1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.17 relative to 1.0.16:
3 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
5 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
6 * bug fix: dynamic extent allocation of nested lists and vectors
7 could leak to otherwise accessible parts.
8 * bug fix: invalid optimization of heap-allocated alien variable
10 * bug fix: fasl header checking is less vulnerable to different
11 platform word lengths.
12 * bug fix: more correct assembler syntax for GNU binutils
13 2.18.50.0.4 support. (thanks to Marijn Schouten)
15 changes in sbcl-1.0.16 relative to 1.0.15:
16 * minor incompatible change: revert the changes to sb-posix's error
17 signaling added in 1.0.14.
18 * minor incompatible change: change PROBE-FILE back to returning
19 NIL whenever we can't get a truename, as was the case before 1.0.14.
20 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
21 inherits from ERROR instead of just CONDITION.
22 * new feature: SB-INTROSPECT can provide source locations for instances
23 as well. (thanks to Tobian Ritterweiler)
24 * optimization: binding special variables now generates smaller code
25 on threaded platforms.
26 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
27 and cases where no :TEST is given but the compiler can infer that
28 the element to search is of type (OR FIXNUM (NOT NUMBER)).
29 * optimization: better LOGNOT on fixnums.
30 * optimization: modular arithmetic for a particular requested width
31 is implemented using a tagged representation unless a better
32 representation is available.
33 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
34 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
35 for wellformedness and returns a specialized vector. (reported by
37 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
38 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
39 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
40 unscheduling timer at the same time as another timer fires.
41 * bug fix: don't reschedule timers for dead threads.
42 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
43 * bug fix: copying output from RUN-PROGRAM to a stream signalled
44 bogus errors if select() was interrupted.
45 * enhancement: add support for fcntl's struct flock to SB-POSIX.
47 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
48 * enhancement: cleaner backtraces for interactive interrupts, as
49 well as other cases where the interesting frames used to be
50 obscured by interrupt handling frames.
51 * enhancement: untracing a whole package using (UNTRACE "FOO") is
52 now supported, and tracing a whole package using (TRACE "FOO") now
53 traces SETF-functions as well.
54 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
55 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
56 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
57 * unreadably printed representation of hash-tables now includes
59 * bug fix: partially fixed #188: type propagation from assignments
60 is now more efficient.
61 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
62 'DOUBLE-FLOAT) are not flushable.
63 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
64 full address of the object, and none of the tag bits.
65 * bug fix: readably printing hash-tables now respects other printer
66 control variables. (reported by Cedric St-Jean)
67 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
68 keyword with MAKE-HASH-TABLE.
69 * bug fix: export SB-POSIX:MKSTEMP.
70 * bug fix: SORT was not interrupt safe.
71 * bug fix: XREF accounts for the last node of each basic-block as
73 * bug fix: MAKE-INSTANCE optimizations interacted badly with
74 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
75 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
76 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
77 create obsolete instances in the case of redefinition or
78 obsoletion of a superclass. (thanks to Andy Hefner)
79 * bug fix: Support for the Alpha architecture has been revived; it had
80 suffered somewhat from lack of maintenance since sbcl-1.0.
81 * improvements to the Windows port:
82 ** The system detects the codepage to use at startup, instead of
83 using the value from the saved core. (thanks to Kei Suzuki)
85 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
86 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
87 (see documentation for details.)
88 * revived support for OpenBSD (contributed by Josh Elsasser)
89 * partially fixed bug #108: ROOM no longer suffers from occasional
90 (AVER (SAP= CURRENT END)) failures.
91 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
92 bodies no longer cause a WARNING to be signalled. (reported by
94 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
95 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
96 the bug in TRUNCATE.) (reported by Michael Weber)
97 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
98 no samples. (reported by Andy Hefner)
99 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
100 no longer appear as (NIL ...) frames in backtraces.
101 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
102 is now actually exported from SB-EXT as documented. (reported by
104 * bug fix: sb-aclrepl now correctly understands how to inspect
105 single-floats on 64-bit platforms where single-floats are not boxed.
106 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
107 yet been finalized. (reported by Levente Meszaros)
108 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
109 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
110 SBCL-specific optimize qualities.
112 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
113 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
114 an executable in the search path, and does so in the child
115 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
116 been removed; it can be found in the CVS history, for somebody who
117 needs that search behavior (see the manual).
118 * minor incompatible change: compiler policy re. weakening type
119 checks has changed: now type checks are weakened only if SAFETY < 2
121 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
122 unparsing of directory pathnames as files. Analogously,
123 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
124 filename to parse into a directory pathname.
125 * enhancement: implicit generic function creation now signals a
126 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
127 users can bind handlers for and muffle around calls to LOAD.
128 * enhancement: RUN-PROGRAM allows unicode arguments and environments
129 to be used (using the default stream external format), and allows
130 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
131 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
132 strings and vectors whose element-type or simplicity is not fully
133 known at compile-time.
134 * optimization: STRING-TRIM and related functions no longer allocate
135 a new string when no trimming needs to be performed. These functions
136 are also faster than before when the input string has been declared
138 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
139 * optimization: READ-LINE is significantly faster for files containing
141 * optimization: non-open coded uses of character comparison operators
142 (e.g. char=) no longer cons when called with more than one parameter
143 on platforms supporting dynamic-extent allocation.
144 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
145 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
146 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
148 * bug fix: some sequence functions elided bounds checking when
150 * bug fix: too liberal weakening of union-type checks when SPEED >
152 * bug fix: more bogus fixnum declarations in ROOM implementation
155 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
156 * new feature: MAKE-HASH-TABLE now experimentally accepts a
157 :SYNCHRONIZED argument, which makes the hash-table safe for
158 concurrent accesses (but not iteration.) See also:
159 SB-EXT:WITH-LOCKED-HASH-TABLE, and
160 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
161 * optimization: CONCATENATE on strings is an order of magnitude faster
162 in code compiled with (> SPEED SPACE).
163 * optimization: SUBSEQ is ~50% faster on lists.
164 * bug fix: bug 417 fixed -- source location reporting is now more robust.
165 * bug fix: SUBSEQ on a list will now correctly signal an error if if
166 END is smaller then START.
167 * bug fix: SB-PROFILE will no longer report extra consing for nested
168 calls to profiled functions.
169 * bug fix: ROOM implementation had bogus fixnum declarations which
170 could cause type-errors when calling ROOM on large images.
171 * bug fix: if file compilation is aborted, the partial fasl is now
172 deleted, and COMPILE-FILE returns NIL as the primary value.
173 * bug fix: number of thread safety issues relating to SBCL's internal
174 hash-table usage have been fixed.
175 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
176 be returned from its body when the values were being returned
177 using unknown-values return convection and the W-P-O was wrapped
178 inside an UNWIND-PROTECT.
179 * bug fix: sb-posix should now compile again under Windows, enabling
182 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
183 * incompatible change: hash-table accessor functions are no longer
184 automatically protected by locks. Concurrent accesses on the same hash-table
185 from multiple threads can give inconsistent results or even corrupt the
186 hash-table completely. Multi-threaded applications should do their own
187 locking at the correct granularity. In the current implementation it is
188 still safe to have multiple readers access the same table, but it's not
189 guaranteed that this property will be maintained in future releases.
190 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
191 and will signal an error at runtime.
192 * enhancement: SB-THREAD package now exports a semaphore interface.
193 * enhancement: CONS can now stack-allocate on x86 and
194 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
196 * enhancement: nested lists can now be stack allocated on
197 platforms providing stack allocation support.
198 * enhancement: dynamic-extent support has been extended to support
199 cases where there are multiple possible sources for the stack
201 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
202 if the mutex is uncontested on Linux.
203 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
204 as specified by the CLHS. (thanks to Tobias Rittweiler)
205 * bug fix: NaN comparison now works on x86-64.
206 * bug fix: CLOSE :ABORT T on a stream with pending output now
208 * bug fix: instances of non-standard metaclasses using standard
209 instance structure protocol sometimes missed the slot type checks
211 * bug fix: known functions can be inlined in high-debug code.
213 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
214 * minor incompatible change: the MSI installer on Windows no longer
215 associates .lisp and .fasl files with the installed SBCL.
216 * minor incompatible change: :UNIX is no longer present in *FEATURES*
217 on Windows. (thanks to Luis Oliviera)
218 * new feature: SB-CLTL2 contrib module now implements
219 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
220 * optimization: scavenging weak pointers is now more efficient,
221 requiring O(1) instead of O(N) per weak pointer to identify
222 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
223 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
224 method bodies using constant slot-names when the first argument is
225 a specializer parameter for the method.
226 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
227 names, when the class of the instance is a direct instance
228 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
229 as slow as the constant slot-name case.)
230 * optimization: member type construction is now O(N) instead
232 * optimization: UNION and NUNION are now O(N+M) for large
233 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
235 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
236 now more readable in environments like Slime which display it.
237 (thanks to Tobias C. Rittweiler)
238 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
239 was able to infer the correct argument type for the object on which
240 the CAS operation was being performed.
241 * bug fix: copy propagation interfered with parallel assignment
242 semantics in local calls. (reported by Paul Khuong)
243 * bug fix: the signed modular fixnum shift compiled to wrong code on
244 x86 and x86-64. (spotted by a slight modification to some of
247 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
248 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
249 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
250 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
252 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
253 and generic function now print the object identity for anonymous
254 instances. (thanks to Attila Lendvai)
255 * enhancement: as an extension to MOP, SBCL now supports SETF of
256 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
257 (thanks to Attila Lendvai)
258 * enhancement: sb-cover has an alternate annotation mode which puts
259 the form annotations at the car of the form, instead of annotating
260 it completely. The benefit of this mode is that it shows better
261 which forms were instrumented by the compiler.
262 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
263 by printer control variables. (reported by Dan Corkill)
264 * bug fix: system leaked memory when delayed output was performed by
265 the OS in smaller chunks then expected. (thanks to David Smith)
266 * bug fix: system leaked memory when file streams were not closed
268 * bug fix: large objects written to slow streams that were modified
269 after the write could end up with the modified state written to
270 the underlying file descriptor.
271 * bug fix: multiple threads operating in parallel on the same stream
272 could cause buffer-overflows.
273 * bug fix: source location information is stored correctly
274 (broken since 1.0.6). This bug would generally show up as the
275 Slime debugger highlighting the wrong form when the sldb-show-source
277 * bug fix: Branch forms are again annotated as branches in the sb-cover
279 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
280 on x86-64. (reported by Gregory Vanuxem)
281 * bug fix: EQUALP could return wrong results for structures with raw
282 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
283 word). (reported by Vjacheslav Fyodorov)
284 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
287 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
288 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
289 atomic compare-and-swap operations on threaded platforms.
290 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
291 allows assigning a global minimum value to optimization qualities
292 (overriding proclamations and declarations).
293 * enhancement: closed over variables can be stack-allocated on x86
295 * performance bug fix: GETHASH and (SETF GETHASH) are once again
297 * optimization: slot definition lookup is now O(1). This speeds up
298 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
299 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
300 * optimization: ASSOC and MEMBER can now be open-coded for all
301 combinations of keyword arguments when second argument is constant
302 and SPEED >= SPACE. In other cases a specialized version is
304 * bug fix: using obsoleted structure instances with TYPEP and
305 generic functions now signals a sensible error.
306 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
307 (reported by Kristoffer Kvello)
308 * bug fix: backtrace construction is now more careful when making
309 lisp-objects from pointers on the stack, to avoid creating bogus
310 objects that can be seen by the GC.
311 * bug fix: defaulting of values in contexts expecting more than 7
312 variables now works on x86-64. (reported by Christopher Laux)
313 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
315 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
316 as the property-list of a symbol.
317 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
318 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
319 situations requested, are once again file-compileable. (reported
322 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
323 * MOP improvement: support for user-defined subclasses of
324 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
325 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
326 the expansion of DEFMETHOD, is responsible for generating a form
327 which creates a list of specializers when evaluated. Additional
328 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
329 debugging and introspective support.
330 * minor incompatible change: the (unsupported) spinlock interface
331 has changed: free spinlock now has the value NIL, and a held spinlock
332 has the owning thread as its value.
333 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
334 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
336 * enhancement: name of a socket-stream is now "a socket" instead of
338 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
339 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
340 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
341 for generic functions. (Thanks to Tobias C. Rittweiler)
342 * optimization: bignum printing speed has been improved by 20-40%
343 (depending on the bignum size.)
344 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
346 * bug fix: the cache used by the CLOS to store precomputed effective
347 methods, slot offsets, and constant return values is now thread and
349 * bug fix: generic function dispatch function updating is now thread
350 and interrupt safe (in the sense that the known issues have been
352 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
353 * bug fix: interning EQL-specializers is now thread and interrupt safe.
354 * bug fix: asdf systems with dependencies to the SB-POSIX or
355 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
356 * bug fix: interrupt safety of applicable method computation has been
359 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
360 * new contrib: sb-cover, an experimental code coverage tool, is included
362 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
363 significantly faster.
364 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
365 produce more efficient code on x86-64 if the compiler can determine
366 that (+ INDEX <constant>) does not require a bounds check and FOO
367 has an element type at least 8 bits wide.
368 * enhancement: a new, experimental synchronous timeout facility is
369 provided. Refer to SB-SYS:WITH-DEADLINE for details.
370 * enhancement: when a symbol name conflict error arises, the
371 conflicting symbols are always printed with a package prefix.
372 (thanks to Kevin Reid)
373 * enhancement: stepping is now once again supported on the SPARC and
374 MIPS platforms. (It is also now more likely to work on CheneyGC
376 * enhancement: sb-sprof can now also track and report accurate call
378 * bug fixes: the treatment of non-standard subclasses of
379 SB-MOP:SPECIALIZER is more correct.
380 * incompatible change: PURIFY no longer copies the data from the
381 dynamic space into the static and read-only spaces on platforms
382 that use the generational garbage collector
383 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
385 * bug fix: GC race condition occasionally resulting in crashes with
386 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
388 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
390 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
391 a STYLE-WARNING to be signalled (regression from 1.0.4.)
392 * bug fix: an asynchronous interrupt could previously leave the
393 system running with GC inhibited.
394 * bug fix: a DECLARE form evaluated at top-level now causes an error
395 rather than silently (or verbosely) returning NIL.
396 * bug fix: trying to dissassemble functions compiled at high DEBUG could
397 result in a error being signalled due to source form lookup errors.
398 (reported by Peter Graves)
400 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
401 * incompatible change: removed writer methods for host-ent-name,
402 host-ent-addresses -- changing the values did not update the DNS
404 * minor incompatible change: changed experimental JOIN-THREAD interface
405 * documentation: the manual now lists reader and writer methods
406 in class slot documentation sections. (thanks to Richard M Kreuter)
407 * documentation: unwinding from asyncronous events has been
408 documented as unsafe.
409 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
410 in multithreaded application code.
411 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
412 platforms. (thanks to James Anderson for the optimization hint)
413 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
415 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
416 variants no longer cons.
417 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
418 their NOT- variants no longer cons.
419 * optimization: EQUAL hash tables no longer use SXHASH for objects
420 of all data types, but instead use an EQL hash for types for which
421 EQUAL is the same as EQL.
422 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
423 are significantly faster.
424 * optimization: new STRING-OUTPUT-STREAM implementation is much
425 faster for large outputs and conses 30% less on average.
426 * enhancement: XREF information is now collected for references made
427 to global variables using SYMBOL-VALUE and a constant argument.
428 * enhancement: SIGINT now causes a specific condition
429 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
430 * bug fix: FILE-POSITION used to signal an error in some cases where
431 ANSI requires it to return NIL.
432 * bug fix: ADJUST-ARRAY is now interrupt-safe.
433 * bug fix: adding and removing fd-handlers is now interrupt-safe.
434 * bug fix: inlined calls to C now ensure 16byte stack alignment on
436 * bug fix: bad type declaration in the CLOS implementation has
437 been fixed. (reported by James Anderson)
438 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
439 (thanks to Lutz Euler)
440 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
441 required by the CLHS. (reported by Samium Gromoff)
442 * bug fix: dead unbound variable references now signal an error.
443 (reported by Marco Monteiro)
444 * bug fix: / with an unused value was being deleted in safe code.
445 (thanks to Marco Monteiro and Kevin Reid)
446 * bug fix: number of characters that can be written onto a single
447 line in a file is unlimited.
448 * bug fix: some GC deadlocks caused by asynchronous interrupts have
449 been fixed by inhibiting interrupts for when GC is disabled.
450 * bug fix: some interrupt safety issues with GETHASH, (SETF
451 GETHASH), CLRHASH and REMHASH have been fixed.
452 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
453 specifier no longer causes infinite recursion.
454 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
455 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
456 * bug fix: an &environment argument with the correct variable information
457 is passed to macros that are expanded during byte compilation.
458 (reported by Samium Gromoff)
459 * bug fix: a logic error was causing the hash function for bignums to
460 have excessive amounts of collisions. (reported by Faré Rideau)
461 * bug fix: modifying the contents of an array could change the return
462 value of SXHASH on that array, which is only allowed for strings
463 and bit vectors. (bug introduced in 0.9.16)
464 * improvement: the x86-64/darwin port now passes all tests (except
465 for the debugger tests) but should still be considered
466 experimental until this is fixed.
467 * improvement: a style-warning is signaled for CASE (etc) clauses with
468 duplicate keys. (thanks to Kevin Reid)
469 * improvement: macos/x86 and macos/x86-64 now use mach exception
470 handlers for dealing with illegal instructions (for trapping and
471 error handling) and memory protection violations (for GC).
473 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
474 * new platform: experimental support for x86-64/darwin (MacOS).
475 * incompatible change: the thread-safe (on most platforms) getaddrinfo
476 and getnameinfo sockets functions are used instead of gethostbyaddr
477 and gethostbyname, on platforms where the newer functions are available.
478 As a result, the ALIASES field of HOST-ENT will always be NIL on these
480 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
481 don't print the startup banner, but behave as if --noinform was passed
482 as a command line argument. (thanks to Kevin Reid)
483 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
484 * optimization: code using alien values with undeclared types is much faster.
485 * optimization: the compiler is now able to open code SEARCH in more cases.
486 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
487 * bug fix: using standardized COMMON-LISP special variables as loop
488 variables no longer signals bogus package lock violations. (reported
490 * bug fix: declaring local loop variables to be of a range-limited type
491 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
492 (reported by Andras Simon)
493 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
494 bugs remain on x86-64.)
495 * bug fix: the #= and ## reader macros now interact reasonably with
496 funcallable instances.
497 * bug fix: type-checks for function arguments were compiled using the
498 compiler policy settings of the wrong lexical environment (in 1.0.2
500 * bug fix: SHADOW accepts characters as string designators, as required
501 by the spec (thanks to Eric Marsden)
502 * bug fix: fixed GC safety issues when foreign functions are called with
503 non-base strings as arguments
504 * bug fix: more consistent error output for fatal-compiler-warnings, like
506 * bug fix: functions with proper names were showing up as NIL in
507 backtraces (reported by Edi Weitz, regression in 1.0.2)
509 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
510 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
511 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
512 produce more efficient code on the x86 if the compiler can determine
513 that (+ INDEX <constant>) does not require a bounds check and FOO
514 has an element type at least 8 bits wide.
515 * bug fix: references to undefined variables are handled the same way
516 in toplevel forms as in the normal compiler
517 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
519 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
520 (thanks to Jon Buller)
521 * improvement: faster compilation times for complex functions
522 * improvement: added readlink support to SB-POSIX (thanks to Richard
525 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
526 * improvement: experimental support for mach exception handling on
527 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
529 * improvement: support for GBK external format.
530 (thanks to Chun Tian (binghe))
531 * improvement: the debugger now displays variables that have been closed
532 over, in code compiled with (DEBUG 2) or higher
533 * improvement: support for executable cores on NetBSD (thanks to
535 * new feature: added a RESTART-FRAME debugger command
536 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
537 be used to provide an implementation for FILE-POSITION on Gray streams
538 (thanks to Eric Marsden)
539 * new feature: add syslog, openlog and closelog support to SB-POSIX
540 (thanks to Richard Kreuter)
541 * optimization: the function call overhead in code compiled with
542 a high DEBUG optimization setting is significantly
543 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
544 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
545 (thanks to Tony Martinez)
546 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
547 evaluated once (reported by Kevin Reid)
548 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
549 works again (thanks to Kilian Sprotte)
550 * bug fix: an error is signaled for tagbodies with duplicate tags
551 (thanks to Stephen Wilson)
552 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
553 * bug fix: Win32 port can now handle foreign code unwinding Lisp
554 stack frames from alien callbacks.
555 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
556 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
557 65k pages (thanks to David Woodhouse)
558 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
560 changes in sbcl-1.0.1 relative to sbcl-1.0:
561 * new platform: FreeBSD/x86-64, including support for threading.
562 * new feature: the compiler stores cross-referencing information
563 about function calls (who-calls), macroexpansion (who-macroexpands)
564 and special variables (who-binds, who-sets, who-references) for code
565 compiled with (< SPACE 3). This information is available through the
566 sb-introspect contrib.
567 * new feature: users may subclass SEQUENCE, and have instances of
568 these classes interoperate with standard Common Lisp functions if
569 a number of methods are defined. (This feature is experimental
570 and the interface subject to change based on feedback from SBCL
571 users and the general community)
572 * improvement: sb-sprof traces call stacks to an arbitrary depth on
573 x86/x86-64, rather than the previous fixed depth of 8
574 * improvement: another pthread back-end of mutex "pthread-futex".
575 * bug fix: non-ascii command-line arguments are processed correctly
576 (thanks to Yaroslav Kavenchuk)
577 * bug fix: non-required arguments were not passed correctly when a method
578 defined using DEFMETHOD was called from a mop-generated method using
579 CALL-NEXT-METHOD (reported by Pascal Costanza)
580 * bug fix: recursion is now permitted in accessors through
581 SLOT-UNBOUND. (reported by Pascal Costanza)
582 * bug fix: an error was signaled at startup if the HOME environment
583 variable was defined, but had an empty value (reported by Peter Van Eynde)
584 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
585 signal a PROGRAM-ERROR, not a BUG.
586 * bug fix: SB-POSIX:READDIR works when built with large file support.
587 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
589 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
590 for better automatic indentation support. (thanks to Matt Pillsbury)
591 * optimization: loading generic functions no longer takes O(n^2) time,
592 proportional to the amount of methods in the generic function
593 (reported by Todd Sabin and Jeremy Brown)
594 * optimization: the FIND and POSITION family of sequence functions
595 are significantly faster on arrays whose element types have been
597 * improvements to the Windows port:
598 ** Intermittent heap corruption problems have been fixed. (thanks
599 to Alastair Bridgewater)
600 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
602 ** Lisp is able to unwind foreign exception frames from alien
603 callbacks. (thanks to Alastair Bridgewater)
605 changes in sbcl-1.0 relative to sbcl-0.9.18:
606 * improvement: experimental support for threading on FreeBSD/x86.
607 (thanks to NIIMI Satoshi)
608 * improvement: runtime option --dynamic-space-size can be used
609 to set the size of the dynamic space reserved on startup.
610 * improvement: floating point modes in effect are now saved in
611 core, and restored on startup.
612 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
613 startup, not time since first call to GET-INTERNAL-REAL-TIME.
614 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
615 threads are not running after *SAVE-HOOKS* have run.
616 * improvement: writes to CLOS instance slots are type-checked in code
617 compiled with (SAFETY 3)
618 * improvement: floating-point exception handling on FreeBSD (thanks to
620 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
621 (thanks to Zach Beane)
622 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
624 * improvement: added support for the Shift-JIS external format.
625 (contributed by NIIMI Satoshi)
626 * improvement: callbacks are supported on Linux/PPC. (thanks to
628 * bug fix: compiler bug triggered by a (non-standard) VALUES
629 declaration in a LET* was fixed. (reported by Kaersten Poeck)
630 * bug fix: file compiler no longer confuses validated and already
631 dumped structurres. (reported by Kaersten Poeck)
632 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
633 fill-pointer signals a type-error as required. (thanks to
635 * bug fix: disassemly of funcallable instances works.
636 * bug fix: single stepping on PPC.
637 * bug fix: fix thread-safety problems in the type system (generally
638 manifesting as nonsensical errors like "STRING is a bad type specifier
639 for sequences" or "The value 1 is not of type FIXNUM")
640 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
641 (reported by Josip Gracin)
642 * bug fix: an error is signaled for attempts to displace arrays with
643 incompatible element types (thanks to Mario Mommer)
644 * bug fix: more correct handling of wide characters in debug info
645 (bug reported by Attila Lendvai and fixed by Juho Snellman)
646 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
647 and don't cause extra consing
648 * optimization: MAP and MAP-INTO are significantly faster on vectors
649 whose elements types have been declared.
650 * Improvements to SB-SPROF:
651 ** Support for allocation profiling
652 ** Reduced profiling overhead, especially for long profiling runs
653 * Improvements to the Windows port:
654 ** floating point exceptions are now reported correctly.
655 ** stack exhaustion detection works partially.
656 ** more accurate GET-INTERNAL-REAL-TIME.
657 ** better RUN-PROGRAM behaviour in case of errors in the child
659 ** PROBE-FILE now simplifies pathnames correctly.
660 ** DIRECTORY now works correctly with :WILD-INFERIORS.
661 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
662 1st 1970 (although time timezone data will be unreliable).
663 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
665 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
666 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
667 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
669 * bug fix: two potential GC deadlocks affecting threaded builds.
670 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
671 returns T (reported by Anton Kazennikov)
672 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
673 with non-variable places
674 * bug fix: remove a race condition in the setting of
675 funcallable-instance functions, this should make threaded CLOS
676 code more stable against memory faults.
677 * bug fix: corruption of specials when unbinding is interrupted by an
678 asynchronous unwind (reported by Hannu Koivisto)
679 * improvement: the debugger will now also display local variables that
680 are only used once, for code compiled with a DEBUG optimization quality
683 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
684 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
685 * incompatible change: External-format support for FFI calls. The
686 SB-ALIEN:C-STRING no longer implies an ASCII
687 external-format. Instead the string is subject to external-format
688 conversion. Additionally return values of type SB-ALIEN:C-STRING
689 are by default of type (SIMPLE-ARRAY CHARACTER), not
690 SIMPLE-BASE-STRING. For an alien type that behaves like the old
691 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
692 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
693 * incompatible change: SB-EXT package no longer contains the
694 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
695 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
697 * incompatible change: the single-stepper is no longer available
698 on Alpha, Mips and Sparc platforms.
699 * minor incompatible change: the direct superclasses of
700 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
701 not (STANDARD-OBJECT FUNCTION). This makes the
702 class-precedence-lists of GENERIC-FUNCTION and
703 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
705 * minor incompatible change: the default stream external format on
706 non-unicode SBCL is no longer detected from the locale, but is always
708 * new feature: Add a version of evaluator that uses an interpreter instead
709 of the compiler. EVAL still uses the compiler by default, to switch it
710 to use the interpreter, set the value of the variable
711 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
712 * minor incompatible change: the single-stepper REPL has been merged
713 with the normal debugger (see the "Stepping" heading of the debugger help
714 for more details). The debugger command STEP will no longer switch
715 to the single-stepper REPL.
716 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
717 for a type now works.
718 * bug fix: loading of default sysinit file works. (thanks to Leonid
720 * bug fix: better detection of circularities in the file-compiler.
721 (reported by Marco Monteiro)
722 * bug fix: the CL pathname functions now work with files that have
723 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
724 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
725 systems with Unix98 pty semantics.
726 * bug fix: ASDF-INSTALL will now work with bsd tar.
727 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
729 * bug fix: timers expiring in dead threads no longer cause a
730 type-error (reported by Paul "Nonny Mouse").
731 * bug fix: thanks to more lightweight single-stepper instrumentation,
732 code compiled with (DEBUG 3) will compile and execute significantly faster,
733 and will have more accurate type-inferencing than before
734 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
735 whose bindings are modified
736 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
737 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
738 ** replace dummy implementations of CL:MACHINE-INSTANCE and
739 CL:SOFTWARE-VERSION with proper version
741 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
742 * feature: implemented the READER-METHOD-CLASS and
743 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
744 as specified by AMOP.
745 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
747 * optimization: faster LOGCOUNT implementation on x86 and x86-64
748 (thanks to Lutz Euler)
749 * optimization: hashing of general arrays and vectors has been
750 improved. (reported by Any Fingerhut)
751 * enhancement: SB-INTROSPECT is now able to find definitions of
752 profiled functions. (thanks to Troels Henriksen)
753 * enhancement: compiler-macro expansion applies now to FUNCALL forms
754 as well, allowing compiler-macros for SETF-functions to expand.
755 * enhancement: step-instrumentation no longer wraps known
756 single-value functions with multiple-value context, allowing
757 better type inference.
758 * fixed bug #337: use of MAKE-METHOD in method combination now works
759 even in the presence of user-defined method classes. (reported by
760 Bruno Haible and Pascal Costanza)
761 * fixed bug #339(c): if there are applicable methods not part of any
762 long-form method-combination group, call INVALID-METHOD-ERROR.
763 (reported by Bruno Haible)
764 * fixed bug #361: the :FUNCTION initarg in the protocol for
765 initialization of methods can now be used to override
766 internally-produced optimized functions. (reported by Bruno
768 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
769 system-provided lambda expression no longer cause warnings about
770 unbound #:|pv-table| symbols.
771 * bug fix: improved the handling of type declarations and the
772 detection of violations for keyword arguments with non-constant
774 * bug fix: potentially erronous calls to PATHNAME and
775 MERGE-PATHNAMES were being flushed in some cases.
776 (reported by Richard Kreuter)
777 * bug fix: compiled calls to TYPEP were mishandling obsolete
778 instances. (reported by James Bielman and Attila Lendvai)
779 * bug fix: format strings with ~> without matching ~< no longer
780 trigger an AVER, but signal an understandable error instead.
781 (reported by Antonio Martinez)
782 * bug fix: specifying an output-file without a pathname-type for
783 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
784 of input-file instead of "fasl". (reported by Robert Dodier)
785 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
786 (reported by James Y Knight).
787 * bug fix: compiler-macros-function did not consider the environment
788 argument for shadowing by local functions.
789 * bug fix: compiler-macros expansion was inhibited by local INLINE
791 * bug fix: inline expansions of known functions were subject to
792 step-instrumentation in high debug policies, leading to problems
794 * bug fix: compiler failed to differentiate between different CONS
796 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
797 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
798 * bug fix: #368: incorrect use of expressed vs. upgraded array
800 * bug fix: #306a: more precise unions of array types.
801 * thread-safety improvements:
802 ** CONDITION-WAIT could return early on Linux, if the thread was
803 interrupted and subsequently continued with SIGCONT.
804 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
806 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
807 * added support for the ucs-2 external format. (contributed by Ivan
809 * minor incompatible change: pretty printing of objects of type
810 (cons symbol) is, in the default pprint-dispatch-table, now
811 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
813 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
814 called later in a class's lifetime, possibly as late as when the
815 first instance of the class is created. Previously,
816 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
817 class became finalizeable.
818 * fixed bug: FILE-POSITION sometimes returned inconsistent results
819 for multibyte external-format streams. (thanks to "vbzoli")
820 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
821 with :ALLOCATION :CLASS inherited from superclasses of the
823 * fixed bug: anonymous classes can now be created using the :NAME
824 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
825 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
826 * fixed bug: core-files saved with :EXECUTABLE T can again be
827 executed when SBCL_HOME isn't set. (reported by James Knight)
828 * fixed bug: toplevel LOCALLY forms with declarations could
829 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
830 * fixed bug: printing from several different threads using different
831 values of *print-case* could cause invalid output, due to
832 some internal special variables of the printer not being bound
833 thread-locally (reported by Max Mikhanosha)
834 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
835 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
836 (reported by Pascal Costanza)
837 * fixed bug: FINALIZE-INHERITANCE is called from
838 REINITIALIZE-INSTANCE on classes when the class has previously
839 been finalized, as required by AMOP.
840 * minor code generation optimizations:
841 ** better register allocation in CLOS dispatching functions
842 ** overflow detection when coercing signed bytes to fixnums on x86-64
843 ** is now implemented with one IMUL instruction instead of three shifts
844 ** more efficient bit-vector access on x86 and x86-64
845 ** more efficient access to raw structure slots on x86 and x86-64
846 * fixed some bugs revealed by Paul Dietz' test suite:
847 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
850 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
851 * feature: thread support on Solaris/x86, and experimental thread support
853 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
855 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
856 default method for SLOT-UNBOUND.
857 * minor incompatible change: prevent the user from specializing the
858 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
859 somewhat counter to the intent of the protocol, I (CSR) think, and
860 additionally it just doesn't work in SBCL as currently
861 implemented, thanks to optimizations (that are always valid for
862 the other three SLOT-VALUEish functions, but not for the setter).
863 * minor incompatible change: the :SB-LDB feature is now enabled by
864 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
865 the low-level debugger.
866 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
867 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
868 down to an SB-SYS:FD-STREAM, allowing direct communication in
869 more cases. Temporary files and pipes are still used as a fallback
871 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
872 on the non-reentrant C functions localtime(3) and gmtime(3).
873 * bug fix: global optimization policy was not visible in LOCALLY and
875 * bug fix: class objects can be used as specializers in methods.
876 (reported by Pascal Costanza)
877 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
878 with a type error. (reported by blitz_ on #lisp)
879 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
880 failed with a type error. (reported by Pascal Bourguignon)
881 * bug fix: merging pathnames against defaults with :DIRECTORY
882 starting with '(:RELATIVE :BACK) should preserve the :BACK.
883 (reported by James Y Knight)
884 * bug fix: saving large (>2GB) cores on x86-64 now works
885 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
886 constant index and a (simple-array (signed-byte 32)) array
887 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
888 error (patch by Robert J. Macomber)
889 * feature: TIME now displays the amount of run-time spent in GC
890 * bug fix: The debugger now does a better job of respecting
891 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
892 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
893 workaround for bug 403.)
894 * bug fix: floating point exception handling now works on FreeBSD
895 (thanks to NIIMI Satoshi)
896 * fixed some bugs revealed by Paul Dietz' test suite:
897 ** MISC.641: LET-conversion were not supposed to work in late
899 * improvements to the Win32/x86 port:
900 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
901 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
903 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
904 * new feature: source path information is generated for macro-expansion
905 errors for use in IDE's like Slime (thanks to Helmut Eller)
906 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
907 * bug fix: compiler does not loop forever on an invalid type in
909 * improvement: compilation of most CLOS applications is significantly
911 * optimization: added a limited bytecode compiler for simple toplevel
912 forms, speeding up compilation and FASL loading
913 * bug fix: the statistical profiler now properly distinguishes anonymous
916 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
917 * minor incompatible change: in sbcl-0.9.11 (but not earlier
918 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
919 into subclasses of the finalized class. Now user calls to
920 FINALIZE-INHERITANCE finalize just the one class, and calls by the
921 system return before any subclasses are finalized.
922 * minor incompatible change: The reader no longer ignores errors
923 regarding non-existent packages in #+ and #- feature tests.
924 * new feature: command line options --no-sysinit, --no-userinit to
925 inhibit loading the corresponding init files
926 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
927 for sending data through UDP sockets (thanks to François-René Rideau)
928 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
929 error is signalled instead (thanks to François-René Rideau)
930 * minor incompatible change: Error signalling behaviour of lexical
931 operations violating package locks has changed slightly. Refer to
932 documentation on package locks for details.
933 * enhancement: EVAL can process IF-expressions without resorting to the
935 * bug fix: Some operations involving SIN, COS, or TAN weren't being
936 constant-folded properly due to missing out-of-line functions.
937 (reported by Mika Pihlajamäki)
938 * bug fix: LISTEN sometimes returned T even in cases where no data was
939 immediately available from the stream
940 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
941 were reversed. (reported by Levente Mészáros)
942 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
943 when given a vector displaced to an adjustable array. (reported
945 * bug fix: derivation of float boundaries from numbers outside the
946 appropriate float range (reported by John Wiseman)
947 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
949 * bug fix: precomputing cache entries for generic functions with
950 some subclasses of specializers as yet invalid does not attempt to
951 fill a cache line with a negative offset. (reported by Levente
953 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
954 allow condition class objects as arguments to DOCUMENTATION and
955 (SETF DOCUMENTATION); only find and set documentation for
956 structure names for the STRUCTURE doc-type. (suggested by Gary
958 * improvements to the Win32/x86 port:
959 ** (user-homedir-pathname) and default initialization file
960 locations now know about the user's "Documents and Settings"
961 directory (thanks to Yaroslav Kavenchuk)
962 ** run-program is implemented (thanks to Mike Thomas)
963 ** sockets support (thanks to Timothy Ritchey)
964 ** better backtrace support (thanks to Alastair Bridgewater)
965 ** sb-grovel supported
966 ** asdf-install and sb-posix work somewhat
967 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
968 * improvements to the Solaris/x86 port:
969 ** works on Solaris 11/Solaris Express
970 ** floating-point exception handling support
971 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
972 * fixed some bugs revealed by Paul Dietz' test suite:
973 ** REMOVE-METHOD returns its generic function argument even when
974 the method is not one of the generic functions' methods.
975 ** objects declared as MEMBER types can be admissible arguments to
977 ** printing characters should simply be printed by the FORMAT ~:C
979 ** compiler failure when compiling functions with hairy constant
980 defaults for optional parameters.
981 ** compiler produces wrong code when MAYBE-INLINE-expanding a
982 function, which is already optimized.
984 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
985 * new platform: experimental support for SBCL x86/Darwin, including
986 MacOS X 10.4.5 on Intel.
987 * new feature: Unicode character names are now known to the system
988 (through CHAR-NAME and NAME-CHAR).
989 * new feature: the filesystem location of SBCL's core file is
990 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
991 * minor incompatible change: the contrib modules SB-POSIX and
992 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
993 this change is to make it easier to distribute
994 location-independent binaries.
995 * bug fix: as implied by AMOP, standardized classes no longer have
996 slots named by external symbols of public packages. (reported by
998 * optimization: calling functions via constant symbols -- as in
999 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1000 via the function object as in (FUNCALL #'FOO).
1001 * optimization: on x86 and x86-64, the calling convention for the
1002 case of dealing with an unknown number of values has been altered
1003 to be friendlier to the prediction heuristics implemented,
1004 particularly with reference to CALL and RET pairing. (thanks to
1005 Alastair Bridgewater)
1006 * optimization: on x86, the code for access to thread-local symbol
1007 values and binding/unbinding of thread-local symbols is smaller.
1008 (thanks to Alastair Bridgewater)
1009 * enhancement: CONSTANTP is now able to determine constantness of
1010 more complex forms, including calls to constant-foldable standardized
1011 functions and some special forms beyond QUOTE.
1012 * fixed bug: occasional GC crashes on Solaris/x86
1013 * optimization: x86-64 supports stack allocation of results of simple
1014 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1015 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1016 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1017 (thanks to James Knight)
1018 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1019 by the standard (thanks to Stephen Compall)
1021 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1022 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1023 be used for bundling the runtime and the core file into one
1024 executable binary. This feature is not currently supported on all SBCL
1025 platforms. (thanks to James Bielman and NIIMI Satoshi)
1026 * new feature: a generational or ephemeral garbage collector is now
1027 the default on the PowerPC platform (both Linux and Darwin). The
1028 old Cheney (stop and copy) collector is a build-time option.
1029 (thanks to Cyrus Harmon, after Raymond Toy)
1030 * minor incompatible change: the method by which SBCL finds its
1031 contributed modules has changed; it no longer relies on symbolic
1032 links from an $SBCL_HOME/systems directory, but searches directly
1033 in the subdirectories of $SBCL_HOME.
1034 * enhancement: the dynamic heap size on the Linux/PPC platform is
1035 markedly larger, even using the older Cheney garbage collector.
1036 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1037 certain complicated string types. (reported by Gary King)
1038 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1039 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1040 * fixed bug: DOCUMENTATION on structures no longer signals an
1041 error if no structure type of the right name exists. (reported by
1043 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1044 applicable non-standard primary or :AROUND method on
1045 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1046 no longer cause unbound variable errors. (reported by Kevin Reid)
1047 * fixed bug: metacircle resolution in cases where methods have slots
1048 added before the slots from STANDARD-METHOD. (reported by Jean
1050 * fixed bug: the Power PC assembly code for calling into lisp
1051 sometimes computed its return address wrongly (depending on the
1052 15th bit of the address, fixed at link time). This used to
1053 manifest itself in a segmentation violation while building PCL.
1054 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1055 many others over the years)
1056 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1057 the stack pointer (treated as a signed integer) is negative.
1058 (thanks to Peter van Eynde)
1060 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1061 * new platform: experimental support for the Windows operating
1062 system has been added. (thanks to Alastair Bridgewater)
1063 * fixed several bugs in and robustified the PPC FFI (including
1064 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1065 * bug fix: allow non-simple string symbol names (reported by Paul
1067 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1069 * bug fix: saving a core corrupted callbacks on x86/x86-64
1070 * bug fix: closed a loophole in metacircularity detection and
1071 grounding in the PCL implementation of CLOS.
1072 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1073 (reported by tomppa on #lisp)
1074 * bug fix: clear the direction flag on Lisp -> C transitions, as
1075 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1076 (reported by Andrej Grozin and Hendrik Maryns)
1077 * optimization: major improvements to GC efficiency on GENCGC platforms
1078 * optimization: faster implementation of EQUAL
1079 * optimization: emit more efficient opcodes for some common
1080 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1081 * optimization: several other minor code-generation improvements on x86-64
1082 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1084 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1085 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1086 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1087 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1088 by ANSI as a generic function, and for consistency (SETF
1089 GENERIC-FUNCTION-NAME) is treated likewise.
1090 * fixed bug #233.b: make constraint propagation notice when a variable
1091 value is changed after it is referenced but before it is used
1092 * fixed bug #296: no more arbitrary behaviour from filenames with
1093 odd characters as --load arguments. (reported by Adam Warner)
1094 * fixed bug #390: :CHARACTER-SET pathname components now work as
1095 expected. (reported by Tim Daly Jr)
1096 * fixed bug #391: complicated :TYPE intersections in slot
1097 definitions no longer cause an error in PCL internals.
1098 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1099 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1100 returning the number of octets which would be written to the
1101 file-stream. (thanks to Robert J. Macomber)
1102 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1103 arguments to the internal stream functions. (thanks to David
1105 * bug fix: stack exhaustion is now handled on the x86/SunOS
1106 platform. (thanks to Daisuke Homma)
1107 * bug fix: more accurate ROOM results on GENCGC platforms
1108 * optimization: improved type inference for arithmetic-for
1109 index variables in LOOP
1110 * optimization: faster floating-point SQRT on x86-64
1111 * fixed some bugs revealed by Paul Dietz' test suite:
1112 ** DOCUMENTATION returns NIL instead of "" for method combinations
1113 that don't have a docstring
1115 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1116 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1117 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1118 therefore are not customizeable by user code (as seems to be at
1119 least permitted and maybe required by AMOP). As a consolation,
1120 however, the SBCL implementation of these functions now calls
1121 REINITIALIZE-INSTANCE as specified by AMOP.
1122 * bug fix: slot-definition documentation is propagated as per ANSI
1123 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1124 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1125 argument, as do eslotd initialization methods. (from Pascal
1126 Costanza's "Closer" project)
1127 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1128 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1130 * bug fix: it is now possible to have more than one subclass of
1131 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1132 (reported by Bruno Haible, Pascal Costanza and others)
1133 * bug fix: the dependent update protocol now works for generic
1134 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1135 and Pascal Costanza)
1136 * bug fix: condition-class instances corresponding to
1137 DEFINE-CONDITION forms are now created eagerly. (reported by
1138 Kalle Olavi Niemitalo on comp.lang.lisp)
1139 * bug fix: floating point printing is more accurate in some
1140 circumstances. (thanks to Simon Alexander)
1141 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1142 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1143 * bug fix: callbacks on OS X now preserve stack-alignment.
1144 (thanks to Cyrus Harmon)
1145 * enhancement: the x86-64 disassembler is much better at
1146 disassembling SSE instructions. (thanks to Lutz Euler)
1147 * enhancement: CHECK-TYPE on a variable now allows type-inference
1148 (thanks to Vincent Arkesteijn)
1149 * optimization: improved performance of EUC-JP external format.
1150 (thanks to NIIMI Satoshi)
1151 * optimization: performance improvements to IO on file streams of
1152 :ELEMENT-TYPE CHARACTER
1153 * optimization: much faster memory allocation on x86-64
1154 * optimization: faster garbage collections (latency and throughput)
1156 * optimization: faster bignum subtraction and division on x86 and x86-64
1157 (thanks to Lutz Euler)
1158 * optimization: more accurate type-inference for loops that use a
1159 floating point index variable or a negative step.
1161 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1162 * bug fix: add a workaround to SBCL looping infinitely at startup on
1163 Linux kernels with apparently buggy implementations of personality().
1164 (thanks to Svein Ove Aas)
1165 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1166 (thanks to David Lichteblau)
1167 * bug fix: local bindings shadow global symbol macros for
1168 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1169 * optimization: non-open coded uses of numeric comparison operators
1170 (e.g. >) no longer cons when called with more than one parameter
1171 on platforms supporting dynamic-extent allocation.
1172 * enhancement: saving cores with foreign code loaded is now supported
1173 on MIPS/Linux in addition to the previously supported platforms.
1174 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1175 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1176 * bug fix: division by zero in sb-sprof when no samples were collected
1177 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1178 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1179 (thanks to Alexander Kjeldaas)
1180 * new feature: ignoring the return values of destructive functions
1181 like NREVERSE signals a compile-time style-warning.
1182 (thanks to Kevin Reid)
1183 * new feature: the alignment of alien structure fields can be
1184 explicitly specified. (thanks to Cyrus Harmon)
1185 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1186 and dump core on SIGQUIT
1188 ** incompatible change: threads do not inherit values of specials
1189 from their parents (see manual)
1190 ** bug fix: threads stacks belonging to dead threads are freed by the
1191 next exiting thread, no need to gc to collect thread stacks anymore
1192 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1193 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1194 ** bug fix: dynamic variable and thread start related gc lossage
1195 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1197 * fixed some bugs revealed by Paul Dietz' test suite:
1198 ** SUBTYPEP is slightly more accurate on heinously complicated
1199 CONS types where some of the members have uncertain (in the
1200 NIL, NIL sense) type relationships to each other.
1202 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1203 no longer purified unless :PURIFY T is explicitly specified.
1204 ** Non-purified cores are significantly smaller than before
1206 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1207 * new feature: timers based on Zach Beane's excellent timer package
1208 * added support for the following external formats: koi8-u,
1209 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1210 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1211 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1212 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1213 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1214 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1215 * incompatible change: a threaded SBCL will no longer revert to
1216 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1217 * bug fix: interrupts are disabled until startup is complete; no
1218 more sigsegvs when receiving a signal to soon
1219 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1221 * bug fix: add a workaround for the memory randomization features in
1222 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1223 workaround will only be in effect on systems with the proc filesystem
1225 * bug fix: printing objects of type HASH-TABLE signals a
1226 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1228 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1229 have been running for over 50 days. (reported by Gilbert Baumann)
1230 * bug fix: the logic for getting names of functions gets less
1231 confused when confronded with alternate-metaclass
1232 funcallable-instances. (reported by Cyrus Harmon)
1233 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1234 consistent, even on internal alternate-metaclass objects.
1235 * bug fix: flush closure information collected by physical
1236 environment analysis prepass before the main pass. (bug reported
1238 * bug fix: compiler pack phase does not modify a hashtable, which is
1239 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1240 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1241 may be expanded. (reported by Alexander <malishev> on c.l.l)
1242 * bug fix: no more style-warnings when compiling code using LAST
1243 under high-SPEED low-SPACE settings. (reported by David Wallin
1245 * fixed some incompatibilities between SBCL's MOP and the MOP
1247 ** the METAOBJECT class is now implemented;
1248 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1249 STANDARD-OBJECT, as required;
1250 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1251 now compatible (as required by VALIDATE-SUPERCLASS); there
1252 remains a consistency requirement over the presence or absence
1253 of the FUNCTION class in the superclass list of finalized
1254 classes; see the manual for more details;
1255 ** the :AROUND method for COMPUTE-SLOTS on
1256 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1257 requested slot ordering.
1259 ** bug fix: parent thread now can be gc'ed even with a live
1261 ** bug fix: binding a special with PROGV to no value is not
1262 the same as that symbol not having been bound (thanks to
1264 ** bug fix: binding specials is thread safe (thanks to
1266 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1267 and SB-SPROF do not die with 'no handler for signal XX in
1268 interrupt_handle_now(..)' anymore
1269 ** bug fix: WITH-TIMEOUT works with multiple running threads
1270 * fixed some bugs revealed by Paul Dietz' test suite:
1271 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1272 the :method-class keyword argument.
1274 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1275 * new port: the Solaris operating system on x86 processors is now
1276 mostly supported, though some rough edges in the environment
1277 remain. (thanks to Daisuke Homma)
1278 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1279 space for its heap. (thanks to Thiemo Seufer)
1280 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1281 closures. (thanks to Thiemo Seufer)
1282 * minor incompatible change: eof selects abort in the debugger.
1283 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1284 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1286 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1287 deprecated, and will go away in a future revision of SBCL.
1288 * minor incompatible change: GC-ON and GC-OFF are no longer
1289 implemented with a counter, it does not matter how many times gc
1290 is switched on or off
1291 * bug fix: discriminating functions for generic function classes
1292 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1293 make invalid assumptions about method precedence order. (reported
1295 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1296 platforms. (thanks to Thiemo Seufer)
1297 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1298 * bug fix: correct pathname printing with printer escaping is on.
1299 (thanks to Kevin Reid)
1300 * bug fix: complex VOP definitions in "user-space" no longer trigger
1301 package locks. (reported by Zach Beane)
1302 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1303 for funcallable-instance functions, and is no different from
1305 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1306 characters in the current readtable. (reported by Nicholas Neuss)
1307 * bug fix: SUBTYPEP on various CONS types returns more a more
1308 accurate acknowledgment of its certainty.
1309 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1310 lists in some cases. This partially fixes bug 384.
1311 * flush all standard streams before prompting in the REPL and the
1313 * bug fix: signal handling and triggering gc do not conflict
1314 directly anymore, in particular a high frequency sb-sprof does
1315 not prevent gc from running
1316 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1317 approximation for timezone and DST information between the
1318 universal time epoch and the smallest negative 32-bit time_t.
1319 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1320 year 1899 so long as the encoded time is non-negative.
1321 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1322 the correct answer for non-constant rotate amounts when performing
1323 an inline 32-bit rotation.
1325 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1326 there is only one thread in the session
1327 ** bug fix: memory leak for streams created in one thread and
1328 written to in another
1329 ** bug fix: lockup when compiled with gcc4
1330 ** bug fix: race that allows the gc to be triggered when gc is
1332 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1333 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1335 ** bug fix: finished threads are gc'ed properly
1336 * fixed some bugs revealed by Paul Dietz' test suite:
1337 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1338 has arguments with a different set of applicable methods from
1339 the orignal arguments.
1340 ** The type error thrown by MAP now has a correct expected-type
1342 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1343 name a compiled function.
1344 ** Three MISC tests where a large bignum was improperly coerced to
1345 a float that couldn't represent that bignum during type
1346 derivation were fixed.
1347 ** SUBTYPEP can now handle types involving the negation of a
1348 list-form FUNCTION type.
1349 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1350 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1351 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1353 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1354 * New feature: Experimental support for bivalent streams: streams
1355 opened with :element-type :default now allow character and binary
1356 (unsigned-byte 8) I/O
1357 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1358 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1360 * bug fix: no more highly sporadic "couldn't check whether ~S is
1361 readable" when reading a stream and an interrupt hits in the middle
1362 of a select system call
1363 * compiler better recognizes complex arrays (reported by Eduardo
1365 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1366 for variables on SBCL built with linkage-tables. (reported by Luis
1368 * various error reporting improvements.
1369 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1370 (Thanks to Thiemo Seufer)
1371 * bug fix: nested reader invokations work correctly
1372 * bug fix: it is possible to have simultaneous references to foreign
1373 code and foreign data with the same name.
1375 ** added x86-64 support
1376 ** incompatible change: the threading api now works with thread
1377 objects instead of thread ids
1378 ** bug fix: threads are protected from signals and interruption when
1379 starting up or going down
1380 ** bug fix: a race where an exiting thread could lose its stack to gc
1381 ** bug fix: don't halt on infinite error in threads if possible
1382 ** fixed numerous gc deadlocks introduced in the pthread merge
1383 ** bug fix: fixed thread safety issues in read and print
1384 ** bug fix: debugger doesn't hang on session lock if interrupted at
1385 an inappropriate moment
1386 ** bug fix: run-program is now thread safe(r)
1387 ** bug fix: inner with-recursive-lock no longer releases the mutex
1388 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1389 * fixed some bugs revealed by Paul Dietz' test suite:
1390 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1391 EXPECTED-TYPE slots filled.
1392 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1394 ** MAKE-CONDITION accepts classes as type-designators.
1395 ** COMPILE may never return NIL.
1396 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1397 range before calling Unix time functions
1399 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1400 * numerous signal handling fixes to increase stability
1401 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1402 * minor incompatible change: we now correctly canonize default
1403 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1404 as per the MOP, rather than the historical (INITARG INITFUNCTION
1405 INITFORM). (reported by Bruno Haible)
1406 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1407 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1408 for more information.
1409 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1410 pathname is a directory pathname.
1411 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1412 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1414 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1415 :METACLASS argument in addition to a class name. (reported by
1416 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1417 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1418 (reported by Bruno Haible; patch from Gerd Moellmann)
1419 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1421 * bug fix: more cleanups to the floating point exception handling on
1422 x86-64 (thanks to James Knight)
1423 * bug fix: the compiler does not try to constant fold calls of
1424 COERCE and BIT-* functions when they return freshly constructed
1425 objects. (reported by Nikodemus Siivola and Paul Dietz)
1426 * optimization: improved the MIPS versions of generic (in the
1427 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1428 * optimization: direct conversion of (unsigned-byte 32) to floats on
1429 the PowerPC platform.
1430 * optimization: structure instances with raw slots now use less
1431 memory, and probably show better memory locality. (thanks to
1433 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1434 are treated as requests for stack allocation on the x86-64,
1435 Alpha-32, PPC and SPARC platforms.
1436 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1437 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1438 (thanks to Tony Martinez)
1440 ** gcing a dead thread can no longer lead to lockups
1441 ** threads block signals until they are set up properly
1442 ** errno is no longer shared by threads
1443 ** interrupt-thread restores the eflags register on x86
1444 ** fixed some lockups due to gc/thread interaction
1445 * fixed some bugs revealed by Paul Dietz' test suite:
1446 ** invalid dotted lists no longer raise a read error when
1447 *READ-SUPPRESS* is T
1448 ** signal an error if a symbol that names a declaration is used
1449 as the name of a type, or vice versa
1450 ** allow using the (declare (typespec var*)) abbreviation for
1451 (declare (type typespec var*)) with all type specifiers
1452 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1453 ** always return NIL from PPRINT-POP when OBJECT is NIL
1454 ** don't signal errors when pretty-printing malformed LABELS,
1455 FLET or MACROLET forms
1456 ** declarations in a DOLIST body are also in scope for the
1458 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1460 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1463 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1464 * fixed cross-compiler leakages that prevented building a 32-bit
1465 target with a 64-bit host compiler.
1466 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1467 opened with :IF-EXISTS :OVERWRITE.
1468 * fixed bug 281: error for an invalid qualifier in a short-form method
1469 combination method is not signalled until the faulty method is called.
1470 * bug fix: iteration variable type inferrer failed to deal with open
1471 intervals. (reported by Alan Shields)
1472 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1473 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1474 or more runtime options were provided to the sbcl binary.
1475 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1476 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1477 * optimization: there's now a fast-path for fixnum arguments in the
1478 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1479 * optimization: the code generated on x86-64 is more compact thanks
1480 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1481 * bug fix: floating-point exception handling now partly works on
1482 x86-64. (Thanks to James Knight)
1483 * improvement to the MIPS backend from Thiemo Seufer: C-style
1484 64-bit long long arguments and return values to alien functions
1486 * fixed some bugs revealed by Paul Dietz' test suite:
1487 ** the type-error signalled from WARN has a filled-in DATUM slot.
1488 ** the type-error required when a stream is not associated with
1489 a file has the stream as its datum.
1490 ** type-errors on single-floats on x86-64 no longer have
1491 :INVALID-OBJECT as the datum
1492 ** the type-errors signalled for invalid function names now have
1493 a correct expected type
1494 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1495 for structures defined with a :TYPE. Documentation strings for
1496 typed structures are no longer immediately discarded
1497 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1498 broadcast streams and synonym streams. FILE-LENGTH now also works
1499 on broadcast streams.
1501 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1502 * incompatible change: the --noprogrammer option, deprecated since
1503 version 0.7.5, has been removed. Please use the equivalent
1504 --disable-debugger option instead.
1505 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1507 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1508 inoperational for a while now) has been completely removed.
1509 * null lexical environments are now printed as #<NULL-LEXENV>,
1510 significantly reducing the amount of clutter in typical backtraces.
1511 * documentation on weak pointers, finalization, and after GC hooks
1512 has been added to the manual.
1513 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1514 as other specialized array types, is much faster. SUBSEQ and
1515 COPY-SEQ on such arrays have also been sped up.
1516 * optimization: EQL is now more efficient when at least other argument
1517 is known to be of type (OR FIXNUM (NOT NUMBER)).
1518 * fixed bug: the runtime is now less vulnerable to changes in the
1519 size of the SBCL object on OS X, and virtual memory is reserved for
1520 all spaces that need to be at a fixed address.
1521 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1522 * fixed bug: finalizers and after GC hooks that cause consing are now
1524 * fixed bug: compiler error messages and summaries are now printed to
1525 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1526 * fixed inference of the upper bound of an iteration variable.
1527 (reported by Rajat Datta).
1528 * fixed bug 211e: calling local functions with duplicated constant
1529 keyword argument no longer causes a bogus style warning about an
1531 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1532 ftype declaration to be disregarded. (reported by Dave Roberts)
1533 * fixed bug 373: caused by erronous compilation of references to alien
1534 variables in the runtime on ppc/darwin.
1535 * fixed bug 376: CONJUGATE type deriver.
1536 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1537 (reported by Baughn on #lisp)
1538 * a cross-compiler bug on non-x86 platforms has been identified and
1539 fixed. (thanks to Bruno Haible)
1540 * improvements to the MIPS runtime code for increased stability.
1541 (thanks to Thiemo Seufer)
1542 * increased the maximimum compact environment size to allow
1543 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1544 * partial workaround for bug 135: don't name the function we're
1545 calling for hairy cases of EVAL, so as not to accumulate one environment
1546 entry per call to EVAL. (reported by Kevin Reid)
1547 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1548 * optimization: type testing for non-vector arrays should be faster.
1549 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1551 * bug fix: setting 31st element of a bit vector to zero did not work
1553 * bug fix: redefining a class definition which failed due to a
1554 previous accessor / function clash now works (but see BUGS entry
1555 #380 for more problems in this area). (thanks to Zach Beane)
1556 * the long form of DEFINE-METHOD-COMBINATION disables method group
1557 checking when given a single method group with pattern *. (thanks
1559 * on x86 compiler supports stack allocation of results of simple
1560 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1561 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1562 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1563 * fixed some bugs related to Unicode integration:
1564 ** the restarts for recovering from input and output encoding
1565 errors only appear when there is in fact such an error to
1567 * fixed some bugs revealed by Paul Dietz' test suite:
1568 ** MISC.549 and similar: late transformation of unsafe type
1569 assertions into derived types caused unexpected code
1571 ** SCALE-FLOAT type deriver is less wrong.
1572 ** type derivers for EXP, LOG and similar functions compute result
1573 types for complex arguments better.
1574 ** (MISC.563) CONJUGATE type deriver works for very restricted
1576 ** out-of-line type testers for character strings are available.
1577 ** EQUAL compiler transform understands specialness of objects
1579 ** accessing double-floats stored on the stack now works on x86-64.
1580 ** debugger internals could sometimes create invalid lispobjs,
1581 resulting in GC crashes.
1582 ** MISC.548: type check weakening can convert required type into
1584 ** initialization forms for bindings are not in scope of free special
1587 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1588 * incompatible change: thread support for non-NPTL systems has
1589 been removed - locking is buggy and unreliable. A threaded
1590 SBCL build will now warn at startup and refuse to create
1591 new threads, unless futex support is detected in the kernel
1592 * incompatible change: the top level REPL now has only an ABORT
1593 restart associated with it, not TOPLEVEL and ABORT as it used to.
1594 TOP and TOPLEVEL are now available as debugger commands for
1595 returning to the top level.
1596 * incompatible change: forms evaluated in the REPL now use the
1597 global optimization policy.
1598 * incompatible change: user- and system-initialization files are
1599 no longer processed with LOAD, but by READ and EVAL; hence the
1600 global optimization policy, startup package, readtable, etc,
1602 * The .fasl file format number has been incremented because of
1603 various incompatible changes.
1604 * internal entry point details and argument counts no longer appear
1605 in backtraces unless explicitly requested by setting
1606 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1607 * built-in and standard functions no longer have names like "top
1608 level local call to FOO".
1609 * fixed bug 32: functions defined in non-null lexical environments
1610 now have more legible printed representation
1611 * fixed bug 33: functions defined in non-null lexical environemnts
1612 are now more amenable to inspection by INSPECT.
1613 * workaround for bug 354: XEPs no longer appear in backtraces unless
1614 explicitly requested.
1615 * fixed bug: receiving the signal which results from attempting to
1616 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1617 SIGBUS on other *BSDs) on architectures where the C stack is also
1618 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1619 to debugger confusion or wild instability. (reported by Cheuksan
1621 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1622 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1623 (reported by Lutz Euler)
1624 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1625 compiled code. (reported by Paul Dietz)
1626 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1627 specialized parameter to a method to something that is not TYPEP
1628 the specializer is now possible.
1629 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1630 face of package deletion.
1631 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1632 pathname. (reported by David Tolpin on c.l.l)
1633 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1634 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1635 STRING-STREAM and FILE-STREAM, respectively.
1636 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1637 than previously; in addition, its test suite is now run on build.
1638 (thanks to Luke Gorrie)
1639 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1640 * optimization: added a immediate representation for single-floats
1642 * fixed some bugs revealed by Paul Dietz' test suite:
1643 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1644 ** attempting to create a package with a colliding nickname causes
1645 correctable errors to be signalled.
1646 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1647 ** several x86-64 backend bugs related to sign-extension of immediate
1650 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1651 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1652 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1654 * fixed loading of multiply forward-referenced layouts.
1655 (thanks to Cheuksan Wang)
1656 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1657 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1658 (see COMPILE-FILE documentation for details of :PRINT option)
1659 * fixed bugs 19 and 317: fixed-format floating point printing is
1660 more accurate. This also fixes a bug reported by Adam Warner
1661 related to the ~@F format directive.
1662 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1664 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1665 dispatch table if the from-char is a dispatch macro character.
1666 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1667 by the wordsize no longer ignore the last word. (reported by Lutz
1669 * fixed bug in type checking of dynamic-extent variables. (reported
1671 * optimization: sequence traversal functions use their freedom to
1672 coerce function designators to functions.
1673 * optimization: code with many calls to CLOS methods specialized on
1674 CLOS classes has had redundant type checks removed.
1675 * optimization: added declarations to speed up operations that access
1676 the internal character database (for example STRING-UPCASE)
1677 * optimization: comparison operations between floats and sufficiently small
1678 fixnums no longer create extra rationals
1679 * fixed some bugs related to Unicode integration:
1680 ** portions of multibyte characters at the end of buffers for
1681 character-based file input are correctly transferred to the
1682 start of the buffer at the next read.
1683 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1684 passing it through to OPEN.
1685 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1686 argument given to internal calls to OPEN.
1687 ** the built-in comment readers (introduced by character sequences
1688 ";" and "#|") are more forgiving to encoding errors; they will
1689 STYLE-WARN and then attempt to resync the stream at a character
1690 boundary. (thanks to Teemu Kalvas)
1691 * fixed some bugs in the x86-64 port:
1692 ** Negative short int return values from c-calls are sign-extended
1694 ** The stack is aligned to 16-bytes for c-calls, as required by
1695 the ABI. (thanks to Cheuksan Wang)
1696 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1697 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1698 for enums. (thanks to Vincent Arkesteijn)
1699 ** Multiple small optimizations and bugfixes for floating point
1701 * fixed some bugs revealed by Paul Dietz' test suite:
1702 ** Space, Tab, Linefeed, Return and Page have the invalid
1703 secondary constituent character trait.
1704 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1706 ** WITH-INPUT-FROM-STRING should only update the index place on
1708 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1709 works more reliably.
1710 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1711 with one-dimensional arrays with fill pointers.
1712 ** TYPE-OF failed on a complex with an integer realpart and a
1714 ** compiler failure during type inference for the code of form
1715 (IF (EQL X (THE ...)) ...) (MISC.535).
1717 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1718 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1719 platform. The port supports 61-bit fixnums, large memory spaces
1720 and reloading shared object files.
1721 * enhancement: saving cores with foreign code loaded is now
1722 supported on ppc/Darwin in addition to the previously supported
1724 * enhancement: the statistical profiler now walks deeper into the
1725 call stack for more meaningful call-graphs and accrued time
1726 reports (x86/x86-64 only). It also now reports time spent in
1728 * enhancement: it is now possible to trace most individual methods
1729 of a generic function in addition to tracing the generic function
1731 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1733 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1734 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1735 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1736 produces an error. (thanks to Vincent Arkesteijn)
1737 * bug fix: NAMESTRING on pathnames with :WILD components in their
1738 directories works correctly. (thanks to Artem V. Andreev)
1739 * fixed bug 125: compiler preserves identity of closures. (reported
1741 * bug fixed: functions with &REST arguments sometimes failed with
1742 "Undefined function" when compiled with (DEBUG 3). (reported by
1744 * bug fix: overflow during compiling of setting element of a bit
1745 vector with constant index and value. (reported by Timmy Douglas)
1746 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1747 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1748 * on x86 compiler partially supports stack allocation of dynamic-extent
1750 * GO and RETURN-FROM do not check the extent of their exit points
1751 when compiled with SAFETY 0.
1752 * fixed some bugs related to Unicode integration:
1753 ** encoding and decoding errors are now much more robustly
1754 handled; it should now be possible to recover even from invalid
1755 input or output to the terminal. (thanks to Teemu Kalvas)
1756 ** provided a first cut at implementing STRING-TO-OCTETS and
1757 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1758 ** altered the SB-MD5 contributed module slightly, changing the
1759 interface just enough for it to be supportable for builds where
1760 lisp characters are not eight bits.
1761 * fixed some bugs revealed by Paul Dietz' test suite:
1762 ** the FORMATTER-generated functions for ~V[ conditionals require
1763 the correct number of arguments.
1764 ** READ-FROM-STRING returns the mandated second value when applied
1765 to displaced strings.
1766 ** the #\Rubout and #\Backspace characters are treated as invalid
1767 constituent characters by the tokenizer.
1769 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1770 * new feature: reloading changed shared object files with
1771 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1772 * new feature: references to foreign variables and functions
1773 can now be compiled and loaded before the corresponding shared
1774 object file is loaded, as long as the foreign definitions are
1775 available at runtime.
1776 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1777 supported. (thanks to Dan Debertin)
1778 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1779 just keywords. (thanks to Vincent Arkesteijn)
1780 * fixed bug #331: structure-class instances corresponding to
1781 DEFSTRUCT forms are now created eagerly.
1782 * fixed bug #345: backtraces from calls to undefined functions work
1783 on x86 as well. Related bug #61 is now also partially fixed on x86
1784 (backtraces from throws to unknown catch tags.)
1785 * bug fix: lambda-list parsing is now stricter vrt. order and number
1786 of lambda-list keywords.
1787 * bug fix: as specified by AMOP, an error is signalled if a
1788 class-option appears multiple times in a DEFCLASS form. (reported
1790 * bug fix: RANDOM can be compiled when the compiler derives the type
1791 of its numeric argument as a disjoint set of small integers.
1792 (reported by Paul Dietz)
1793 * bug fix: starting a core saved with shared objects loaded when
1794 those objects are not available no longer causes threaded SBCL to
1795 hang. (reported by Sean Ross)
1796 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1798 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1799 argument, which must always be NIL. (reported by Kalle Niemitalo)
1800 * bug fix: printing 1.0d+23 no longer results in an error.
1801 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1802 * bug fix: structure slot setters preserve evaluation order. (thanks
1804 * fixed some bugs related to Unicode integration:
1805 ** RUN-PROGRAM can allow its child to take input from a Lisp
1806 stream. (reported by Stefan Scholl)
1807 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1808 locales. (reported by Ken Causey)
1809 * fixed some bugs revealed by Paul Dietz' test suite:
1810 ** INCF, DECF and REMF evaluate their place form as specified in
1812 ** FORMATTER expands ~{ ~} iteration directives with V or #
1813 parameters correctly.
1814 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1815 consequent uses no arguments correctly.
1816 ** the system has a partial understanding of the (COMPLEX RATIO)
1818 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1821 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1822 * new feature: a build-time option (controlled by the :SB-UNICODE
1823 keyword feature, enabled by default) for building the system with
1824 support for the entire 21-bit character space defined by the
1826 * new feature: the system now has rudimentary external-format
1827 support; the primary user-visible change at this time is that
1828 characters with the high bit set (such as lower-case-e-acute) will
1829 print correctly to a terminal in a UTF-8 environment.
1830 * minor incompatible change: BASE-CHAR no longer names a class;
1831 however, CHARACTER continues to do so, as required by ANSI.
1832 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1833 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1834 has been moved to the SB-EXT package (temporarily re-exported from
1836 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1837 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1839 * fixed bug #351: better error-handlind and reporting for malformed
1841 * fixed bug #350: bignum-printing is now more memory-efficient,
1842 allowing printing of very large bignums, eg. (expt 2 10000000).
1843 (reported by Bruno Haible)
1844 * fixed bug #302: better primitive-type selection for intersection
1846 * fixed bug #308: non-graphic characters now all have names, as
1847 required. (reported by Bruno Haible)
1848 * bug fix: redefining a class with different superclasses now correctly
1849 removes it from the direct-subclasses of its previous superclasses.
1850 (reported by David Morse)
1851 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1852 new value now works. (reported by Bruno Haible)
1853 * bug fix: correct canonicalization of multiple non-standard slot
1854 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1855 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1856 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1857 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1858 now exists, an signals an error.
1859 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1860 during macroexpansion for non-iteration variables as well. (reported
1861 by Bruno Haible for CMUCL)
1862 * bug fix: Cyclic structures and unprintable objects in compiler
1863 messages no longer cause errors. (reported by Bruno Haible)
1864 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1865 and READ-FROM-STRING all now return a primary value of NIL if
1866 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1867 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1868 (reported by Bruno Haible for CMUCL)
1869 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1870 do not designate a condition. (reported by Bruno Haible for
1872 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1873 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1874 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1875 permitting resolution in favour of any of the conflicting symbols.
1876 (reported by Bruno Haible for CMUCL)
1877 * bug fix: EQUAL compiler optimizations is less aggressive on
1878 strings which can potentially compare true despite having distinct
1879 specialized array element types.
1880 * bug fix: unit enumerations can be defined without dividing by
1881 zero. (thanks to Vincent Arkesteijn)
1882 * FORMAT compile-time argument count checking has been enhanced.
1883 (report from Bruno Haible for CMUCL)
1884 * a partial workaround for the bug 262: the compiler does not try to
1885 inline-expand a local function doing RETURN-FROM from a deleted
1886 BLOCK. (thanks to Peter Denno for the bug report and to David
1887 Wragg for the simple test case)
1888 * fixed some bugs revealed by Paul Dietz' test suite:
1889 ** RENAME-PACKAGE allows all package designators as new package
1891 ** constraint propagation and lambda variable substitution are
1892 more cautious in dealing with partially deleted code.
1893 ** compiler sometimes lost reoptimization passes.
1894 ** CERROR, given a condition as condition designator, passes its
1895 remaining arguments to the continue format control without
1897 ** Case-altering FORMAT directives work correctly on non-ASCII
1899 ** The REAL type specifier handles bounds outside the floating
1900 point ranges without signalling FLOATING-POINT-OVERFLOW.
1901 ** Functions with IR1-transformations can create intercomponent
1902 references to global functions.
1903 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1905 ** FORMAT ~R treats a nil value for its first parameter correctly.
1907 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1908 * enhancement: saving cores with foreign code loaded is now
1909 supported on x86/NetBSD and sparc/Linux in addition to the previously
1910 supported platforms.
1911 * bug fix: on some platforms repeated installations caused multiple
1912 copies of HTML documentation to be installed -- should not happen
1913 any more. (reported by Stefan Scholl)
1914 * bug fix: parsing self-recursive alien record types multiple times
1915 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1916 original patch by Helmut Eller for CMUCL)
1917 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1918 (thanks to Richard Kreuter)
1919 * bug fix: defining classes whose accessors are methods on existing
1920 generic functions in other (locked) packages no longer signals
1921 bogus package lock violations. (reported by François-René Rideau)
1922 * bug fix: special variables as DEFMETHOD parameters no longer have
1923 associated bogus type declarations. (reported by David Wragg and
1925 * bug fix: read-write consistency on streams of element-type
1926 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1927 * bug fix: redefiniton of the only method of a generic function with
1928 no DEFGENERIC no longer emits a full WARNING. In addition,
1929 redefinition of generic functions with no DEFGENERIC to an
1930 incompatible lambda list now signals an error. (thanks to Zach
1932 * bug fix: DEFGENERIC now works even when there's a function of the
1933 same name in an enclosing lexical environment. (thanks to Zach
1935 * fixed compiler failure, caused by instrumenting code during
1936 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1937 * optimization: added loop analysis and improved register allocation
1938 * fixed some bugs revealed by Paul Dietz' test suite:
1939 ** POSITION on displaced vectors with non-zero displacement
1940 returns the right answer.
1941 ** (SIMPLE-STRING) is a valid type specifier for sequence
1943 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1945 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1946 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1948 ** The justification version of the FORMAT ~< directive treats
1949 non-zero minpad parameter correctly.
1951 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1952 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1953 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1954 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1955 the supported interface.
1956 * new feature: Single-stepping of code compiled with DEBUG 2 or
1957 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1958 * new feature: saving cores with foreign code loaded is now
1959 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1960 Timothy Moore's work for CMUCL)
1961 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1962 parameters to * instead of NIL if no initform is supplied.
1963 (reported by Johan Bockgaard)
1964 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1965 correctly when FUNCALL appears as the car of the form. Note:
1966 despite this FUNCALL forms are not currently subject to
1967 compiler-macro expansion. (port of Raymond Toy's fix for the
1968 same from CMUCL, reported by Johan Bockgaard)
1969 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1970 (thanks for Teemu Kalvas)
1971 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1972 case instead of terminating the iteration (thanks for Julian
1973 Squires, Sean Champ and Raymond Toy)
1974 * bug fix: incorrect expansion of defgeneric that caused a style
1975 warning. (thanks for Zach Beane)
1976 * bug fix: slot accessor effective method computation works properly
1977 for classes with multiple non-standard applicable methods on
1978 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1979 * on x86 compiler supports stack allocation of results of LIST and
1980 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1981 CMUCL implementation by Gerd Moellmann)
1982 * fixed some bugs revealed by Paul Dietz' test suite:
1983 ** FORMAT strings with both the ~<~:;~> form of the justification
1984 directive and pretty-printing directives cause an error.
1986 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1987 * incompatible change: the internal functions
1988 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1989 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1990 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1991 instead of the old functions.
1992 * new feature: on platforms where "dladdr" is available foreign
1993 function names now appear in backtraces. (based on Helmut Eller's
1995 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1996 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1998 * documentation: ASDF manual is now distributed with SBCL.
1999 * bug fix: SBCL can now load its contributed modules with REQUIRE
2000 even if the system-provided entries have been removed from
2001 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2003 * bug fix: backtraces involving undefined functions or assembly
2004 routines are more informative. (thanks to Brian Downing)
2005 * bug fix: mutually referent alien structures now work correctly.
2006 (reported by Rick Taube)
2007 * bug fix: structures defined by WITH-ALIEN can be referred to
2008 within other definitions in the same WITH-ALIEN.
2009 * bug fix: division operators (MOD, TRUNCATE and the like) with
2010 constant zero divisors and integer dividends no longer generate
2012 * bug fix: provide default methods for INPUT-STREAM-P and
2013 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2014 * bug fix: improve the detection and resolution of MOP metacycles
2015 (where effective method computation depends on the generic function
2016 having its effective method computed). (reported by Bruno Haible)
2017 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2018 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2019 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2020 * bug fix: pathnames with relative directory components can be
2021 represented relative to default pathnames.
2022 * optimization: in taking the GCD of bignums, reduce the two bignums
2023 to approximately the same size (using Euclid's algorithm) before
2024 applying the more sophisticated binary GCD. (thanks to Juho
2026 * optimization: COUNT on bitvectors now operates word-at-a-time.
2027 * optimization: ASH with a positive, but not necessarily constant,
2028 (leftwards) shift, when in a modular context, is compiled to a
2030 * fixed some bugs revealed by Paul Dietz' test suite:
2031 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2032 if the corresponding argument is NIL.
2033 ** FORMAT directives accepting a minpad parameter treat negative
2035 ** literal commas as character parameters to FORMAT directives are
2037 ** literal spaces directly after ~<Newline> directives within a
2038 format-logical-block (~:< ~@:>) do not induce :FILL-style
2039 conditional newlines.
2040 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2041 as specified: it no longer includes conditional newlines).
2042 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2044 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2045 :PER-LINE-PREFIX argument does not evaluate to a string.
2047 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2048 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2049 the manual for details; for now, package locks can be disabled by
2050 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2051 if no major problems are found then it is likely that they will be
2052 compiled in unconditionally.
2053 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2054 unsupported operators on all platforms. To load a shared library
2055 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2056 object file, link it into a shared library outside of SBCL and
2057 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2059 * bug workaround: ROOM T and the GC were not working together
2060 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2061 reporting facility aren't preserved. That reporting has been
2062 disabled, so now until and unless someone figures out how to make it
2063 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2064 * minor incompatible change: as threatened around sbcl-0.8.0, the
2065 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2066 The symbols are also exported from SB-PCL for backwards
2067 compatibility, but more so than before SB-PCL should be treated as
2068 an implementation-internal package.
2069 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2071 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2072 It is known as of this release not to work on the Alpha, however.
2073 * fixed bug #167: errors signalled due to illegal syntax in method
2074 bodies are now more legible.
2075 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2076 designators and can hence be used with TYPEP.
2077 * fixed bug #333: CHECK-TYPE now ensures that the type error
2078 signalled, if any, has the right object to be accessed by
2079 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2080 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2081 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2083 * fixed bug #334: programmatic addition of slots using specialized
2084 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2085 and :ALLOCATION :CLASS slots.
2086 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2087 float exponents. (rereported by Peter Seibel)
2088 * fixed a bug: #\Space (and other whitespace characters) are no
2089 longer considered to be macro characters in standard syntax by
2090 GET-MACRO-CHARACTER.
2091 * fixed bug: initialization of condition class metaobjects no longer
2092 causes an instance of the condition to be created. (reported by
2094 * fixed bug: it is now possible to have slots such that
2095 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2096 description is neither :INSTANCE nor :CLASS.
2097 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2098 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2099 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2100 non-standard SLOT-DEFINITION-ALLOCATION values.
2101 * fixed bugs: various race conditions handling exiting threads.
2102 CL-PPCRE's thread torture test now passes.
2103 * fixed arguably-a-bug: GC time proportional to number of threads in
2104 system even when most of them are idle
2105 * optimization: improved performance of BIT and SBIT on bit-vectors.
2106 * .fasl file incompatibility: The fasl file version number has
2107 been incremented because of changes associated with package locks.
2109 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2110 * minor incompatible change: the system no longer provides
2111 optimization hints (conditions which are TYPEP
2112 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2113 mode; these hints are still emitted when higher SPEED optimization
2115 * new contrib module: a sampling profiler (profiling by statistical
2116 sampling, rather than by instrumenting functions) is available as
2117 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2118 * the behaviour of the standard function ED is now customizeable by
2119 third parties through a hook variable: see ED's documentation
2120 string for information on the protocol.
2121 * the compiler no longer emits efficiency notes for (FUNCALL X)
2122 when the type of X is uncertain under default optimization
2124 * fixed bug 276: mutating a binding of a specialized parameter to a
2125 method to something that is not TYPEP the specializer is now
2127 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2128 exhibit the required behaviour on the x86 platform. (thanks to
2129 Peter van Eynde, Eric Marsden and Bruno Haible)
2130 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2131 even for difficult arguments. (reported by Peter Graves)
2132 * fixed bug 141a: the backquote printer now descends quoted
2134 * fixed another bug in backquote printing: no more destructive
2135 modification of the form's list structure. (reported by Brian
2137 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2138 move between its address being taken and the call to
2139 interrupt_thread, fixing a crashing race condition.
2140 * the SB-POSIX contrib implementation has been adjusted so that it
2141 no longer exhibits ridiculously poor performance when constructing
2142 instances corresponding to C structs.
2144 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2145 * minor incompatible change: the sb-grovel contrib now treats C
2146 structures as alien (in the sense of SB-ALIEN) objects rather than
2147 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2148 has implications for memory management of client code
2149 (sb-grovel-returned objects must now be manually managed) and for
2150 type safety (alien objects now have full types).
2151 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2152 used to control emission of compiler diagnostics, rather than the
2153 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2154 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2155 quality should be considered deprecated.
2156 * install.sh now installs the user manual as well
2157 * (not quite a new documentable feature, but worth considering in
2158 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2159 beginnings of a semantically meaningful condition hierarchy is
2160 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2162 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2163 designator as the defaults argument.
2164 * fixed bug: Displaced arrays whose displaced-to array has become
2165 too small now cause ARRAY-DIMENSION to signal an error, providing
2166 for safer bounds-checking. (reported by Bruno Haible)
2167 * fixed bug: DEFCLASS slot definitions with identical :READER and
2168 :WRITER names now signal a reasonable error. (reported by Thomas
2170 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2172 * fixed bug: Invalid slot specification errors now print correctly.
2173 (thanks to Zach Beane)
2174 * fixed bug 320: Shared to local slot value transfers in class
2175 redefinitions now happen corrently from superclasses as
2176 well. (reported by Bruno Haible)
2177 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2179 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2180 improper lists correctly. (reported by Bruno Haible)
2181 * fixed bug 313: source-transform for <fun-name> was erroneously
2182 applied to a call of a value of a variable with name <fun-name>.
2183 (reported by Antonio Menezes Leitao)
2184 * fixed bug 307: The obsolete instance protocol ensures that
2185 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2186 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2187 remove dead unknown-values globs from the middle of the stack.
2188 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2189 easier to resume long computations after using *BREAK-ON-SIGNALS*
2190 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2191 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2192 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2193 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2194 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2195 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2197 * fixed a (fairly theoretical) bug in string printing: if
2198 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2199 does not have array-element-type equal to the most general string
2201 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2202 function name. (thanks to Nikodemus Siivola)
2203 * fixed bug: compile-time format string checker failed on
2204 non-closed ~{. (reported by Thomas F Burdick)
2205 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2206 don't warn on legal loop code involving more than one aggregate
2207 boolean. (thanks to Nikodemus Siivola)
2208 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2209 in keyword position to MAKE-PACKAGE was not being reported as
2211 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2212 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2214 * fixed bugs 280 and 312: the checking for multiple definitions in a
2215 file is less likely to become confused by uses of inline
2217 * fixed bug: the #S reader macro performs the keyword coercion
2218 specified for slot names. (reported by Kalle Niemitalo)
2219 * fixed bug: lambda lists may contain symbols whose names start with
2220 & but are not lambda-list-keywords; their occurrence triggers a
2222 * fixed bug 321: define-method-combination argument lambda lists do
2223 not cause invalid code to be generated when &optional or &aux
2224 variables have default values. (reported by Bruno Haible)
2225 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2226 have CLOS classes; this appears also to have fixed reported
2227 instability in stack exhaustion detection.
2228 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2229 MUFFLE-WARNING when no associated restart is present is now
2231 * optimization: rearranged the expansion of various defining macros
2232 so that each expands into only one top-level form in a
2233 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2235 * optimization: used a previously little-used slot in symbols to
2236 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2238 * fixed some bugs revealed by Paul Dietz' test suite:
2239 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2241 ** Fixed an optimization bug related to inheritance of initforms
2242 from local to shared slots.
2243 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2244 ** CAST optimizer forgot to flush argument derived type.
2245 ** print/read consistency on floats is now orders of magnitude
2246 more likely. (thanks also to Bruno Haible for a similar report
2248 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2249 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2250 for objects of type REAL. Make it so.
2251 ** SXHASH is specified (infelicitously) to respect similarity,
2252 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2253 it so. (thanks to Markus Ziegler)
2254 ** on the Alpha, the compiler succeeds in compiling functions
2255 returning a known number of arguments greater than 63.
2256 ** fixed handling of invalid NIL arguments in keyword position
2258 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2260 ** provided a readably-printable representation for RANDOM-STATE
2262 ** ensured that pathnames, where they have a namestring, always
2263 print using #P"..." syntax.
2265 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2266 * Support for the forthcoming 2.0 version of the NetBSD kernel
2267 running on x86 hardware has been added. (thanks to Perry
2268 E. Metzger most immediately, and others for their past work)
2269 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2270 current release version; SBCL's OpenBSD support had been broken
2271 since about the time of OpenBSD's switch to ELF binary format.)
2272 * [placeholder for DX summary]
2273 ** user code with &REST lists declared dynamic-extent, under high
2274 speed or space and low safety and debug optimization policy.
2275 * The manual has been converted to Texinfo format and the debugger
2276 chapter from the cmucl manual has been added.
2277 * A facility has been added to extract documentation strings from
2278 sbcl and store them as Texinfo-formatted snippets for inclusion in
2279 the manual (via Texinfo's @include directive)
2280 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2281 the test case to Dave Roberts)
2282 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2283 pointers. (reported by Sean Ross)
2284 * bug fix: PROFILE output is printed nicely even for large numerical
2285 values. (thanks to Zach Beane)
2286 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2287 greater than 32 handle EOF correctly.
2288 * bug fix: on X86 an immediate argument of the IMUL instruction is
2289 correctly printed in disassembly. (thanks to Lutz Euler)
2290 * bug fix: class slots in redefined classes preserve their old
2291 values. (thanks to Bruno Haible and Nikodemus Siivola)
2292 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2293 succeeds. (reported by Marco Baringer)
2294 * fixed some bugs revealed by Paul Dietz' test suite:
2295 ** READ-SEQUENCE now works on ECHO-STREAMs.
2296 ** RATIONALIZE works more according to its specification. (thanks
2299 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2300 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2301 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2302 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2303 should matter to you only if you rebind the printer control
2304 variables and then find you want different bindings in the
2305 debugger than in the ordinary execution of your program.)
2306 * The runtime build system has been tweaked to support building
2307 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2308 assembler and linker. (thanks to Nikodemus Siivola)
2309 * Unbound, undefined, undeclared variables now trigger full
2310 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2311 more often programmer error than deliberate exploitation of undefined
2313 * optimization: the hash algorithm for strings has changed to one
2314 that is less vulnerable to spurious collisions. (thanks to Juho
2316 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2317 needless bounds checking. (thanks to Juho Snellman)
2318 * optimization: implemented multiplication as a modular
2319 (UNSIGNED-BYTE 32) operation on the PPC backend.
2320 * fixed some bugs revealed by Paul Dietz' test suite:
2321 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2323 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2324 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2325 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2327 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2329 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2331 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2332 their output stream on EOF from read.
2333 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2334 have been read to end-of-file.
2335 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2337 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2338 description of determination of which consecutive characters
2340 ** Printing the "Space" character with escaping on now yields "#\\ ",
2341 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2342 ** Reading floating-point numbers with *READ-BASE* set to a number
2343 less than 10 works correctly.
2344 ** Reading floating-point numbers with *READ-BASE* set to a number
2345 more than 10 works correctly.
2346 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2347 the readtable currently in effect.
2349 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2350 * minor incompatible change: parsing of namestrings on a physical
2351 (Unix) host has changed; numbers after the final #\. in a
2352 namestring are no longer interpreted as a version field. This is
2353 intented to be largely invisible to the user, except that the
2354 meaning of the namestring "*.*.*" has changed: it now refers to a
2355 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2356 should usually be replaced by
2357 (make-pathname :name :wild :type :wild :version :wild)
2358 with the added benefit that this is more likely to be portable.
2359 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2360 option to OPEN now signals an error if the file being opened
2361 exists; this may have an impact on existing code.
2362 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2363 signals received, on the PowerPC platforms (both Linux and
2364 Darwin). (thanks to Pierre Mai for pointing out the location of
2366 * several fixes on OS X: The system now builds and runs cleanly on
2367 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2368 (thanks to Brian Mastenbrook)
2369 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2370 second-resolution: integer multiples of 1/3600 between -24 and 24.
2371 (thanks to Vincent Arkesteijn)
2372 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2373 type when called with 1 argument; PEEK-CHAR checked type of
2374 PEEK-TYPE only after having read first character from a
2375 stream. (reported by Peter Graves)
2376 * bug fix: the garbage collector now has much better locality
2377 behaviour, and in particular no longer treats all memory as being
2378 exhausted when many small objects point to each other in a deeply
2380 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2381 recognized as being TYPEP their class.
2382 * bug fix: the PUSHNEW documentation string has been corrected.
2383 (thanks to Vincent Arkesteijn)
2384 * bug fix: defaulting of the value for the last of an atypically
2385 large number of multiple values being bound was not being
2386 performed correctly on the Alpha or PPC platforms
2387 * optimization: implemented multiplication as a modular
2388 (UNSIGNED-BYTE 32) operation on the x86 backend.
2389 * optimization: SEARCH on simple-base-strings can now be open-coded.
2390 (see also contrib/compiler-extras.lisp for inspiration for
2391 teaching the compiler about the Boyer-Moore algorithm).
2392 * value, returned by MAX (and MIN) called with several EQUALP, but
2393 not EQL, arguments now does not depend on compiler settings.
2394 * fixed some bugs revealed by Paul Dietz' test suite:
2395 ** in stack analysis liveness information is propagated from
2396 non-local entry points.
2397 ** pathwise CAST removing failed when the CAST node did not start
2399 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2400 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2402 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2403 host is already defined.
2404 ** RENAME-FILE works on streams instead of signalling an internal
2406 ** PEEK-CHAR uses the current readtable when determining whether
2407 or not a character is whitespace.
2408 ** MERGE-PATHNAMES handles the case when the pathname does not
2409 specify a name while the default-pathname specifies a version.
2410 ** Pathnames now stand a better chance of respecting print/read
2412 ** Attempting to use standardized file system operators with a
2413 pathname with invalid :DIRECTORY components signals a
2415 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2417 ** DIRECTORY on logical pathnames is more correct.
2418 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2419 signal a TYPE-ERROR if their argument is not a stream.
2420 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2421 designator argument does not designate a stream.
2422 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2423 examining the synonym.
2424 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2426 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2427 element-type larger than ([UN]SIGNED-BYTE 32).
2429 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2430 * When built with the :SB-FUTEX feature, threaded builds now take
2431 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2432 for faster/more reliable mutex and condition variable support.
2433 * Incompatible change (but one you probably shouldn't have been using
2434 anyway): the interface and code for arbitrating between multiple
2435 threads in the same user session has been redesigned.
2436 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2437 error when called without an explicit environment argument.
2438 (thanks to Nikodemus Siivola)
2439 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2440 confused about their position. (thanks to Adam Warner and Gerd
2442 * bug fix: Pretty printing backquoted forms with unquotations in the
2443 argument list position of various code constructs such as LAMBDA
2444 now works correctly. (reported by Paul Dietz)
2445 * bug fix: Pretty printing unquotations no longer loses all
2446 stream position information.
2447 * optimization: performance of string output streams is now less
2448 poor for multiple small sequence writes.
2449 * optimization: performance of CSUBTYPEP in the presence of complex
2450 expressions involving CONS and NOT many times has been improved.
2451 (reported by Paul Dietz)
2452 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2454 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2455 streams for the REPL, for the debugger, and so on. (thanks to
2457 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2458 * fixed some bugs revealed by Paul Dietz' test suite:
2459 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2460 is ignored if it is too small, rather than propagating through
2461 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2462 ** extremely complex negations of CONS types were not being
2463 sufficiently canonicalized, leading to inconsistencies in
2465 ** VALUES tranformer lost derived type.
2467 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2468 * fixed a bootstrapping bug: the build process no longer assumes
2469 that the various BOOLE-related constants have the same value in
2470 host and target lisps. (noted by Paul Dietz' test suite on an
2471 SBCL binary built from CLISP)
2472 * The system can now be dynamically linked on the MIPS platform,
2473 which enables dynamic loading of foreign code from Lisp. (thanks
2474 to Ralf Baechle for discussions on the MIPS ABI)
2475 * The system now records debugging information for its own source
2476 files in a filesystem-position-independent manner, relative to
2477 the "SYS" logical host.
2478 * fixed a compiler bug: MV-LET convertion did not check references
2479 to the "max args" entry point. (reported by Brian Downing)
2480 * tweaked disassembly notes to be less confident about proclaiming
2481 some instruction as an LRA. (thanks to Brian Downing)
2482 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2483 now each have their own history, command character, and other
2484 characteristics. (thanks to David Lichteblau)
2485 * fixed some bugs revealed by Paul Dietz' test suite:
2486 ** compiler failure in compiling LOGAND expressions including a
2488 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2489 shift greater than 32.
2490 ** FLUSH-DEST did not mark blocks for type check regeneration.
2491 ** HANDLER-CASE failed to accept declarations in handler clauses
2492 in some circumstances.
2494 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2495 * New code in contrib/sb-introspect (still probably not entirely
2496 stable yet) provides some support for smart Lisp development
2497 environments like SLIME.
2498 * The conditions signalled for errors occurring when loading .fasl
2499 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2500 in a way which should help ASDF recover gracefully.
2501 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2502 stuff has been cleaned up. If you code contrib/ stuff, this might
2503 affect you, and you can look at contrib/README, contrib/STANDARDS,
2504 and/or the 0.8.4.27 diff to check.
2505 * In full calls the compiler now does not generate checks for declared
2506 argument types for all arguments.
2507 * various threading fixes
2508 ** and some experimental patches which didn't make it into
2509 the main tree for this release, but which are shipped in
2510 contrib/experimental-thread.patch as a possible fix for some
2511 failures (deadlock, spinning...) in GC-intensive multithreaded
2513 * fixed PPC build problem (source code incompatibility of different
2514 library versions): added offsetof() hackery which attempts to divine
2515 where glibc maintainers put uc_mcontext today
2516 * fixed bug 282: compiler does not trust type assertions while passing
2517 arguments to a full call.
2518 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2519 &OPTIONAL VALUES type parameter.
2520 * fix bug 214: algorithm for noting rejected templates is now more
2521 similar to that of template seletion. (also reported by rydis on
2523 * fixed bug 141b: printing backquoted information readably and prettily
2524 inserts a space where necessary.
2525 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2526 (CHARACTER 10) are now reported as errors, rather than propagated
2527 as unknown types. (reported by piso on #lisp)
2528 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2529 with values NIL and :ERROR. (thanks to Milan Zamazal)
2530 * fixed bug 191c: CLOS now does proper keyword argument checking as
2531 described in CLHS 7.6.5 and 7.6.5.1.
2532 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2533 longer raise an error; further, using a list as a for-as-arithmetic
2534 counter now raises a meaningful error.
2535 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2536 now understood by sequence creation functions such as MAKE-SEQUENCE
2538 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2539 a STRING-INPUT-STREAM.
2540 * compiler enhancement: SIGNUM is now better able to derive the type
2542 * type declarations inside WITH-SLOTS are checked. (reported by
2544 * fixed some bugs revealed by Paul Dietz' test suite:
2545 ** incorrect optimization of TRUNCATE for a positive first
2546 argument and negative second.
2547 ** compiler failure in let-convertion during flushing dead code.
2548 ** compiler failure while deriving type of TRUNCATE on an
2549 interval, containing 0.
2550 ** ASH of a negative bignum by a negative bignum count now returns
2552 ** intersection of CONS types now canonicalizes properly, fixing
2553 inconsistencies in SUBTYPEP.
2555 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2556 * incompatible change: The --disable-debugger command line
2557 option now clobbers the debugger at a more fundamental
2558 level, by redefining #'INVOKE-DEBUGGER instead of by
2559 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2560 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2561 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2562 enter the debugger REPL and then suffer recursive errors
2563 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2564 while under the new system, BREAK in --disable-debugger mode
2565 terminates the system just as an unhandled error would.
2566 * fixed compiler performance when processing loops with a step >1;
2567 * bug fix: DOCUMENTATION now retrieves generic function
2568 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2569 support has been systematized, and now supports the methods
2570 specified by ANSI, along with a default method and a method for
2571 slot documentation. (reported by Nathan Froyd)
2572 * bug fix: effective methods associated with a generic function are
2573 no longer cached over a change of that generic function's method
2574 combination. (reported by Andreas Fuchs)
2575 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2576 names a non-existent directory. (thanks to Andreas Fuchs)
2577 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2578 the PPC platform, lead to stack corruption; this has been fixed.
2579 (reported by Rainer Joswig)
2580 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2581 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2582 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2583 arithmetic in general, is now much more reliable on the PPC
2585 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2586 platform now returns the right answer.
2587 * optimization: restored some effective method precomputation in
2588 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2589 precomputation is now tunable.
2590 * optimization: compiler-internal data structure use has been
2591 reviewed, and changes have been made that should improve the
2592 performance of the compiler by about 20%.
2593 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2594 simple-base-strings and simple-bit-vectors is improved.
2595 * optimization: the optimization of 32-bit logical and arithmetic
2596 functions introduced in version 0.8.3 on the x86 has been
2597 implemented on the mips, ppc and sparc platforms; an
2598 implementation of the same facility, but for 64-bit arithmetic,
2599 has been added for the alpha.
2600 * microoptimization: the compiler is better able to make use of the
2601 x86 LEA instruction for multiplication by constants.
2602 * bug fix: in some situations compiler did not report usage of
2603 generic arithmetic in (SPEED 3) policy.
2604 * bug 145b fix: compiler used wrong type specifier while converting
2605 MEMBER-types to numeric.
2606 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2608 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2609 index argument now works properly on the Alpha platform.
2610 * bug fix: floating point exception treatment on the Alpha platform
2612 * bug fix: FILE-POSITION works much better on string input and
2613 output streams. (thanks to Nikodemus Siivola)
2614 * bug fix: many threading/garbage collection symptoms sorted.
2615 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2616 might be pseudo-atomic.
2617 * internal change: Stopping for GC is now done with signals not
2618 ptrace. GC is now done in whichever thread wanted it, instead of
2620 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2622 * bug fix: result form in DO is not contained in the implicit
2624 * incompatible change: ICR structure is changed; the value part of
2625 CONTINUATION is now called LVAR; corresponding functions are
2626 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2627 * added type deriver for ISQRT (thanks to Robert E. Brown).
2628 * bug fix: better support for loading from the command line when an
2629 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2631 * fixed some bugs revealed by Paul Dietz' test suite:
2632 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2633 ** ROUND and FROUND now give the right answer when given very
2634 small float arguments.
2635 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2637 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2638 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2639 ** LCM should return a non-negative integer.
2640 ** PARSE-INTEGER returned the index of a terminator instead of the
2641 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2642 ** PARSE-INTEGER returned an incorrect index being applied to a
2644 ** LCM with two arguments of 0 returns 0 rather than signalling
2646 ** unsigned addition of a 32-bit constant with the high bit set no
2647 longer causes an internal compiler error on the x86.
2648 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2649 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2650 with negative last argument.
2651 ** byte specifiers with zero size and position no longer cause
2652 an error during type derivation.
2653 ** bignum multiplication on the Alpha platform now returns the
2655 * porting: The system now builds on SuSE AMD64, although it still
2656 generates a 32-bit binary.
2657 * .fasl file incompatibility: The fasl file version number has
2658 been incremented (because of the changes to internal compiler
2659 data structures referred to above).
2661 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2662 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2663 more accurately, on the Darwin kernel running on PowerPC hardware.
2664 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2665 * Compiler code deletion notes now signal a condition of type
2666 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2667 an associated MUFFLE-WARNING restart.
2668 * The compiler now performs limited argument count validation of
2669 constant format strings in FORMAT, and where appropriate in ERROR,
2670 CERROR and WARN. (thanks to Gerd Moellmann)
2671 * New ASDF-INSTALL contrib can be used for automatic download and
2672 installation of third-party Lisp code from CCLAN or other sites
2674 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2675 which forces another thread to execute a function supplied by the
2677 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2678 now accept and act upon their :ELEMENT-TYPE keyword argument.
2679 (reported by Martin Atzmueller, Edi Weitz)
2680 * bug fix: FILE-POSITION now accepts position designators up to
2681 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2682 is the greater. (thanks to Patrik Nordebo)
2683 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2684 Kalle Olavi Niemitalo)
2685 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2686 printable. (reported by Eric Marsden)
2687 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2688 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2689 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2690 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2691 (thanks to Stig Sandoe)
2692 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2693 UNREAD-CHAR now correctly includes the unread character in the
2694 target sequence. (thanks to Gerd Moellmann)
2695 * bug fix in threaded builds: the system can now be suspended and
2696 resumed by shell job control with minimal disruption.
2697 * bug fixes in times and timezones >2038AD
2698 * better handling of "where is GNU make?" problem in build scripts
2699 (thanks to Nikodemus Siivola)
2700 * new optimization: inside a named function any reference to a
2701 function with the same name is considered to be a self-reference;
2702 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2703 optimization quality.
2704 * new optimization on x86: logical functions and + now have
2705 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2706 used when the result is truncated to 32 bits.
2707 * VALUES declaration is partially enabled.
2708 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2709 * fixed some bugs revealed by Paul Dietz' test suite:
2710 ** The system now obeys the constraint imposed by
2711 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2712 types form a lattice under type intersection.
2713 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2714 ** ASSOC now ignores NIL elements in an alist.
2715 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2716 and (1+ MOST-POSITIVE-FIXNUM) answers.
2717 ** The addition of a method with invalid qualifiers to a generic
2718 function does not cause an error to be signalled immediately;
2719 a warning is signalled, and the error is generated only on
2720 calling the generic function.
2721 * changed .fasl file version number, in part to add type codes for
2722 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2723 obscure ANSI requirements
2725 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2726 * fixed bug 148: failure to inline-expand a local function left
2727 garbage, confusing the compiler.
2728 * fixed bugs 3cd: structure slot readers perform type check if the
2729 slot can have an invalid value (i.e. it is either not initialized
2730 or can be written with a less specific slot writer).
2731 * bug fix: the compiler now traps array references to elements off
2732 the end of an array; previously, the bounds checking in some
2733 circumstances could go off-by-one.
2734 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2736 * type declarations for array element types now obey the description
2737 on the CLHS page "Declaration TYPE", as per discussions on
2738 sbcl-help around 2003-05-08. This means that a declaration
2739 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2740 declaration, all references to BAR will be asserted or assumed
2741 (with THE, so dependent on compiler policy) to involve objects of
2742 type FOO. Note that no such declaration is implied in
2743 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2744 * declared types of functions from the "Conditions"
2745 chapter. (reported by Paul Dietz)
2746 * bug fix: CERROR accepts a function as its first argument.
2747 * bug fix: NTH an NTHCDR accept a bignum as index
2748 arguments. (reported by Adam Warner)
2749 * optimization: character compare routines now optimize comparing
2750 against a constant character. (reported by Gilbert Baumann)
2751 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2752 argument now works properly on the MIPS platform.
2753 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2754 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2755 declared to have a functional type, does not cause an error
2757 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2758 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2759 * fixed bug: if last continuation of a deleted block has a
2760 destination, this destination should be deleted too. (reported by
2762 * fixed a bug in the bootstrap process: the host compiler's values
2763 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2764 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2765 test case from Patrik Nordebo)
2766 * improved the ability of the disassembler on the PPC platform to
2767 provide helpful disassembly notes.
2768 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2769 the class in more cases than previously.
2770 * bug fix: FILE-POSITION now understands :START and :END for
2771 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2772 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2773 8) data. (thanks to David Lichteblau)
2774 * bug fix: it is possible to add a method to a generic function
2775 without lambda list.
2776 * bug fix: reader failed to signal END-OF-FILE inside an
2777 object representation. (reported by Nikodemus Siivola)
2778 * fixed some bugs revealed by Paul Dietz' test suite:
2779 ** LAST and [N]BUTLAST should accept a bignum.
2780 ** condition slot accessors are methods.
2781 ** (VECTOR NIL) is a subtype of STRING.
2783 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2784 * minor incompatible change: some nonsensical specialized lambda
2785 lists (used in DEFMETHOD) which were previously ignored now signal
2787 * minor incompatible change: the system is now aware of the types of
2788 variables in the COMMON-LISP package, and will signal errors for
2789 most violations of these type constraints (where previously they
2790 were silently accepted).
2791 * minor incompatible change: COMPILE-FILE now uses the freedom
2792 afforded (ANSI 3.2.2.3) to use derived function types for
2793 functions defined in the same file. This also permits the system
2794 to warn on static type mismatches and function
2795 redefinition. (Currently it does not work with high DEBUG level.)
2796 * minor incompatible change: VALUES declaration is disabled.
2797 * When issuing notes, the compiler now signals a condition of type
2798 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2799 restart for use in user handlers. It is expected that the
2800 COMPILER-NOTE condition will eventually become a condition
2801 supertype to a hierarchy of note types, which will then be
2802 handleable in a similar fashion. However, at the moment, no such
2803 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2804 but it's an implementation detail, not a classification for the
2806 * Changes in type checking closed the following bugs:
2807 ** type checking of unused values (192b, 194d, 203);
2808 ** template selection based on unsafe type assertions (192c, 236);
2809 ** type checking in branches (194bc).
2810 * A short form of VALUES type specifier has ANSI meaning (it has
2811 increased the number of situations when SBCL cannot perform type
2813 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2814 keywords or constants is permissible.
2815 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2816 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2817 classes. (thanks to Antonio Martinez)
2818 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2819 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2820 argument to a pretty-print stream. Instead, it leaves any such
2821 operations to DESCRIBE-OBJECT methods.
2822 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2824 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2825 lambda lists. (thanks to David Lichteblau)
2826 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2827 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2828 (thanks to Gerd Moellmann)
2829 * fixed bug 246: increased compilation speed of long
2830 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2832 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2834 * DEFCONSTANT now throws a condition of type
2835 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2836 constant to a non-EQL value; CONTINUE and ABORT restarts
2837 respectively change and preserve the value.
2838 * fixed bug 63: The code walker, part of the implementation of CLOS,
2839 is now better at handling symbol macros.
2840 * bug fix: There is no longer an internal implementation type named
2841 CL:LENGTH. (reported by Raymond Toy)
2842 * bug fix: In macro-like defining macros/special operators the
2843 implicit block does not enclose the lambda list.
2844 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2845 enclosed in parentheses) are not suitable as type specifiers, and
2846 their use properly signals an error now.
2847 * bug fix: An argument count mismatch for a type specifier in code
2848 being compiled no longer causes an unhandled error at compile
2849 time, but signals a compile-time warning.
2850 * fixed simple vector readable printing
2851 * bug fix: DESCRIBE takes more care over whether the class
2852 precedence list slot of a class is bound before accessing it.
2853 (reported by Markus Krummenacker)
2854 * bug fix: FORMATTER can successfully compile pretty-printer format
2855 strings which use variants of the ~* directive inside.
2856 * bug fix: SEARCH now applies its TEST predicate to the elements of
2857 the arguments in the correct order. (thanks to Wolfhard Buss)
2858 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2859 in type inference. (thanks to Robert E. Brown)
2860 * bug fix: Reading in symbols with an explicit package name of ""
2861 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2862 * fixed some bugs revealed by Paul Dietz' test suite:
2863 ** NIL is now allowed as a structure slot name.
2864 ** Arbitrary numbers, not just REALs, are allowed in certain
2865 circumstances in LOOP for-as-arithmetic clauses.
2866 ** Multiple class redefinitions before slot access no longer
2867 causes a type error.
2868 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2869 association between the name and a class.
2870 ** Generic functions with non-standard method-combination and over
2871 five methods all of which return constants no longer return NIL
2872 after the first few invocations. (thanks to Gerd Moellmann)
2873 ** CALL-NEXT-METHOD with no arguments now passes the original
2874 values of the arguments, even in the presence of assignment.
2875 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2877 ** DISASSEMBLE works with closures and funcallable instances.
2878 ** ADD-METHOD now returns the generic function, not the new method.
2879 ** FIND-METHOD signals an error if the lengths of the specializers
2880 is incompatible with the generic function, even if the ERRORP
2882 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2883 which its argument is a member.
2884 ** DEFCLASS only redefines the class named by its class-name
2885 argument if that name is the proper name of the class;
2886 otherwise, it creates a new class.
2887 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2888 of the UNBOUND-SLOT condition to the name of the slot.
2889 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2890 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2891 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2892 SLOT-MAKUNBOUND in the specified fashion.
2894 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2895 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2896 cross-compilation host. As a consequence, we can now bootstrap our
2897 way up to SBCL starting with a bare gcc toolchain and human-readable
2898 source code (first the source to CLISP, then the source to SBCL).
2899 * A contributed module containing a partial implementation of the
2900 simple-streams interface has been included. (thanks to Rudi
2902 * A contributed module implementing the RFC1321 Message Digest
2903 Algorithm, known as MD5, has been included.
2904 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2905 no longer has any effect, as the code controlled by this feature
2906 has been deleted. (As far as we know, no-one has ever built using
2907 this feature, and its semantics were confused in any case).
2908 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2909 work on conditions (as required by the ANSI specification),
2910 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2911 expected behaviour on conditions. Users should note, however,
2912 that such behaviour is not required by the ANSI specification,
2913 and so use of this behaviour may render their code unportable.
2914 * fixed some bugs revealed by Paul Dietz' test suite:
2915 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2917 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2918 ** SXHASH on condition objects no longer returns NIL.
2919 ** :ALLOCATION :CLASS slots are better treated; their values are
2920 updated on class redefinition, and initforms inherited from
2921 superclasses are applied.
2922 ** REMOVE-METHOD returns its generic function argument even when
2923 no method was removed.
2924 ** SHARED-INITIALIZE now initializes the values of the requested
2925 slots, including those with :ALLOCATION :CLASS.
2926 ** ALLOCATE-INSTANCE now works on structure classes defined via
2927 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2929 ** SLOT-EXISTS-P now works on conditions, as well as structures
2931 ** MAKE-LOAD-FORM now has the required methods on
2932 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2933 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2934 keyword, and now implements the SLOT-NAMES argument.
2935 ** methods with &OPTIONAL arguments no longer allow too many
2936 arguments to be passed in the call without error.
2937 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2938 option is consistent with the required arguments of the generic
2939 function lambda list.
2940 * bug fix: REQUIRE accepts a string designator. (Thanks to
2942 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2943 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2944 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2945 * bug fix: The compiler now checks for duplicated variables in macro
2947 * bug fix: SETQ on globals returns the correct value.
2948 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2949 not a COMPILER-ERROR (followed by some other strange error on
2950 choosing the CONTINUE restart).
2951 * bug fix: make.sh and friends are now more consistent in the way that
2952 they look for GNU "make".
2954 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2955 * experimental native threads support (on x86 Linux >=2.4 only).
2956 This is not compiled in by default: you need to add :SB-THREAD to
2957 the target features. See the "Beyond ANSI" chapter of the manual
2959 * fix for longstanding nonANSIism: The old distinction between
2960 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2961 The return value from CL:FIND-CLASS is now a CLOS class, and
2962 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2963 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2964 * An interface to the MetaObject Protocol, as described in Kiczales,
2965 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2966 MIT Press, 1991, is available from the SB-MOP package.
2967 * incompatible change: the SB-PCL package should now be considered
2968 a private implementation detail, and no longer a semi-private MOP
2970 * minor incompatible change: due to rearrangement for threads, the
2971 control stack and binding stack are now allocated at arbitrary
2972 addresses instead of being hardcoded per-port. Users affected by
2973 this probably have to be doing advanced things with shared
2974 libraries, and will know who they are.
2975 * minor incompatible change: Previously, all --eval forms used were
2976 processed with READ before any of them were processed with EVAL.
2977 Now each --eval form is processed with both READ and EVAL before
2978 the next --eval form is processed. (Thus package operations like
2979 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2980 work as the user might reasonably expect.)
2981 * minor incompatible change: *STANDARD-INPUT* is now only an
2982 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2984 * minor incompatible change: Y-OR-N-P is now character-oriented,
2985 not line oriented. Also, YES-OR-NO-P now works without errors.
2986 (thanks to Antonio Martinez)
2987 * sb-aclrepl module improvements: an integrated inspector, added
2988 repl features, and a bug fix to :trace command.
2989 * Known functions, which cannot be open coded by the backend, are
2990 considered to be able to check types of their arguments. (fixing
2991 a bug report by Nathan J. Froyd)
2992 * fixed a bug in computing method discriminating functions: It is
2993 now possible to define methods specialized on classes which have
2994 forward-referenced superclasses. (thanks to Gerd Moellmann)
2995 * fixed evaluation order in optional entries (reported by Gilbert
2997 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2998 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2999 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3000 required-by-AMOP NAME argument, as well as CLASS and
3001 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3002 * fixed bug 20: DEFMETHOD can define methods using names that are
3003 not the proper names of classes to designate class specializers.
3004 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3005 Unix file descriptors, instead of blowing up. (thanks to Antonio
3007 * Garbage collection refactoring: user-visible change is that a
3008 call to the GC function during WITHOUT-GCING will not do garbage
3009 collection until the end of the WITHOUT-GCING. If you were doing
3010 this you were probably losing anyway.
3011 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3012 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3013 TYPEP the latter but not the former.
3014 * The compiler issues a full WARNING for calls to undefined functions
3015 with names from the CL package.
3016 * MAP-INTO for a vector destination is open coded. (reported by
3017 Brian Downing on c.l.l)
3018 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3019 documentation string.
3020 * fixed some bugs revealed by Paul Dietz' test suite:
3021 ** COPY-ALIST now signals an error if its argument is a dotted
3023 ** Condition slots are now accessed more correctly in the presence
3024 of multiple initargs for a given slot.
3025 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3026 exclude restarts of the same name associated with a different
3028 ** DEFCLASS of forward-referenced classes with another
3029 forward-referenced class in the superclasses list no longer
3031 ** Condition slots are now initialized once each, not multiple
3032 times. (thanks to Gerd Moellmann)
3033 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3035 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3036 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3037 arguments contain duplicated elements.
3038 ** RESTART-CASE understands local macros.
3039 ** RESTART-CASE associates exactly its own restarts with a condition.
3040 ** ENDP in safe mode checks its argument to be of type LIST.
3041 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3042 in question is unbound.
3043 ** Optimized MAKE-INSTANCE functions no longer cause internal
3044 assertion failures in the presence of duplicate initargs.
3045 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3046 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3047 propagated even in the ctor optimized implementation of
3049 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3051 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3052 * a better implementation of SXHASH on (simple) bit vectors,
3053 measured both in execution speed and in distribution of results
3054 over the positive fixnums, has been installed. Likewise, a better
3055 implementation of EQUAL for simple bit vectors is now available.
3056 * fixed CEILING optimization for a divisor of form 2^k.
3057 * fixed bug 240 (emitting extra style warnings "using the lexical
3058 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3059 by Antonio Martinez)
3060 * fixed SXHASH, giving different results for NIL depending on type
3061 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3062 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3063 variable. (found by Rolf Wester)
3064 * SBCL does not ignore type declarations for special
3065 variables. (reported by rif on c.l.l 2003-03-05)
3066 * some bug fixes in contrib/sb-aclrepl/
3067 * fixed some bugs revealed by Paul Dietz' test suite:
3068 ** a bug in the CONS type specifier, whereby the CAR and CDR
3069 types got intertwined, has been fixed;
3070 ** the type system is now able to reason about the interaction
3071 between INTEGER and RATIO types more completely;
3072 ** APPEND, [N]REVERSE and NRECONC check that those their
3073 arguments, which must be proper lists, are really so;
3074 ** An array specialized to be unable to hold elements has been
3075 implemented, as required -- yes, really -- by ANSI;
3076 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3077 on malformed property lists;
3079 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3080 * incompatible packaging change: in line with Unix convention,
3081 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3082 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3083 * REQUIRE and PROVIDE are now optionally capable of doing something
3084 useful. See the documentation string for REQUIRE.
3085 * infrastructure for a managed SBCL contrib system: contributed
3086 modules in this release include:
3087 ** the ASDF system definition facility;
3088 ** an interface to the BSD Sockets API;
3089 ** an ACL-like convenience interface to the repl;
3090 (thanks to Kevin Rosenberg)
3091 ** an implementation of ROTATE-BYTE, with efficient implementation
3093 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3094 gives a closer approximation to the right answer than previously.
3095 (thanks to Raymond Toy)
3096 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3097 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3098 optional environment arguments, as required by ANSI.
3099 * fixed bugs in other functions taking environment objects, allowing
3100 calls with an explicit NIL environment argument to be compiled
3102 * fixed bug 228: primary return values from
3103 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3104 COMPILE or FUNCTION.
3105 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3106 :TYPE will no longer signal errors on innocuous objects.
3107 * fixed bug 231b: SETQ is better at respecting type declarations in
3108 the lexical environment.
3109 * fixed a bug in DEFCLASS: classes named by symbols with no or
3110 unprintable packages can now be defined.
3111 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3112 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3113 * fixed compiler failure related to checking types of functions.
3114 (reported by Robert E. Brown)
3115 * the compiler is now much more consistent in its error-checking
3116 treatment of bounding index arguments to sequence functions: in
3117 (SAFETY 3) code, errors will be signalled in almost all cases if
3118 invalid sequence bounding indices are passed to functions defined
3119 by ANSI to operate on sequences.
3120 * fixed a bug in the build procedure: documentation of SBCL-specific
3121 packages is now preserved and available in the final Lisp image.
3122 * lifted FDEFINITION lookup out of loops in the implementation of
3123 many list operations. (thanks to Robert E. Brown)
3124 * fixed a bug in the reader: the #n# reader macro now works for
3125 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3126 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3127 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3128 E. Brown and Tony Martinez respectively)
3129 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3130 SLOT-BOUNDP: in particular, it is now able to optimize them much
3131 better, and is now not vulnerable to having packages renamed.
3132 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3133 always called when a slot is not present in an instance. (thanks
3135 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3136 longer signal an error on unreading a character following EOF on
3137 the previous constituent stream. (thanks to Tony Martinez)
3138 * fixed some bugs revealed by Paul Dietz' test suite:
3139 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3140 not just nonnegative fixnums;
3141 ** the logical bit-array operators such as BIT-AND now accept an
3142 explicit NIL for their "opt-arg" argument (to indicate a
3143 freshly-consed result bit-array);
3144 ** ELT now signals an error on an invalid sequence index in safe
3146 ** the type system is now cleverer about negations of numeric
3147 types, and consequently understands the BIGNUM and RATIO types
3149 ** the type system is now cleverer about the interaction between
3150 INTEGER and RATIO types: while bugs still remain, many more
3151 cases are accurately computed;
3152 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3153 if it is in the last clause;
3154 ** CONSTANTLY now correctly returns a side-effect-free function in
3156 ** DECLARE is no longer treated as a special-operator; in
3157 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3158 * incremented fasl file version number due to the change in the
3159 DEFSTRUCT-SLOT-DESCRIPTION structure.
3161 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3162 * minor incompatible change: code processed by the "interpreter" or
3163 EVAL now has a compilation optimization policy of (DEBUG 2)
3164 (changed from (DEBUG 1)) to improve debuggability of interactive
3165 development, and to allow the use of the debug RETURN command in
3167 * an experimental implementation of the RETURN command for the
3168 debugger has been included. (thanks to Frederik Kuivinen)
3169 * fixed bug 62: constraints were not propagated into a loop.
3170 * fixed bug in embedded calls of SORT (reported and investigated by
3172 * fixed some bugs revealed by Paul Dietz' test suite:
3173 ** printing and reading of arrays with some dimensions having
3174 length 0 (thanks to Gerd Moellmann);
3175 ** BOA constructor with &AUX argument without a default value does
3176 not cause a type error;
3177 ** CONSTANTP now returns true for all self-evaluating objects.
3179 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3180 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3181 accessors that are related by inheritance, as specified in the
3182 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3183 Valtteri Vuorikoski)
3184 * The compiler is now able to inline functions that were defined in
3185 a complex lexical environment (e.g. inside a MACROLET).
3186 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3188 * Support for the upcoming FreeBSD-5.0 release has been included.
3189 (thanks to Dag-Erling Smorgrav)
3190 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3191 effect when it is not in a toplevel context.
3192 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3193 stand a better chance of being correct. (thanks to Gerd
3195 * fixed bug in COERCE, which now signals an error on coercing a
3196 rational to a bounded real type which excludes the expected
3198 * The compiler is now able to derive types more accurately from the
3199 COERCE and COMPILE functions.
3200 * fixed bug 223: functional binding is considered to be constant
3201 only for symbols in the CL package.
3202 * fixed bug 231: SETQ did not check the type of a variable being set
3203 (reported by Robert E. Brown)
3204 * A new optimization for MAKE-INSTANCE has been included, fixing
3205 various bugs (including relating to :ALLOCATION :CLASS slots and
3206 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3208 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3209 ** As required by ANSI, LOOP now disallows anonymous collection
3210 clauses such as COLLECT I in conjunction with aggregate boolean
3211 clauses such as THEREIS (= I 1);
3212 ** LOOP now signals an error when any variable is reused in the
3213 same loop (including the potentially useful construct analogous
3214 to WITH A = 1 WITH A = (1+ A);
3215 ** IT is only a special loop symbol within the first clause of a
3216 conditional loop clause;
3217 ** LOOP with a typed iteration variable over a hashtable now
3218 signals a type error iff it should.
3219 * fixed some other bugs revealed by Paul Dietz' test suite:
3220 ** FILE-STREAM now names the class previously known as FD-STREAM;
3221 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3222 argument) no longer signals an error;
3223 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3224 of the slot symbol, rather than using the current package
3225 ((:CONC-NAME "") continues to intern the slot's name in the
3227 * incremented fasl file version number, because of the incompatible
3228 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3229 of the new implementation of DEFINE-COMPILER-MACRO.
3231 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3232 * Support for building SBCL for MIPS platforms running in
3233 little-endian mode has now been checked in, and basic
3234 functionality on said platforms verified.
3235 * minor incompatible change: PCL now records the pathname of a file
3236 in which methods and the like are defined, rather than its
3238 * minor incompatible change: TRUENAME now considers the truename of
3239 a file naming a directory to be the pathname with :DIRECTORY
3240 component indicating that directory.
3241 * minor incompatible change: a NAMED clause in the extended form of
3242 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3243 reason for the previous behaviour is unclear.
3244 * more systematization and improvement of CLOS and MOP conformance
3245 in PCL (thanks to Gerd Moellman and Pierre Mai):
3246 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3248 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3249 primary methods with no specializers;
3250 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3252 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3253 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3254 FORWARD-REFERENCED-CLASSes; error reporting on
3255 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3257 ** SXHASH on CLOS instances now uses a slot internal to the
3258 instance to return different numbers on distinct instances,
3259 while preserving the same return value through invocations of
3261 ** DEFMETHOD signals errors when methods with longer incongruent
3262 lambda lists are added to generic functions;
3263 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3264 CLASS, as specified in AMOP;
3265 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3266 on the order returned by the primary method for classes of
3267 class STANDARD-CLASS;
3268 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3269 * fixed some bugs shown by Paul Dietz' test suite:
3270 ** DOLIST puts its body in TAGBODY;
3271 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3273 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3274 value producing form;
3275 ** if more variables are given to PROGV than values, extra
3276 variables are bound and made to have no value;
3277 ** NSUBSTITUTE on list arguments gets the right answer with
3279 ** ELT signals an error of type TYPE-ERROR when the index argument
3280 is not a valid sequence index;
3281 ** LOOP signals (at macroexpansion time) an error of type
3282 PROGRAM-ERROR when duplicate variable names are found;
3283 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3284 ** FUNCALL of special-operators now cause an error of type
3286 ** PSETQ now works as required in the presence of side-effecting
3287 symbol-macro places;
3288 ** NCONC accepts any object as its last argument;
3289 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3291 ** loop-for-as-package does not require a package to be explicitely
3293 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3295 * fixed bug 166: compiler preserves "there is a way to go"
3296 invariant when deleting code.
3297 * fixed bug 172: macro lambda lists with required arguments after
3298 &REST arguments now cause an error to be signalled. (thanks to
3300 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3301 bug. (thanks to Matthew Danish)
3302 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3304 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3305 arguments are lexically rebound. (thanks to Gerd Moellmann and
3307 * fixed bug 194: error messages are now more informative when there
3308 is no primary method applicable in a call to a generic
3309 function. (thanks to Gerd Moellmann)
3310 * fixed bug in command line argument checking (thanks to Julian
3312 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3313 (thanks to Matthew Danish)
3314 * incremented fasl file version number, because of the
3315 SXHASH-related changes in the layout of CLOS data structures
3317 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3318 * minor incompatible change: The runtime (the Unix executable named
3319 "sbcl") is now much pickier about the .core files it will load.
3320 Essentially it now requires .core files to descend from the same
3321 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3322 as the runtime does. (The intent is to prevent the crashes which
3323 can occur, and which can even be reported as mysterious failures,
3324 when people patch the sources or change the build parameters
3325 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3326 sbcl and .core files.)
3327 * fixed bug: VALUES-LIST is no longer optimized away.
3328 * fixed bug 142: The FFI conversion of C string values to Lisp
3329 string values no longer conses excessively. (thanks to Nathan
3330 Froyd porting Raymond Toy's fix to CMU CL)
3331 * began to systematize and improve MOP conformance in PCL (thanks to
3332 Nathan Froyd, Gerd Moellman and Pierre Mai):
3333 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3335 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3336 ** FINALIZE-INHERITANCE is now called on class finalization;
3337 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3338 argument precedence order.
3339 * fixed bug 202: The compiler no longer fails on functions whose
3340 derived types contradict their declared type.
3341 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3342 so it can be non-toplevel.
3343 * The fasl file version number has changed (because of the new
3344 implementation of DEFMACRO).
3345 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3346 safe code, that any length requirement by their type-specifier
3347 argument is valid. The exceptions to this are described in bug
3349 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3350 functions now check the types of their inputs as required by ANSI.
3351 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3352 introduced symbol is DECLAREd to be SPECIAL.
3353 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3354 * fixed bug: :COUNT argument to sequence functions may be negative
3355 * fixed bug: body of DO-SYMBOLS may contain declarations
3356 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3357 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3358 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3359 symbol macro only once
3360 * fixed printing of call frame when argument list is unavailable
3361 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3362 * compiler no longer signals WARNING on unknown keyword
3365 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3366 * A beta-quality port to the mips architecture running Linux,
3367 based on the old CMUCL backend, has been made. It has been tested
3368 on a big-endian kernel, and works sufficiently well to be able to
3369 rebuild itself; it has not been tested in little-endian mode.
3370 * fixed an inconsistency between gencgc.c and purify.c which made
3371 dumping/loading .core files unreliable
3372 * fixed bug 120a: The compiler now deals correctly with IFs where
3373 the consequent is the same as the alternative, instead of
3374 misderiving the return type. (thanks to Alexey Dejneka)
3375 * fixed bug 113: Logical pathnames are now dumpable (the logical
3376 host is resolved at load-time, throwing an error if it is not
3378 * fixed bug 174: FORMAT's error message is slightly clearer when a
3379 non-printing character is used in a format directive.
3380 * fixed several bugs in compiler checking of type declarations, i.e.
3381 violations of the Python "declarations are assertions" principle
3382 (thanks to Alexey Dejneka)
3383 * fixed several bugs in PCL's error checking (thanks to Gerd
3385 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3387 * fixed bug in compilation of functions as first class values
3388 (thanks to Antonio Martinez-Shotton)
3389 * The compiler's handling TYPE-ERRORs which it can prove will
3390 inevitably happen at runtime has been cleaned up and corrected
3391 in several ways. (thanks to Alexey Dejneka)
3392 * improved argument type checking for various basic arithmetic
3393 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3394 lavished on them in the past that they can be compiled in many
3395 ways in different special cases
3396 * fixed bug 181: compiler checks validity of user supplied type
3398 * cleaned up code flushing in optimization: Function calls which
3399 should signal errors for safety purposes (e.g. which ANSI says
3400 should signal errors when their arguments are of incorrect type)
3401 are no longer optimized away.
3402 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3403 * incremented fasl file version number, because changes in the
3404 implementation of sequence functions like COERCE caused
3405 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3406 in old inline expansions) to become undefined. (Actually these
3407 changes were later undone, so we might very well be binary
3408 compatible with 0.7.7 after all, but leaving the version number
3409 incremented seemed like the simplest and most conservative
3412 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3413 * An alpha-quality port to the parisc architecture running Linux,
3414 based on the old CMUCL backend, has been made. This, even more so
3415 than the other backends, should be considered still a work in
3416 progress; known problems include that the Linux kernel in 64-bit
3417 mode does not propagate the correct sigcontext structure to
3418 userspace, and consequently SBCL on a parisc64 kernel will not
3420 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3421 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3422 suppressed.) Also now that the compiler is looking at declarations
3423 in the environment, it checks optimization declarations as well,
3424 and suppresses inlining when (> DEBUG SPEED).
3425 * More fixes have been made to treatment of floating point exception
3426 treatment and other Unix signals. In particular, floating point
3427 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3428 * The detection and handling of control stack exhaustion (infinite
3429 or very deeply nested recursion) has changed. Stack exhaustion
3430 detection is now done by write-protecting pages at the OS level
3431 and applies at all optimization settings; when found, a
3432 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3433 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3434 be caught using IGNORE-ERRORS.
3435 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3436 correctly when transforming recursive LABELS functions to LETs.
3437 (thanks to Alexey Dejneka)
3438 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3439 that are names of constants or global variables.
3440 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3441 alien routines with docstrings.
3442 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3443 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3445 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3446 (thanks to Eric Marsden)
3447 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3448 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3449 * Bug fix: The SPARC backend can now compile functions involving
3450 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3451 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3452 to a functions expecting a PATHNAME-DESIGNATOR.
3453 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3454 lambda lists. (thanks to Alexey Dejneka)
3455 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3456 functions so that EXPORTing the name of the class doesn't cause
3457 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3458 to Antonio Martinez for reporting this)
3459 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3460 constant-folded without failing an assertion. (thanks to Einar
3461 Floystad Dorum for reporting this)
3462 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3463 types are now dealt with more correctly by the compiler.
3464 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3465 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3466 behaviour analogous to RENAME-FILE. This puts its behaviour more
3467 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3469 * The fasl file version number has changed again. (because of the
3470 bug fix involving the names of PCL MAKE-INSTANCE functions)
3472 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3473 * bug fix: Floating point exceptions are treated much more
3474 consistently on the x86/Linux and PPC/Linux platforms.
3475 * Array initialization with :INITIAL-ELEMENT is now much faster for
3476 cases when the compiler cannot open code the array creation, but
3477 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3478 array accesses have also seen a speed increase.
3479 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3480 specified. (This was at the root of some bad interactions between
3481 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3483 * bug fix: Internal error arguments for undefined functions are now
3484 computed correctly on the PPC/Linux platform.
3485 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3486 Raymond Toy's patch for CMU CL)
3487 * Support for the Solaris 9 operating environment has been included
3488 (thanks to Daniel Merritt)
3489 * A very ugly but hopefully complete draft of the missing FFI chapter
3490 of the manual has been created by reformatting the corresponding
3491 CMU CL manual chapter into (currently very ugly and incoherent)
3492 DocBook and bringing it up to date for SBCL behavior. Thus, the
3493 manual is now essentially complete, at least by my extreme
3494 once-and-only-once standards, whereby it's acceptable to refer to
3495 the doc strings of SB-EXT functions as the primary documentation.
3496 * The fasl file version number has changed again, due to cleanup of
3497 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3498 is no longer a static symbol.)
3500 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3501 * SBCL now builds with OpenMCL (version 0.12) as the
3502 cross-compilation host; also, more progress has been made toward
3503 bootstrapping under CLISP.
3504 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3506 * bug 158 fixed: The compiler can now deal with integer loop
3507 increments different from 1; fixing this turned out also to fix
3509 * bug 169 fixed: no more bogus warnings about using lexical bindings
3510 despite the presence of perfectly good SPECIAL declarations (thanks
3511 to David Lichteblau)
3512 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3513 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3514 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3516 * bug 180 fixed: Method combination specifications no longer ignore
3517 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3518 * bug fix: Structure type predicate functions now check their argument
3519 count as they should.
3520 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3521 correctly. (thanks to Pierre Mai)
3522 * minor incompatible change: The --noprogrammer option is deprecated
3523 in favor of the new --disable-debugger option, which is very similar.
3524 (The major difference is that it takes effect at a slightly different
3525 time at startup, causing handling of errors in --sysinit and
3526 --userinit files will be affected differently.) The
3527 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3528 been added to allow this functionality to be controlled from ordinary
3529 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3530 maintainers, who might want to run non-interactive scripts to
3531 build SBCL cores which will later be used interactively.)
3532 * minor incompatible change: The LOAD function no longer, when given
3533 a wild pathname to load, loads all files matching that pathname.
3534 Instead, an error of type FILE-ERROR is signalled.
3536 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3537 * bug 147 fixed: The compiler preserves its block link/count
3538 invariants more correctly now so that it doesn't crash. (thanks
3540 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3542 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3543 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3544 no-value-for-this-character as NIL (as specified by ANSI).
3545 * HOST-NAMESTRING on physical pathnames now returns a string that is
3546 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3547 (thanks to Christophe Rhodes)
3548 * The Alpha port handles icache flushing more correctly. (thanks to
3550 * More progress has been made toward bootstrapping under CLISP. (thanks
3551 to Christophe Rhodes)
3552 * The fasl file format has changed again, because dynamic loading
3553 on OpenBSD (which has non-ELF object files) motivated some cleanups
3554 in the way that foreign symbols are transformed and passed around.
3555 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3556 is no longer treated as whitespace by the reader, but instead as
3557 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3558 |A
\7fB|, instead of A as it used to.
3560 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3561 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3562 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3563 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3564 this as of 0.7.1.45, but was left out of the previous news section
3565 (thanks to Dan Barlow)
3566 * SBCL now runs on the Solaris operating system on SPARC architectures
3567 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3568 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3569 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3570 * SPARC backend cleanups, allowing builds of cores optimized for V8
3571 and V9 SPARCS, and also emission of code targeted to a particular
3572 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3574 * SBCL is closer to bootstrapping under CLISP, thanks to various
3575 fixes by Christophe Rhodes.
3576 * The fasl file format has changed again, to allow the compiler's
3577 INFO database to support symbol macros.
3578 * The user manual (in doc/) is formatted into HTML more nicely.
3579 (thanks to coreythomas)
3580 * The system is smarter about SUBTYPEP relationships, especially
3581 those involving NOT types (including types such as ATOM which are
3582 represented internally using NOT types). Thus SUBTYPEP is less
3583 likely to return (VALUES NIL NIL) in general, and in particular
3584 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3586 * The fasl file format has changed again, because the internal
3587 representation of types now includes a new slot to support the new
3588 SUBTYPEP-of-NOT-types logic.
3589 * (not a change in the main branch of SBCL, but a related prototype
3590 which can hopefully be merged into the main branch of SBCL in the
3591 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3592 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3593 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3594 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3595 dispatch macro character combinations. (thanks to Alexey Dejneka)
3596 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3597 that superficially look like logical namestrings correctly.
3598 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3600 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3601 default supplied-p) for &optional and &key arguments. (thanks to
3604 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3605 * incompatible change: The compiler is now less aggressive about
3606 tail call optimization, doing it only when (> SPACE DEBUG) or
3607 (> SPEED DEBUG). (This is an incompatible change because there are
3608 programs which relied on the old CMU-CL-style behavior to optimize
3609 away their unbounded recursion which will now die of stack overflow.)
3610 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3611 for non-GENCGC systems has been increased to 20M (since that
3612 seems much closer to the likely performance optimum for modern
3613 systems than the old 4M value was)
3614 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3615 and *DEBUG-PRINT-LEVEL*
3616 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3617 of CMU CL's support for SPARC, and various endianness and other
3618 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3619 * new syntactic sugar for the Unix command line: --load foo.bar is now
3620 an alternate notation for --eval '(load "foo.bar")'.
3622 ** The system now detects stack overflow and handles it gracefully,
3623 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3624 optimization settings. (This is a good thing in general, and
3625 its introduction in this version should be particularly timely
3626 for anyone whose code fails because of suppression of tail
3628 ** The system now hunts for the C variable "environ" in a more
3629 devious way, to avoid segfaults when the C library version differs
3630 between compile time and run time. (thanks to Christophe Rhodes)
3631 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3632 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3633 ** The compiler no longer issues bogus style warnings for undefined
3634 classes in the same source file as the DEFCLASSes which defined
3635 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3637 ** fixes in CONDITION class precedence list for undefined function
3638 errors (thanks to Alexei Dejneka)
3639 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3640 correctly. (thanks to Dan Barlow)
3641 ** portability fixes aiming at bootstrapping under CLISP (thanks
3642 to Dave McDonald and Christophe Rhodes)
3643 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3644 ** fixes in type translation and and type inference (thanks to
3646 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3647 ** various fixes in the new ports (thanks to Dan Barlow)
3648 * several changes related to debugging:
3649 ** suppression of tail recursion, as noted above
3650 ** stack overflow detection, as noted above
3651 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3652 is now the default. (For some time encapsulation has been more
3653 reliable than the breakpoint-based :ENCAPSULATE NIL
3654 implementation, at least on X86 systems; and I just noticed that
3655 encapsulation also seems closer to the spirit of the ANSI
3658 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3660 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3661 up properly again. (There was a packaging bug in 0.7.0 which
3662 left their definitions in SB-SYS::LOAD-FOREIGN and
3663 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3664 vital for most things which interface to C-level interfaces,
3665 like extensions working with sockets or databases or
3666 Perl-compatible regexes or whatever, and the need to fix
3667 this bug is the main reason that 0.7.1 was released so
3669 ** DEFGENERIC is now choosier about the methods it redefines, so that
3670 reLOADing a previously-LOADed file containing DEFGENERICs does
3671 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3672 cycle now works as it should. (thanks to Alexey Dejneka)
3673 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3674 fixed by Christophe Rhodes. (He actually submitted this patch
3675 months ago, and I delayed until after 0.7.0.)
3676 ** Bug 111 (internal compiler confusion about runtime checks on
3677 FUNCTION types) was fixed by Alexey Dejneka.
3678 * Some internal cleanups (getting rid of variables which aren't
3679 needed now that the byte interpreter is gone) caused the fasl
3680 file format number to change again.
3682 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3683 * major incompatible change: The default fasl file extension, i.e. the
3684 default extension for files produced by COMPILE-FILE, has changed
3685 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3687 ** There are many changes in the implementation of the compiler.
3688 SBCL is now essentially a compiler-only implementation of ANSI
3689 Common Lisp. EVAL still "interprets" a few special cases, but
3690 almost all the interesting cases are handled by creating
3691 a LAMBDA expression, calling COMPILE on it, then calling
3692 FUNCALL on the result.
3693 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3694 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3695 Since the code is newer, there might still be some new bugs
3696 (though not as many as before Martin Atzmueller's fixes:-). But
3697 the new code is substantially simpler and clearer, and hopefully
3698 any remaining bugs will be simpler, less fundamental, and more
3699 fixable then the bugs in the old code.
3700 ** The revised compiler is still a little unsteady on its feet.
3702 *** The debugging information it produces (particularly the names
3703 of FUNCTION objects) is sometimes much less useful than what
3704 the old compiler produced.
3705 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3706 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3707 within a MACROLET) has been temporarily weakened.
3708 ** There are new compiler optimizations for various functions:
3709 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3710 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3711 *** the math functions TRUNCATE, FLOOR, and CEILING
3712 *** the function-of-all-trades COERCE
3713 Mostly these should be transparent, but there's one
3714 potentially-annoying problem (bug 117): when the compiler
3715 inline-expands a function and does type analysis on the result,
3716 it can create control paths which have type mismatches, and
3717 when it can't prove that those control paths aren't taken,
3718 it will issue WARNINGs about the type mismatches. This is
3719 a particular problem in practice for the new sequence functions.
3720 It's not clear how this should be fixed, and for now, a
3721 workaround is given in the entry for 117 in the BUGS file.
3722 ** (Because of the interaction between the two previous items --
3723 occasional inlining problems and new inline expansions -- some
3724 of the new sequence function optimizations won't really kick in
3725 completely until debugging information, and then inlining, are
3726 straightened out in some future version.)
3727 * minor incompatible changes:
3728 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3729 DIRECTORY no longer implicitly promotes NIL slots of its
3730 pathname argument to :WILD. In particular, when you ask for the
3731 contents of a directory (which you used to be able to do without
3732 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3733 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3734 ** changes in behavior that ANSI explicitly defines to be
3735 implementation dependent:
3736 *** The new compiler-only implementation still conforms with ANSI,
3737 but acts a little different than before. Besides the obvious
3738 changes in performance tradeoffs (that the cost per form passed
3739 to EVAL has gone up, and the cost per form executed by EVAL
3740 has gone down), the behavior of the system changes a little
3741 because there are no longer any interpreted function objects.
3742 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3743 e.g. doing COMPILE on the output of interactive DEFUN is
3745 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3747 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3748 has changed from (:CL) to NIL.
3749 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3750 e.g. #\Newline and #\Tab, don't have names specified in the
3751 ANSI Common Lisp standard, is now based on their ASCII symbolic
3752 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3753 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3754 are no longer used for output.
3755 ** changes in internal implementation constants:
3756 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3757 4 million. (If your application spends a lot of time GCing and
3758 you have a lot of RAM, you might want to experiment with
3759 increasing it even more.)
3760 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3761 However, almost all old code should still continue to work without
3762 immediate update, as SB-C-CALL is now a (deprecated) nickname
3764 ** Old operator names in the style DEF-FOO are now deprecated in
3765 favor of new corresponding names DEFINE-FOO, for consistency with
3766 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3767 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3768 internal symbols, but a few supported extensions like
3769 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3770 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3771 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3772 etc. as you get deeper into recursive calls to the debugger
3773 command loop, instead of the old "5]", "5]]", "5]]]"
3774 sequence. (I was motivated to do this when squabbles between
3775 ILISP and SBCL left me very deeply nested in the debugger. In the
3776 short term, this change will probably provoke more ILISP/SBCL
3777 squabbles, but hopefully it will be an improvement in the long run.)
3778 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3779 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3780 function, since declaiming return types involving aliens is
3781 (1) annoyingly messy to do by hand and (2) vital to efficient
3782 compilation of code which calls such functions.
3783 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3784 longer reexported by the SB-EXT package. They're solely useful
3785 for alien code, so it seems more logical that you should get
3786 them from the SB-ALIEN package, not in SB-EXT.
3787 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3788 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3789 features. Instead, the code that they used to control is always
3790 built into the system.
3791 * many other bug fixes
3792 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3793 advantage of the new EVAL-WHEN stuff and to clean them up in
3794 general, and they are now more ANSI-compliant in a number of
3795 ways. Martin Atzmueller is responsible for a lot of this.
3796 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3798 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3799 *** correct ERROR type for various file operations
3800 *** some fixes for Lisp streams
3801 *** DEFMETHOD syntax checking
3802 *** changing old weird representation of debug information as
3803 strings (which, among their other deficiencies, don't transform
3804 correctly when you rename packages, and don't change their
3805 print representation when you change things like *PACKAGE*
3806 and *PRINT-LENGTH*) to symbols and lists of symbols
3807 He also made several improvements and fixed several bugs in DESCRIBE.
3808 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3810 *** misbehavior of WRITE-STRING/WRITE-LINE
3811 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3812 and several other LOOP problems as well
3813 *** DIRECTORY when similar filenames are present
3814 *** DEFGENERIC with :METHOD options
3815 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3816 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3817 *** argument ordering in FIND with :TEST option
3818 *** mishandled package designator argument in APROPOS-LIST
3819 *** various problems in the backquote readmacro
3820 *** a bug in APROPOS
3821 *** probably some others that I'm not describing very well here,
3822 since the CVS log documents them by reference to sbcl-devel
3823 messages, and the SourceForge archives aren't working well.:-(
3824 ** Dan Barlow improved the Alpha port (and is making progress on the
3825 PPC port, for those of you who think different).
3826 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3827 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3828 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3829 continues to work on the SPARC port (for those of you in a position
3830 to look down upon our little PC-compatible boxes from a great height).
3831 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3832 and uses that copy, rather than the current dynamic value, when
3833 it's trying to decide whether to truncate output. Thus e.g.
3834 (let ((*print-lines* 50))
3835 (pprint-logical-block (stream nil)
3837 (let ((*print-lines* 8))
3838 (print (aref possiblybigthings i) stream)))))
3839 should now truncate the logical block only at 50 lines, instead of
3840 often truncating it at 8 lines, as it did before.
3841 * The doc/cmucl/ directory, containing old CMU CL documentation
3842 from the time of the fork, is no longer part of the base system.
3843 SourceForge has shut down its anonymous FTP service, and with it
3844 my original plan for distributing the old CMU CL documentation
3845 there. For now, if you need these files you can download an old
3846 SBCL source release and extract them from it.
3847 * The fasl file version number changed again, for dozens of reasons,
3848 some of which are apparent above.
3850 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3851 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3852 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3853 patch, so that DISASSEMBLE output is much nicer.
3854 * The code in the SB-PROFILE package now seems reasonably stable.
3855 I still haven't decided what the final interface should look like
3856 (I'd like PROFILE to interact cleanly with TRACE, since both
3857 facilities use function encapsulation) but if you have a need
3858 for profiling now, you can probably use it successfully with
3859 the current CMU-CL-style interface.
3860 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3861 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3862 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3863 link now returns the dangling link itself, and for similar
3864 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3865 link itself. (In these cases the old code signalled an error and
3866 looped endlessly, respectively.) Thus, DIRECTORY now works even
3867 in the presence of dangling and cyclic symbolic links.
3868 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3869 is now a supported extension again, since the consensus on
3870 sbcl-devel was that it can be useful for ordinary development
3871 work, not just for debugging SBCL itself.
3872 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3873 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3874 that currently-defined functions might be redefined later with
3875 different return types.
3876 * Hash tables can be printed readably, as inspired by CMU CL code
3877 of Eric Marsden and SBCL code of Martin Atzmueller.
3878 * better error handling in CLOS method combination, thanks to
3879 Martin Atzmueller porting Pierre Mai's CMU CL patches
3880 * more overflow fixes for >16Mbyte I/O buffers
3881 * A bug in READ has been fixed, so that now a single Ctrl-D
3882 character suffices to cause end-of-file on character streams.
3883 In particular, now you only need one Ctrl-D at the command
3884 line (not two) to exit SBCL.
3885 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3887 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3888 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3889 does the right thing.
3890 * The implementation of some type tests, especially for CONDITION
3891 types, is now tidier and maybe faster, due to CMU CL code
3892 originally by Douglas Crosher, ported by Martin Atzmueller.
3893 * Some math functions have been fixed, and there are new
3894 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3895 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3896 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3897 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3898 likely to be built into sbcl-0.7.0.)
3899 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3900 CL, ported to SBCL by Martin Atzmueller)
3901 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3902 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3903 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3904 * There's a new slam.sh hack to shorten the edit/compile/debug
3905 cycle for low-level changes to SBCL itself, and a new
3906 :SB-AFTER-XC-CORE target feature to control the generation of
3907 the after-xc.core file needed by slam.sh.
3908 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3909 COMPILE-FILE is no longer supported, so that now every function
3910 gets an entry point, so that block compilation looks a little
3911 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3912 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3913 returns the number of bytes consed since the system started,
3914 rather than the number consed since the first time the function
3915 was called. (The new definition parallels ANSI functions like
3916 CL:GET-INTERNAL-RUN-TIME.)
3917 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3918 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3919 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3920 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3921 * Fasl file version numbers are now independent of the target CPU,
3922 since historically most system changes which required version
3923 number changes have affected all CPUs equally. Similarly,
3924 the byte fasl file version is now equal to the ordinary
3927 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3928 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3929 is no longer recognized. I apologize for this, because it was
3930 listed in SB-EXT as a supported extension, but I found that
3931 its existing behavior was poorly specified, as well as incorrectly
3932 specified, and it looked like too much of a mess to straighten it
3933 out. I have enough on my hands trying to get ANSI stuff to work..
3934 * many patches ported from CMU CL by Martin Atzmueller, with
3935 half a dozen bug fixes in pretty-printing and the debugger, and
3936 half a dozen others elsewhere
3937 * fixed bug 13: Floating point infinities are now supported again.
3938 They might still be a little bit flaky, but thanks to bug reports
3939 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3940 as flaky as they were.
3941 * The --noprogrammer command line option is now supported. (Its
3942 behavior is slightly different in detail from what the old man
3943 page claimed it would do, but it's still appropriate under the
3944 same circumstances that the man page talks about.)
3945 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3946 are now supported, and enabled by default. Thus, the compiler can
3947 handle many floating point and complex operations much less
3948 inefficiently. (Thus e.g. you can implement a complex FFT
3950 * The compiler now detects type mismatches between DECLAIM FTYPE
3951 and DEFUN better, and implements CHECK-TYPE more correctly, and
3952 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3953 thanks to patches from Martin Atzmueller.
3954 * various fixes to make the cross-compiler more portable to
3955 ANSI-conforming-but-different cross-compilation hosts (notably
3956 Lispworks for Windows, following bug reports from Arthur Lemmens)
3957 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3958 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3959 fixed thanks to Pierre Mai's CMU CL patches.
3960 * fixes to keep the system from overflowing internal counters when
3961 it tries to use i/o buffers larger than 16M bytes
3962 * fixed bug 45a: Various internal functions required to support
3963 complex special functions have been merged from CMU CL sources.
3964 (When I was first setting up SBCL, I misunderstood a compile-time
3965 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3966 * improved support for type intersection and union, fixing bug 12
3967 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3968 more obscure bugs as well
3969 * some steps toward byte-compiling non-performance-critical
3970 parts of the system, courtesy of patches from Martin Atzmueller
3971 * Christophe Rhodes has made some debian packages of sbcl at
3972 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3973 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3974 stable, but are nonetheless usable. When he's ready, I'd be happy
3975 to add them to the SourceForge "File Releases" section. (And if
3976 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3977 * new fasl file format version number (because of changes in
3978 internal representation of (OR ..) types to accommodate the new
3979 support for (AND ..) types, among other things)
3981 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3982 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3984 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3986 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3987 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3988 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3989 * DESCRIBE now works on structure objects again.
3990 * Most function call argument type mismatches are now handled as
3991 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3992 know whether the function will be redefined before the call is
3993 executed. (The compiler could flag local calls with full WARNINGs,
3994 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3995 it doesn't keep track of enough information to know whether calls
3996 are local in this sense.)
3997 * Compiler output is now more verbose, with messages truncated
3998 later than before. (There should be some supported way for users
3999 to override the default verbosity, but I haven't decided how to
4000 provide it yet, so this behavior is still controlled by the internal
4001 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4002 src/compiler/ir1util.lisp.)
4003 * Fasl file format version numbers have increased again, because
4004 support for the Gray streams extension changes the layout of the
4005 system's STREAM objects.
4006 * The Gray subclassable streams extension now works, thanks to a
4007 patch from Martin Atzmueller.
4008 * The full LOAD-FOREIGN extension (not just the primitive
4009 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4010 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4011 but like most other programs, it defaults to copying the Unix
4012 environment from the original process instead of starting the
4013 new process in an empty environment.
4014 * Extensions which manipulate the Unix environment now support
4015 an :ENVIRONMENT keyword option which doesn't smash case or
4016 do other bad things. The CMU-CL-style :ENV option is retained
4017 for porting convenience.
4018 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4019 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4021 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4023 * A patch from Martin Atzmueller seems to have solved the SIGINT
4024 problem, and as far as we know, signal-handling now works cleanly.
4025 (If you find any new bugs, please report them!)
4026 * The system no longer defaults Lisp source file names to types
4027 ".l", ".cl", or ".lsp", but only to ".lisp".
4028 * The compiler no longer uses special default file extensions for
4029 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4030 seems to expect a single default extension for all compiled code,
4031 and there's no compelling reason to try to stretch the standard
4032 to allow two different extensions.) Instead, byte-compiled files
4033 default to the same extension as native-compiled files.
4034 * Fasl file format version numbers have increased again, because
4035 a rearrangement of internal implementation packages made some
4036 dumped symbols in old fasl files unreadable in new cores.
4037 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4038 many fewer weird special cases.
4039 * Bug #17 (differing COMPILE-FILE behavior between logical and
4040 physical pathnames) has been fixed, and some related misbehavior too,
4041 thanks to a patch from Martin Atzmueller.
4042 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4043 by Tim Moore, ported to SBCL by Martin Atzmueller.
4044 * Martin Atzmueller fixed several filesystem-related problems,
4045 including bug #36, in part by porting CMU CL patches, which were
4046 written in part by Paul Werkowski.
4047 * More compiler warnings in src/runtime/ are gone, thanks to
4048 more patches from Martin Atzmueller.
4049 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4052 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4054 * DESCRIBE now works on CONDITION objects.
4055 * The debugger now handles errors which arise when trying to print
4056 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4058 * The build system now uses an additional file, customize-target-features.lisp,
4059 to allow local modifications to the target *FEATURES* list. (The point of
4060 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4061 debugging features enabled, without having to worry about propagating it
4062 into everyone's system when I do a "cvs update".) When no
4063 customize-target-features.lisp file exists, the target *FEATURES* list
4064 should be constructed the same way as before.
4065 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4066 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4067 warn about behavior which is undefined under the ANSI spec. Note: This
4068 is specified by ANSI, but it's not very popular with programmers.
4069 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4070 macro in the SBCL sources for an example of a workaround which you
4071 might use to make portable ANSI-standard code which does what you want.
4072 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4073 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4074 when it's used as a non-toplevel form. (This is required in order
4075 to implement the DEFCONSTANT-EQX macro.)
4076 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4077 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4078 funny things, due to bugs in the implementation of EVAL-WHEN
4079 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4080 non-toplevel DEFCONSTANTs are uncommon.)
4081 * The core file version number and fasl file version number have been
4082 incremented, because the old noncompliant DEFCONSTANT behavior involved
4083 calling functions which no longer exist, and because I also took the
4084 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4085 * fixed bug 1 (error handling before read-eval-print loop starts), and
4086 redid debugger restarts and related debugger commands somewhat while
4088 ** The QUIT debugger command is gone, since it did something
4089 rather different than the SB-EXT:QUIT command, and since it never
4090 worked properly outside the main toplevel read/eval/print loop.
4091 Invoking the new TOPLEVEL restart provides the same functionality.
4092 ** The GO debugger command is also gone, since you can just invoke
4093 the CONTINUE restart directly instead.
4094 ** The TOP debugger command is also gone, since it's redundant with the
4095 FRAME 0 command, and since it interfered with abbreviations for the
4097 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4099 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4100 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4101 some porting work by Martin Atzmueller)
4102 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4103 with a boatload of patches which he ported to SBCL. Now that those
4105 ** The system tries to make sure that its low-priority messages
4106 are prefixed by semicolons, to help people who like to use
4107 syntax highlighting in their ILISP buffer. (This patch
4108 was originally due to Raymond Toy.)
4109 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4110 patches originally written by Raymond Toy.
4111 ** The compiler understands coercion between single-value and
4112 multiple-VALUES type expressions better, getting rid of some very
4113 weird behavior, thanks to patches originally by Robert MacLachlan
4114 and Douglas Crosher.
4115 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4116 lambda lists, thanks to a patch originally by Pierre Mai.
4117 ** The system no longer bogusly warns about "abbreviated type
4119 ** The compiler gets less confused by inlining and RETURN-FROM,
4120 thanks to some patches originally by Tim Moore.
4121 ** The system no longer hangs when dumping circular lists to fasl
4122 files, thanks to a patch originally from Douglas Crosher.
4123 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4124 undefined function error.
4125 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4126 in structure slot initforms) for now, documented workaround instead:-|
4127 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4128 * fixed bug 5: added stubs for various Gray stream functions called
4129 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4130 installed, at least appropriate type errors are generated
4131 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4132 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4133 consistently in DEFMETHOD forms.
4134 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4135 it doesn't seem to affect SBCL after all
4136 * The C runtime system now builds with better optimization and many
4137 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4139 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4141 * The system is now under CVS at SourceForge (instead of the
4142 CVS repository on my home machine).
4143 * The new signal handling code has been tweaked to treat register
4144 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4145 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4146 downstream, e.g. in debug-int.lisp, has implicit dependencies
4147 on the unsignedness of integer representation of machine words,
4148 and that caused the system to bomb out with infinite regress
4149 when trying to recover from type errors involving signed values,
4150 e.g. (BUTLAST '(1 2 3) -1).)
4151 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4152 (This was one of the bugs Peter Van Eynde reported back in July.)
4153 * The system now uses code inspired by Colin Walters' O(N)
4154 implementation of MAP (from the cmucl-imp@cons.org mailing
4155 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4156 inline the MAP operation, and there is more than one
4157 sequence argument to the MAP call (so that it can't just
4158 do ETYPECASE once and for all based on the type of the
4159 single sequence argument). (The old non-inline implementation
4160 of the general M-argument sequence-of-length-N case required
4161 O(M*N*N) time when any of the sequence arguments were LISTs.)
4162 * The QUIT :UNIX-CODE keyword argument has been renamed to
4163 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4164 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4166 * Raymond Wiker's patches to port dynamic loading from Linux to
4167 FreeBSD have been added.
4168 * The BUGS file is now more nearly up to date, thanks in large part
4169 to Martin Atzmueller's review of it.
4170 * The debugger now flushes standard output streams before it begins
4171 its output ("debugger invoked" and so forth).
4172 * The core version number and fasl file version number have both
4173 been incremented, because of incompatible changes in the layout
4175 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4176 used to not be called for a saved Lisp image.)
4177 * Martin Atzmueller's version of a patch to fix a compiler crash,
4178 as posted on sbcl-devel 13 September 2000, has been installed.
4179 * Instead of installing Martin Atzmueller's patch for the
4180 compiler transform for SUBSEQ, I deleted the compiler transform,
4181 and transforms for some similar consing operations.
4182 * A bug in signal handling which kept TRACE from working on OpenBSD
4184 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4185 away by constant folding
4186 * The system now defines its address space constants in one place
4187 (in the Lisp sources), and propagates them automatically elsewhere
4188 (through GENESIS and the sbcl.h file). Therefore, patching the
4189 address map is less unnecessarily tedious and error-prone. The
4190 Lisp names of address space constants have also been systematized.
4191 * CVS tags like dollar-Header-dollar have been removed from
4192 the sources, because they have never saved me trouble and
4193 they've been source of trouble working with patches and other
4194 diff-related operations.
4195 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4196 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4198 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4200 * The system has been ported to OpenBSD.
4201 * The system now compiles with a simple "sh make.sh" on the systems
4202 that it's supported on. I.e., now you no longer need to tweak
4203 text in the target-features.lisp-expr and symlinks in src/runtime/
4204 by hand, the make.sh takes care of it for you.
4205 * The system is no longer so grossly inefficient when compiling code
4206 involving vectors implemented as general (not simple) vectors (VECTOR T),
4207 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4208 which dares to use the various sequence functions on non-simple
4209 vectors, takes less of a performance hit.
4210 * There is now a primitive type predicate VECTOR-T-P
4211 to test for the (VECTOR T) type, so that e.g.
4212 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4213 can now be compiled with some semblance of efficiency. (The old code
4214 turned the type declaration into a full call to %TYPEP at runtime!)
4215 * AREF on (VECTOR T) is still not fast, since it's still compiled
4216 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4217 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4218 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4219 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4220 without making full calls to SUBTYPEP at runtime.
4221 (Some analogous efficiency issues for non-simple vectors specialized to
4222 element types other than T, or for non-simple multidimensional arrays,
4223 have not been addressed. They could almost certainly be handled the
4224 same way if anyone is motivated to do so.)
4225 * The changes in array handling break binary compatibility, so
4226 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4227 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4228 instead of (VALUES T T).
4229 * By following the instructions that Dan Barlow posted to sbcl-devel
4230 on 2 July 2000, I was able to enable primitive dynamic object
4231 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4232 functionality is not implemented (since it calls ld to resolve
4233 library references automatically, requiring RUN-PROGRAM for its
4234 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4235 try to resolve library references) is now supported.
4236 * The system now flushes the standard output streams when it terminates,
4237 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4238 them at several other probably-convenient times, e.g. in each pass of
4239 the toplevel read-eval-print loop, and after evaluating a form given
4240 as an "--eval" command-line option. (These changes were motivated by a
4241 discussion of stream flushing issues on cmucl-imp in August 2000.)
4242 * The source transform for TYPEP of array types no longer assumes
4243 that an array whose element type is a not-yet-defined type
4244 is implemented as an array of T, but instead punts, so that the
4245 type will be interpreted at runtime.
4246 * There is now some support for cross-compiling in make.sh: each of
4247 the phases of make.sh has its own script. (This should be transparent
4248 to people doing ordinary, non-cross-compile builds.)
4249 * Since my laptop doesn't have hundreds of megabytes of memory like
4250 my desktop machine, I became more motivated to do some items on
4251 my to-do list in order to reduce the size of the system a little:
4252 ** Arrange for various needed-only-at-cold-init things to be
4253 uninterned after cold init. To support this, those things have
4254 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4255 symbols with such names are now uninterned after cold init).
4256 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4257 fasl files for cold load.
4258 ** Remove the old compiler structure pooling code (which used to
4259 be conditional on the target feature :SB-ALLOC) completely.
4260 ** Redo the representation of some data in cold init to be more compact.
4261 (I also looked into supporting byte compiled code at bootstrap time,
4262 which would probably reduce the size of the system a lot, but that
4263 looked too complicated, so I punted for now.)
4264 * The maximum signal nesting depth in the src/runtime/ support code has
4265 been reduced from 4096 to 256. (I don't know any reason for the very
4266 large old value. If the new smaller value turns out to break something,
4267 I'll probably just bump it back up.)
4268 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4270 * Many, many bugs reported by Peter Van Eynde have been added to
4271 the BUGS list; some have even been fixed.
4272 * While enabling dynamic object file loading, I tried to make the
4273 code easier to understand, renaming various functions and variables
4274 with less ambiguous names, and changing some function calling
4275 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4276 * While trying to figure out how to do the OpenBSD port, I tried to
4277 clean up some of the code in src/runtime/. In particular, I dropped
4278 support for non-POSIX signal handling, added various comments,
4279 tweaked the code to reduce the number of compilation warnings, and
4280 renamed some files to increase consistency.
4281 * To support the new automatic configuration functionality in make.sh,
4282 the source file target-features.lisp-expr has been replaced with the
4283 source file base-target-features.lisp-expr and the machine-generated
4284 file local-target-features.lisp-expr.
4285 * fixed a stupid quoting error in make.sh so that using CMU CL
4286 "lisp -batch" as cross-compilation host works again
4288 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4290 * DESCRIBE no longer tries to call itself recursively to describe
4291 bound/fbound values, so that it no longer fails on symbols which are
4292 bound to themselves (like keywords, T, and NIL).
4293 * DESCRIBE now works on generic functions.
4294 * The printer now prints less-screwed-up representations of closures
4295 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4296 * A private symbol is used instead of the :EMPTY keyword previously
4297 used to mark empty slots in hash tables. Thus
4298 (DEFVAR *HT* (MAKE-HASH-TABLE))
4299 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4300 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4301 now does what ANSI says that it should. (You can still get
4302 similar noncompliant behavior if bang on the hash table
4303 implementation with all the symbols you get back from
4304 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4305 This breaks binary compatibility, since tests for equality to
4306 :EMPTY are wired into things like the macroexpansion of
4307 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4309 * There's now a minimal placeholder implementation for CL:STEP,
4310 as required by ANSI.
4311 * An obscure bug in the interaction of the normal compiler, the byte
4312 compiler, inlining, and structure predicates has been patched
4313 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4314 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4316 * Missing ordinary arguments in a macro call are now detected even
4317 when the macro lambda list contains &KEY or &REST.
4318 * The debugger no longer complains about encountering the top of the
4319 stack when you type "FRAME 0" to explicitly instruct it to go to
4320 the top of the stack. And it now prints the frame you request even
4321 if it's the current frame (instead of saying "You are here.").
4322 * As specified by ANSI, the system now always prints keywords
4323 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4325 * The default initial SIZE of HASH-TABLEs is now smaller.
4326 * Type information from CLOS class dispatch is now propagated
4327 into DEFMETHOD bodies, so that e.g.
4328 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4330 is now basically equivalent to
4331 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4332 (DECLARE (TYPE SINGLE-FLOAT X))
4334 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4335 operation, without having to do run-time type dispatch.
4336 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4337 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4338 * Since I don't seem to be making big file reorganizations very often
4339 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4340 is overflowing my ability to conveniently back them up), I've finally
4341 checked the system into CVS. (The CVS repository is on my home system,
4342 not at SourceForge -- putting it on SourceForge might come later.)
4343 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4344 high-level GC-NOTIFY-FOO functions send their output. (There's
4345 still very little control of where low-level verbose GC functions
4346 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4347 less than it used to -- the GC-NOTIFY-FOO functions are now under
4348 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4349 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4350 in only one place in the source code, and propagates it automatically
4351 everywhere that it's needed. Thus e.g. when I bump the version from
4352 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4353 * The C source files now include boilerplate legalese and documentation
4354 at the head of each file (just as the Lisp source files already did).
4355 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4356 to his page will be replaced with a link to his new CLiki service.
4358 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4360 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4361 * The build process now looks for GNU make under the default name "gmake",
4362 instead of "make" as it used to. If GNU make is not available as "gmake"
4363 on your system, you can change this default behavior by setting the
4364 GNUMAKE environment variable.
4365 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4366 can build without error under CMU CL.
4368 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4370 * There is now a partial SBCL user manual (with some new text and some
4371 text cribbed from the CMU CL manual).
4372 * The beginnings of a profiler have been added (starting with the
4373 CMU CL profiler and simplifying and cleaning up). Eventually the
4374 main interface should be through the TRACE macro, but for now,
4375 it's still accessed through vaguely CMU-CL-style functions and macros
4376 exported from the package SB-PROFILE.
4377 * Some problems left over from porting CMU CL to the new
4378 cross-compilation bootstrap process have been cleaned up:
4379 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4380 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4381 part when getting rid of the compiler *BACKEND* stuff.)
4382 ** The value of *NULL-TYPE* was screwed up, because it was
4383 being initialized before the type system knew the final
4384 definition of the 'NULL type. This screwed up several key
4385 optimizations in the compiler, causing inefficiency in all sorts
4386 of places. (I found it because I wanted to understand why
4387 GET-INTERNAL-RUN-TIME was consing.)
4388 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4389 PROCLAIM FTYPE information. Unfortunately this broke binary
4390 compatibility again, since now the forms output by DEFGENERIC
4391 to refer to functions which didn't exist in 0.6.3.
4392 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4393 can use the new (as of 0.6.3) transform for SOME into MAP into
4395 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4396 values to (INTEGER 0 1000000), so that the time code will no longer
4397 occasionally get blown up by Linux returning 1000000 microseconds
4398 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4399 its output conform to the ANSI spec. (Alas, this makes its output
4400 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4401 * A full call to MAP NIL with a single sequence argument no longer conses.
4402 * fixes to problems pointed out by Martin Atzmueller:
4403 * The manual page no longer talks about multiprocessing as though
4404 it were currently supported.
4405 * The ILISP support patches have been removed from the distribution,
4406 because as of version 5.10.1, ILISP now supports SBCL without us
4407 having to maintain patches.
4408 * added a modified version of Raymond Toy's recent CMU CL patch for
4409 EQUALP comparison of HASH-TABLE
4411 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4413 * The system still can't cross-compile itself with
4414 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4415 entails), but at least it can compile more of itself that way
4416 than it used to be able to, and various buglets which were uncovered
4417 by trying to cross-compile itself that way have now been fixed.
4418 * This release breaks binary compatibility again. This time
4419 at least I've incremented the FASL file format version to 2, so that the
4420 problem can be detected reliably instead of just causing weird errors.
4421 * various new style warnings:
4422 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4423 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4424 ** using the lexical binding of a variable named in the *FOO* style
4425 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4426 as specified by ANSI.
4427 * *RANDOM-STATE* is no longer automatically initialized from
4428 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4429 default behavior of the system is to repeat its behavior every time
4430 it's run. If you'd like to change this behavior, you can always
4431 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4432 old convention there was no comparably easy way to get the system to
4433 repeat its behavior every time it was run.
4434 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4435 FTYPE declarations has been removed, in favor of their ANSI
4436 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4437 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4438 the types of their sequence arguments can be determined at compile time.
4439 This is done through a new open code expansion for MAP which eliminates
4440 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4441 argument types can be determined at compile time.
4442 * The optimizer now transforms COERCE into an identity operation when it
4443 can prove that the coerced object is already of the correct type. (This
4444 can be a win for machine generated code, including the output of other
4445 optimization transforms, such as the MAP transform above.)
4446 * Credit information has been moved from source file headers into CREDITS.
4447 * Source file headers have been made more standard.
4448 * The CASE macro now compiles without complaining even when it has
4451 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4453 * (Note that the way that the PCL macroexpansions were rewritten
4454 to accommodate the change in DEFGENERIC below breaks binary
4455 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4456 not run under sbcl-0.6.2. Once we get out of alpha releases,
4457 i.e. hit release 1.0.0, we'll probably try to maintain binary
4458 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4459 and sbcl-1.4.4. Until then, however, it might be fairly common
4460 for maintenance releases to break binary compatibility.)
4461 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4463 * The reporting of unhandled signals has been changed to print some
4464 explanatory text as well as the report form. (Previously only
4465 the report form was printed.)
4466 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4467 it defines, so that the compiler no longer issues undefined function
4468 warnings for compiled-but-not-yet-loaded generic functions.
4469 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4470 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4471 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4472 section of the ANSI spec are supported.
4473 * The "" nickname for the "KEYWORD" package has been removed.
4474 The reader still handles symbol tokens which begin with a package marker
4475 as keywords, but it doesn't expose its mechanism for doing so in the
4476 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4477 * The system now issues STYLE-WARNINGs for contradictory TYPE
4478 proclamations. (Warnings for contradictory FTYPE proclamations would
4479 be nice too, but those can't be done usefully unless the type system
4480 is made smarter about FUNCTION types.)
4481 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4482 systematized, so that "*target-*.lisp is supposed to exist only on the
4483 target and imply that there's a related file which exists on the
4484 host, and *host-*.lisp is supposed to exist only on the host and imply
4485 that there's a related file which exists on the target. This involves a
4486 lot of renaming. Hopefully the acute confusion caused by the renaming
4487 will be justified by the reduction in chronic confusion..
4488 ** runtime-type.lisp -> early-target-type.lisp
4489 ** target-type.lisp -> late-target-type.lisp
4490 ** early-host-format.lisp -> early-format.lisp
4491 ** late-host-format.lisp -> late-format.lisp
4492 ** host-error.lisp -> misc-error.lisp
4493 ** early-error.lisp -> early-target-error.lisp
4494 ** late-error.lisp -> late-target-error.lisp
4495 ** host-defboot.lisp -> early-defboot.lisp
4496 ** code/misc.lisp -> code/target-misc.lisp
4497 ** code/host-misc.lisp -> code/misc.lisp
4498 ** code/numbers.lisp -> code/target-numbers.lisp
4499 ** code/early-numbers.lisp -> numbers.lisp
4500 ** early-host-type.lisp -> early-type.lisp
4501 ** late-host-type.lisp -> late-type.lisp
4502 ** host-typep.lisp -> typep.lisp
4503 ** load.lisp -> target-load.lisp
4504 ** host-load.lisp -> load.lisp
4505 ** host-disassem.lisp -> disassem.lisp
4506 ** host-insts.lisp -> insts.lisp
4507 ** byte-comp.lisp -> target-byte-comp.lisp
4508 ** host-byte-comp.lisp -> byte-comp.lisp
4509 ** host-signal.lisp -> signal.lisp
4510 ** host-defstruct.lisp -> defstruct.lisp
4511 ** late-target-type.lisp -> deftypes-for-target.lisp
4512 Furthermore, several other previously target-only files foo.lisp (e.g.
4513 hash-table.lisp and random.lisp) have been split into a target-and-host
4514 foo.lisp file and a target-only target-foo.lisp file, with their key type
4515 definitions in the target-and-host part, so that the cross-compiler will
4516 know more about target types.
4517 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4518 gone away. In their place are various *BACKEND-FOO* variables
4519 corresponding to the slots of the old structure.
4520 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4521 into the target SBCL has been fixed.
4522 * The chill.lisp system for loading cold code into a running SBCL
4524 * Support for the CMU CL "scavenger hook" extension has been removed.
4525 (It was undocumented and unused in the CMU CL sources that SBCL was
4526 derived from, and stale in sbcl-0.6.1.)
4527 * Various errors in the cross-compiler type system were detected
4528 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4529 (enabling various consistency checks). Many of them were fixed,
4530 but some hard problems remain, so the compiler is back to
4531 running without *TYPE-SYSTEM-INITIALIZED* for now.
4532 * As part of the cross-compiler type system cleanup, I implemented
4533 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4534 * I have started adding UNCROSS calls throughout the type system
4535 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4536 on cross-compiler input files will be able to go away, and various
4538 * CONSTANTP now returns true for quoted forms (as explicitly required
4541 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4543 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4544 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4545 (MAKE-LIST -1). In the longer term, it ought to become true
4546 that declarations are assertions even at SAFETY 1. For now, it's not
4547 quite true even at SAFETY 3, but it's at least more nearly true..
4548 (Note that this change seems to increases the size of the system by
4549 O(5%) and to decrease the speed of the compiler by 20% or more.)
4550 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4551 for the problem of printing dozens of lines of distracting information
4552 about low-level system machinery as part of the top stack frame
4553 on entry to the debugger when an undefined function was called.
4554 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4555 is not reset to COMMON-LISP-USER.
4556 * Compilation of stuff related to dyncount.lisp has been made conditional
4557 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4558 smaller. The various dyncount-related symbols have been moved into
4559 a new "SB-DYNCOUNT" package.
4560 * tty-inspect.lisp has been renamed to inspect.lisp.
4561 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4562 feature has gone away. (When we eventually port to other flavors of
4563 libc and/or Unix, we'll try to make the differences between flavors
4564 invisible at the user level.)
4565 * Various other *FEATURES* tags, and/or their associated conditionals,
4566 have been removed if obsolescent, or given better documentation, or
4567 sometimes given more-mnemonic names.
4569 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4571 * tidied up "make.sh" script
4572 * tidied up system directory structure
4573 * better "clean.sh" behavior
4574 * added doc/FOR-CMUCL-DEVELOPERS
4575 * many many small tweaks to output format, e.g. removing possibly-confusing
4576 trailing #\. character in DESCRIBE-INSTANCE
4577 * (EQUALP #\A 'A) no longer signals an error.
4578 * new hashing code, including EQUALP hashing
4579 * tidied up Lisp initialization and toplevel
4580 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4581 * command line argument processing
4582 * added POSIX-GETENV function to deal with Unix-ish environment variables
4583 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4584 terminating SBCL on EOF
4585 * non-verbose GC by default
4586 * There is no more "sbcl" shell script; the sbcl file is now the C
4587 runtime executable (just like CMU CL).
4588 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4590 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4591 symbol and package information as Lisp native symbol and package objects
4592 instead of strings naming symbols and strings naming packages. This way,
4593 whenever packages are renamed (as in warm init), debug information is
4594 transformed along with everything else.
4595 * tweaked the optimization policy declarations which control the building
4596 of SBCL itself. Now, among other things, the system no longer saves
4597 source location debugging information. (This helps two problems at once
4598 by reducing SBCL size and by keeping SBCL from trying to look for its
4599 sources -- which may not exist -- when reporting errors.)
4600 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4601 debugging and testing purposes
4602 * cleaned up printing, making the printer call PRINT-OBJECT for
4603 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4604 methods, giving nearly-ANSI behavior
4605 * converted almost all special variables to use *FOO* naming convention
4606 * deleted PARSE-TIME functionality, since it can be done portably
4607 * moved some files out of cold init into warm init
4608 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4610 * regularized formatting of source files
4611 * added an install.sh script
4612 * fixed ridiculous memory usage of cross-compiler by making
4613 compiler/alloc.lisp not try to do pooling unless it can hook
4614 itself into the GC of the cross-compilation host. Now the system
4615 builds nicely on my old laptop.
4616 * added :SB-ALLOC in target-features.lisp-expr
4617 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4618 was not implemented)
4619 * re-did condition handling and note reporting in the compiler. Notes
4620 are no longer handled by signalling conditions. Style warnings
4621 and warnings are handled more correctly and reported in such a way
4622 that it's easy to find one or the other in your output (so that you
4623 can e.g. figure out which of many problems caused COMPILE-FILE to
4625 * changed the severity of several compiler warnings from full WARNING
4626 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4627 compiler note reporting so that it doesn't use the condition system
4628 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4629 COMPILE-FILE command)
4630 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4631 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4632 no longer supported, since their implementation was deeply intertwingled
4633 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4634 * removed bogus "support" for compiler macros named (SETF FOO), and
4635 removed the compiler macro for SETF INFO (but only after making a fool
4636 of myself on the cmucl-imp mailing list by posting a bogus patch for
4637 DEFINE-COMPILER-MACRO..)
4638 * Compiled files containing forms which have side effects on the Lisp
4639 reader (such as DEFPACKAGE forms) are now handled more correctly.
4640 (Compiler queuing of top level lambdas has been suppressed by setting
4641 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4642 * deleted various currently-unused source files, e.g. gengc.lisp. They
4643 may be added back at some point e.g. when porting to other architectures,
4644 but until they are it's distracting to distribute them and to try to
4646 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4647 there were so many of them they're just distractions, and UNCROSS is
4648 known to be able to handle the current sources
4649 * moved PROFILE functionality into TRACE, so that it will be clear
4650 how the wrapping and unwrapping of functions when you profile them
4651 interacts with the wrapping and unwrapping of functions when you
4652 trace them. (Actually, the functionality isn't there yet, but at least
4653 the interface specification is there. Hopefully, the functionality will
4654 arrive with some future maintenance release.)
4655 * removed host-oops.lisp
4656 * changed signature of QUIT function to allow UNIX-CODE argument
4657 * fixed READ-SEQUENCE bug
4658 * tweaked verbose GC output so that it looks more like the progress
4659 output that ANSI specifies for functions like LOAD
4660 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4661 * added <http://sbcl.sourceforge.com> to the banner information printed by