 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
505 (thanks to Tony Martinez)
506 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
507 evaluated once (reported by Kevin Reid)
508 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
509 works again (thanks to Kilian Sprotte)
510 * bug fix: an error is signaled for tagbodies with duplicate tags
511 (thanks to Stephen Wilson)
512 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
513 * bug fix: Win32 port can now handle foreign code unwinding Lisp
514 stack frames from alien callbacks.
515 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
516 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
517 65k pages (thanks to David Woodhouse)
518 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
520 changes in sbcl-1.0.1 relative to sbcl-1.0:
521 * new platform: FreeBSD/x86-64, including support for threading.
522 * new feature: the compiler stores cross-referencing information
523 about function calls (who-calls), macroexpansion (who-macroexpands)
524 and special variables (who-binds, who-sets, who-references) for code
525 compiled with (< SPACE 3). This information is available through the
526 sb-introspect contrib.
527 * new feature: users may subclass SEQUENCE, and have instances of
528 these classes interoperate with standard Common Lisp functions if
529 a number of methods are defined. (This feature is experimental
530 and the interface subject to change based on feedback from SBCL
531 users and the general community)
532 * improvement: sb-sprof traces call stacks to an arbitrary depth on
533 x86/x86-64, rather than the previous fixed depth of 8
534 * improvement: another pthread back-end of mutex "pthread-futex".
535 * bug fix: non-ascii command-line arguments are processed correctly
536 (thanks to Yaroslav Kavenchuk)
537 * bug fix: non-required arguments were not passed correctly when a method
538 defined using DEFMETHOD was called from a mop-generated method using
539 CALL-NEXT-METHOD (reported by Pascal Costanza)
540 * bug fix: recursion is now permitted in accessors through
541 SLOT-UNBOUND. (reported by Pascal Costanza)
542 * bug fix: an error was signaled at startup if the HOME environment
543 variable was defined, but had an empty value (reported by Peter Van Eynde)
544 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
545 signal a PROGRAM-ERROR, not a BUG.
546 * bug fix: SB-POSIX:READDIR works when built with large file support.
547 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
549 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
550 for better automatic indentation support. (thanks to Matt Pillsbury)
551 * optimization: loading generic functions no longer takes O(n^2) time,
552 proportional to the amount of methods in the generic function
553 (reported by Todd Sabin and Jeremy Brown)
554 * optimization: the FIND and POSITION family of sequence functions
555 are significantly faster on arrays whose element types have been
557 * improvements to the Windows port:
558 ** Intermittent heap corruption problems have been fixed. (thanks
559 to Alastair Bridgewater)
560 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
562 ** Lisp is able to unwind foreign exception frames from alien
563 callbacks. (thanks to Alastair Bridgewater)
565 changes in sbcl-1.0 relative to sbcl-0.9.18:
566 * improvement: experimental support for threading on FreeBSD/x86.
567 (thanks to NIIMI Satoshi)
568 * improvement: runtime option --dynamic-space-size can be used
569 to set the size of the dynamic space reserved on startup.
570 * improvement: floating point modes in effect are now saved in
571 core, and restored on startup.
572 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
573 startup, not time since first call to GET-INTERNAL-REAL-TIME.
574 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
575 threads are not running after *SAVE-HOOKS* have run.
576 * improvement: writes to CLOS instance slots are type-checked in code
577 compiled with (SAFETY 3)
578 * improvement: floating-point exception handling on FreeBSD (thanks to
580 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
581 (thanks to Zach Beane)
582 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
584 * improvement: added support for the Shift-JIS external format.
585 (contributed by NIIMI Satoshi)
586 * improvement: callbacks are supported on Linux/PPC. (thanks to
588 * bug fix: compiler bug triggered by a (non-standard) VALUES
589 declaration in a LET* was fixed. (reported by Kaersten Poeck)
590 * bug fix: file compiler no longer confuses validated and already
591 dumped structurres. (reported by Kaersten Poeck)
592 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
593 fill-pointer signals a type-error as required. (thanks to
595 * bug fix: disassemly of funcallable instances works.
596 * bug fix: single stepping on PPC.
597 * bug fix: fix thread-safety problems in the type system (generally
598 manifesting as nonsensical errors like "STRING is a bad type specifier
599 for sequences" or "The value 1 is not of type FIXNUM")
600 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
601 (reported by Josip Gracin)
602 * bug fix: an error is signaled for attempts to displace arrays with
603 incompatible element types (thanks to Mario Mommer)
604 * bug fix: more correct handling of wide characters in debug info
605 (bug reported by Attila Lendvai and fixed by Juho Snellman)
606 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
607 and don't cause extra consing
608 * optimization: MAP and MAP-INTO are significantly faster on vectors
609 whose elements types have been declared.
610 * Improvements to SB-SPROF:
611 ** Support for allocation profiling
612 ** Reduced profiling overhead, especially for long profiling runs
613 * Improvements to the Windows port:
614 ** floating point exceptions are now reported correctly.
615 ** stack exhaustion detection works partially.
616 ** more accurate GET-INTERNAL-REAL-TIME.
617 ** better RUN-PROGRAM behaviour in case of errors in the child
619 ** PROBE-FILE now simplifies pathnames correctly.
620 ** DIRECTORY now works correctly with :WILD-INFERIORS.
621 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
622 1st 1970 (although time timezone data will be unreliable).
623 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
625 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
626 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
627 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
629 * bug fix: two potential GC deadlocks affecting threaded builds.
630 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
631 returns T (reported by Anton Kazennikov)
632 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
633 with non-variable places
634 * bug fix: remove a race condition in the setting of
635 funcallable-instance functions, this should make threaded CLOS
636 code more stable against memory faults.
637 * bug fix: corruption of specials when unbinding is interrupted by an
638 asynchronous unwind (reported by Hannu Koivisto)
639 * improvement: the debugger will now also display local variables that
640 are only used once, for code compiled with a DEBUG optimization quality
643 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
644 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
645 * incompatible change: External-format support for FFI calls. The
646 SB-ALIEN:C-STRING no longer implies an ASCII
647 external-format. Instead the string is subject to external-format
648 conversion. Additionally return values of type SB-ALIEN:C-STRING
649 are by default of type (SIMPLE-ARRAY CHARACTER), not
650 SIMPLE-BASE-STRING. For an alien type that behaves like the old
651 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
652 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
653 * incompatible change: SB-EXT package no longer contains the
654 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
655 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
657 * incompatible change: the single-stepper is no longer available
658 on Alpha, Mips and Sparc platforms.
659 * minor incompatible change: the direct superclasses of
660 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
661 not (STANDARD-OBJECT FUNCTION). This makes the
662 class-precedence-lists of GENERIC-FUNCTION and
663 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
665 * minor incompatible change: the default stream external format on
666 non-unicode SBCL is no longer detected from the locale, but is always
668 * new feature: Add a version of evaluator that uses an interpreter instead
669 of the compiler. EVAL still uses the compiler by default, to switch it
670 to use the interpreter, set the value of the variable
671 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
672 * minor incompatible change: the single-stepper REPL has been merged
673 with the normal debugger (see the "Stepping" heading of the debugger help
674 for more details). The debugger command STEP will no longer switch
675 to the single-stepper REPL.
676 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
677 for a type now works.
678 * bug fix: loading of default sysinit file works. (thanks to Leonid
680 * bug fix: better detection of circularities in the file-compiler.
681 (reported by Marco Monteiro)
682 * bug fix: the CL pathname functions now work with files that have
683 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
684 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
685 systems with Unix98 pty semantics.
686 * bug fix: ASDF-INSTALL will now work with bsd tar.
687 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
689 * bug fix: timers expiring in dead threads no longer cause a
690 type-error (reported by Paul "Nonny Mouse").
691 * bug fix: thanks to more lightweight single-stepper instrumentation,
692 code compiled with (DEBUG 3) will compile and execute significantly faster,
693 and will have more accurate type-inferencing than before
694 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
695 whose bindings are modified
696 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
697 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
698 ** replace dummy implementations of CL:MACHINE-INSTANCE and
699 CL:SOFTWARE-VERSION with proper version
701 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
702 * feature: implemented the READER-METHOD-CLASS and
703 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
704 as specified by AMOP.
705 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
707 * optimization: faster LOGCOUNT implementation on x86 and x86-64
708 (thanks to Lutz Euler)
709 * optimization: hashing of general arrays and vectors has been
710 improved. (reported by Any Fingerhut)
711 * enhancement: SB-INTROSPECT is now able to find definitions of
712 profiled functions. (thanks to Troels Henriksen)
713 * enhancement: compiler-macro expansion applies now to FUNCALL forms
714 as well, allowing compiler-macros for SETF-functions to expand.
715 * enhancement: step-instrumentation no longer wraps known
716 single-value functions with multiple-value context, allowing
717 better type inference.
718 * fixed bug #337: use of MAKE-METHOD in method combination now works
719 even in the presence of user-defined method classes. (reported by
720 Bruno Haible and Pascal Costanza)
721 * fixed bug #339(c): if there are applicable methods not part of any
722 long-form method-combination group, call INVALID-METHOD-ERROR.
723 (reported by Bruno Haible)
724 * fixed bug #361: the :FUNCTION initarg in the protocol for
725 initialization of methods can now be used to override
726 internally-produced optimized functions. (reported by Bruno
728 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
729 system-provided lambda expression no longer cause warnings about
730 unbound #:|pv-table| symbols.
731 * bug fix: improved the handling of type declarations and the
732 detection of violations for keyword arguments with non-constant
734 * bug fix: potentially erronous calls to PATHNAME and
735 MERGE-PATHNAMES were being flushed in some cases.
736 (reported by Richard Kreuter)
737 * bug fix: compiled calls to TYPEP were mishandling obsolete
738 instances. (reported by James Bielman and Attila Lendvai)
739 * bug fix: format strings with ~> without matching ~< no longer
740 trigger an AVER, but signal an understandable error instead.
741 (reported by Antonio Martinez)
742 * bug fix: specifying an output-file without a pathname-type for
743 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
744 of input-file instead of "fasl". (reported by Robert Dodier)
745 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
746 (reported by James Y Knight).
747 * bug fix: compiler-macros-function did not consider the environment
748 argument for shadowing by local functions.
749 * bug fix: compiler-macros expansion was inhibited by local INLINE
751 * bug fix: inline expansions of known functions were subject to
752 step-instrumentation in high debug policies, leading to problems
754 * bug fix: compiler failed to differentiate between different CONS
756 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
757 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
758 * bug fix: #368: incorrect use of expressed vs. upgraded array
760 * bug fix: #306a: more precise unions of array types.
761 * thread-safety improvements:
762 ** CONDITION-WAIT could return early on Linux, if the thread was
763 interrupted and subsequently continued with SIGCONT.
764 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
766 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
767 * added support for the ucs-2 external format. (contributed by Ivan
769 * minor incompatible change: pretty printing of objects of type
770 (cons symbol) is, in the default pprint-dispatch-table, now
771 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
773 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
774 called later in a class's lifetime, possibly as late as when the
775 first instance of the class is created. Previously,
776 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
777 class became finalizeable.
778 * fixed bug: FILE-POSITION sometimes returned inconsistent results
779 for multibyte external-format streams. (thanks to "vbzoli")
780 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
781 with :ALLOCATION :CLASS inherited from superclasses of the
783 * fixed bug: anonymous classes can now be created using the :NAME
784 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
785 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
786 * fixed bug: core-files saved with :EXECUTABLE T can again be
787 executed when SBCL_HOME isn't set. (reported by James Knight)
788 * fixed bug: toplevel LOCALLY forms with declarations could
789 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
790 * fixed bug: printing from several different threads using different
791 values of *print-case* could cause invalid output, due to
792 some internal special variables of the printer not being bound
793 thread-locally (reported by Max Mikhanosha)
794 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
795 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
796 (reported by Pascal Costanza)
797 * fixed bug: FINALIZE-INHERITANCE is called from
798 REINITIALIZE-INSTANCE on classes when the class has previously
799 been finalized, as required by AMOP.
800 * minor code generation optimizations:
801 ** better register allocation in CLOS dispatching functions
802 ** overflow detection when coercing signed bytes to fixnums on x86-64
803 ** is now implemented with one IMUL instruction instead of three shifts
804 ** more efficient bit-vector access on x86 and x86-64
805 ** more efficient access to raw structure slots on x86 and x86-64
806 * fixed some bugs revealed by Paul Dietz' test suite:
807 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
810 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
811 * feature: thread support on Solaris/x86, and experimental thread support
813 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
815 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
816 default method for SLOT-UNBOUND.
817 * minor incompatible change: prevent the user from specializing the
818 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
819 somewhat counter to the intent of the protocol, I (CSR) think, and
820 additionally it just doesn't work in SBCL as currently
821 implemented, thanks to optimizations (that are always valid for
822 the other three SLOT-VALUEish functions, but not for the setter).
823 * minor incompatible change: the :SB-LDB feature is now enabled by
824 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
825 the low-level debugger.
826 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
827 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
828 down to an SB-SYS:FD-STREAM, allowing direct communication in
829 more cases. Temporary files and pipes are still used as a fallback
831 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
832 on the non-reentrant C functions localtime(3) and gmtime(3).
833 * bug fix: global optimization policy was not visible in LOCALLY and
835 * bug fix: class objects can be used as specializers in methods.
836 (reported by Pascal Costanza)
837 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
838 with a type error. (reported by blitz_ on #lisp)
839 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
840 failed with a type error. (reported by Pascal Bourguignon)
841 * bug fix: merging pathnames against defaults with :DIRECTORY
842 starting with '(:RELATIVE :BACK) should preserve the :BACK.
843 (reported by James Y Knight)
844 * bug fix: saving large (>2GB) cores on x86-64 now works
845 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
846 constant index and a (simple-array (signed-byte 32)) array
847 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
848 error (patch by Robert J. Macomber)
849 * feature: TIME now displays the amount of run-time spent in GC
850 * bug fix: The debugger now does a better job of respecting
851 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
852 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
853 workaround for bug 403.)
854 * bug fix: floating point exception handling now works on FreeBSD
855 (thanks to NIIMI Satoshi)
856 * fixed some bugs revealed by Paul Dietz' test suite:
857 ** MISC.641: LET-conversion were not supposed to work in late
859 * improvements to the Win32/x86 port:
860 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
861 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
863 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
864 * new feature: source path information is generated for macro-expansion
865 errors for use in IDE's like Slime (thanks to Helmut Eller)
866 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
867 * bug fix: compiler does not loop forever on an invalid type in
869 * improvement: compilation of most CLOS applications is significantly
871 * optimization: added a limited bytecode compiler for simple toplevel
872 forms, speeding up compilation and FASL loading
873 * bug fix: the statistical profiler now properly distinguishes anonymous
876 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
877 * minor incompatible change: in sbcl-0.9.11 (but not earlier
878 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
879 into subclasses of the finalized class. Now user calls to
880 FINALIZE-INHERITANCE finalize just the one class, and calls by the
881 system return before any subclasses are finalized.
882 * minor incompatible change: The reader no longer ignores errors
883 regarding non-existent packages in #+ and #- feature tests.
884 * new feature: command line options --no-sysinit, --no-userinit to
885 inhibit loading the corresponding init files
886 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
887 for sending data through UDP sockets (thanks to François-René Rideau)
888 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
889 error is signalled instead (thanks to François-René Rideau)
890 * minor incompatible change: Error signalling behaviour of lexical
891 operations violating package locks has changed slightly. Refer to
892 documentation on package locks for details.
893 * enhancement: EVAL can process IF-expressions without resorting to the
895 * bug fix: Some operations involving SIN, COS, or TAN weren't being
896 constant-folded properly due to missing out-of-line functions.
897 (reported by Mika Pihlajamäki)
898 * bug fix: LISTEN sometimes returned T even in cases where no data was
899 immediately available from the stream
900 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
901 were reversed. (reported by Levente Mészáros)
902 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
903 when given a vector displaced to an adjustable array. (reported
905 * bug fix: derivation of float boundaries from numbers outside the
906 appropriate float range (reported by John Wiseman)
907 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
909 * bug fix: precomputing cache entries for generic functions with
910 some subclasses of specializers as yet invalid does not attempt to
911 fill a cache line with a negative offset. (reported by Levente
913 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
914 allow condition class objects as arguments to DOCUMENTATION and
915 (SETF DOCUMENTATION); only find and set documentation for
916 structure names for the STRUCTURE doc-type. (suggested by Gary
918 * improvements to the Win32/x86 port:
919 ** (user-homedir-pathname) and default initialization file
920 locations now know about the user's "Documents and Settings"
921 directory (thanks to Yaroslav Kavenchuk)
922 ** run-program is implemented (thanks to Mike Thomas)
923 ** sockets support (thanks to Timothy Ritchey)
924 ** better backtrace support (thanks to Alastair Bridgewater)
925 ** sb-grovel supported
926 ** asdf-install and sb-posix work somewhat
927 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
928 * improvements to the Solaris/x86 port:
929 ** works on Solaris 11/Solaris Express
930 ** floating-point exception handling support
931 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
932 * fixed some bugs revealed by Paul Dietz' test suite:
933 ** REMOVE-METHOD returns its generic function argument even when
934 the method is not one of the generic functions' methods.
935 ** objects declared as MEMBER types can be admissible arguments to
937 ** printing characters should simply be printed by the FORMAT ~:C
939 ** compiler failure when compiling functions with hairy constant
940 defaults for optional parameters.
941 ** compiler produces wrong code when MAYBE-INLINE-expanding a
942 function, which is already optimized.
944 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
945 * new platform: experimental support for SBCL x86/Darwin, including
946 MacOS X 10.4.5 on Intel.
947 * new feature: Unicode character names are now known to the system
948 (through CHAR-NAME and NAME-CHAR).
949 * new feature: the filesystem location of SBCL's core file is
950 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
951 * minor incompatible change: the contrib modules SB-POSIX and
952 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
953 this change is to make it easier to distribute
954 location-independent binaries.
955 * bug fix: as implied by AMOP, standardized classes no longer have
956 slots named by external symbols of public packages. (reported by
958 * optimization: calling functions via constant symbols -- as in
959 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
960 via the function object as in (FUNCALL #'FOO).
961 * optimization: on x86 and x86-64, the calling convention for the
962 case of dealing with an unknown number of values has been altered
963 to be friendlier to the prediction heuristics implemented,
964 particularly with reference to CALL and RET pairing. (thanks to
965 Alastair Bridgewater)
966 * optimization: on x86, the code for access to thread-local symbol
967 values and binding/unbinding of thread-local symbols is smaller.
968 (thanks to Alastair Bridgewater)
969 * enhancement: CONSTANTP is now able to determine constantness of
970 more complex forms, including calls to constant-foldable standardized
971 functions and some special forms beyond QUOTE.
972 * fixed bug: occasional GC crashes on Solaris/x86
973 * optimization: x86-64 supports stack allocation of results of simple
974 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
975 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
976 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
977 (thanks to James Knight)
978 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
979 by the standard (thanks to Stephen Compall)
981 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
982 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
983 be used for bundling the runtime and the core file into one
984 executable binary. This feature is not currently supported on all SBCL
985 platforms. (thanks to James Bielman and NIIMI Satoshi)
986 * new feature: a generational or ephemeral garbage collector is now
987 the default on the PowerPC platform (both Linux and Darwin). The
988 old Cheney (stop and copy) collector is a build-time option.
989 (thanks to Cyrus Harmon, after Raymond Toy)
990 * minor incompatible change: the method by which SBCL finds its
991 contributed modules has changed; it no longer relies on symbolic
992 links from an $SBCL_HOME/systems directory, but searches directly
993 in the subdirectories of $SBCL_HOME.
994 * enhancement: the dynamic heap size on the Linux/PPC platform is
995 markedly larger, even using the older Cheney garbage collector.
996 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
997 certain complicated string types. (reported by Gary King)
998 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
999 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1000 * fixed bug: DOCUMENTATION on structures no longer signals an
1001 error if no structure type of the right name exists. (reported by
1003 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1004 applicable non-standard primary or :AROUND method on
1005 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1006 no longer cause unbound variable errors. (reported by Kevin Reid)
1007 * fixed bug: metacircle resolution in cases where methods have slots
1008 added before the slots from STANDARD-METHOD. (reported by Jean
1010 * fixed bug: the Power PC assembly code for calling into lisp
1011 sometimes computed its return address wrongly (depending on the
1012 15th bit of the address, fixed at link time). This used to
1013 manifest itself in a segmentation violation while building PCL.
1014 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1015 many others over the years)
1016 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1017 the stack pointer (treated as a signed integer) is negative.
1018 (thanks to Peter van Eynde)
1020 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1021 * new platform: experimental support for the Windows operating
1022 system has been added. (thanks to Alastair Bridgewater)
1023 * fixed several bugs in and robustified the PPC FFI (including
1024 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1025 * bug fix: allow non-simple string symbol names (reported by Paul
1027 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1029 * bug fix: saving a core corrupted callbacks on x86/x86-64
1030 * bug fix: closed a loophole in metacircularity detection and
1031 grounding in the PCL implementation of CLOS.
1032 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1033 (reported by tomppa on #lisp)
1034 * bug fix: clear the direction flag on Lisp -> C transitions, as
1035 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1036 (reported by Andrej Grozin and Hendrik Maryns)
1037 * optimization: major improvements to GC efficiency on GENCGC platforms
1038 * optimization: faster implementation of EQUAL
1039 * optimization: emit more efficient opcodes for some common
1040 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1041 * optimization: several other minor code-generation improvements on x86-64
1042 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1044 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1045 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1046 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1047 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1048 by ANSI as a generic function, and for consistency (SETF
1049 GENERIC-FUNCTION-NAME) is treated likewise.
1050 * fixed bug #233.b: make constraint propagation notice when a variable
1051 value is changed after it is referenced but before it is used
1052 * fixed bug #296: no more arbitrary behaviour from filenames with
1053 odd characters as --load arguments. (reported by Adam Warner)
1054 * fixed bug #390: :CHARACTER-SET pathname components now work as
1055 expected. (reported by Tim Daly Jr)
1056 * fixed bug #391: complicated :TYPE intersections in slot
1057 definitions no longer cause an error in PCL internals.
1058 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1059 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1060 returning the number of octets which would be written to the
1061 file-stream. (thanks to Robert J. Macomber)
1062 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1063 arguments to the internal stream functions. (thanks to David
1065 * bug fix: stack exhaustion is now handled on the x86/SunOS
1066 platform. (thanks to Daisuke Homma)
1067 * bug fix: more accurate ROOM results on GENCGC platforms
1068 * optimization: improved type inference for arithmetic-for
1069 index variables in LOOP
1070 * optimization: faster floating-point SQRT on x86-64
1071 * fixed some bugs revealed by Paul Dietz' test suite:
1072 ** DOCUMENTATION returns NIL instead of "" for method combinations
1073 that don't have a docstring
1075 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1076 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1077 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1078 therefore are not customizeable by user code (as seems to be at
1079 least permitted and maybe required by AMOP). As a consolation,
1080 however, the SBCL implementation of these functions now calls
1081 REINITIALIZE-INSTANCE as specified by AMOP.
1082 * bug fix: slot-definition documentation is propagated as per ANSI
1083 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1084 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1085 argument, as do eslotd initialization methods. (from Pascal
1086 Costanza's "Closer" project)
1087 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1088 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1090 * bug fix: it is now possible to have more than one subclass of
1091 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1092 (reported by Bruno Haible, Pascal Costanza and others)
1093 * bug fix: the dependent update protocol now works for generic
1094 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1095 and Pascal Costanza)
1096 * bug fix: condition-class instances corresponding to
1097 DEFINE-CONDITION forms are now created eagerly. (reported by
1098 Kalle Olavi Niemitalo on comp.lang.lisp)
1099 * bug fix: floating point printing is more accurate in some
1100 circumstances. (thanks to Simon Alexander)
1101 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1102 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1103 * bug fix: callbacks on OS X now preserve stack-alignment.
1104 (thanks to Cyrus Harmon)
1105 * enhancement: the x86-64 disassembler is much better at
1106 disassembling SSE instructions. (thanks to Lutz Euler)
1107 * enhancement: CHECK-TYPE on a variable now allows type-inference
1108 (thanks to Vincent Arkesteijn)
1109 * optimization: improved performance of EUC-JP external format.
1110 (thanks to NIIMI Satoshi)
1111 * optimization: performance improvements to IO on file streams of
1112 :ELEMENT-TYPE CHARACTER
1113 * optimization: much faster memory allocation on x86-64
1114 * optimization: faster garbage collections (latency and throughput)
1116 * optimization: faster bignum subtraction and division on x86 and x86-64
1117 (thanks to Lutz Euler)
1118 * optimization: more accurate type-inference for loops that use a
1119 floating point index variable or a negative step.
1121 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1122 * bug fix: add a workaround to SBCL looping infinitely at startup on
1123 Linux kernels with apparently buggy implementations of personality().
1124 (thanks to Svein Ove Aas)
1125 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1126 (thanks to David Lichteblau)
1127 * bug fix: local bindings shadow global symbol macros for
1128 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1129 * optimization: non-open coded uses of numeric comparison operators
1130 (e.g. >) no longer cons when called with more than one parameter
1131 on platforms supporting dynamic-extent allocation.
1132 * enhancement: saving cores with foreign code loaded is now supported
1133 on MIPS/Linux in addition to the previously supported platforms.
1134 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1135 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1136 * bug fix: division by zero in sb-sprof when no samples were collected
1137 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1138 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1139 (thanks to Alexander Kjeldaas)
1140 * new feature: ignoring the return values of destructive functions
1141 like NREVERSE signals a compile-time style-warning.
1142 (thanks to Kevin Reid)
1143 * new feature: the alignment of alien structure fields can be
1144 explicitly specified. (thanks to Cyrus Harmon)
1145 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1146 and dump core on SIGQUIT
1148 ** incompatible change: threads do not inherit values of specials
1149 from their parents (see manual)
1150 ** bug fix: threads stacks belonging to dead threads are freed by the
1151 next exiting thread, no need to gc to collect thread stacks anymore
1152 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1153 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1154 ** bug fix: dynamic variable and thread start related gc lossage
1155 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1157 * fixed some bugs revealed by Paul Dietz' test suite:
1158 ** SUBTYPEP is slightly more accurate on heinously complicated
1159 CONS types where some of the members have uncertain (in the
1160 NIL, NIL sense) type relationships to each other.
1162 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1163 no longer purified unless :PURIFY T is explicitly specified.
1164 ** Non-purified cores are significantly smaller than before
1166 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1167 * new feature: timers based on Zach Beane's excellent timer package
1168 * added support for the following external formats: koi8-u,
1169 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1170 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1171 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1172 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1173 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1174 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1175 * incompatible change: a threaded SBCL will no longer revert to
1176 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1177 * bug fix: interrupts are disabled until startup is complete; no
1178 more sigsegvs when receiving a signal to soon
1179 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1181 * bug fix: add a workaround for the memory randomization features in
1182 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1183 workaround will only be in effect on systems with the proc filesystem
1185 * bug fix: printing objects of type HASH-TABLE signals a
1186 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1188 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1189 have been running for over 50 days. (reported by Gilbert Baumann)
1190 * bug fix: the logic for getting names of functions gets less
1191 confused when confronded with alternate-metaclass
1192 funcallable-instances. (reported by Cyrus Harmon)
1193 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1194 consistent, even on internal alternate-metaclass objects.
1195 * bug fix: flush closure information collected by physical
1196 environment analysis prepass before the main pass. (bug reported
1198 * bug fix: compiler pack phase does not modify a hashtable, which is
1199 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1200 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1201 may be expanded. (reported by Alexander <malishev> on c.l.l)
1202 * bug fix: no more style-warnings when compiling code using LAST
1203 under high-SPEED low-SPACE settings. (reported by David Wallin
1205 * fixed some incompatibilities between SBCL's MOP and the MOP
1207 ** the METAOBJECT class is now implemented;
1208 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1209 STANDARD-OBJECT, as required;
1210 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1211 now compatible (as required by VALIDATE-SUPERCLASS); there
1212 remains a consistency requirement over the presence or absence
1213 of the FUNCTION class in the superclass list of finalized
1214 classes; see the manual for more details;
1215 ** the :AROUND method for COMPUTE-SLOTS on
1216 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1217 requested slot ordering.
1219 ** bug fix: parent thread now can be gc'ed even with a live
1221 ** bug fix: binding a special with PROGV to no value is not
1222 the same as that symbol not having been bound (thanks to
1224 ** bug fix: binding specials is thread safe (thanks to
1226 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1227 and SB-SPROF do not die with 'no handler for signal XX in
1228 interrupt_handle_now(..)' anymore
1229 ** bug fix: WITH-TIMEOUT works with multiple running threads
1230 * fixed some bugs revealed by Paul Dietz' test suite:
1231 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1232 the :method-class keyword argument.
1234 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1235 * new port: the Solaris operating system on x86 processors is now
1236 mostly supported, though some rough edges in the environment
1237 remain. (thanks to Daisuke Homma)
1238 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1239 space for its heap. (thanks to Thiemo Seufer)
1240 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1241 closures. (thanks to Thiemo Seufer)
1242 * minor incompatible change: eof selects abort in the debugger.
1243 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1244 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1246 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1247 deprecated, and will go away in a future revision of SBCL.
1248 * minor incompatible change: GC-ON and GC-OFF are no longer
1249 implemented with a counter, it does not matter how many times gc
1250 is switched on or off
1251 * bug fix: discriminating functions for generic function classes
1252 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1253 make invalid assumptions about method precedence order. (reported
1255 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1256 platforms. (thanks to Thiemo Seufer)
1257 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1258 * bug fix: correct pathname printing with printer escaping is on.
1259 (thanks to Kevin Reid)
1260 * bug fix: complex VOP definitions in "user-space" no longer trigger
1261 package locks. (reported by Zach Beane)
1262 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1263 for funcallable-instance functions, and is no different from
1265 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1266 characters in the current readtable. (reported by Nicholas Neuss)
1267 * bug fix: SUBTYPEP on various CONS types returns more a more
1268 accurate acknowledgment of its certainty.
1269 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1270 lists in some cases. This partially fixes bug 384.
1271 * flush all standard streams before prompting in the REPL and the
1273 * bug fix: signal handling and triggering gc do not conflict
1274 directly anymore, in particular a high frequency sb-sprof does
1275 not prevent gc from running
1276 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1277 approximation for timezone and DST information between the
1278 universal time epoch and the smallest negative 32-bit time_t.
1279 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1280 year 1899 so long as the encoded time is non-negative.
1281 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1282 the correct answer for non-constant rotate amounts when performing
1283 an inline 32-bit rotation.
1285 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1286 there is only one thread in the session
1287 ** bug fix: memory leak for streams created in one thread and
1288 written to in another
1289 ** bug fix: lockup when compiled with gcc4
1290 ** bug fix: race that allows the gc to be triggered when gc is
1292 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1293 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1295 ** bug fix: finished threads are gc'ed properly
1296 * fixed some bugs revealed by Paul Dietz' test suite:
1297 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1298 has arguments with a different set of applicable methods from
1299 the orignal arguments.
1300 ** The type error thrown by MAP now has a correct expected-type
1302 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1303 name a compiled function.
1304 ** Three MISC tests where a large bignum was improperly coerced to
1305 a float that couldn't represent that bignum during type
1306 derivation were fixed.
1307 ** SUBTYPEP can now handle types involving the negation of a
1308 list-form FUNCTION type.
1309 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1310 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1311 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1313 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1314 * New feature: Experimental support for bivalent streams: streams
1315 opened with :element-type :default now allow character and binary
1316 (unsigned-byte 8) I/O
1317 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1318 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1320 * bug fix: no more highly sporadic "couldn't check whether ~S is
1321 readable" when reading a stream and an interrupt hits in the middle
1322 of a select system call
1323 * compiler better recognizes complex arrays (reported by Eduardo
1325 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1326 for variables on SBCL built with linkage-tables. (reported by Luis
1328 * various error reporting improvements.
1329 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1330 (Thanks to Thiemo Seufer)
1331 * bug fix: nested reader invokations work correctly
1332 * bug fix: it is possible to have simultaneous references to foreign
1333 code and foreign data with the same name.
1335 ** added x86-64 support
1336 ** incompatible change: the threading api now works with thread
1337 objects instead of thread ids
1338 ** bug fix: threads are protected from signals and interruption when
1339 starting up or going down
1340 ** bug fix: a race where an exiting thread could lose its stack to gc
1341 ** bug fix: don't halt on infinite error in threads if possible
1342 ** fixed numerous gc deadlocks introduced in the pthread merge
1343 ** bug fix: fixed thread safety issues in read and print
1344 ** bug fix: debugger doesn't hang on session lock if interrupted at
1345 an inappropriate moment
1346 ** bug fix: run-program is now thread safe(r)
1347 ** bug fix: inner with-recursive-lock no longer releases the mutex
1348 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1349 * fixed some bugs revealed by Paul Dietz' test suite:
1350 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1351 EXPECTED-TYPE slots filled.
1352 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1354 ** MAKE-CONDITION accepts classes as type-designators.
1355 ** COMPILE may never return NIL.
1356 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1357 range before calling Unix time functions
1359 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1360 * numerous signal handling fixes to increase stability
1361 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1362 * minor incompatible change: we now correctly canonize default
1363 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1364 as per the MOP, rather than the historical (INITARG INITFUNCTION
1365 INITFORM). (reported by Bruno Haible)
1366 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1367 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1368 for more information.
1369 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1370 pathname is a directory pathname.
1371 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1372 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1374 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1375 :METACLASS argument in addition to a class name. (reported by
1376 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1377 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1378 (reported by Bruno Haible; patch from Gerd Moellmann)
1379 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1381 * bug fix: more cleanups to the floating point exception handling on
1382 x86-64 (thanks to James Knight)
1383 * bug fix: the compiler does not try to constant fold calls of
1384 COERCE and BIT-* functions when they return freshly constructed
1385 objects. (reported by Nikodemus Siivola and Paul Dietz)
1386 * optimization: improved the MIPS versions of generic (in the
1387 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1388 * optimization: direct conversion of (unsigned-byte 32) to floats on
1389 the PowerPC platform.
1390 * optimization: structure instances with raw slots now use less
1391 memory, and probably show better memory locality. (thanks to
1393 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1394 are treated as requests for stack allocation on the x86-64,
1395 Alpha-32, PPC and SPARC platforms.
1396 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1397 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1398 (thanks to Tony Martinez)
1400 ** gcing a dead thread can no longer lead to lockups
1401 ** threads block signals until they are set up properly
1402 ** errno is no longer shared by threads
1403 ** interrupt-thread restores the eflags register on x86
1404 ** fixed some lockups due to gc/thread interaction
1405 * fixed some bugs revealed by Paul Dietz' test suite:
1406 ** invalid dotted lists no longer raise a read error when
1407 *READ-SUPPRESS* is T
1408 ** signal an error if a symbol that names a declaration is used
1409 as the name of a type, or vice versa
1410 ** allow using the (declare (typespec var*)) abbreviation for
1411 (declare (type typespec var*)) with all type specifiers
1412 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1413 ** always return NIL from PPRINT-POP when OBJECT is NIL
1414 ** don't signal errors when pretty-printing malformed LABELS,
1415 FLET or MACROLET forms
1416 ** declarations in a DOLIST body are also in scope for the
1418 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1420 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1423 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1424 * fixed cross-compiler leakages that prevented building a 32-bit
1425 target with a 64-bit host compiler.
1426 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1427 opened with :IF-EXISTS :OVERWRITE.
1428 * fixed bug 281: error for an invalid qualifier in a short-form method
1429 combination method is not signalled until the faulty method is called.
1430 * bug fix: iteration variable type inferrer failed to deal with open
1431 intervals. (reported by Alan Shields)
1432 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1433 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1434 or more runtime options were provided to the sbcl binary.
1435 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1436 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1437 * optimization: there's now a fast-path for fixnum arguments in the
1438 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1439 * optimization: the code generated on x86-64 is more compact thanks
1440 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1441 * bug fix: floating-point exception handling now partly works on
1442 x86-64. (Thanks to James Knight)
1443 * improvement to the MIPS backend from Thiemo Seufer: C-style
1444 64-bit long long arguments and return values to alien functions
1446 * fixed some bugs revealed by Paul Dietz' test suite:
1447 ** the type-error signalled from WARN has a filled-in DATUM slot.
1448 ** the type-error required when a stream is not associated with
1449 a file has the stream as its datum.
1450 ** type-errors on single-floats on x86-64 no longer have
1451 :INVALID-OBJECT as the datum
1452 ** the type-errors signalled for invalid function names now have
1453 a correct expected type
1454 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1455 for structures defined with a :TYPE. Documentation strings for
1456 typed structures are no longer immediately discarded
1457 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1458 broadcast streams and synonym streams. FILE-LENGTH now also works
1459 on broadcast streams.
1461 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1462 * incompatible change: the --noprogrammer option, deprecated since
1463 version 0.7.5, has been removed. Please use the equivalent
1464 --disable-debugger option instead.
1465 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1467 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1468 inoperational for a while now) has been completely removed.
1469 * null lexical environments are now printed as #<NULL-LEXENV>,
1470 significantly reducing the amount of clutter in typical backtraces.
1471 * documentation on weak pointers, finalization, and after GC hooks
1472 has been added to the manual.
1473 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1474 as other specialized array types, is much faster. SUBSEQ and
1475 COPY-SEQ on such arrays have also been sped up.
1476 * optimization: EQL is now more efficient when at least other argument
1477 is known to be of type (OR FIXNUM (NOT NUMBER)).
1478 * fixed bug: the runtime is now less vulnerable to changes in the
1479 size of the SBCL object on OS X, and virtual memory is reserved for
1480 all spaces that need to be at a fixed address.
1481 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1482 * fixed bug: finalizers and after GC hooks that cause consing are now
1484 * fixed bug: compiler error messages and summaries are now printed to
1485 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1486 * fixed inference of the upper bound of an iteration variable.
1487 (reported by Rajat Datta).
1488 * fixed bug 211e: calling local functions with duplicated constant
1489 keyword argument no longer causes a bogus style warning about an
1491 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1492 ftype declaration to be disregarded. (reported by Dave Roberts)
1493 * fixed bug 373: caused by erronous compilation of references to alien
1494 variables in the runtime on ppc/darwin.
1495 * fixed bug 376: CONJUGATE type deriver.
1496 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1497 (reported by Baughn on #lisp)
1498 * a cross-compiler bug on non-x86 platforms has been identified and
1499 fixed. (thanks to Bruno Haible)
1500 * improvements to the MIPS runtime code for increased stability.
1501 (thanks to Thiemo Seufer)
1502 * increased the maximimum compact environment size to allow
1503 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1504 * partial workaround for bug 135: don't name the function we're
1505 calling for hairy cases of EVAL, so as not to accumulate one environment
1506 entry per call to EVAL. (reported by Kevin Reid)
1507 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1508 * optimization: type testing for non-vector arrays should be faster.
1509 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1511 * bug fix: setting 31st element of a bit vector to zero did not work
1513 * bug fix: redefining a class definition which failed due to a
1514 previous accessor / function clash now works (but see BUGS entry
1515 #380 for more problems in this area). (thanks to Zach Beane)
1516 * the long form of DEFINE-METHOD-COMBINATION disables method group
1517 checking when given a single method group with pattern *. (thanks
1519 * on x86 compiler supports stack allocation of results of simple
1520 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1521 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1522 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1523 * fixed some bugs related to Unicode integration:
1524 ** the restarts for recovering from input and output encoding
1525 errors only appear when there is in fact such an error to
1527 * fixed some bugs revealed by Paul Dietz' test suite:
1528 ** MISC.549 and similar: late transformation of unsafe type
1529 assertions into derived types caused unexpected code
1531 ** SCALE-FLOAT type deriver is less wrong.
1532 ** type derivers for EXP, LOG and similar functions compute result
1533 types for complex arguments better.
1534 ** (MISC.563) CONJUGATE type deriver works for very restricted
1536 ** out-of-line type testers for character strings are available.
1537 ** EQUAL compiler transform understands specialness of objects
1539 ** accessing double-floats stored on the stack now works on x86-64.
1540 ** debugger internals could sometimes create invalid lispobjs,
1541 resulting in GC crashes.
1542 ** MISC.548: type check weakening can convert required type into
1544 ** initialization forms for bindings are not in scope of free special
1547 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1548 * incompatible change: thread support for non-NPTL systems has
1549 been removed - locking is buggy and unreliable. A threaded
1550 SBCL build will now warn at startup and refuse to create
1551 new threads, unless futex support is detected in the kernel
1552 * incompatible change: the top level REPL now has only an ABORT
1553 restart associated with it, not TOPLEVEL and ABORT as it used to.
1554 TOP and TOPLEVEL are now available as debugger commands for
1555 returning to the top level.
1556 * incompatible change: forms evaluated in the REPL now use the
1557 global optimization policy.
1558 * incompatible change: user- and system-initialization files are
1559 no longer processed with LOAD, but by READ and EVAL; hence the
1560 global optimization policy, startup package, readtable, etc,
1562 * The .fasl file format number has been incremented because of
1563 various incompatible changes.
1564 * internal entry point details and argument counts no longer appear
1565 in backtraces unless explicitly requested by setting
1566 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1567 * built-in and standard functions no longer have names like "top
1568 level local call to FOO".
1569 * fixed bug 32: functions defined in non-null lexical environments
1570 now have more legible printed representation
1571 * fixed bug 33: functions defined in non-null lexical environemnts
1572 are now more amenable to inspection by INSPECT.
1573 * workaround for bug 354: XEPs no longer appear in backtraces unless
1574 explicitly requested.
1575 * fixed bug: receiving the signal which results from attempting to
1576 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1577 SIGBUS on other *BSDs) on architectures where the C stack is also
1578 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1579 to debugger confusion or wild instability. (reported by Cheuksan
1581 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1582 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1583 (reported by Lutz Euler)
1584 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1585 compiled code. (reported by Paul Dietz)
1586 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1587 specialized parameter to a method to something that is not TYPEP
1588 the specializer is now possible.
1589 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1590 face of package deletion.
1591 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1592 pathname. (reported by David Tolpin on c.l.l)
1593 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1594 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1595 STRING-STREAM and FILE-STREAM, respectively.
1596 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1597 than previously; in addition, its test suite is now run on build.
1598 (thanks to Luke Gorrie)
1599 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1600 * optimization: added a immediate representation for single-floats
1602 * fixed some bugs revealed by Paul Dietz' test suite:
1603 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1604 ** attempting to create a package with a colliding nickname causes
1605 correctable errors to be signalled.
1606 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1607 ** several x86-64 backend bugs related to sign-extension of immediate
1610 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1611 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1612 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1614 * fixed loading of multiply forward-referenced layouts.
1615 (thanks to Cheuksan Wang)
1616 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1617 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1618 (see COMPILE-FILE documentation for details of :PRINT option)
1619 * fixed bugs 19 and 317: fixed-format floating point printing is
1620 more accurate. This also fixes a bug reported by Adam Warner
1621 related to the ~@F format directive.
1622 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1624 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1625 dispatch table if the from-char is a dispatch macro character.
1626 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1627 by the wordsize no longer ignore the last word. (reported by Lutz
1629 * fixed bug in type checking of dynamic-extent variables. (reported
1631 * optimization: sequence traversal functions use their freedom to
1632 coerce function designators to functions.
1633 * optimization: code with many calls to CLOS methods specialized on
1634 CLOS classes has had redundant type checks removed.
1635 * optimization: added declarations to speed up operations that access
1636 the internal character database (for example STRING-UPCASE)
1637 * optimization: comparison operations between floats and sufficiently small
1638 fixnums no longer create extra rationals
1639 * fixed some bugs related to Unicode integration:
1640 ** portions of multibyte characters at the end of buffers for
1641 character-based file input are correctly transferred to the
1642 start of the buffer at the next read.
1643 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1644 passing it through to OPEN.
1645 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1646 argument given to internal calls to OPEN.
1647 ** the built-in comment readers (introduced by character sequences
1648 ";" and "#|") are more forgiving to encoding errors; they will
1649 STYLE-WARN and then attempt to resync the stream at a character
1650 boundary. (thanks to Teemu Kalvas)
1651 * fixed some bugs in the x86-64 port:
1652 ** Negative short int return values from c-calls are sign-extended
1654 ** The stack is aligned to 16-bytes for c-calls, as required by
1655 the ABI. (thanks to Cheuksan Wang)
1656 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1657 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1658 for enums. (thanks to Vincent Arkesteijn)
1659 ** Multiple small optimizations and bugfixes for floating point
1661 * fixed some bugs revealed by Paul Dietz' test suite:
1662 ** Space, Tab, Linefeed, Return and Page have the invalid
1663 secondary constituent character trait.
1664 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1666 ** WITH-INPUT-FROM-STRING should only update the index place on
1668 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1669 works more reliably.
1670 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1671 with one-dimensional arrays with fill pointers.
1672 ** TYPE-OF failed on a complex with an integer realpart and a
1674 ** compiler failure during type inference for the code of form
1675 (IF (EQL X (THE ...)) ...) (MISC.535).
1677 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1678 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1679 platform. The port supports 61-bit fixnums, large memory spaces
1680 and reloading shared object files.
1681 * enhancement: saving cores with foreign code loaded is now
1682 supported on ppc/Darwin in addition to the previously supported
1684 * enhancement: the statistical profiler now walks deeper into the
1685 call stack for more meaningful call-graphs and accrued time
1686 reports (x86/x86-64 only). It also now reports time spent in
1688 * enhancement: it is now possible to trace most individual methods
1689 of a generic function in addition to tracing the generic function
1691 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1693 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1694 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1695 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1696 produces an error. (thanks to Vincent Arkesteijn)
1697 * bug fix: NAMESTRING on pathnames with :WILD components in their
1698 directories works correctly. (thanks to Artem V. Andreev)
1699 * fixed bug 125: compiler preserves identity of closures. (reported
1701 * bug fixed: functions with &REST arguments sometimes failed with
1702 "Undefined function" when compiled with (DEBUG 3). (reported by
1704 * bug fix: overflow during compiling of setting element of a bit
1705 vector with constant index and value. (reported by Timmy Douglas)
1706 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1707 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1708 * on x86 compiler partially supports stack allocation of dynamic-extent
1710 * GO and RETURN-FROM do not check the extent of their exit points
1711 when compiled with SAFETY 0.
1712 * fixed some bugs related to Unicode integration:
1713 ** encoding and decoding errors are now much more robustly
1714 handled; it should now be possible to recover even from invalid
1715 input or output to the terminal. (thanks to Teemu Kalvas)
1716 ** provided a first cut at implementing STRING-TO-OCTETS and
1717 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1718 ** altered the SB-MD5 contributed module slightly, changing the
1719 interface just enough for it to be supportable for builds where
1720 lisp characters are not eight bits.
1721 * fixed some bugs revealed by Paul Dietz' test suite:
1722 ** the FORMATTER-generated functions for ~V[ conditionals require
1723 the correct number of arguments.
1724 ** READ-FROM-STRING returns the mandated second value when applied
1725 to displaced strings.
1726 ** the #\Rubout and #\Backspace characters are treated as invalid
1727 constituent characters by the tokenizer.
1729 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1730 * new feature: reloading changed shared object files with
1731 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1732 * new feature: references to foreign variables and functions
1733 can now be compiled and loaded before the corresponding shared
1734 object file is loaded, as long as the foreign definitions are
1735 available at runtime.
1736 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1737 supported. (thanks to Dan Debertin)
1738 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1739 just keywords. (thanks to Vincent Arkesteijn)
1740 * fixed bug #331: structure-class instances corresponding to
1741 DEFSTRUCT forms are now created eagerly.
1742 * fixed bug #345: backtraces from calls to undefined functions work
1743 on x86 as well. Related bug #61 is now also partially fixed on x86
1744 (backtraces from throws to unknown catch tags.)
1745 * bug fix: lambda-list parsing is now stricter vrt. order and number
1746 of lambda-list keywords.
1747 * bug fix: as specified by AMOP, an error is signalled if a
1748 class-option appears multiple times in a DEFCLASS form. (reported
1750 * bug fix: RANDOM can be compiled when the compiler derives the type
1751 of its numeric argument as a disjoint set of small integers.
1752 (reported by Paul Dietz)
1753 * bug fix: starting a core saved with shared objects loaded when
1754 those objects are not available no longer causes threaded SBCL to
1755 hang. (reported by Sean Ross)
1756 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1758 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1759 argument, which must always be NIL. (reported by Kalle Niemitalo)
1760 * bug fix: printing 1.0d+23 no longer results in an error.
1761 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1762 * bug fix: structure slot setters preserve evaluation order. (thanks
1764 * fixed some bugs related to Unicode integration:
1765 ** RUN-PROGRAM can allow its child to take input from a Lisp
1766 stream. (reported by Stefan Scholl)
1767 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1768 locales. (reported by Ken Causey)
1769 * fixed some bugs revealed by Paul Dietz' test suite:
1770 ** INCF, DECF and REMF evaluate their place form as specified in
1772 ** FORMATTER expands ~{ ~} iteration directives with V or #
1773 parameters correctly.
1774 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1775 consequent uses no arguments correctly.
1776 ** the system has a partial understanding of the (COMPLEX RATIO)
1778 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1781 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1782 * new feature: a build-time option (controlled by the :SB-UNICODE
1783 keyword feature, enabled by default) for building the system with
1784 support for the entire 21-bit character space defined by the
1786 * new feature: the system now has rudimentary external-format
1787 support; the primary user-visible change at this time is that
1788 characters with the high bit set (such as lower-case-e-acute) will
1789 print correctly to a terminal in a UTF-8 environment.
1790 * minor incompatible change: BASE-CHAR no longer names a class;
1791 however, CHARACTER continues to do so, as required by ANSI.
1792 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1793 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1794 has been moved to the SB-EXT package (temporarily re-exported from
1796 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1797 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1799 * fixed bug #351: better error-handlind and reporting for malformed
1801 * fixed bug #350: bignum-printing is now more memory-efficient,
1802 allowing printing of very large bignums, eg. (expt 2 10000000).
1803 (reported by Bruno Haible)
1804 * fixed bug #302: better primitive-type selection for intersection
1806 * fixed bug #308: non-graphic characters now all have names, as
1807 required. (reported by Bruno Haible)
1808 * bug fix: redefining a class with different superclasses now correctly
1809 removes it from the direct-subclasses of its previous superclasses.
1810 (reported by David Morse)
1811 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1812 new value now works. (reported by Bruno Haible)
1813 * bug fix: correct canonicalization of multiple non-standard slot
1814 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1815 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1816 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1817 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1818 now exists, an signals an error.
1819 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1820 during macroexpansion for non-iteration variables as well. (reported
1821 by Bruno Haible for CMUCL)
1822 * bug fix: Cyclic structures and unprintable objects in compiler
1823 messages no longer cause errors. (reported by Bruno Haible)
1824 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1825 and READ-FROM-STRING all now return a primary value of NIL if
1826 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1827 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1828 (reported by Bruno Haible for CMUCL)
1829 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1830 do not designate a condition. (reported by Bruno Haible for
1832 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1833 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1834 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1835 permitting resolution in favour of any of the conflicting symbols.
1836 (reported by Bruno Haible for CMUCL)
1837 * bug fix: EQUAL compiler optimizations is less aggressive on
1838 strings which can potentially compare true despite having distinct
1839 specialized array element types.
1840 * bug fix: unit enumerations can be defined without dividing by
1841 zero. (thanks to Vincent Arkesteijn)
1842 * FORMAT compile-time argument count checking has been enhanced.
1843 (report from Bruno Haible for CMUCL)
1844 * a partial workaround for the bug 262: the compiler does not try to
1845 inline-expand a local function doing RETURN-FROM from a deleted
1846 BLOCK. (thanks to Peter Denno for the bug report and to David
1847 Wragg for the simple test case)
1848 * fixed some bugs revealed by Paul Dietz' test suite:
1849 ** RENAME-PACKAGE allows all package designators as new package
1851 ** constraint propagation and lambda variable substitution are
1852 more cautious in dealing with partially deleted code.
1853 ** compiler sometimes lost reoptimization passes.
1854 ** CERROR, given a condition as condition designator, passes its
1855 remaining arguments to the continue format control without
1857 ** Case-altering FORMAT directives work correctly on non-ASCII
1859 ** The REAL type specifier handles bounds outside the floating
1860 point ranges without signalling FLOATING-POINT-OVERFLOW.
1861 ** Functions with IR1-transformations can create intercomponent
1862 references to global functions.
1863 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1865 ** FORMAT ~R treats a nil value for its first parameter correctly.
1867 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1868 * enhancement: saving cores with foreign code loaded is now
1869 supported on x86/NetBSD and sparc/Linux in addition to the previously
1870 supported platforms.
1871 * bug fix: on some platforms repeated installations caused multiple
1872 copies of HTML documentation to be installed -- should not happen
1873 any more. (reported by Stefan Scholl)
1874 * bug fix: parsing self-recursive alien record types multiple times
1875 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1876 original patch by Helmut Eller for CMUCL)
1877 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1878 (thanks to Richard Kreuter)
1879 * bug fix: defining classes whose accessors are methods on existing
1880 generic functions in other (locked) packages no longer signals
1881 bogus package lock violations. (reported by François-René Rideau)
1882 * bug fix: special variables as DEFMETHOD parameters no longer have
1883 associated bogus type declarations. (reported by David Wragg and
1885 * bug fix: read-write consistency on streams of element-type
1886 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1887 * bug fix: redefiniton of the only method of a generic function with
1888 no DEFGENERIC no longer emits a full WARNING. In addition,
1889 redefinition of generic functions with no DEFGENERIC to an
1890 incompatible lambda list now signals an error. (thanks to Zach
1892 * bug fix: DEFGENERIC now works even when there's a function of the
1893 same name in an enclosing lexical environment. (thanks to Zach
1895 * fixed compiler failure, caused by instrumenting code during
1896 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1897 * optimization: added loop analysis and improved register allocation
1898 * fixed some bugs revealed by Paul Dietz' test suite:
1899 ** POSITION on displaced vectors with non-zero displacement
1900 returns the right answer.
1901 ** (SIMPLE-STRING) is a valid type specifier for sequence
1903 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1905 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1906 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1908 ** The justification version of the FORMAT ~< directive treats
1909 non-zero minpad parameter correctly.
1911 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1912 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1913 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1914 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1915 the supported interface.
1916 * new feature: Single-stepping of code compiled with DEBUG 2 or
1917 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1918 * new feature: saving cores with foreign code loaded is now
1919 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1920 Timothy Moore's work for CMUCL)
1921 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1922 parameters to * instead of NIL if no initform is supplied.
1923 (reported by Johan Bockgaard)
1924 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1925 correctly when FUNCALL appears as the car of the form. Note:
1926 despite this FUNCALL forms are not currently subject to
1927 compiler-macro expansion. (port of Raymond Toy's fix for the
1928 same from CMUCL, reported by Johan Bockgaard)
1929 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1930 (thanks for Teemu Kalvas)
1931 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1932 case instead of terminating the iteration (thanks for Julian
1933 Squires, Sean Champ and Raymond Toy)
1934 * bug fix: incorrect expansion of defgeneric that caused a style
1935 warning. (thanks for Zach Beane)
1936 * bug fix: slot accessor effective method computation works properly
1937 for classes with multiple non-standard applicable methods on
1938 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1939 * on x86 compiler supports stack allocation of results of LIST and
1940 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1941 CMUCL implementation by Gerd Moellmann)
1942 * fixed some bugs revealed by Paul Dietz' test suite:
1943 ** FORMAT strings with both the ~<~:;~> form of the justification
1944 directive and pretty-printing directives cause an error.
1946 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1947 * incompatible change: the internal functions
1948 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1949 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1950 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1951 instead of the old functions.
1952 * new feature: on platforms where "dladdr" is available foreign
1953 function names now appear in backtraces. (based on Helmut Eller's
1955 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1956 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1958 * documentation: ASDF manual is now distributed with SBCL.
1959 * bug fix: SBCL can now load its contributed modules with REQUIRE
1960 even if the system-provided entries have been removed from
1961 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1963 * bug fix: backtraces involving undefined functions or assembly
1964 routines are more informative. (thanks to Brian Downing)
1965 * bug fix: mutually referent alien structures now work correctly.
1966 (reported by Rick Taube)
1967 * bug fix: structures defined by WITH-ALIEN can be referred to
1968 within other definitions in the same WITH-ALIEN.
1969 * bug fix: division operators (MOD, TRUNCATE and the like) with
1970 constant zero divisors and integer dividends no longer generate
1972 * bug fix: provide default methods for INPUT-STREAM-P and
1973 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1974 * bug fix: improve the detection and resolution of MOP metacycles
1975 (where effective method computation depends on the generic function
1976 having its effective method computed). (reported by Bruno Haible)
1977 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1978 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1979 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1980 * bug fix: pathnames with relative directory components can be
1981 represented relative to default pathnames.
1982 * optimization: in taking the GCD of bignums, reduce the two bignums
1983 to approximately the same size (using Euclid's algorithm) before
1984 applying the more sophisticated binary GCD. (thanks to Juho
1986 * optimization: COUNT on bitvectors now operates word-at-a-time.
1987 * optimization: ASH with a positive, but not necessarily constant,
1988 (leftwards) shift, when in a modular context, is compiled to a
1990 * fixed some bugs revealed by Paul Dietz' test suite:
1991 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1992 if the corresponding argument is NIL.
1993 ** FORMAT directives accepting a minpad parameter treat negative
1995 ** literal commas as character parameters to FORMAT directives are
1997 ** literal spaces directly after ~<Newline> directives within a
1998 format-logical-block (~:< ~@:>) do not induce :FILL-style
1999 conditional newlines.
2000 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2001 as specified: it no longer includes conditional newlines).
2002 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2004 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2005 :PER-LINE-PREFIX argument does not evaluate to a string.
2007 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2008 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2009 the manual for details; for now, package locks can be disabled by
2010 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2011 if no major problems are found then it is likely that they will be
2012 compiled in unconditionally.
2013 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2014 unsupported operators on all platforms. To load a shared library
2015 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2016 object file, link it into a shared library outside of SBCL and
2017 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2019 * bug workaround: ROOM T and the GC were not working together
2020 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2021 reporting facility aren't preserved. That reporting has been
2022 disabled, so now until and unless someone figures out how to make it
2023 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2024 * minor incompatible change: as threatened around sbcl-0.8.0, the
2025 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2026 The symbols are also exported from SB-PCL for backwards
2027 compatibility, but more so than before SB-PCL should be treated as
2028 an implementation-internal package.
2029 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2031 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2032 It is known as of this release not to work on the Alpha, however.
2033 * fixed bug #167: errors signalled due to illegal syntax in method
2034 bodies are now more legible.
2035 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2036 designators and can hence be used with TYPEP.
2037 * fixed bug #333: CHECK-TYPE now ensures that the type error
2038 signalled, if any, has the right object to be accessed by
2039 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2040 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2041 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2043 * fixed bug #334: programmatic addition of slots using specialized
2044 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2045 and :ALLOCATION :CLASS slots.
2046 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2047 float exponents. (rereported by Peter Seibel)
2048 * fixed a bug: #\Space (and other whitespace characters) are no
2049 longer considered to be macro characters in standard syntax by
2050 GET-MACRO-CHARACTER.
2051 * fixed bug: initialization of condition class metaobjects no longer
2052 causes an instance of the condition to be created. (reported by
2054 * fixed bug: it is now possible to have slots such that
2055 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2056 description is neither :INSTANCE nor :CLASS.
2057 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2058 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2059 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2060 non-standard SLOT-DEFINITION-ALLOCATION values.
2061 * fixed bugs: various race conditions handling exiting threads.
2062 CL-PPCRE's thread torture test now passes.
2063 * fixed arguably-a-bug: GC time proportional to number of threads in
2064 system even when most of them are idle
2065 * optimization: improved performance of BIT and SBIT on bit-vectors.
2066 * .fasl file incompatibility: The fasl file version number has
2067 been incremented because of changes associated with package locks.
2069 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2070 * minor incompatible change: the system no longer provides
2071 optimization hints (conditions which are TYPEP
2072 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2073 mode; these hints are still emitted when higher SPEED optimization
2075 * new contrib module: a sampling profiler (profiling by statistical
2076 sampling, rather than by instrumenting functions) is available as
2077 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2078 * the behaviour of the standard function ED is now customizeable by
2079 third parties through a hook variable: see ED's documentation
2080 string for information on the protocol.
2081 * the compiler no longer emits efficiency notes for (FUNCALL X)
2082 when the type of X is uncertain under default optimization
2084 * fixed bug 276: mutating a binding of a specialized parameter to a
2085 method to something that is not TYPEP the specializer is now
2087 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2088 exhibit the required behaviour on the x86 platform. (thanks to
2089 Peter van Eynde, Eric Marsden and Bruno Haible)
2090 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2091 even for difficult arguments. (reported by Peter Graves)
2092 * fixed bug 141a: the backquote printer now descends quoted
2094 * fixed another bug in backquote printing: no more destructive
2095 modification of the form's list structure. (reported by Brian
2097 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2098 move between its address being taken and the call to
2099 interrupt_thread, fixing a crashing race condition.
2100 * the SB-POSIX contrib implementation has been adjusted so that it
2101 no longer exhibits ridiculously poor performance when constructing
2102 instances corresponding to C structs.
2104 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2105 * minor incompatible change: the sb-grovel contrib now treats C
2106 structures as alien (in the sense of SB-ALIEN) objects rather than
2107 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2108 has implications for memory management of client code
2109 (sb-grovel-returned objects must now be manually managed) and for
2110 type safety (alien objects now have full types).
2111 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2112 used to control emission of compiler diagnostics, rather than the
2113 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2114 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2115 quality should be considered deprecated.
2116 * install.sh now installs the user manual as well
2117 * (not quite a new documentable feature, but worth considering in
2118 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2119 beginnings of a semantically meaningful condition hierarchy is
2120 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2122 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2123 designator as the defaults argument.
2124 * fixed bug: Displaced arrays whose displaced-to array has become
2125 too small now cause ARRAY-DIMENSION to signal an error, providing
2126 for safer bounds-checking. (reported by Bruno Haible)
2127 * fixed bug: DEFCLASS slot definitions with identical :READER and
2128 :WRITER names now signal a reasonable error. (reported by Thomas
2130 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2132 * fixed bug: Invalid slot specification errors now print correctly.
2133 (thanks to Zach Beane)
2134 * fixed bug 320: Shared to local slot value transfers in class
2135 redefinitions now happen corrently from superclasses as
2136 well. (reported by Bruno Haible)
2137 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2139 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2140 improper lists correctly. (reported by Bruno Haible)
2141 * fixed bug 313: source-transform for <fun-name> was erroneously
2142 applied to a call of a value of a variable with name <fun-name>.
2143 (reported by Antonio Menezes Leitao)
2144 * fixed bug 307: The obsolete instance protocol ensures that
2145 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2146 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2147 remove dead unknown-values globs from the middle of the stack.
2148 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2149 easier to resume long computations after using *BREAK-ON-SIGNALS*
2150 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2151 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2152 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2153 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2154 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2155 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2157 * fixed a (fairly theoretical) bug in string printing: if
2158 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2159 does not have array-element-type equal to the most general string
2161 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2162 function name. (thanks to Nikodemus Siivola)
2163 * fixed bug: compile-time format string checker failed on
2164 non-closed ~{. (reported by Thomas F Burdick)
2165 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2166 don't warn on legal loop code involving more than one aggregate
2167 boolean. (thanks to Nikodemus Siivola)
2168 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2169 in keyword position to MAKE-PACKAGE was not being reported as
2171 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2172 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2174 * fixed bugs 280 and 312: the checking for multiple definitions in a
2175 file is less likely to become confused by uses of inline
2177 * fixed bug: the #S reader macro performs the keyword coercion
2178 specified for slot names. (reported by Kalle Niemitalo)
2179 * fixed bug: lambda lists may contain symbols whose names start with
2180 & but are not lambda-list-keywords; their occurrence triggers a
2182 * fixed bug 321: define-method-combination argument lambda lists do
2183 not cause invalid code to be generated when &optional or &aux
2184 variables have default values. (reported by Bruno Haible)
2185 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2186 have CLOS classes; this appears also to have fixed reported
2187 instability in stack exhaustion detection.
2188 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2189 MUFFLE-WARNING when no associated restart is present is now
2191 * optimization: rearranged the expansion of various defining macros
2192 so that each expands into only one top-level form in a
2193 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2195 * optimization: used a previously little-used slot in symbols to
2196 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2198 * fixed some bugs revealed by Paul Dietz' test suite:
2199 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2201 ** Fixed an optimization bug related to inheritance of initforms
2202 from local to shared slots.
2203 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2204 ** CAST optimizer forgot to flush argument derived type.
2205 ** print/read consistency on floats is now orders of magnitude
2206 more likely. (thanks also to Bruno Haible for a similar report
2208 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2209 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2210 for objects of type REAL. Make it so.
2211 ** SXHASH is specified (infelicitously) to respect similarity,
2212 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2213 it so. (thanks to Markus Ziegler)
2214 ** on the Alpha, the compiler succeeds in compiling functions
2215 returning a known number of arguments greater than 63.
2216 ** fixed handling of invalid NIL arguments in keyword position
2218 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2220 ** provided a readably-printable representation for RANDOM-STATE
2222 ** ensured that pathnames, where they have a namestring, always
2223 print using #P"..." syntax.
2225 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2226 * Support for the forthcoming 2.0 version of the NetBSD kernel
2227 running on x86 hardware has been added. (thanks to Perry
2228 E. Metzger most immediately, and others for their past work)
2229 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2230 current release version; SBCL's OpenBSD support had been broken
2231 since about the time of OpenBSD's switch to ELF binary format.)
2232 * [placeholder for DX summary]
2233 ** user code with &REST lists declared dynamic-extent, under high
2234 speed or space and low safety and debug optimization policy.
2235 * The manual has been converted to Texinfo format and the debugger
2236 chapter from the cmucl manual has been added.
2237 * A facility has been added to extract documentation strings from
2238 sbcl and store them as Texinfo-formatted snippets for inclusion in
2239 the manual (via Texinfo's @include directive)
2240 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2241 the test case to Dave Roberts)
2242 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2243 pointers. (reported by Sean Ross)
2244 * bug fix: PROFILE output is printed nicely even for large numerical
2245 values. (thanks to Zach Beane)
2246 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2247 greater than 32 handle EOF correctly.
2248 * bug fix: on X86 an immediate argument of the IMUL instruction is
2249 correctly printed in disassembly. (thanks to Lutz Euler)
2250 * bug fix: class slots in redefined classes preserve their old
2251 values. (thanks to Bruno Haible and Nikodemus Siivola)
2252 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2253 succeeds. (reported by Marco Baringer)
2254 * fixed some bugs revealed by Paul Dietz' test suite:
2255 ** READ-SEQUENCE now works on ECHO-STREAMs.
2256 ** RATIONALIZE works more according to its specification. (thanks
2259 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2260 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2261 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2262 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2263 should matter to you only if you rebind the printer control
2264 variables and then find you want different bindings in the
2265 debugger than in the ordinary execution of your program.)
2266 * The runtime build system has been tweaked to support building
2267 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2268 assembler and linker. (thanks to Nikodemus Siivola)
2269 * Unbound, undefined, undeclared variables now trigger full
2270 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2271 more often programmer error than deliberate exploitation of undefined
2273 * optimization: the hash algorithm for strings has changed to one
2274 that is less vulnerable to spurious collisions. (thanks to Juho
2276 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2277 needless bounds checking. (thanks to Juho Snellman)
2278 * optimization: implemented multiplication as a modular
2279 (UNSIGNED-BYTE 32) operation on the PPC backend.
2280 * fixed some bugs revealed by Paul Dietz' test suite:
2281 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2283 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2284 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2285 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2287 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2289 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2291 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2292 their output stream on EOF from read.
2293 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2294 have been read to end-of-file.
2295 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2297 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2298 description of determination of which consecutive characters
2300 ** Printing the "Space" character with escaping on now yields "#\\ ",
2301 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2302 ** Reading floating-point numbers with *READ-BASE* set to a number
2303 less than 10 works correctly.
2304 ** Reading floating-point numbers with *READ-BASE* set to a number
2305 more than 10 works correctly.
2306 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2307 the readtable currently in effect.
2309 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2310 * minor incompatible change: parsing of namestrings on a physical
2311 (Unix) host has changed; numbers after the final #\. in a
2312 namestring are no longer interpreted as a version field. This is
2313 intented to be largely invisible to the user, except that the
2314 meaning of the namestring "*.*.*" has changed: it now refers to a
2315 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2316 should usually be replaced by
2317 (make-pathname :name :wild :type :wild :version :wild)
2318 with the added benefit that this is more likely to be portable.
2319 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2320 option to OPEN now signals an error if the file being opened
2321 exists; this may have an impact on existing code.
2322 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2323 signals received, on the PowerPC platforms (both Linux and
2324 Darwin). (thanks to Pierre Mai for pointing out the location of
2326 * several fixes on OS X: The system now builds and runs cleanly on
2327 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2328 (thanks to Brian Mastenbrook)
2329 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2330 second-resolution: integer multiples of 1/3600 between -24 and 24.
2331 (thanks to Vincent Arkesteijn)
2332 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2333 type when called with 1 argument; PEEK-CHAR checked type of
2334 PEEK-TYPE only after having read first character from a
2335 stream. (reported by Peter Graves)
2336 * bug fix: the garbage collector now has much better locality
2337 behaviour, and in particular no longer treats all memory as being
2338 exhausted when many small objects point to each other in a deeply
2340 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2341 recognized as being TYPEP their class.
2342 * bug fix: the PUSHNEW documentation string has been corrected.
2343 (thanks to Vincent Arkesteijn)
2344 * bug fix: defaulting of the value for the last of an atypically
2345 large number of multiple values being bound was not being
2346 performed correctly on the Alpha or PPC platforms
2347 * optimization: implemented multiplication as a modular
2348 (UNSIGNED-BYTE 32) operation on the x86 backend.
2349 * optimization: SEARCH on simple-base-strings can now be open-coded.
2350 (see also contrib/compiler-extras.lisp for inspiration for
2351 teaching the compiler about the Boyer-Moore algorithm).
2352 * value, returned by MAX (and MIN) called with several EQUALP, but
2353 not EQL, arguments now does not depend on compiler settings.
2354 * fixed some bugs revealed by Paul Dietz' test suite:
2355 ** in stack analysis liveness information is propagated from
2356 non-local entry points.
2357 ** pathwise CAST removing failed when the CAST node did not start
2359 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2360 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2362 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2363 host is already defined.
2364 ** RENAME-FILE works on streams instead of signalling an internal
2366 ** PEEK-CHAR uses the current readtable when determining whether
2367 or not a character is whitespace.
2368 ** MERGE-PATHNAMES handles the case when the pathname does not
2369 specify a name while the default-pathname specifies a version.
2370 ** Pathnames now stand a better chance of respecting print/read
2372 ** Attempting to use standardized file system operators with a
2373 pathname with invalid :DIRECTORY components signals a
2375 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2377 ** DIRECTORY on logical pathnames is more correct.
2378 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2379 signal a TYPE-ERROR if their argument is not a stream.
2380 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2381 designator argument does not designate a stream.
2382 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2383 examining the synonym.
2384 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2386 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2387 element-type larger than ([UN]SIGNED-BYTE 32).
2389 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2390 * When built with the :SB-FUTEX feature, threaded builds now take
2391 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2392 for faster/more reliable mutex and condition variable support.
2393 * Incompatible change (but one you probably shouldn't have been using
2394 anyway): the interface and code for arbitrating between multiple
2395 threads in the same user session has been redesigned.
2396 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2397 error when called without an explicit environment argument.
2398 (thanks to Nikodemus Siivola)
2399 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2400 confused about their position. (thanks to Adam Warner and Gerd
2402 * bug fix: Pretty printing backquoted forms with unquotations in the
2403 argument list position of various code constructs such as LAMBDA
2404 now works correctly. (reported by Paul Dietz)
2405 * bug fix: Pretty printing unquotations no longer loses all
2406 stream position information.
2407 * optimization: performance of string output streams is now less
2408 poor for multiple small sequence writes.
2409 * optimization: performance of CSUBTYPEP in the presence of complex
2410 expressions involving CONS and NOT many times has been improved.
2411 (reported by Paul Dietz)
2412 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2414 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2415 streams for the REPL, for the debugger, and so on. (thanks to
2417 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2418 * fixed some bugs revealed by Paul Dietz' test suite:
2419 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2420 is ignored if it is too small, rather than propagating through
2421 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2422 ** extremely complex negations of CONS types were not being
2423 sufficiently canonicalized, leading to inconsistencies in
2425 ** VALUES tranformer lost derived type.
2427 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2428 * fixed a bootstrapping bug: the build process no longer assumes
2429 that the various BOOLE-related constants have the same value in
2430 host and target lisps. (noted by Paul Dietz' test suite on an
2431 SBCL binary built from CLISP)
2432 * The system can now be dynamically linked on the MIPS platform,
2433 which enables dynamic loading of foreign code from Lisp. (thanks
2434 to Ralf Baechle for discussions on the MIPS ABI)
2435 * The system now records debugging information for its own source
2436 files in a filesystem-position-independent manner, relative to
2437 the "SYS" logical host.
2438 * fixed a compiler bug: MV-LET convertion did not check references
2439 to the "max args" entry point. (reported by Brian Downing)
2440 * tweaked disassembly notes to be less confident about proclaiming
2441 some instruction as an LRA. (thanks to Brian Downing)
2442 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2443 now each have their own history, command character, and other
2444 characteristics. (thanks to David Lichteblau)
2445 * fixed some bugs revealed by Paul Dietz' test suite:
2446 ** compiler failure in compiling LOGAND expressions including a
2448 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2449 shift greater than 32.
2450 ** FLUSH-DEST did not mark blocks for type check regeneration.
2451 ** HANDLER-CASE failed to accept declarations in handler clauses
2452 in some circumstances.
2454 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2455 * New code in contrib/sb-introspect (still probably not entirely
2456 stable yet) provides some support for smart Lisp development
2457 environments like SLIME.
2458 * The conditions signalled for errors occurring when loading .fasl
2459 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2460 in a way which should help ASDF recover gracefully.
2461 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2462 stuff has been cleaned up. If you code contrib/ stuff, this might
2463 affect you, and you can look at contrib/README, contrib/STANDARDS,
2464 and/or the 0.8.4.27 diff to check.
2465 * In full calls the compiler now does not generate checks for declared
2466 argument types for all arguments.
2467 * various threading fixes
2468 ** and some experimental patches which didn't make it into
2469 the main tree for this release, but which are shipped in
2470 contrib/experimental-thread.patch as a possible fix for some
2471 failures (deadlock, spinning...) in GC-intensive multithreaded
2473 * fixed PPC build problem (source code incompatibility of different
2474 library versions): added offsetof() hackery which attempts to divine
2475 where glibc maintainers put uc_mcontext today
2476 * fixed bug 282: compiler does not trust type assertions while passing
2477 arguments to a full call.
2478 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2479 &OPTIONAL VALUES type parameter.
2480 * fix bug 214: algorithm for noting rejected templates is now more
2481 similar to that of template seletion. (also reported by rydis on
2483 * fixed bug 141b: printing backquoted information readably and prettily
2484 inserts a space where necessary.
2485 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2486 (CHARACTER 10) are now reported as errors, rather than propagated
2487 as unknown types. (reported by piso on #lisp)
2488 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2489 with values NIL and :ERROR. (thanks to Milan Zamazal)
2490 * fixed bug 191c: CLOS now does proper keyword argument checking as
2491 described in CLHS 7.6.5 and 7.6.5.1.
2492 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2493 longer raise an error; further, using a list as a for-as-arithmetic
2494 counter now raises a meaningful error.
2495 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2496 now understood by sequence creation functions such as MAKE-SEQUENCE
2498 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2499 a STRING-INPUT-STREAM.
2500 * compiler enhancement: SIGNUM is now better able to derive the type
2502 * type declarations inside WITH-SLOTS are checked. (reported by
2504 * fixed some bugs revealed by Paul Dietz' test suite:
2505 ** incorrect optimization of TRUNCATE for a positive first
2506 argument and negative second.
2507 ** compiler failure in let-convertion during flushing dead code.
2508 ** compiler failure while deriving type of TRUNCATE on an
2509 interval, containing 0.
2510 ** ASH of a negative bignum by a negative bignum count now returns
2512 ** intersection of CONS types now canonicalizes properly, fixing
2513 inconsistencies in SUBTYPEP.
2515 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2516 * incompatible change: The --disable-debugger command line
2517 option now clobbers the debugger at a more fundamental
2518 level, by redefining #'INVOKE-DEBUGGER instead of by
2519 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2520 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2521 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2522 enter the debugger REPL and then suffer recursive errors
2523 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2524 while under the new system, BREAK in --disable-debugger mode
2525 terminates the system just as an unhandled error would.
2526 * fixed compiler performance when processing loops with a step >1;
2527 * bug fix: DOCUMENTATION now retrieves generic function
2528 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2529 support has been systematized, and now supports the methods
2530 specified by ANSI, along with a default method and a method for
2531 slot documentation. (reported by Nathan Froyd)
2532 * bug fix: effective methods associated with a generic function are
2533 no longer cached over a change of that generic function's method
2534 combination. (reported by Andreas Fuchs)
2535 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2536 names a non-existent directory. (thanks to Andreas Fuchs)
2537 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2538 the PPC platform, lead to stack corruption; this has been fixed.
2539 (reported by Rainer Joswig)
2540 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2541 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2542 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2543 arithmetic in general, is now much more reliable on the PPC
2545 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2546 platform now returns the right answer.
2547 * optimization: restored some effective method precomputation in
2548 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2549 precomputation is now tunable.
2550 * optimization: compiler-internal data structure use has been
2551 reviewed, and changes have been made that should improve the
2552 performance of the compiler by about 20%.
2553 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2554 simple-base-strings and simple-bit-vectors is improved.
2555 * optimization: the optimization of 32-bit logical and arithmetic
2556 functions introduced in version 0.8.3 on the x86 has been
2557 implemented on the mips, ppc and sparc platforms; an
2558 implementation of the same facility, but for 64-bit arithmetic,
2559 has been added for the alpha.
2560 * microoptimization: the compiler is better able to make use of the
2561 x86 LEA instruction for multiplication by constants.
2562 * bug fix: in some situations compiler did not report usage of
2563 generic arithmetic in (SPEED 3) policy.
2564 * bug 145b fix: compiler used wrong type specifier while converting
2565 MEMBER-types to numeric.
2566 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2568 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2569 index argument now works properly on the Alpha platform.
2570 * bug fix: floating point exception treatment on the Alpha platform
2572 * bug fix: FILE-POSITION works much better on string input and
2573 output streams. (thanks to Nikodemus Siivola)
2574 * bug fix: many threading/garbage collection symptoms sorted.
2575 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2576 might be pseudo-atomic.
2577 * internal change: Stopping for GC is now done with signals not
2578 ptrace. GC is now done in whichever thread wanted it, instead of
2580 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2582 * bug fix: result form in DO is not contained in the implicit
2584 * incompatible change: ICR structure is changed; the value part of
2585 CONTINUATION is now called LVAR; corresponding functions are
2586 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2587 * added type deriver for ISQRT (thanks to Robert E. Brown).
2588 * bug fix: better support for loading from the command line when an
2589 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2591 * fixed some bugs revealed by Paul Dietz' test suite:
2592 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2593 ** ROUND and FROUND now give the right answer when given very
2594 small float arguments.
2595 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2597 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2598 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2599 ** LCM should return a non-negative integer.
2600 ** PARSE-INTEGER returned the index of a terminator instead of the
2601 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2602 ** PARSE-INTEGER returned an incorrect index being applied to a
2604 ** LCM with two arguments of 0 returns 0 rather than signalling
2606 ** unsigned addition of a 32-bit constant with the high bit set no
2607 longer causes an internal compiler error on the x86.
2608 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2609 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2610 with negative last argument.
2611 ** byte specifiers with zero size and position no longer cause
2612 an error during type derivation.
2613 ** bignum multiplication on the Alpha platform now returns the
2615 * porting: The system now builds on SuSE AMD64, although it still
2616 generates a 32-bit binary.
2617 * .fasl file incompatibility: The fasl file version number has
2618 been incremented (because of the changes to internal compiler
2619 data structures referred to above).
2621 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2622 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2623 more accurately, on the Darwin kernel running on PowerPC hardware.
2624 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2625 * Compiler code deletion notes now signal a condition of type
2626 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2627 an associated MUFFLE-WARNING restart.
2628 * The compiler now performs limited argument count validation of
2629 constant format strings in FORMAT, and where appropriate in ERROR,
2630 CERROR and WARN. (thanks to Gerd Moellmann)
2631 * New ASDF-INSTALL contrib can be used for automatic download and
2632 installation of third-party Lisp code from CCLAN or other sites
2634 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2635 which forces another thread to execute a function supplied by the
2637 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2638 now accept and act upon their :ELEMENT-TYPE keyword argument.
2639 (reported by Martin Atzmueller, Edi Weitz)
2640 * bug fix: FILE-POSITION now accepts position designators up to
2641 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2642 is the greater. (thanks to Patrik Nordebo)
2643 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2644 Kalle Olavi Niemitalo)
2645 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2646 printable. (reported by Eric Marsden)
2647 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2648 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2649 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2650 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2651 (thanks to Stig Sandoe)
2652 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2653 UNREAD-CHAR now correctly includes the unread character in the
2654 target sequence. (thanks to Gerd Moellmann)
2655 * bug fix in threaded builds: the system can now be suspended and
2656 resumed by shell job control with minimal disruption.
2657 * bug fixes in times and timezones >2038AD
2658 * better handling of "where is GNU make?" problem in build scripts
2659 (thanks to Nikodemus Siivola)
2660 * new optimization: inside a named function any reference to a
2661 function with the same name is considered to be a self-reference;
2662 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2663 optimization quality.
2664 * new optimization on x86: logical functions and + now have
2665 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2666 used when the result is truncated to 32 bits.
2667 * VALUES declaration is partially enabled.
2668 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2669 * fixed some bugs revealed by Paul Dietz' test suite:
2670 ** The system now obeys the constraint imposed by
2671 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2672 types form a lattice under type intersection.
2673 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2674 ** ASSOC now ignores NIL elements in an alist.
2675 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2676 and (1+ MOST-POSITIVE-FIXNUM) answers.
2677 ** The addition of a method with invalid qualifiers to a generic
2678 function does not cause an error to be signalled immediately;
2679 a warning is signalled, and the error is generated only on
2680 calling the generic function.
2681 * changed .fasl file version number, in part to add type codes for
2682 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2683 obscure ANSI requirements
2685 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2686 * fixed bug 148: failure to inline-expand a local function left
2687 garbage, confusing the compiler.
2688 * fixed bugs 3cd: structure slot readers perform type check if the
2689 slot can have an invalid value (i.e. it is either not initialized
2690 or can be written with a less specific slot writer).
2691 * bug fix: the compiler now traps array references to elements off
2692 the end of an array; previously, the bounds checking in some
2693 circumstances could go off-by-one.
2694 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2696 * type declarations for array element types now obey the description
2697 on the CLHS page "Declaration TYPE", as per discussions on
2698 sbcl-help around 2003-05-08. This means that a declaration
2699 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2700 declaration, all references to BAR will be asserted or assumed
2701 (with THE, so dependent on compiler policy) to involve objects of
2702 type FOO. Note that no such declaration is implied in
2703 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2704 * declared types of functions from the "Conditions"
2705 chapter. (reported by Paul Dietz)
2706 * bug fix: CERROR accepts a function as its first argument.
2707 * bug fix: NTH an NTHCDR accept a bignum as index
2708 arguments. (reported by Adam Warner)
2709 * optimization: character compare routines now optimize comparing
2710 against a constant character. (reported by Gilbert Baumann)
2711 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2712 argument now works properly on the MIPS platform.
2713 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2714 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2715 declared to have a functional type, does not cause an error
2717 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2718 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2719 * fixed bug: if last continuation of a deleted block has a
2720 destination, this destination should be deleted too. (reported by
2722 * fixed a bug in the bootstrap process: the host compiler's values
2723 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2724 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2725 test case from Patrik Nordebo)
2726 * improved the ability of the disassembler on the PPC platform to
2727 provide helpful disassembly notes.
2728 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2729 the class in more cases than previously.
2730 * bug fix: FILE-POSITION now understands :START and :END for
2731 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2732 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2733 8) data. (thanks to David Lichteblau)
2734 * bug fix: it is possible to add a method to a generic function
2735 without lambda list.
2736 * bug fix: reader failed to signal END-OF-FILE inside an
2737 object representation. (reported by Nikodemus Siivola)
2738 * fixed some bugs revealed by Paul Dietz' test suite:
2739 ** LAST and [N]BUTLAST should accept a bignum.
2740 ** condition slot accessors are methods.
2741 ** (VECTOR NIL) is a subtype of STRING.
2743 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2744 * minor incompatible change: some nonsensical specialized lambda
2745 lists (used in DEFMETHOD) which were previously ignored now signal
2747 * minor incompatible change: the system is now aware of the types of
2748 variables in the COMMON-LISP package, and will signal errors for
2749 most violations of these type constraints (where previously they
2750 were silently accepted).
2751 * minor incompatible change: COMPILE-FILE now uses the freedom
2752 afforded (ANSI 3.2.2.3) to use derived function types for
2753 functions defined in the same file. This also permits the system
2754 to warn on static type mismatches and function
2755 redefinition. (Currently it does not work with high DEBUG level.)
2756 * minor incompatible change: VALUES declaration is disabled.
2757 * When issuing notes, the compiler now signals a condition of type
2758 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2759 restart for use in user handlers. It is expected that the
2760 COMPILER-NOTE condition will eventually become a condition
2761 supertype to a hierarchy of note types, which will then be
2762 handleable in a similar fashion. However, at the moment, no such
2763 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2764 but it's an implementation detail, not a classification for the
2766 * Changes in type checking closed the following bugs:
2767 ** type checking of unused values (192b, 194d, 203);
2768 ** template selection based on unsafe type assertions (192c, 236);
2769 ** type checking in branches (194bc).
2770 * A short form of VALUES type specifier has ANSI meaning (it has
2771 increased the number of situations when SBCL cannot perform type
2773 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2774 keywords or constants is permissible.
2775 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2776 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2777 classes. (thanks to Antonio Martinez)
2778 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2779 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2780 argument to a pretty-print stream. Instead, it leaves any such
2781 operations to DESCRIBE-OBJECT methods.
2782 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2784 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2785 lambda lists. (thanks to David Lichteblau)
2786 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2787 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2788 (thanks to Gerd Moellmann)
2789 * fixed bug 246: increased compilation speed of long
2790 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2792 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2794 * DEFCONSTANT now throws a condition of type
2795 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2796 constant to a non-EQL value; CONTINUE and ABORT restarts
2797 respectively change and preserve the value.
2798 * fixed bug 63: The code walker, part of the implementation of CLOS,
2799 is now better at handling symbol macros.
2800 * bug fix: There is no longer an internal implementation type named
2801 CL:LENGTH. (reported by Raymond Toy)
2802 * bug fix: In macro-like defining macros/special operators the
2803 implicit block does not enclose the lambda list.
2804 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2805 enclosed in parentheses) are not suitable as type specifiers, and
2806 their use properly signals an error now.
2807 * bug fix: An argument count mismatch for a type specifier in code
2808 being compiled no longer causes an unhandled error at compile
2809 time, but signals a compile-time warning.
2810 * fixed simple vector readable printing
2811 * bug fix: DESCRIBE takes more care over whether the class
2812 precedence list slot of a class is bound before accessing it.
2813 (reported by Markus Krummenacker)
2814 * bug fix: FORMATTER can successfully compile pretty-printer format
2815 strings which use variants of the ~* directive inside.
2816 * bug fix: SEARCH now applies its TEST predicate to the elements of
2817 the arguments in the correct order. (thanks to Wolfhard Buss)
2818 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2819 in type inference. (thanks to Robert E. Brown)
2820 * bug fix: Reading in symbols with an explicit package name of ""
2821 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2822 * fixed some bugs revealed by Paul Dietz' test suite:
2823 ** NIL is now allowed as a structure slot name.
2824 ** Arbitrary numbers, not just REALs, are allowed in certain
2825 circumstances in LOOP for-as-arithmetic clauses.
2826 ** Multiple class redefinitions before slot access no longer
2827 causes a type error.
2828 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2829 association between the name and a class.
2830 ** Generic functions with non-standard method-combination and over
2831 five methods all of which return constants no longer return NIL
2832 after the first few invocations. (thanks to Gerd Moellmann)
2833 ** CALL-NEXT-METHOD with no arguments now passes the original
2834 values of the arguments, even in the presence of assignment.
2835 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2837 ** DISASSEMBLE works with closures and funcallable instances.
2838 ** ADD-METHOD now returns the generic function, not the new method.
2839 ** FIND-METHOD signals an error if the lengths of the specializers
2840 is incompatible with the generic function, even if the ERRORP
2842 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2843 which its argument is a member.
2844 ** DEFCLASS only redefines the class named by its class-name
2845 argument if that name is the proper name of the class;
2846 otherwise, it creates a new class.
2847 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2848 of the UNBOUND-SLOT condition to the name of the slot.
2849 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2850 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2851 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2852 SLOT-MAKUNBOUND in the specified fashion.
2854 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2855 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2856 cross-compilation host. As a consequence, we can now bootstrap our
2857 way up to SBCL starting with a bare gcc toolchain and human-readable
2858 source code (first the source to CLISP, then the source to SBCL).
2859 * A contributed module containing a partial implementation of the
2860 simple-streams interface has been included. (thanks to Rudi
2862 * A contributed module implementing the RFC1321 Message Digest
2863 Algorithm, known as MD5, has been included.
2864 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2865 no longer has any effect, as the code controlled by this feature
2866 has been deleted. (As far as we know, no-one has ever built using
2867 this feature, and its semantics were confused in any case).
2868 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2869 work on conditions (as required by the ANSI specification),
2870 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2871 expected behaviour on conditions. Users should note, however,
2872 that such behaviour is not required by the ANSI specification,
2873 and so use of this behaviour may render their code unportable.
2874 * fixed some bugs revealed by Paul Dietz' test suite:
2875 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2877 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2878 ** SXHASH on condition objects no longer returns NIL.
2879 ** :ALLOCATION :CLASS slots are better treated; their values are
2880 updated on class redefinition, and initforms inherited from
2881 superclasses are applied.
2882 ** REMOVE-METHOD returns its generic function argument even when
2883 no method was removed.
2884 ** SHARED-INITIALIZE now initializes the values of the requested
2885 slots, including those with :ALLOCATION :CLASS.
2886 ** ALLOCATE-INSTANCE now works on structure classes defined via
2887 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2889 ** SLOT-EXISTS-P now works on conditions, as well as structures
2891 ** MAKE-LOAD-FORM now has the required methods on
2892 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2893 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2894 keyword, and now implements the SLOT-NAMES argument.
2895 ** methods with &OPTIONAL arguments no longer allow too many
2896 arguments to be passed in the call without error.
2897 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2898 option is consistent with the required arguments of the generic
2899 function lambda list.
2900 * bug fix: REQUIRE accepts a string designator. (Thanks to
2902 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2903 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2904 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2905 * bug fix: The compiler now checks for duplicated variables in macro
2907 * bug fix: SETQ on globals returns the correct value.
2908 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2909 not a COMPILER-ERROR (followed by some other strange error on
2910 choosing the CONTINUE restart).
2911 * bug fix: make.sh and friends are now more consistent in the way that
2912 they look for GNU "make".
2914 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2915 * experimental native threads support (on x86 Linux >=2.4 only).
2916 This is not compiled in by default: you need to add :SB-THREAD to
2917 the target features. See the "Beyond ANSI" chapter of the manual
2919 * fix for longstanding nonANSIism: The old distinction between
2920 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2921 The return value from CL:FIND-CLASS is now a CLOS class, and
2922 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2923 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2924 * An interface to the MetaObject Protocol, as described in Kiczales,
2925 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2926 MIT Press, 1991, is available from the SB-MOP package.
2927 * incompatible change: the SB-PCL package should now be considered
2928 a private implementation detail, and no longer a semi-private MOP
2930 * minor incompatible change: due to rearrangement for threads, the
2931 control stack and binding stack are now allocated at arbitrary
2932 addresses instead of being hardcoded per-port. Users affected by
2933 this probably have to be doing advanced things with shared
2934 libraries, and will know who they are.
2935 * minor incompatible change: Previously, all --eval forms used were
2936 processed with READ before any of them were processed with EVAL.
2937 Now each --eval form is processed with both READ and EVAL before
2938 the next --eval form is processed. (Thus package operations like
2939 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2940 work as the user might reasonably expect.)
2941 * minor incompatible change: *STANDARD-INPUT* is now only an
2942 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2944 * minor incompatible change: Y-OR-N-P is now character-oriented,
2945 not line oriented. Also, YES-OR-NO-P now works without errors.
2946 (thanks to Antonio Martinez)
2947 * sb-aclrepl module improvements: an integrated inspector, added
2948 repl features, and a bug fix to :trace command.
2949 * Known functions, which cannot be open coded by the backend, are
2950 considered to be able to check types of their arguments. (fixing
2951 a bug report by Nathan J. Froyd)
2952 * fixed a bug in computing method discriminating functions: It is
2953 now possible to define methods specialized on classes which have
2954 forward-referenced superclasses. (thanks to Gerd Moellmann)
2955 * fixed evaluation order in optional entries (reported by Gilbert
2957 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2958 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2959 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2960 required-by-AMOP NAME argument, as well as CLASS and
2961 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2962 * fixed bug 20: DEFMETHOD can define methods using names that are
2963 not the proper names of classes to designate class specializers.
2964 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2965 Unix file descriptors, instead of blowing up. (thanks to Antonio
2967 * Garbage collection refactoring: user-visible change is that a
2968 call to the GC function during WITHOUT-GCING will not do garbage
2969 collection until the end of the WITHOUT-GCING. If you were doing
2970 this you were probably losing anyway.
2971 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2972 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2973 TYPEP the latter but not the former.
2974 * The compiler issues a full WARNING for calls to undefined functions
2975 with names from the CL package.
2976 * MAP-INTO for a vector destination is open coded. (reported by
2977 Brian Downing on c.l.l)
2978 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2979 documentation string.
2980 * fixed some bugs revealed by Paul Dietz' test suite:
2981 ** COPY-ALIST now signals an error if its argument is a dotted
2983 ** Condition slots are now accessed more correctly in the presence
2984 of multiple initargs for a given slot.
2985 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2986 exclude restarts of the same name associated with a different
2988 ** DEFCLASS of forward-referenced classes with another
2989 forward-referenced class in the superclasses list no longer
2991 ** Condition slots are now initialized once each, not multiple
2992 times. (thanks to Gerd Moellmann)
2993 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2995 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2996 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2997 arguments contain duplicated elements.
2998 ** RESTART-CASE understands local macros.
2999 ** RESTART-CASE associates exactly its own restarts with a condition.
3000 ** ENDP in safe mode checks its argument to be of type LIST.
3001 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3002 in question is unbound.
3003 ** Optimized MAKE-INSTANCE functions no longer cause internal
3004 assertion failures in the presence of duplicate initargs.
3005 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3006 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3007 propagated even in the ctor optimized implementation of
3009 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3011 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3012 * a better implementation of SXHASH on (simple) bit vectors,
3013 measured both in execution speed and in distribution of results
3014 over the positive fixnums, has been installed. Likewise, a better
3015 implementation of EQUAL for simple bit vectors is now available.
3016 * fixed CEILING optimization for a divisor of form 2^k.
3017 * fixed bug 240 (emitting extra style warnings "using the lexical
3018 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3019 by Antonio Martinez)
3020 * fixed SXHASH, giving different results for NIL depending on type
3021 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3022 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3023 variable. (found by Rolf Wester)
3024 * SBCL does not ignore type declarations for special
3025 variables. (reported by rif on c.l.l 2003-03-05)
3026 * some bug fixes in contrib/sb-aclrepl/
3027 * fixed some bugs revealed by Paul Dietz' test suite:
3028 ** a bug in the CONS type specifier, whereby the CAR and CDR
3029 types got intertwined, has been fixed;
3030 ** the type system is now able to reason about the interaction
3031 between INTEGER and RATIO types more completely;
3032 ** APPEND, [N]REVERSE and NRECONC check that those their
3033 arguments, which must be proper lists, are really so;
3034 ** An array specialized to be unable to hold elements has been
3035 implemented, as required -- yes, really -- by ANSI;
3036 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3037 on malformed property lists;
3039 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3040 * incompatible packaging change: in line with Unix convention,
3041 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3042 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3043 * REQUIRE and PROVIDE are now optionally capable of doing something
3044 useful. See the documentation string for REQUIRE.
3045 * infrastructure for a managed SBCL contrib system: contributed
3046 modules in this release include:
3047 ** the ASDF system definition facility;
3048 ** an interface to the BSD Sockets API;
3049 ** an ACL-like convenience interface to the repl;
3050 (thanks to Kevin Rosenberg)
3051 ** an implementation of ROTATE-BYTE, with efficient implementation
3053 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3054 gives a closer approximation to the right answer than previously.
3055 (thanks to Raymond Toy)
3056 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3057 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3058 optional environment arguments, as required by ANSI.
3059 * fixed bugs in other functions taking environment objects, allowing
3060 calls with an explicit NIL environment argument to be compiled
3062 * fixed bug 228: primary return values from
3063 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3064 COMPILE or FUNCTION.
3065 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3066 :TYPE will no longer signal errors on innocuous objects.
3067 * fixed bug 231b: SETQ is better at respecting type declarations in
3068 the lexical environment.
3069 * fixed a bug in DEFCLASS: classes named by symbols with no or
3070 unprintable packages can now be defined.
3071 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3072 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3073 * fixed compiler failure related to checking types of functions.
3074 (reported by Robert E. Brown)
3075 * the compiler is now much more consistent in its error-checking
3076 treatment of bounding index arguments to sequence functions: in
3077 (SAFETY 3) code, errors will be signalled in almost all cases if
3078 invalid sequence bounding indices are passed to functions defined
3079 by ANSI to operate on sequences.
3080 * fixed a bug in the build procedure: documentation of SBCL-specific
3081 packages is now preserved and available in the final Lisp image.
3082 * lifted FDEFINITION lookup out of loops in the implementation of
3083 many list operations. (thanks to Robert E. Brown)
3084 * fixed a bug in the reader: the #n# reader macro now works for
3085 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3086 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3087 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3088 E. Brown and Tony Martinez respectively)
3089 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3090 SLOT-BOUNDP: in particular, it is now able to optimize them much
3091 better, and is now not vulnerable to having packages renamed.
3092 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3093 always called when a slot is not present in an instance. (thanks
3095 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3096 longer signal an error on unreading a character following EOF on
3097 the previous constituent stream. (thanks to Tony Martinez)
3098 * fixed some bugs revealed by Paul Dietz' test suite:
3099 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3100 not just nonnegative fixnums;
3101 ** the logical bit-array operators such as BIT-AND now accept an
3102 explicit NIL for their "opt-arg" argument (to indicate a
3103 freshly-consed result bit-array);
3104 ** ELT now signals an error on an invalid sequence index in safe
3106 ** the type system is now cleverer about negations of numeric
3107 types, and consequently understands the BIGNUM and RATIO types
3109 ** the type system is now cleverer about the interaction between
3110 INTEGER and RATIO types: while bugs still remain, many more
3111 cases are accurately computed;
3112 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3113 if it is in the last clause;
3114 ** CONSTANTLY now correctly returns a side-effect-free function in
3116 ** DECLARE is no longer treated as a special-operator; in
3117 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3118 * incremented fasl file version number due to the change in the
3119 DEFSTRUCT-SLOT-DESCRIPTION structure.
3121 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3122 * minor incompatible change: code processed by the "interpreter" or
3123 EVAL now has a compilation optimization policy of (DEBUG 2)
3124 (changed from (DEBUG 1)) to improve debuggability of interactive
3125 development, and to allow the use of the debug RETURN command in
3127 * an experimental implementation of the RETURN command for the
3128 debugger has been included. (thanks to Frederik Kuivinen)
3129 * fixed bug 62: constraints were not propagated into a loop.
3130 * fixed bug in embedded calls of SORT (reported and investigated by
3132 * fixed some bugs revealed by Paul Dietz' test suite:
3133 ** printing and reading of arrays with some dimensions having
3134 length 0 (thanks to Gerd Moellmann);
3135 ** BOA constructor with &AUX argument without a default value does
3136 not cause a type error;
3137 ** CONSTANTP now returns true for all self-evaluating objects.
3139 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3140 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3141 accessors that are related by inheritance, as specified in the
3142 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3143 Valtteri Vuorikoski)
3144 * The compiler is now able to inline functions that were defined in
3145 a complex lexical environment (e.g. inside a MACROLET).
3146 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3148 * Support for the upcoming FreeBSD-5.0 release has been included.
3149 (thanks to Dag-Erling Smorgrav)
3150 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3151 effect when it is not in a toplevel context.
3152 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3153 stand a better chance of being correct. (thanks to Gerd
3155 * fixed bug in COERCE, which now signals an error on coercing a
3156 rational to a bounded real type which excludes the expected
3158 * The compiler is now able to derive types more accurately from the
3159 COERCE and COMPILE functions.
3160 * fixed bug 223: functional binding is considered to be constant
3161 only for symbols in the CL package.
3162 * fixed bug 231: SETQ did not check the type of a variable being set
3163 (reported by Robert E. Brown)
3164 * A new optimization for MAKE-INSTANCE has been included, fixing
3165 various bugs (including relating to :ALLOCATION :CLASS slots and
3166 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3168 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3169 ** As required by ANSI, LOOP now disallows anonymous collection
3170 clauses such as COLLECT I in conjunction with aggregate boolean
3171 clauses such as THEREIS (= I 1);
3172 ** LOOP now signals an error when any variable is reused in the
3173 same loop (including the potentially useful construct analogous
3174 to WITH A = 1 WITH A = (1+ A);
3175 ** IT is only a special loop symbol within the first clause of a
3176 conditional loop clause;
3177 ** LOOP with a typed iteration variable over a hashtable now
3178 signals a type error iff it should.
3179 * fixed some other bugs revealed by Paul Dietz' test suite:
3180 ** FILE-STREAM now names the class previously known as FD-STREAM;
3181 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3182 argument) no longer signals an error;
3183 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3184 of the slot symbol, rather than using the current package
3185 ((:CONC-NAME "") continues to intern the slot's name in the
3187 * incremented fasl file version number, because of the incompatible
3188 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3189 of the new implementation of DEFINE-COMPILER-MACRO.
3191 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3192 * Support for building SBCL for MIPS platforms running in
3193 little-endian mode has now been checked in, and basic
3194 functionality on said platforms verified.
3195 * minor incompatible change: PCL now records the pathname of a file
3196 in which methods and the like are defined, rather than its
3198 * minor incompatible change: TRUENAME now considers the truename of
3199 a file naming a directory to be the pathname with :DIRECTORY
3200 component indicating that directory.
3201 * minor incompatible change: a NAMED clause in the extended form of
3202 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3203 reason for the previous behaviour is unclear.
3204 * more systematization and improvement of CLOS and MOP conformance
3205 in PCL (thanks to Gerd Moellman and Pierre Mai):
3206 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3208 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3209 primary methods with no specializers;
3210 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3212 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3213 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3214 FORWARD-REFERENCED-CLASSes; error reporting on
3215 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3217 ** SXHASH on CLOS instances now uses a slot internal to the
3218 instance to return different numbers on distinct instances,
3219 while preserving the same return value through invocations of
3221 ** DEFMETHOD signals errors when methods with longer incongruent
3222 lambda lists are added to generic functions;
3223 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3224 CLASS, as specified in AMOP;
3225 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3226 on the order returned by the primary method for classes of
3227 class STANDARD-CLASS;
3228 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3229 * fixed some bugs shown by Paul Dietz' test suite:
3230 ** DOLIST puts its body in TAGBODY;
3231 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3233 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3234 value producing form;
3235 ** if more variables are given to PROGV than values, extra
3236 variables are bound and made to have no value;
3237 ** NSUBSTITUTE on list arguments gets the right answer with
3239 ** ELT signals an error of type TYPE-ERROR when the index argument
3240 is not a valid sequence index;
3241 ** LOOP signals (at macroexpansion time) an error of type
3242 PROGRAM-ERROR when duplicate variable names are found;
3243 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3244 ** FUNCALL of special-operators now cause an error of type
3246 ** PSETQ now works as required in the presence of side-effecting
3247 symbol-macro places;
3248 ** NCONC accepts any object as its last argument;
3249 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3251 ** loop-for-as-package does not require a package to be explicitely
3253 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3255 * fixed bug 166: compiler preserves "there is a way to go"
3256 invariant when deleting code.
3257 * fixed bug 172: macro lambda lists with required arguments after
3258 &REST arguments now cause an error to be signalled. (thanks to
3260 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3261 bug. (thanks to Matthew Danish)
3262 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3264 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3265 arguments are lexically rebound. (thanks to Gerd Moellmann and
3267 * fixed bug 194: error messages are now more informative when there
3268 is no primary method applicable in a call to a generic
3269 function. (thanks to Gerd Moellmann)
3270 * fixed bug in command line argument checking (thanks to Julian
3272 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3273 (thanks to Matthew Danish)
3274 * incremented fasl file version number, because of the
3275 SXHASH-related changes in the layout of CLOS data structures
3277 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3278 * minor incompatible change: The runtime (the Unix executable named
3279 "sbcl") is now much pickier about the .core files it will load.
3280 Essentially it now requires .core files to descend from the same
3281 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3282 as the runtime does. (The intent is to prevent the crashes which
3283 can occur, and which can even be reported as mysterious failures,
3284 when people patch the sources or change the build parameters
3285 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3286 sbcl and .core files.)
3287 * fixed bug: VALUES-LIST is no longer optimized away.
3288 * fixed bug 142: The FFI conversion of C string values to Lisp
3289 string values no longer conses excessively. (thanks to Nathan
3290 Froyd porting Raymond Toy's fix to CMU CL)
3291 * began to systematize and improve MOP conformance in PCL (thanks to
3292 Nathan Froyd, Gerd Moellman and Pierre Mai):
3293 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3295 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3296 ** FINALIZE-INHERITANCE is now called on class finalization;
3297 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3298 argument precedence order.
3299 * fixed bug 202: The compiler no longer fails on functions whose
3300 derived types contradict their declared type.
3301 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3302 so it can be non-toplevel.
3303 * The fasl file version number has changed (because of the new
3304 implementation of DEFMACRO).
3305 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3306 safe code, that any length requirement by their type-specifier
3307 argument is valid. The exceptions to this are described in bug
3309 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3310 functions now check the types of their inputs as required by ANSI.
3311 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3312 introduced symbol is DECLAREd to be SPECIAL.
3313 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3314 * fixed bug: :COUNT argument to sequence functions may be negative
3315 * fixed bug: body of DO-SYMBOLS may contain declarations
3316 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3317 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3318 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3319 symbol macro only once
3320 * fixed printing of call frame when argument list is unavailable
3321 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3322 * compiler no longer signals WARNING on unknown keyword
3325 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3326 * A beta-quality port to the mips architecture running Linux,
3327 based on the old CMUCL backend, has been made. It has been tested
3328 on a big-endian kernel, and works sufficiently well to be able to
3329 rebuild itself; it has not been tested in little-endian mode.
3330 * fixed an inconsistency between gencgc.c and purify.c which made
3331 dumping/loading .core files unreliable
3332 * fixed bug 120a: The compiler now deals correctly with IFs where
3333 the consequent is the same as the alternative, instead of
3334 misderiving the return type. (thanks to Alexey Dejneka)
3335 * fixed bug 113: Logical pathnames are now dumpable (the logical
3336 host is resolved at load-time, throwing an error if it is not
3338 * fixed bug 174: FORMAT's error message is slightly clearer when a
3339 non-printing character is used in a format directive.
3340 * fixed several bugs in compiler checking of type declarations, i.e.
3341 violations of the Python "declarations are assertions" principle
3342 (thanks to Alexey Dejneka)
3343 * fixed several bugs in PCL's error checking (thanks to Gerd
3345 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3347 * fixed bug in compilation of functions as first class values
3348 (thanks to Antonio Martinez-Shotton)
3349 * The compiler's handling TYPE-ERRORs which it can prove will
3350 inevitably happen at runtime has been cleaned up and corrected
3351 in several ways. (thanks to Alexey Dejneka)
3352 * improved argument type checking for various basic arithmetic
3353 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3354 lavished on them in the past that they can be compiled in many
3355 ways in different special cases
3356 * fixed bug 181: compiler checks validity of user supplied type
3358 * cleaned up code flushing in optimization: Function calls which
3359 should signal errors for safety purposes (e.g. which ANSI says
3360 should signal errors when their arguments are of incorrect type)
3361 are no longer optimized away.
3362 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3363 * incremented fasl file version number, because changes in the
3364 implementation of sequence functions like COERCE caused
3365 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3366 in old inline expansions) to become undefined. (Actually these
3367 changes were later undone, so we might very well be binary
3368 compatible with 0.7.7 after all, but leaving the version number
3369 incremented seemed like the simplest and most conservative
3372 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3373 * An alpha-quality port to the parisc architecture running Linux,
3374 based on the old CMUCL backend, has been made. This, even more so
3375 than the other backends, should be considered still a work in
3376 progress; known problems include that the Linux kernel in 64-bit
3377 mode does not propagate the correct sigcontext structure to
3378 userspace, and consequently SBCL on a parisc64 kernel will not
3380 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3381 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3382 suppressed.) Also now that the compiler is looking at declarations
3383 in the environment, it checks optimization declarations as well,
3384 and suppresses inlining when (> DEBUG SPEED).
3385 * More fixes have been made to treatment of floating point exception
3386 treatment and other Unix signals. In particular, floating point
3387 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3388 * The detection and handling of control stack exhaustion (infinite
3389 or very deeply nested recursion) has changed. Stack exhaustion
3390 detection is now done by write-protecting pages at the OS level
3391 and applies at all optimization settings; when found, a
3392 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3393 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3394 be caught using IGNORE-ERRORS.
3395 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3396 correctly when transforming recursive LABELS functions to LETs.
3397 (thanks to Alexey Dejneka)
3398 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3399 that are names of constants or global variables.
3400 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3401 alien routines with docstrings.
3402 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3403 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3405 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3406 (thanks to Eric Marsden)
3407 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3408 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3409 * Bug fix: The SPARC backend can now compile functions involving
3410 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3411 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3412 to a functions expecting a PATHNAME-DESIGNATOR.
3413 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3414 lambda lists. (thanks to Alexey Dejneka)
3415 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3416 functions so that EXPORTing the name of the class doesn't cause
3417 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3418 to Antonio Martinez for reporting this)
3419 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3420 constant-folded without failing an assertion. (thanks to Einar
3421 Floystad Dorum for reporting this)
3422 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3423 types are now dealt with more correctly by the compiler.
3424 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3425 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3426 behaviour analogous to RENAME-FILE. This puts its behaviour more
3427 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3429 * The fasl file version number has changed again. (because of the
3430 bug fix involving the names of PCL MAKE-INSTANCE functions)
3432 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3433 * bug fix: Floating point exceptions are treated much more
3434 consistently on the x86/Linux and PPC/Linux platforms.
3435 * Array initialization with :INITIAL-ELEMENT is now much faster for
3436 cases when the compiler cannot open code the array creation, but
3437 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3438 array accesses have also seen a speed increase.
3439 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3440 specified. (This was at the root of some bad interactions between
3441 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3443 * bug fix: Internal error arguments for undefined functions are now
3444 computed correctly on the PPC/Linux platform.
3445 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3446 Raymond Toy's patch for CMU CL)
3447 * Support for the Solaris 9 operating environment has been included
3448 (thanks to Daniel Merritt)
3449 * A very ugly but hopefully complete draft of the missing FFI chapter
3450 of the manual has been created by reformatting the corresponding
3451 CMU CL manual chapter into (currently very ugly and incoherent)
3452 DocBook and bringing it up to date for SBCL behavior. Thus, the
3453 manual is now essentially complete, at least by my extreme
3454 once-and-only-once standards, whereby it's acceptable to refer to
3455 the doc strings of SB-EXT functions as the primary documentation.
3456 * The fasl file version number has changed again, due to cleanup of
3457 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3458 is no longer a static symbol.)
3460 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3461 * SBCL now builds with OpenMCL (version 0.12) as the
3462 cross-compilation host; also, more progress has been made toward
3463 bootstrapping under CLISP.
3464 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3466 * bug 158 fixed: The compiler can now deal with integer loop
3467 increments different from 1; fixing this turned out also to fix
3469 * bug 169 fixed: no more bogus warnings about using lexical bindings
3470 despite the presence of perfectly good SPECIAL declarations (thanks
3471 to David Lichteblau)
3472 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3473 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3474 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3476 * bug 180 fixed: Method combination specifications no longer ignore
3477 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3478 * bug fix: Structure type predicate functions now check their argument
3479 count as they should.
3480 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3481 correctly. (thanks to Pierre Mai)
3482 * minor incompatible change: The --noprogrammer option is deprecated
3483 in favor of the new --disable-debugger option, which is very similar.
3484 (The major difference is that it takes effect at a slightly different
3485 time at startup, causing handling of errors in --sysinit and
3486 --userinit files will be affected differently.) The
3487 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3488 been added to allow this functionality to be controlled from ordinary
3489 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3490 maintainers, who might want to run non-interactive scripts to
3491 build SBCL cores which will later be used interactively.)
3492 * minor incompatible change: The LOAD function no longer, when given
3493 a wild pathname to load, loads all files matching that pathname.
3494 Instead, an error of type FILE-ERROR is signalled.
3496 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3497 * bug 147 fixed: The compiler preserves its block link/count
3498 invariants more correctly now so that it doesn't crash. (thanks
3500 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3502 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3503 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3504 no-value-for-this-character as NIL (as specified by ANSI).
3505 * HOST-NAMESTRING on physical pathnames now returns a string that is
3506 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3507 (thanks to Christophe Rhodes)
3508 * The Alpha port handles icache flushing more correctly. (thanks to
3510 * More progress has been made toward bootstrapping under CLISP. (thanks
3511 to Christophe Rhodes)
3512 * The fasl file format has changed again, because dynamic loading
3513 on OpenBSD (which has non-ELF object files) motivated some cleanups
3514 in the way that foreign symbols are transformed and passed around.
3515 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3516 is no longer treated as whitespace by the reader, but instead as
3517 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3518 |A
\7fB|, instead of A as it used to.
3520 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3521 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3522 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3523 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3524 this as of 0.7.1.45, but was left out of the previous news section
3525 (thanks to Dan Barlow)
3526 * SBCL now runs on the Solaris operating system on SPARC architectures
3527 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3528 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3529 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3530 * SPARC backend cleanups, allowing builds of cores optimized for V8
3531 and V9 SPARCS, and also emission of code targeted to a particular
3532 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3534 * SBCL is closer to bootstrapping under CLISP, thanks to various
3535 fixes by Christophe Rhodes.
3536 * The fasl file format has changed again, to allow the compiler's
3537 INFO database to support symbol macros.
3538 * The user manual (in doc/) is formatted into HTML more nicely.
3539 (thanks to coreythomas)
3540 * The system is smarter about SUBTYPEP relationships, especially
3541 those involving NOT types (including types such as ATOM which are
3542 represented internally using NOT types). Thus SUBTYPEP is less
3543 likely to return (VALUES NIL NIL) in general, and in particular
3544 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3546 * The fasl file format has changed again, because the internal
3547 representation of types now includes a new slot to support the new
3548 SUBTYPEP-of-NOT-types logic.
3549 * (not a change in the main branch of SBCL, but a related prototype
3550 which can hopefully be merged into the main branch of SBCL in the
3551 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3552 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3553 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3554 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3555 dispatch macro character combinations. (thanks to Alexey Dejneka)
3556 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3557 that superficially look like logical namestrings correctly.
3558 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3560 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3561 default supplied-p) for &optional and &key arguments. (thanks to
3564 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3565 * incompatible change: The compiler is now less aggressive about
3566 tail call optimization, doing it only when (> SPACE DEBUG) or
3567 (> SPEED DEBUG). (This is an incompatible change because there are
3568 programs which relied on the old CMU-CL-style behavior to optimize
3569 away their unbounded recursion which will now die of stack overflow.)
3570 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3571 for non-GENCGC systems has been increased to 20M (since that
3572 seems much closer to the likely performance optimum for modern
3573 systems than the old 4M value was)
3574 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3575 and *DEBUG-PRINT-LEVEL*
3576 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3577 of CMU CL's support for SPARC, and various endianness and other
3578 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3579 * new syntactic sugar for the Unix command line: --load foo.bar is now
3580 an alternate notation for --eval '(load "foo.bar")'.
3582 ** The system now detects stack overflow and handles it gracefully,
3583 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3584 optimization settings. (This is a good thing in general, and
3585 its introduction in this version should be particularly timely
3586 for anyone whose code fails because of suppression of tail
3588 ** The system now hunts for the C variable "environ" in a more
3589 devious way, to avoid segfaults when the C library version differs
3590 between compile time and run time. (thanks to Christophe Rhodes)
3591 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3592 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3593 ** The compiler no longer issues bogus style warnings for undefined
3594 classes in the same source file as the DEFCLASSes which defined
3595 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3597 ** fixes in CONDITION class precedence list for undefined function
3598 errors (thanks to Alexei Dejneka)
3599 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3600 correctly. (thanks to Dan Barlow)
3601 ** portability fixes aiming at bootstrapping under CLISP (thanks
3602 to Dave McDonald and Christophe Rhodes)
3603 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3604 ** fixes in type translation and and type inference (thanks to
3606 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3607 ** various fixes in the new ports (thanks to Dan Barlow)
3608 * several changes related to debugging:
3609 ** suppression of tail recursion, as noted above
3610 ** stack overflow detection, as noted above
3611 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3612 is now the default. (For some time encapsulation has been more
3613 reliable than the breakpoint-based :ENCAPSULATE NIL
3614 implementation, at least on X86 systems; and I just noticed that
3615 encapsulation also seems closer to the spirit of the ANSI
3618 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3620 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3621 up properly again. (There was a packaging bug in 0.7.0 which
3622 left their definitions in SB-SYS::LOAD-FOREIGN and
3623 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3624 vital for most things which interface to C-level interfaces,
3625 like extensions working with sockets or databases or
3626 Perl-compatible regexes or whatever, and the need to fix
3627 this bug is the main reason that 0.7.1 was released so
3629 ** DEFGENERIC is now choosier about the methods it redefines, so that
3630 reLOADing a previously-LOADed file containing DEFGENERICs does
3631 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3632 cycle now works as it should. (thanks to Alexey Dejneka)
3633 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3634 fixed by Christophe Rhodes. (He actually submitted this patch
3635 months ago, and I delayed until after 0.7.0.)
3636 ** Bug 111 (internal compiler confusion about runtime checks on
3637 FUNCTION types) was fixed by Alexey Dejneka.
3638 * Some internal cleanups (getting rid of variables which aren't
3639 needed now that the byte interpreter is gone) caused the fasl
3640 file format number to change again.
3642 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3643 * major incompatible change: The default fasl file extension, i.e. the
3644 default extension for files produced by COMPILE-FILE, has changed
3645 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3647 ** There are many changes in the implementation of the compiler.
3648 SBCL is now essentially a compiler-only implementation of ANSI
3649 Common Lisp. EVAL still "interprets" a few special cases, but
3650 almost all the interesting cases are handled by creating
3651 a LAMBDA expression, calling COMPILE on it, then calling
3652 FUNCALL on the result.
3653 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3654 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3655 Since the code is newer, there might still be some new bugs
3656 (though not as many as before Martin Atzmueller's fixes:-). But
3657 the new code is substantially simpler and clearer, and hopefully
3658 any remaining bugs will be simpler, less fundamental, and more
3659 fixable then the bugs in the old code.
3660 ** The revised compiler is still a little unsteady on its feet.
3662 *** The debugging information it produces (particularly the names
3663 of FUNCTION objects) is sometimes much less useful than what
3664 the old compiler produced.
3665 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3666 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3667 within a MACROLET) has been temporarily weakened.
3668 ** There are new compiler optimizations for various functions:
3669 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3670 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3671 *** the math functions TRUNCATE, FLOOR, and CEILING
3672 *** the function-of-all-trades COERCE
3673 Mostly these should be transparent, but there's one
3674 potentially-annoying problem (bug 117): when the compiler
3675 inline-expands a function and does type analysis on the result,
3676 it can create control paths which have type mismatches, and
3677 when it can't prove that those control paths aren't taken,
3678 it will issue WARNINGs about the type mismatches. This is
3679 a particular problem in practice for the new sequence functions.
3680 It's not clear how this should be fixed, and for now, a
3681 workaround is given in the entry for 117 in the BUGS file.
3682 ** (Because of the interaction between the two previous items --
3683 occasional inlining problems and new inline expansions -- some
3684 of the new sequence function optimizations won't really kick in
3685 completely until debugging information, and then inlining, are
3686 straightened out in some future version.)
3687 * minor incompatible changes:
3688 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3689 DIRECTORY no longer implicitly promotes NIL slots of its
3690 pathname argument to :WILD. In particular, when you ask for the
3691 contents of a directory (which you used to be able to do without
3692 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3693 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3694 ** changes in behavior that ANSI explicitly defines to be
3695 implementation dependent:
3696 *** The new compiler-only implementation still conforms with ANSI,
3697 but acts a little different than before. Besides the obvious
3698 changes in performance tradeoffs (that the cost per form passed
3699 to EVAL has gone up, and the cost per form executed by EVAL
3700 has gone down), the behavior of the system changes a little
3701 because there are no longer any interpreted function objects.
3702 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3703 e.g. doing COMPILE on the output of interactive DEFUN is
3705 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3707 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3708 has changed from (:CL) to NIL.
3709 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3710 e.g. #\Newline and #\Tab, don't have names specified in the
3711 ANSI Common Lisp standard, is now based on their ASCII symbolic
3712 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3713 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3714 are no longer used for output.
3715 ** changes in internal implementation constants:
3716 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3717 4 million. (If your application spends a lot of time GCing and
3718 you have a lot of RAM, you might want to experiment with
3719 increasing it even more.)
3720 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3721 However, almost all old code should still continue to work without
3722 immediate update, as SB-C-CALL is now a (deprecated) nickname
3724 ** Old operator names in the style DEF-FOO are now deprecated in
3725 favor of new corresponding names DEFINE-FOO, for consistency with
3726 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3727 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3728 internal symbols, but a few supported extensions like
3729 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3730 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3731 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3732 etc. as you get deeper into recursive calls to the debugger
3733 command loop, instead of the old "5]", "5]]", "5]]]"
3734 sequence. (I was motivated to do this when squabbles between
3735 ILISP and SBCL left me very deeply nested in the debugger. In the
3736 short term, this change will probably provoke more ILISP/SBCL
3737 squabbles, but hopefully it will be an improvement in the long run.)
3738 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3739 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3740 function, since declaiming return types involving aliens is
3741 (1) annoyingly messy to do by hand and (2) vital to efficient
3742 compilation of code which calls such functions.
3743 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3744 longer reexported by the SB-EXT package. They're solely useful
3745 for alien code, so it seems more logical that you should get
3746 them from the SB-ALIEN package, not in SB-EXT.
3747 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3748 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3749 features. Instead, the code that they used to control is always
3750 built into the system.
3751 * many other bug fixes
3752 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3753 advantage of the new EVAL-WHEN stuff and to clean them up in
3754 general, and they are now more ANSI-compliant in a number of
3755 ways. Martin Atzmueller is responsible for a lot of this.
3756 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3758 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3759 *** correct ERROR type for various file operations
3760 *** some fixes for Lisp streams
3761 *** DEFMETHOD syntax checking
3762 *** changing old weird representation of debug information as
3763 strings (which, among their other deficiencies, don't transform
3764 correctly when you rename packages, and don't change their
3765 print representation when you change things like *PACKAGE*
3766 and *PRINT-LENGTH*) to symbols and lists of symbols
3767 He also made several improvements and fixed several bugs in DESCRIBE.
3768 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3770 *** misbehavior of WRITE-STRING/WRITE-LINE
3771 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3772 and several other LOOP problems as well
3773 *** DIRECTORY when similar filenames are present
3774 *** DEFGENERIC with :METHOD options
3775 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3776 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3777 *** argument ordering in FIND with :TEST option
3778 *** mishandled package designator argument in APROPOS-LIST
3779 *** various problems in the backquote readmacro
3780 *** a bug in APROPOS
3781 *** probably some others that I'm not describing very well here,
3782 since the CVS log documents them by reference to sbcl-devel
3783 messages, and the SourceForge archives aren't working well.:-(
3784 ** Dan Barlow improved the Alpha port (and is making progress on the
3785 PPC port, for those of you who think different).
3786 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3787 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3788 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3789 continues to work on the SPARC port (for those of you in a position
3790 to look down upon our little PC-compatible boxes from a great height).
3791 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3792 and uses that copy, rather than the current dynamic value, when
3793 it's trying to decide whether to truncate output. Thus e.g.
3794 (let ((*print-lines* 50))
3795 (pprint-logical-block (stream nil)
3797 (let ((*print-lines* 8))
3798 (print (aref possiblybigthings i) stream)))))
3799 should now truncate the logical block only at 50 lines, instead of
3800 often truncating it at 8 lines, as it did before.
3801 * The doc/cmucl/ directory, containing old CMU CL documentation
3802 from the time of the fork, is no longer part of the base system.
3803 SourceForge has shut down its anonymous FTP service, and with it
3804 my original plan for distributing the old CMU CL documentation
3805 there. For now, if you need these files you can download an old
3806 SBCL source release and extract them from it.
3807 * The fasl file version number changed again, for dozens of reasons,
3808 some of which are apparent above.
3810 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3811 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3812 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3813 patch, so that DISASSEMBLE output is much nicer.
3814 * The code in the SB-PROFILE package now seems reasonably stable.
3815 I still haven't decided what the final interface should look like
3816 (I'd like PROFILE to interact cleanly with TRACE, since both
3817 facilities use function encapsulation) but if you have a need
3818 for profiling now, you can probably use it successfully with
3819 the current CMU-CL-style interface.
3820 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3821 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3822 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3823 link now returns the dangling link itself, and for similar
3824 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3825 link itself. (In these cases the old code signalled an error and
3826 looped endlessly, respectively.) Thus, DIRECTORY now works even
3827 in the presence of dangling and cyclic symbolic links.
3828 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3829 is now a supported extension again, since the consensus on
3830 sbcl-devel was that it can be useful for ordinary development
3831 work, not just for debugging SBCL itself.
3832 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3833 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3834 that currently-defined functions might be redefined later with
3835 different return types.
3836 * Hash tables can be printed readably, as inspired by CMU CL code
3837 of Eric Marsden and SBCL code of Martin Atzmueller.
3838 * better error handling in CLOS method combination, thanks to
3839 Martin Atzmueller porting Pierre Mai's CMU CL patches
3840 * more overflow fixes for >16Mbyte I/O buffers
3841 * A bug in READ has been fixed, so that now a single Ctrl-D
3842 character suffices to cause end-of-file on character streams.
3843 In particular, now you only need one Ctrl-D at the command
3844 line (not two) to exit SBCL.
3845 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3847 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3848 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3849 does the right thing.
3850 * The implementation of some type tests, especially for CONDITION
3851 types, is now tidier and maybe faster, due to CMU CL code
3852 originally by Douglas Crosher, ported by Martin Atzmueller.
3853 * Some math functions have been fixed, and there are new
3854 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3855 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3856 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3857 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3858 likely to be built into sbcl-0.7.0.)
3859 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3860 CL, ported to SBCL by Martin Atzmueller)
3861 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3862 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3863 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3864 * There's a new slam.sh hack to shorten the edit/compile/debug
3865 cycle for low-level changes to SBCL itself, and a new
3866 :SB-AFTER-XC-CORE target feature to control the generation of
3867 the after-xc.core file needed by slam.sh.
3868 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3869 COMPILE-FILE is no longer supported, so that now every function
3870 gets an entry point, so that block compilation looks a little
3871 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3872 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3873 returns the number of bytes consed since the system started,
3874 rather than the number consed since the first time the function
3875 was called. (The new definition parallels ANSI functions like
3876 CL:GET-INTERNAL-RUN-TIME.)
3877 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3878 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3879 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3880 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3881 * Fasl file version numbers are now independent of the target CPU,
3882 since historically most system changes which required version
3883 number changes have affected all CPUs equally. Similarly,
3884 the byte fasl file version is now equal to the ordinary
3887 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3888 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3889 is no longer recognized. I apologize for this, because it was
3890 listed in SB-EXT as a supported extension, but I found that
3891 its existing behavior was poorly specified, as well as incorrectly
3892 specified, and it looked like too much of a mess to straighten it
3893 out. I have enough on my hands trying to get ANSI stuff to work..
3894 * many patches ported from CMU CL by Martin Atzmueller, with
3895 half a dozen bug fixes in pretty-printing and the debugger, and
3896 half a dozen others elsewhere
3897 * fixed bug 13: Floating point infinities are now supported again.
3898 They might still be a little bit flaky, but thanks to bug reports
3899 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3900 as flaky as they were.
3901 * The --noprogrammer command line option is now supported. (Its
3902 behavior is slightly different in detail from what the old man
3903 page claimed it would do, but it's still appropriate under the
3904 same circumstances that the man page talks about.)
3905 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3906 are now supported, and enabled by default. Thus, the compiler can
3907 handle many floating point and complex operations much less
3908 inefficiently. (Thus e.g. you can implement a complex FFT
3910 * The compiler now detects type mismatches between DECLAIM FTYPE
3911 and DEFUN better, and implements CHECK-TYPE more correctly, and
3912 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3913 thanks to patches from Martin Atzmueller.
3914 * various fixes to make the cross-compiler more portable to
3915 ANSI-conforming-but-different cross-compilation hosts (notably
3916 Lispworks for Windows, following bug reports from Arthur Lemmens)
3917 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3918 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3919 fixed thanks to Pierre Mai's CMU CL patches.
3920 * fixes to keep the system from overflowing internal counters when
3921 it tries to use i/o buffers larger than 16M bytes
3922 * fixed bug 45a: Various internal functions required to support
3923 complex special functions have been merged from CMU CL sources.
3924 (When I was first setting up SBCL, I misunderstood a compile-time
3925 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3926 * improved support for type intersection and union, fixing bug 12
3927 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3928 more obscure bugs as well
3929 * some steps toward byte-compiling non-performance-critical
3930 parts of the system, courtesy of patches from Martin Atzmueller
3931 * Christophe Rhodes has made some debian packages of sbcl at
3932 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3933 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3934 stable, but are nonetheless usable. When he's ready, I'd be happy
3935 to add them to the SourceForge "File Releases" section. (And if
3936 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3937 * new fasl file format version number (because of changes in
3938 internal representation of (OR ..) types to accommodate the new
3939 support for (AND ..) types, among other things)
3941 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3942 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3944 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3946 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3947 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3948 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3949 * DESCRIBE now works on structure objects again.
3950 * Most function call argument type mismatches are now handled as
3951 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3952 know whether the function will be redefined before the call is
3953 executed. (The compiler could flag local calls with full WARNINGs,
3954 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3955 it doesn't keep track of enough information to know whether calls
3956 are local in this sense.)
3957 * Compiler output is now more verbose, with messages truncated
3958 later than before. (There should be some supported way for users
3959 to override the default verbosity, but I haven't decided how to
3960 provide it yet, so this behavior is still controlled by the internal
3961 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3962 src/compiler/ir1util.lisp.)
3963 * Fasl file format version numbers have increased again, because
3964 support for the Gray streams extension changes the layout of the
3965 system's STREAM objects.
3966 * The Gray subclassable streams extension now works, thanks to a
3967 patch from Martin Atzmueller.
3968 * The full LOAD-FOREIGN extension (not just the primitive
3969 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3970 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3971 but like most other programs, it defaults to copying the Unix
3972 environment from the original process instead of starting the
3973 new process in an empty environment.
3974 * Extensions which manipulate the Unix environment now support
3975 an :ENVIRONMENT keyword option which doesn't smash case or
3976 do other bad things. The CMU-CL-style :ENV option is retained
3977 for porting convenience.
3978 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3979 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3981 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3983 * A patch from Martin Atzmueller seems to have solved the SIGINT
3984 problem, and as far as we know, signal-handling now works cleanly.
3985 (If you find any new bugs, please report them!)
3986 * The system no longer defaults Lisp source file names to types
3987 ".l", ".cl", or ".lsp", but only to ".lisp".
3988 * The compiler no longer uses special default file extensions for
3989 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3990 seems to expect a single default extension for all compiled code,
3991 and there's no compelling reason to try to stretch the standard
3992 to allow two different extensions.) Instead, byte-compiled files
3993 default to the same extension as native-compiled files.
3994 * Fasl file format version numbers have increased again, because
3995 a rearrangement of internal implementation packages made some
3996 dumped symbols in old fasl files unreadable in new cores.
3997 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3998 many fewer weird special cases.
3999 * Bug #17 (differing COMPILE-FILE behavior between logical and
4000 physical pathnames) has been fixed, and some related misbehavior too,
4001 thanks to a patch from Martin Atzmueller.
4002 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4003 by Tim Moore, ported to SBCL by Martin Atzmueller.
4004 * Martin Atzmueller fixed several filesystem-related problems,
4005 including bug #36, in part by porting CMU CL patches, which were
4006 written in part by Paul Werkowski.
4007 * More compiler warnings in src/runtime/ are gone, thanks to
4008 more patches from Martin Atzmueller.
4009 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4012 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4014 * DESCRIBE now works on CONDITION objects.
4015 * The debugger now handles errors which arise when trying to print
4016 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4018 * The build system now uses an additional file, customize-target-features.lisp,
4019 to allow local modifications to the target *FEATURES* list. (The point of
4020 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4021 debugging features enabled, without having to worry about propagating it
4022 into everyone's system when I do a "cvs update".) When no
4023 customize-target-features.lisp file exists, the target *FEATURES* list
4024 should be constructed the same way as before.
4025 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4026 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4027 warn about behavior which is undefined under the ANSI spec. Note: This
4028 is specified by ANSI, but it's not very popular with programmers.
4029 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4030 macro in the SBCL sources for an example of a workaround which you
4031 might use to make portable ANSI-standard code which does what you want.
4032 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4033 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4034 when it's used as a non-toplevel form. (This is required in order
4035 to implement the DEFCONSTANT-EQX macro.)
4036 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4037 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4038 funny things, due to bugs in the implementation of EVAL-WHEN
4039 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4040 non-toplevel DEFCONSTANTs are uncommon.)
4041 * The core file version number and fasl file version number have been
4042 incremented, because the old noncompliant DEFCONSTANT behavior involved
4043 calling functions which no longer exist, and because I also took the
4044 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4045 * fixed bug 1 (error handling before read-eval-print loop starts), and
4046 redid debugger restarts and related debugger commands somewhat while
4048 ** The QUIT debugger command is gone, since it did something
4049 rather different than the SB-EXT:QUIT command, and since it never
4050 worked properly outside the main toplevel read/eval/print loop.
4051 Invoking the new TOPLEVEL restart provides the same functionality.
4052 ** The GO debugger command is also gone, since you can just invoke
4053 the CONTINUE restart directly instead.
4054 ** The TOP debugger command is also gone, since it's redundant with the
4055 FRAME 0 command, and since it interfered with abbreviations for the
4057 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4059 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4060 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4061 some porting work by Martin Atzmueller)
4062 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4063 with a boatload of patches which he ported to SBCL. Now that those
4065 ** The system tries to make sure that its low-priority messages
4066 are prefixed by semicolons, to help people who like to use
4067 syntax highlighting in their ILISP buffer. (This patch
4068 was originally due to Raymond Toy.)
4069 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4070 patches originally written by Raymond Toy.
4071 ** The compiler understands coercion between single-value and
4072 multiple-VALUES type expressions better, getting rid of some very
4073 weird behavior, thanks to patches originally by Robert MacLachlan
4074 and Douglas Crosher.
4075 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4076 lambda lists, thanks to a patch originally by Pierre Mai.
4077 ** The system no longer bogusly warns about "abbreviated type
4079 ** The compiler gets less confused by inlining and RETURN-FROM,
4080 thanks to some patches originally by Tim Moore.
4081 ** The system no longer hangs when dumping circular lists to fasl
4082 files, thanks to a patch originally from Douglas Crosher.
4083 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4084 undefined function error.
4085 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4086 in structure slot initforms) for now, documented workaround instead:-|
4087 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4088 * fixed bug 5: added stubs for various Gray stream functions called
4089 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4090 installed, at least appropriate type errors are generated
4091 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4092 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4093 consistently in DEFMETHOD forms.
4094 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4095 it doesn't seem to affect SBCL after all
4096 * The C runtime system now builds with better optimization and many
4097 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4099 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4101 * The system is now under CVS at SourceForge (instead of the
4102 CVS repository on my home machine).
4103 * The new signal handling code has been tweaked to treat register
4104 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4105 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4106 downstream, e.g. in debug-int.lisp, has implicit dependencies
4107 on the unsignedness of integer representation of machine words,
4108 and that caused the system to bomb out with infinite regress
4109 when trying to recover from type errors involving signed values,
4110 e.g. (BUTLAST '(1 2 3) -1).)
4111 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4112 (This was one of the bugs Peter Van Eynde reported back in July.)
4113 * The system now uses code inspired by Colin Walters' O(N)
4114 implementation of MAP (from the cmucl-imp@cons.org mailing
4115 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4116 inline the MAP operation, and there is more than one
4117 sequence argument to the MAP call (so that it can't just
4118 do ETYPECASE once and for all based on the type of the
4119 single sequence argument). (The old non-inline implementation
4120 of the general M-argument sequence-of-length-N case required
4121 O(M*N*N) time when any of the sequence arguments were LISTs.)
4122 * The QUIT :UNIX-CODE keyword argument has been renamed to
4123 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4124 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4126 * Raymond Wiker's patches to port dynamic loading from Linux to
4127 FreeBSD have been added.
4128 * The BUGS file is now more nearly up to date, thanks in large part
4129 to Martin Atzmueller's review of it.
4130 * The debugger now flushes standard output streams before it begins
4131 its output ("debugger invoked" and so forth).
4132 * The core version number and fasl file version number have both
4133 been incremented, because of incompatible changes in the layout
4135 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4136 used to not be called for a saved Lisp image.)
4137 * Martin Atzmueller's version of a patch to fix a compiler crash,
4138 as posted on sbcl-devel 13 September 2000, has been installed.
4139 * Instead of installing Martin Atzmueller's patch for the
4140 compiler transform for SUBSEQ, I deleted the compiler transform,
4141 and transforms for some similar consing operations.
4142 * A bug in signal handling which kept TRACE from working on OpenBSD
4144 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4145 away by constant folding
4146 * The system now defines its address space constants in one place
4147 (in the Lisp sources), and propagates them automatically elsewhere
4148 (through GENESIS and the sbcl.h file). Therefore, patching the
4149 address map is less unnecessarily tedious and error-prone. The
4150 Lisp names of address space constants have also been systematized.
4151 * CVS tags like dollar-Header-dollar have been removed from
4152 the sources, because they have never saved me trouble and
4153 they've been source of trouble working with patches and other
4154 diff-related operations.
4155 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4156 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4158 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4160 * The system has been ported to OpenBSD.
4161 * The system now compiles with a simple "sh make.sh" on the systems
4162 that it's supported on. I.e., now you no longer need to tweak
4163 text in the target-features.lisp-expr and symlinks in src/runtime/
4164 by hand, the make.sh takes care of it for you.
4165 * The system is no longer so grossly inefficient when compiling code
4166 involving vectors implemented as general (not simple) vectors (VECTOR T),
4167 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4168 which dares to use the various sequence functions on non-simple
4169 vectors, takes less of a performance hit.
4170 * There is now a primitive type predicate VECTOR-T-P
4171 to test for the (VECTOR T) type, so that e.g.
4172 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4173 can now be compiled with some semblance of efficiency. (The old code
4174 turned the type declaration into a full call to %TYPEP at runtime!)
4175 * AREF on (VECTOR T) is still not fast, since it's still compiled
4176 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4177 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4178 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4179 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4180 without making full calls to SUBTYPEP at runtime.
4181 (Some analogous efficiency issues for non-simple vectors specialized to
4182 element types other than T, or for non-simple multidimensional arrays,
4183 have not been addressed. They could almost certainly be handled the
4184 same way if anyone is motivated to do so.)
4185 * The changes in array handling break binary compatibility, so
4186 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4187 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4188 instead of (VALUES T T).
4189 * By following the instructions that Dan Barlow posted to sbcl-devel
4190 on 2 July 2000, I was able to enable primitive dynamic object
4191 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4192 functionality is not implemented (since it calls ld to resolve
4193 library references automatically, requiring RUN-PROGRAM for its
4194 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4195 try to resolve library references) is now supported.
4196 * The system now flushes the standard output streams when it terminates,
4197 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4198 them at several other probably-convenient times, e.g. in each pass of
4199 the toplevel read-eval-print loop, and after evaluating a form given
4200 as an "--eval" command-line option. (These changes were motivated by a
4201 discussion of stream flushing issues on cmucl-imp in August 2000.)
4202 * The source transform for TYPEP of array types no longer assumes
4203 that an array whose element type is a not-yet-defined type
4204 is implemented as an array of T, but instead punts, so that the
4205 type will be interpreted at runtime.
4206 * There is now some support for cross-compiling in make.sh: each of
4207 the phases of make.sh has its own script. (This should be transparent
4208 to people doing ordinary, non-cross-compile builds.)
4209 * Since my laptop doesn't have hundreds of megabytes of memory like
4210 my desktop machine, I became more motivated to do some items on
4211 my to-do list in order to reduce the size of the system a little:
4212 ** Arrange for various needed-only-at-cold-init things to be
4213 uninterned after cold init. To support this, those things have
4214 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4215 symbols with such names are now uninterned after cold init).
4216 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4217 fasl files for cold load.
4218 ** Remove the old compiler structure pooling code (which used to
4219 be conditional on the target feature :SB-ALLOC) completely.
4220 ** Redo the representation of some data in cold init to be more compact.
4221 (I also looked into supporting byte compiled code at bootstrap time,
4222 which would probably reduce the size of the system a lot, but that
4223 looked too complicated, so I punted for now.)
4224 * The maximum signal nesting depth in the src/runtime/ support code has
4225 been reduced from 4096 to 256. (I don't know any reason for the very
4226 large old value. If the new smaller value turns out to break something,
4227 I'll probably just bump it back up.)
4228 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4230 * Many, many bugs reported by Peter Van Eynde have been added to
4231 the BUGS list; some have even been fixed.
4232 * While enabling dynamic object file loading, I tried to make the
4233 code easier to understand, renaming various functions and variables
4234 with less ambiguous names, and changing some function calling
4235 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4236 * While trying to figure out how to do the OpenBSD port, I tried to
4237 clean up some of the code in src/runtime/. In particular, I dropped
4238 support for non-POSIX signal handling, added various comments,
4239 tweaked the code to reduce the number of compilation warnings, and
4240 renamed some files to increase consistency.
4241 * To support the new automatic configuration functionality in make.sh,
4242 the source file target-features.lisp-expr has been replaced with the
4243 source file base-target-features.lisp-expr and the machine-generated
4244 file local-target-features.lisp-expr.
4245 * fixed a stupid quoting error in make.sh so that using CMU CL
4246 "lisp -batch" as cross-compilation host works again
4248 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4250 * DESCRIBE no longer tries to call itself recursively to describe
4251 bound/fbound values, so that it no longer fails on symbols which are
4252 bound to themselves (like keywords, T, and NIL).
4253 * DESCRIBE now works on generic functions.
4254 * The printer now prints less-screwed-up representations of closures
4255 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4256 * A private symbol is used instead of the :EMPTY keyword previously
4257 used to mark empty slots in hash tables. Thus
4258 (DEFVAR *HT* (MAKE-HASH-TABLE))
4259 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4260 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4261 now does what ANSI says that it should. (You can still get
4262 similar noncompliant behavior if bang on the hash table
4263 implementation with all the symbols you get back from
4264 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4265 This breaks binary compatibility, since tests for equality to
4266 :EMPTY are wired into things like the macroexpansion of
4267 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4269 * There's now a minimal placeholder implementation for CL:STEP,
4270 as required by ANSI.
4271 * An obscure bug in the interaction of the normal compiler, the byte
4272 compiler, inlining, and structure predicates has been patched
4273 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4274 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4276 * Missing ordinary arguments in a macro call are now detected even
4277 when the macro lambda list contains &KEY or &REST.
4278 * The debugger no longer complains about encountering the top of the
4279 stack when you type "FRAME 0" to explicitly instruct it to go to
4280 the top of the stack. And it now prints the frame you request even
4281 if it's the current frame (instead of saying "You are here.").
4282 * As specified by ANSI, the system now always prints keywords
4283 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4285 * The default initial SIZE of HASH-TABLEs is now smaller.
4286 * Type information from CLOS class dispatch is now propagated
4287 into DEFMETHOD bodies, so that e.g.
4288 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4290 is now basically equivalent to
4291 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4292 (DECLARE (TYPE SINGLE-FLOAT X))
4294 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4295 operation, without having to do run-time type dispatch.
4296 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4297 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4298 * Since I don't seem to be making big file reorganizations very often
4299 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4300 is overflowing my ability to conveniently back them up), I've finally
4301 checked the system into CVS. (The CVS repository is on my home system,
4302 not at SourceForge -- putting it on SourceForge might come later.)
4303 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4304 high-level GC-NOTIFY-FOO functions send their output. (There's
4305 still very little control of where low-level verbose GC functions
4306 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4307 less than it used to -- the GC-NOTIFY-FOO functions are now under
4308 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4309 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4310 in only one place in the source code, and propagates it automatically
4311 everywhere that it's needed. Thus e.g. when I bump the version from
4312 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4313 * The C source files now include boilerplate legalese and documentation
4314 at the head of each file (just as the Lisp source files already did).
4315 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4316 to his page will be replaced with a link to his new CLiki service.
4318 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4320 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4321 * The build process now looks for GNU make under the default name "gmake",
4322 instead of "make" as it used to. If GNU make is not available as "gmake"
4323 on your system, you can change this default behavior by setting the
4324 GNUMAKE environment variable.
4325 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4326 can build without error under CMU CL.
4328 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4330 * There is now a partial SBCL user manual (with some new text and some
4331 text cribbed from the CMU CL manual).
4332 * The beginnings of a profiler have been added (starting with the
4333 CMU CL profiler and simplifying and cleaning up). Eventually the
4334 main interface should be through the TRACE macro, but for now,
4335 it's still accessed through vaguely CMU-CL-style functions and macros
4336 exported from the package SB-PROFILE.
4337 * Some problems left over from porting CMU CL to the new
4338 cross-compilation bootstrap process have been cleaned up:
4339 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4340 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4341 part when getting rid of the compiler *BACKEND* stuff.)
4342 ** The value of *NULL-TYPE* was screwed up, because it was
4343 being initialized before the type system knew the final
4344 definition of the 'NULL type. This screwed up several key
4345 optimizations in the compiler, causing inefficiency in all sorts
4346 of places. (I found it because I wanted to understand why
4347 GET-INTERNAL-RUN-TIME was consing.)
4348 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4349 PROCLAIM FTYPE information. Unfortunately this broke binary
4350 compatibility again, since now the forms output by DEFGENERIC
4351 to refer to functions which didn't exist in 0.6.3.
4352 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4353 can use the new (as of 0.6.3) transform for SOME into MAP into
4355 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4356 values to (INTEGER 0 1000000), so that the time code will no longer
4357 occasionally get blown up by Linux returning 1000000 microseconds
4358 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4359 its output conform to the ANSI spec. (Alas, this makes its output
4360 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4361 * A full call to MAP NIL with a single sequence argument no longer conses.
4362 * fixes to problems pointed out by Martin Atzmueller:
4363 * The manual page no longer talks about multiprocessing as though
4364 it were currently supported.
4365 * The ILISP support patches have been removed from the distribution,
4366 because as of version 5.10.1, ILISP now supports SBCL without us
4367 having to maintain patches.
4368 * added a modified version of Raymond Toy's recent CMU CL patch for
4369 EQUALP comparison of HASH-TABLE
4371 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4373 * The system still can't cross-compile itself with
4374 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4375 entails), but at least it can compile more of itself that way
4376 than it used to be able to, and various buglets which were uncovered
4377 by trying to cross-compile itself that way have now been fixed.
4378 * This release breaks binary compatibility again. This time
4379 at least I've incremented the FASL file format version to 2, so that the
4380 problem can be detected reliably instead of just causing weird errors.
4381 * various new style warnings:
4382 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4383 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4384 ** using the lexical binding of a variable named in the *FOO* style
4385 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4386 as specified by ANSI.
4387 * *RANDOM-STATE* is no longer automatically initialized from
4388 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4389 default behavior of the system is to repeat its behavior every time
4390 it's run. If you'd like to change this behavior, you can always
4391 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4392 old convention there was no comparably easy way to get the system to
4393 repeat its behavior every time it was run.
4394 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4395 FTYPE declarations has been removed, in favor of their ANSI
4396 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4397 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4398 the types of their sequence arguments can be determined at compile time.
4399 This is done through a new open code expansion for MAP which eliminates
4400 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4401 argument types can be determined at compile time.
4402 * The optimizer now transforms COERCE into an identity operation when it
4403 can prove that the coerced object is already of the correct type. (This
4404 can be a win for machine generated code, including the output of other
4405 optimization transforms, such as the MAP transform above.)
4406 * Credit information has been moved from source file headers into CREDITS.
4407 * Source file headers have been made more standard.
4408 * The CASE macro now compiles without complaining even when it has
4411 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4413 * (Note that the way that the PCL macroexpansions were rewritten
4414 to accommodate the change in DEFGENERIC below breaks binary
4415 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4416 not run under sbcl-0.6.2. Once we get out of alpha releases,
4417 i.e. hit release 1.0.0, we'll probably try to maintain binary
4418 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4419 and sbcl-1.4.4. Until then, however, it might be fairly common
4420 for maintenance releases to break binary compatibility.)
4421 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4423 * The reporting of unhandled signals has been changed to print some
4424 explanatory text as well as the report form. (Previously only
4425 the report form was printed.)
4426 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4427 it defines, so that the compiler no longer issues undefined function
4428 warnings for compiled-but-not-yet-loaded generic functions.
4429 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4430 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4431 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4432 section of the ANSI spec are supported.
4433 * The "" nickname for the "KEYWORD" package has been removed.
4434 The reader still handles symbol tokens which begin with a package marker
4435 as keywords, but it doesn't expose its mechanism for doing so in the
4436 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4437 * The system now issues STYLE-WARNINGs for contradictory TYPE
4438 proclamations. (Warnings for contradictory FTYPE proclamations would
4439 be nice too, but those can't be done usefully unless the type system
4440 is made smarter about FUNCTION types.)
4441 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4442 systematized, so that "*target-*.lisp is supposed to exist only on the
4443 target and imply that there's a related file which exists on the
4444 host, and *host-*.lisp is supposed to exist only on the host and imply
4445 that there's a related file which exists on the target. This involves a
4446 lot of renaming. Hopefully the acute confusion caused by the renaming
4447 will be justified by the reduction in chronic confusion..
4448 ** runtime-type.lisp -> early-target-type.lisp
4449 ** target-type.lisp -> late-target-type.lisp
4450 ** early-host-format.lisp -> early-format.lisp
4451 ** late-host-format.lisp -> late-format.lisp
4452 ** host-error.lisp -> misc-error.lisp
4453 ** early-error.lisp -> early-target-error.lisp
4454 ** late-error.lisp -> late-target-error.lisp
4455 ** host-defboot.lisp -> early-defboot.lisp
4456 ** code/misc.lisp -> code/target-misc.lisp
4457 ** code/host-misc.lisp -> code/misc.lisp
4458 ** code/numbers.lisp -> code/target-numbers.lisp
4459 ** code/early-numbers.lisp -> numbers.lisp
4460 ** early-host-type.lisp -> early-type.lisp
4461 ** late-host-type.lisp -> late-type.lisp
4462 ** host-typep.lisp -> typep.lisp
4463 ** load.lisp -> target-load.lisp
4464 ** host-load.lisp -> load.lisp
4465 ** host-disassem.lisp -> disassem.lisp
4466 ** host-insts.lisp -> insts.lisp
4467 ** byte-comp.lisp -> target-byte-comp.lisp
4468 ** host-byte-comp.lisp -> byte-comp.lisp
4469 ** host-signal.lisp -> signal.lisp
4470 ** host-defstruct.lisp -> defstruct.lisp
4471 ** late-target-type.lisp -> deftypes-for-target.lisp
4472 Furthermore, several other previously target-only files foo.lisp (e.g.
4473 hash-table.lisp and random.lisp) have been split into a target-and-host
4474 foo.lisp file and a target-only target-foo.lisp file, with their key type
4475 definitions in the target-and-host part, so that the cross-compiler will
4476 know more about target types.
4477 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4478 gone away. In their place are various *BACKEND-FOO* variables
4479 corresponding to the slots of the old structure.
4480 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4481 into the target SBCL has been fixed.
4482 * The chill.lisp system for loading cold code into a running SBCL
4484 * Support for the CMU CL "scavenger hook" extension has been removed.
4485 (It was undocumented and unused in the CMU CL sources that SBCL was
4486 derived from, and stale in sbcl-0.6.1.)
4487 * Various errors in the cross-compiler type system were detected
4488 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4489 (enabling various consistency checks). Many of them were fixed,
4490 but some hard problems remain, so the compiler is back to
4491 running without *TYPE-SYSTEM-INITIALIZED* for now.
4492 * As part of the cross-compiler type system cleanup, I implemented
4493 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4494 * I have started adding UNCROSS calls throughout the type system
4495 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4496 on cross-compiler input files will be able to go away, and various
4498 * CONSTANTP now returns true for quoted forms (as explicitly required
4501 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4503 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4504 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4505 (MAKE-LIST -1). In the longer term, it ought to become true
4506 that declarations are assertions even at SAFETY 1. For now, it's not
4507 quite true even at SAFETY 3, but it's at least more nearly true..
4508 (Note that this change seems to increases the size of the system by
4509 O(5%) and to decrease the speed of the compiler by 20% or more.)
4510 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4511 for the problem of printing dozens of lines of distracting information
4512 about low-level system machinery as part of the top stack frame
4513 on entry to the debugger when an undefined function was called.
4514 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4515 is not reset to COMMON-LISP-USER.
4516 * Compilation of stuff related to dyncount.lisp has been made conditional
4517 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4518 smaller. The various dyncount-related symbols have been moved into
4519 a new "SB-DYNCOUNT" package.
4520 * tty-inspect.lisp has been renamed to inspect.lisp.
4521 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4522 feature has gone away. (When we eventually port to other flavors of
4523 libc and/or Unix, we'll try to make the differences between flavors
4524 invisible at the user level.)
4525 * Various other *FEATURES* tags, and/or their associated conditionals,
4526 have been removed if obsolescent, or given better documentation, or
4527 sometimes given more-mnemonic names.
4529 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4531 * tidied up "make.sh" script
4532 * tidied up system directory structure
4533 * better "clean.sh" behavior
4534 * added doc/FOR-CMUCL-DEVELOPERS
4535 * many many small tweaks to output format, e.g. removing possibly-confusing
4536 trailing #\. character in DESCRIBE-INSTANCE
4537 * (EQUALP #\A 'A) no longer signals an error.
4538 * new hashing code, including EQUALP hashing
4539 * tidied up Lisp initialization and toplevel
4540 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4541 * command line argument processing
4542 * added POSIX-GETENV function to deal with Unix-ish environment variables
4543 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4544 terminating SBCL on EOF
4545 * non-verbose GC by default
4546 * There is no more "sbcl" shell script; the sbcl file is now the C
4547 runtime executable (just like CMU CL).
4548 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4550 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4551 symbol and package information as Lisp native symbol and package objects
4552 instead of strings naming symbols and strings naming packages. This way,
4553 whenever packages are renamed (as in warm init), debug information is
4554 transformed along with everything else.
4555 * tweaked the optimization policy declarations which control the building
4556 of SBCL itself. Now, among other things, the system no longer saves
4557 source location debugging information. (This helps two problems at once
4558 by reducing SBCL size and by keeping SBCL from trying to look for its
4559 sources -- which may not exist -- when reporting errors.)
4560 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4561 debugging and testing purposes
4562 * cleaned up printing, making the printer call PRINT-OBJECT for
4563 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4564 methods, giving nearly-ANSI behavior
4565 * converted almost all special variables to use *FOO* naming convention
4566 * deleted PARSE-TIME functionality, since it can be done portably
4567 * moved some files out of cold init into warm init
4568 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4570 * regularized formatting of source files
4571 * added an install.sh script
4572 * fixed ridiculous memory usage of cross-compiler by making
4573 compiler/alloc.lisp not try to do pooling unless it can hook
4574 itself into the GC of the cross-compilation host. Now the system
4575 builds nicely on my old laptop.
4576 * added :SB-ALLOC in target-features.lisp-expr
4577 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4578 was not implemented)
4579 * re-did condition handling and note reporting in the compiler. Notes
4580 are no longer handled by signalling conditions. Style warnings
4581 and warnings are handled more correctly and reported in such a way
4582 that it's easy to find one or the other in your output (so that you
4583 can e.g. figure out which of many problems caused COMPILE-FILE to
4585 * changed the severity of several compiler warnings from full WARNING
4586 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4587 compiler note reporting so that it doesn't use the condition system
4588 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4589 COMPILE-FILE command)
4590 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4591 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4592 no longer supported, since their implementation was deeply intertwingled
4593 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4594 * removed bogus "support" for compiler macros named (SETF FOO), and
4595 removed the compiler macro for SETF INFO (but only after making a fool
4596 of myself on the cmucl-imp mailing list by posting a bogus patch for
4597 DEFINE-COMPILER-MACRO..)
4598 * Compiled files containing forms which have side effects on the Lisp
4599 reader (such as DEFPACKAGE forms) are now handled more correctly.
4600 (Compiler queuing of top level lambdas has been suppressed by setting
4601 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4602 * deleted various currently-unused source files, e.g. gengc.lisp. They
4603 may be added back at some point e.g. when porting to other architectures,
4604 but until they are it's distracting to distribute them and to try to
4606 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4607 there were so many of them they're just distractions, and UNCROSS is
4608 known to be able to handle the current sources
4609 * moved PROFILE functionality into TRACE, so that it will be clear
4610 how the wrapping and unwrapping of functions when you profile them
4611 interacts with the wrapping and unwrapping of functions when you
4612 trace them. (Actually, the functionality isn't there yet, but at least
4613 the interface specification is there. Hopefully, the functionality will
4614 arrive with some future maintenance release.)
4615 * removed host-oops.lisp
4616 * changed signature of QUIT function to allow UNIX-CODE argument
4617 * fixed READ-SEQUENCE bug
4618 * tweaked verbose GC output so that it looks more like the progress
4619 output that ANSI specifies for functions like LOAD
4620 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4621 * added <http://sbcl.sourceforge.com> to the banner information printed by