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 * optimization: two argument forms of LAST are upto ~10% faster
7 in normal SPEED policies.
8 * bug fix: LAST when always returned the whole list when given a bignum
9 as the second argument.
10 * bug fix: dynamic extent allocation of nested lists and vectors
11 could leak to otherwise accessible parts.
12 * bug fix: invalid optimization of heap-allocated alien variable
14 * bug fix: fasl header checking is less vulnerable to different
15 platform word lengths.
16 * bug fix: more correct assembler syntax for GNU binutils
17 2.18.50.0.4 support. (thanks to Marijn Schouten)
19 changes in sbcl-1.0.16 relative to 1.0.15:
20 * minor incompatible change: revert the changes to sb-posix's error
21 signaling added in 1.0.14.
22 * minor incompatible change: change PROBE-FILE back to returning
23 NIL whenever we can't get a truename, as was the case before 1.0.14.
24 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
25 inherits from ERROR instead of just CONDITION.
26 * new feature: SB-INTROSPECT can provide source locations for instances
27 as well. (thanks to Tobian Ritterweiler)
28 * optimization: binding special variables now generates smaller code
29 on threaded platforms.
30 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
31 and cases where no :TEST is given but the compiler can infer that
32 the element to search is of type (OR FIXNUM (NOT NUMBER)).
33 * optimization: better LOGNOT on fixnums.
34 * optimization: modular arithmetic for a particular requested width
35 is implemented using a tagged representation unless a better
36 representation is available.
37 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
38 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
39 for wellformedness and returns a specialized vector. (reported by
41 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
42 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
43 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
44 unscheduling timer at the same time as another timer fires.
45 * bug fix: don't reschedule timers for dead threads.
46 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
47 * bug fix: copying output from RUN-PROGRAM to a stream signalled
48 bogus errors if select() was interrupted.
49 * enhancement: add support for fcntl's struct flock to SB-POSIX.
51 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
52 * enhancement: cleaner backtraces for interactive interrupts, as
53 well as other cases where the interesting frames used to be
54 obscured by interrupt handling frames.
55 * enhancement: untracing a whole package using (UNTRACE "FOO") is
56 now supported, and tracing a whole package using (TRACE "FOO") now
57 traces SETF-functions as well.
58 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
59 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
60 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
61 * unreadably printed representation of hash-tables now includes
63 * bug fix: partially fixed #188: type propagation from assignments
64 is now more efficient.
65 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
66 'DOUBLE-FLOAT) are not flushable.
67 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
68 full address of the object, and none of the tag bits.
69 * bug fix: readably printing hash-tables now respects other printer
70 control variables. (reported by Cedric St-Jean)
71 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
72 keyword with MAKE-HASH-TABLE.
73 * bug fix: export SB-POSIX:MKSTEMP.
74 * bug fix: SORT was not interrupt safe.
75 * bug fix: XREF accounts for the last node of each basic-block as
77 * bug fix: MAKE-INSTANCE optimizations interacted badly with
78 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
79 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
80 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
81 create obsolete instances in the case of redefinition or
82 obsoletion of a superclass. (thanks to Andy Hefner)
83 * bug fix: Support for the Alpha architecture has been revived; it had
84 suffered somewhat from lack of maintenance since sbcl-1.0.
85 * improvements to the Windows port:
86 ** The system detects the codepage to use at startup, instead of
87 using the value from the saved core. (thanks to Kei Suzuki)
89 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
90 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
91 (see documentation for details.)
92 * revived support for OpenBSD (contributed by Josh Elsasser)
93 * partially fixed bug #108: ROOM no longer suffers from occasional
94 (AVER (SAP= CURRENT END)) failures.
95 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
96 bodies no longer cause a WARNING to be signalled. (reported by
98 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
99 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
100 the bug in TRUNCATE.) (reported by Michael Weber)
101 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
102 no samples. (reported by Andy Hefner)
103 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
104 no longer appear as (NIL ...) frames in backtraces.
105 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
106 is now actually exported from SB-EXT as documented. (reported by
108 * bug fix: sb-aclrepl now correctly understands how to inspect
109 single-floats on 64-bit platforms where single-floats are not boxed.
110 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
111 yet been finalized. (reported by Levente Meszaros)
112 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
113 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
114 SBCL-specific optimize qualities.
116 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
117 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
118 an executable in the search path, and does so in the child
119 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
120 been removed; it can be found in the CVS history, for somebody who
121 needs that search behavior (see the manual).
122 * minor incompatible change: compiler policy re. weakening type
123 checks has changed: now type checks are weakened only if SAFETY < 2
125 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
126 unparsing of directory pathnames as files. Analogously,
127 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
128 filename to parse into a directory pathname.
129 * enhancement: implicit generic function creation now signals a
130 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
131 users can bind handlers for and muffle around calls to LOAD.
132 * enhancement: RUN-PROGRAM allows unicode arguments and environments
133 to be used (using the default stream external format), and allows
134 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
135 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
136 strings and vectors whose element-type or simplicity is not fully
137 known at compile-time.
138 * optimization: STRING-TRIM and related functions no longer allocate
139 a new string when no trimming needs to be performed. These functions
140 are also faster than before when the input string has been declared
142 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
143 * optimization: READ-LINE is significantly faster for files containing
145 * optimization: non-open coded uses of character comparison operators
146 (e.g. char=) no longer cons when called with more than one parameter
147 on platforms supporting dynamic-extent allocation.
148 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
149 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
150 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
152 * bug fix: some sequence functions elided bounds checking when
154 * bug fix: too liberal weakening of union-type checks when SPEED >
156 * bug fix: more bogus fixnum declarations in ROOM implementation
159 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
160 * new feature: MAKE-HASH-TABLE now experimentally accepts a
161 :SYNCHRONIZED argument, which makes the hash-table safe for
162 concurrent accesses (but not iteration.) See also:
163 SB-EXT:WITH-LOCKED-HASH-TABLE, and
164 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
165 * optimization: CONCATENATE on strings is an order of magnitude faster
166 in code compiled with (> SPEED SPACE).
167 * optimization: SUBSEQ is ~50% faster on lists.
168 * bug fix: bug 417 fixed -- source location reporting is now more robust.
169 * bug fix: SUBSEQ on a list will now correctly signal an error if if
170 END is smaller then START.
171 * bug fix: SB-PROFILE will no longer report extra consing for nested
172 calls to profiled functions.
173 * bug fix: ROOM implementation had bogus fixnum declarations which
174 could cause type-errors when calling ROOM on large images.
175 * bug fix: if file compilation is aborted, the partial fasl is now
176 deleted, and COMPILE-FILE returns NIL as the primary value.
177 * bug fix: number of thread safety issues relating to SBCL's internal
178 hash-table usage have been fixed.
179 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
180 be returned from its body when the values were being returned
181 using unknown-values return convection and the W-P-O was wrapped
182 inside an UNWIND-PROTECT.
183 * bug fix: sb-posix should now compile again under Windows, enabling
186 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
187 * incompatible change: hash-table accessor functions are no longer
188 automatically protected by locks. Concurrent accesses on the same hash-table
189 from multiple threads can give inconsistent results or even corrupt the
190 hash-table completely. Multi-threaded applications should do their own
191 locking at the correct granularity. In the current implementation it is
192 still safe to have multiple readers access the same table, but it's not
193 guaranteed that this property will be maintained in future releases.
194 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
195 and will signal an error at runtime.
196 * enhancement: SB-THREAD package now exports a semaphore interface.
197 * enhancement: CONS can now stack-allocate on x86 and
198 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
200 * enhancement: nested lists can now be stack allocated on
201 platforms providing stack allocation support.
202 * enhancement: dynamic-extent support has been extended to support
203 cases where there are multiple possible sources for the stack
205 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
206 if the mutex is uncontested on Linux.
207 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
208 as specified by the CLHS. (thanks to Tobias Rittweiler)
209 * bug fix: NaN comparison now works on x86-64.
210 * bug fix: CLOSE :ABORT T on a stream with pending output now
212 * bug fix: instances of non-standard metaclasses using standard
213 instance structure protocol sometimes missed the slot type checks
215 * bug fix: known functions can be inlined in high-debug code.
217 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
218 * minor incompatible change: the MSI installer on Windows no longer
219 associates .lisp and .fasl files with the installed SBCL.
220 * minor incompatible change: :UNIX is no longer present in *FEATURES*
221 on Windows. (thanks to Luis Oliviera)
222 * new feature: SB-CLTL2 contrib module now implements
223 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
224 * optimization: scavenging weak pointers is now more efficient,
225 requiring O(1) instead of O(N) per weak pointer to identify
226 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
227 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
228 method bodies using constant slot-names when the first argument is
229 a specializer parameter for the method.
230 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
231 names, when the class of the instance is a direct instance
232 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
233 as slow as the constant slot-name case.)
234 * optimization: member type construction is now O(N) instead
236 * optimization: UNION and NUNION are now O(N+M) for large
237 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
239 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
240 now more readable in environments like Slime which display it.
241 (thanks to Tobias C. Rittweiler)
242 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
243 was able to infer the correct argument type for the object on which
244 the CAS operation was being performed.
245 * bug fix: copy propagation interfered with parallel assignment
246 semantics in local calls. (reported by Paul Khuong)
247 * bug fix: the signed modular fixnum shift compiled to wrong code on
248 x86 and x86-64. (spotted by a slight modification to some of
251 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
252 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
253 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
254 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
256 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
257 and generic function now print the object identity for anonymous
258 instances. (thanks to Attila Lendvai)
259 * enhancement: as an extension to MOP, SBCL now supports SETF of
260 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
261 (thanks to Attila Lendvai)
262 * enhancement: sb-cover has an alternate annotation mode which puts
263 the form annotations at the car of the form, instead of annotating
264 it completely. The benefit of this mode is that it shows better
265 which forms were instrumented by the compiler.
266 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
267 by printer control variables. (reported by Dan Corkill)
268 * bug fix: system leaked memory when delayed output was performed by
269 the OS in smaller chunks then expected. (thanks to David Smith)
270 * bug fix: system leaked memory when file streams were not closed
272 * bug fix: large objects written to slow streams that were modified
273 after the write could end up with the modified state written to
274 the underlying file descriptor.
275 * bug fix: multiple threads operating in parallel on the same stream
276 could cause buffer-overflows.
277 * bug fix: source location information is stored correctly
278 (broken since 1.0.6). This bug would generally show up as the
279 Slime debugger highlighting the wrong form when the sldb-show-source
281 * bug fix: Branch forms are again annotated as branches in the sb-cover
283 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
284 on x86-64. (reported by Gregory Vanuxem)
285 * bug fix: EQUALP could return wrong results for structures with raw
286 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
287 word). (reported by Vjacheslav Fyodorov)
288 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
291 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
292 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
293 atomic compare-and-swap operations on threaded platforms.
294 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
295 allows assigning a global minimum value to optimization qualities
296 (overriding proclamations and declarations).
297 * enhancement: closed over variables can be stack-allocated on x86
299 * performance bug fix: GETHASH and (SETF GETHASH) are once again
301 * optimization: slot definition lookup is now O(1). This speeds up
302 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
303 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
304 * optimization: ASSOC and MEMBER can now be open-coded for all
305 combinations of keyword arguments when second argument is constant
306 and SPEED >= SPACE. In other cases a specialized version is
308 * bug fix: using obsoleted structure instances with TYPEP and
309 generic functions now signals a sensible error.
310 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
311 (reported by Kristoffer Kvello)
312 * bug fix: backtrace construction is now more careful when making
313 lisp-objects from pointers on the stack, to avoid creating bogus
314 objects that can be seen by the GC.
315 * bug fix: defaulting of values in contexts expecting more than 7
316 variables now works on x86-64. (reported by Christopher Laux)
317 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
319 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
320 as the property-list of a symbol.
321 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
322 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
323 situations requested, are once again file-compileable. (reported
326 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
327 * MOP improvement: support for user-defined subclasses of
328 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
329 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
330 the expansion of DEFMETHOD, is responsible for generating a form
331 which creates a list of specializers when evaluated. Additional
332 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
333 debugging and introspective support.
334 * minor incompatible change: the (unsupported) spinlock interface
335 has changed: free spinlock now has the value NIL, and a held spinlock
336 has the owning thread as its value.
337 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
338 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
340 * enhancement: name of a socket-stream is now "a socket" instead of
342 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
343 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
344 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
345 for generic functions. (Thanks to Tobias C. Rittweiler)
346 * optimization: bignum printing speed has been improved by 20-40%
347 (depending on the bignum size.)
348 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
350 * bug fix: the cache used by the CLOS to store precomputed effective
351 methods, slot offsets, and constant return values is now thread and
353 * bug fix: generic function dispatch function updating is now thread
354 and interrupt safe (in the sense that the known issues have been
356 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
357 * bug fix: interning EQL-specializers is now thread and interrupt safe.
358 * bug fix: asdf systems with dependencies to the SB-POSIX or
359 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
360 * bug fix: interrupt safety of applicable method computation has been
363 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
364 * new contrib: sb-cover, an experimental code coverage tool, is included
366 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
367 significantly faster.
368 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
369 produce more efficient code on x86-64 if the compiler can determine
370 that (+ INDEX <constant>) does not require a bounds check and FOO
371 has an element type at least 8 bits wide.
372 * enhancement: a new, experimental synchronous timeout facility is
373 provided. Refer to SB-SYS:WITH-DEADLINE for details.
374 * enhancement: when a symbol name conflict error arises, the
375 conflicting symbols are always printed with a package prefix.
376 (thanks to Kevin Reid)
377 * enhancement: stepping is now once again supported on the SPARC and
378 MIPS platforms. (It is also now more likely to work on CheneyGC
380 * enhancement: sb-sprof can now also track and report accurate call
382 * bug fixes: the treatment of non-standard subclasses of
383 SB-MOP:SPECIALIZER is more correct.
384 * incompatible change: PURIFY no longer copies the data from the
385 dynamic space into the static and read-only spaces on platforms
386 that use the generational garbage collector
387 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
389 * bug fix: GC race condition occasionally resulting in crashes with
390 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
392 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
394 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
395 a STYLE-WARNING to be signalled (regression from 1.0.4.)
396 * bug fix: an asynchronous interrupt could previously leave the
397 system running with GC inhibited.
398 * bug fix: a DECLARE form evaluated at top-level now causes an error
399 rather than silently (or verbosely) returning NIL.
400 * bug fix: trying to dissassemble functions compiled at high DEBUG could
401 result in a error being signalled due to source form lookup errors.
402 (reported by Peter Graves)
404 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
405 * incompatible change: removed writer methods for host-ent-name,
406 host-ent-addresses -- changing the values did not update the DNS
408 * minor incompatible change: changed experimental JOIN-THREAD interface
409 * documentation: the manual now lists reader and writer methods
410 in class slot documentation sections. (thanks to Richard M Kreuter)
411 * documentation: unwinding from asyncronous events has been
412 documented as unsafe.
413 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
414 in multithreaded application code.
415 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
416 platforms. (thanks to James Anderson for the optimization hint)
417 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
419 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
420 variants no longer cons.
421 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
422 their NOT- variants no longer cons.
423 * optimization: EQUAL hash tables no longer use SXHASH for objects
424 of all data types, but instead use an EQL hash for types for which
425 EQUAL is the same as EQL.
426 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
427 are significantly faster.
428 * optimization: new STRING-OUTPUT-STREAM implementation is much
429 faster for large outputs and conses 30% less on average.
430 * enhancement: XREF information is now collected for references made
431 to global variables using SYMBOL-VALUE and a constant argument.
432 * enhancement: SIGINT now causes a specific condition
433 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
434 * bug fix: FILE-POSITION used to signal an error in some cases where
435 ANSI requires it to return NIL.
436 * bug fix: ADJUST-ARRAY is now interrupt-safe.
437 * bug fix: adding and removing fd-handlers is now interrupt-safe.
438 * bug fix: inlined calls to C now ensure 16byte stack alignment on
440 * bug fix: bad type declaration in the CLOS implementation has
441 been fixed. (reported by James Anderson)
442 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
443 (thanks to Lutz Euler)
444 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
445 required by the CLHS. (reported by Samium Gromoff)
446 * bug fix: dead unbound variable references now signal an error.
447 (reported by Marco Monteiro)
448 * bug fix: / with an unused value was being deleted in safe code.
449 (thanks to Marco Monteiro and Kevin Reid)
450 * bug fix: number of characters that can be written onto a single
451 line in a file is unlimited.
452 * bug fix: some GC deadlocks caused by asynchronous interrupts have
453 been fixed by inhibiting interrupts for when GC is disabled.
454 * bug fix: some interrupt safety issues with GETHASH, (SETF
455 GETHASH), CLRHASH and REMHASH have been fixed.
456 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
457 specifier no longer causes infinite recursion.
458 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
459 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
460 * bug fix: an &environment argument with the correct variable information
461 is passed to macros that are expanded during byte compilation.
462 (reported by Samium Gromoff)
463 * bug fix: a logic error was causing the hash function for bignums to
464 have excessive amounts of collisions. (reported by Faré Rideau)
465 * bug fix: modifying the contents of an array could change the return
466 value of SXHASH on that array, which is only allowed for strings
467 and bit vectors. (bug introduced in 0.9.16)
468 * improvement: the x86-64/darwin port now passes all tests (except
469 for the debugger tests) but should still be considered
470 experimental until this is fixed.
471 * improvement: a style-warning is signaled for CASE (etc) clauses with
472 duplicate keys. (thanks to Kevin Reid)
473 * improvement: macos/x86 and macos/x86-64 now use mach exception
474 handlers for dealing with illegal instructions (for trapping and
475 error handling) and memory protection violations (for GC).
477 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
478 * new platform: experimental support for x86-64/darwin (MacOS).
479 * incompatible change: the thread-safe (on most platforms) getaddrinfo
480 and getnameinfo sockets functions are used instead of gethostbyaddr
481 and gethostbyname, on platforms where the newer functions are available.
482 As a result, the ALIASES field of HOST-ENT will always be NIL on these
484 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
485 don't print the startup banner, but behave as if --noinform was passed
486 as a command line argument. (thanks to Kevin Reid)
487 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
488 * optimization: code using alien values with undeclared types is much faster.
489 * optimization: the compiler is now able to open code SEARCH in more cases.
490 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
491 * bug fix: using standardized COMMON-LISP special variables as loop
492 variables no longer signals bogus package lock violations. (reported
494 * bug fix: declaring local loop variables to be of a range-limited type
495 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
496 (reported by Andras Simon)
497 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
498 bugs remain on x86-64.)
499 * bug fix: the #= and ## reader macros now interact reasonably with
500 funcallable instances.
501 * bug fix: type-checks for function arguments were compiled using the
502 compiler policy settings of the wrong lexical environment (in 1.0.2
504 * bug fix: SHADOW accepts characters as string designators, as required
505 by the spec (thanks to Eric Marsden)
506 * bug fix: fixed GC safety issues when foreign functions are called with
507 non-base strings as arguments
508 * bug fix: more consistent error output for fatal-compiler-warnings, like
510 * bug fix: functions with proper names were showing up as NIL in
511 backtraces (reported by Edi Weitz, regression in 1.0.2)
513 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
514 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
515 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
516 produce more efficient code on the x86 if the compiler can determine
517 that (+ INDEX <constant>) does not require a bounds check and FOO
518 has an element type at least 8 bits wide.
519 * bug fix: references to undefined variables are handled the same way
520 in toplevel forms as in the normal compiler
521 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
523 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
524 (thanks to Jon Buller)
525 * improvement: faster compilation times for complex functions
526 * improvement: added readlink support to SB-POSIX (thanks to Richard
529 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
530 * improvement: experimental support for mach exception handling on
531 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
533 * improvement: support for GBK external format.
534 (thanks to Chun Tian (binghe))
535 * improvement: the debugger now displays variables that have been closed
536 over, in code compiled with (DEBUG 2) or higher
537 * improvement: support for executable cores on NetBSD (thanks to
539 * new feature: added a RESTART-FRAME debugger command
540 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
541 be used to provide an implementation for FILE-POSITION on Gray streams
542 (thanks to Eric Marsden)
543 * new feature: add syslog, openlog and closelog support to SB-POSIX
544 (thanks to Richard Kreuter)
545 * optimization: the function call overhead in code compiled with
546 a high DEBUG optimization setting is significantly
547 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
548 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
549 (thanks to Tony Martinez)
550 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
551 evaluated once (reported by Kevin Reid)
552 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
553 works again (thanks to Kilian Sprotte)
554 * bug fix: an error is signaled for tagbodies with duplicate tags
555 (thanks to Stephen Wilson)
556 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
557 * bug fix: Win32 port can now handle foreign code unwinding Lisp
558 stack frames from alien callbacks.
559 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
560 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
561 65k pages (thanks to David Woodhouse)
562 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
564 changes in sbcl-1.0.1 relative to sbcl-1.0:
565 * new platform: FreeBSD/x86-64, including support for threading.
566 * new feature: the compiler stores cross-referencing information
567 about function calls (who-calls), macroexpansion (who-macroexpands)
568 and special variables (who-binds, who-sets, who-references) for code
569 compiled with (< SPACE 3). This information is available through the
570 sb-introspect contrib.
571 * new feature: users may subclass SEQUENCE, and have instances of
572 these classes interoperate with standard Common Lisp functions if
573 a number of methods are defined. (This feature is experimental
574 and the interface subject to change based on feedback from SBCL
575 users and the general community)
576 * improvement: sb-sprof traces call stacks to an arbitrary depth on
577 x86/x86-64, rather than the previous fixed depth of 8
578 * improvement: another pthread back-end of mutex "pthread-futex".
579 * bug fix: non-ascii command-line arguments are processed correctly
580 (thanks to Yaroslav Kavenchuk)
581 * bug fix: non-required arguments were not passed correctly when a method
582 defined using DEFMETHOD was called from a mop-generated method using
583 CALL-NEXT-METHOD (reported by Pascal Costanza)
584 * bug fix: recursion is now permitted in accessors through
585 SLOT-UNBOUND. (reported by Pascal Costanza)
586 * bug fix: an error was signaled at startup if the HOME environment
587 variable was defined, but had an empty value (reported by Peter Van Eynde)
588 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
589 signal a PROGRAM-ERROR, not a BUG.
590 * bug fix: SB-POSIX:READDIR works when built with large file support.
591 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
593 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
594 for better automatic indentation support. (thanks to Matt Pillsbury)
595 * optimization: loading generic functions no longer takes O(n^2) time,
596 proportional to the amount of methods in the generic function
597 (reported by Todd Sabin and Jeremy Brown)
598 * optimization: the FIND and POSITION family of sequence functions
599 are significantly faster on arrays whose element types have been
601 * improvements to the Windows port:
602 ** Intermittent heap corruption problems have been fixed. (thanks
603 to Alastair Bridgewater)
604 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
606 ** Lisp is able to unwind foreign exception frames from alien
607 callbacks. (thanks to Alastair Bridgewater)
609 changes in sbcl-1.0 relative to sbcl-0.9.18:
610 * improvement: experimental support for threading on FreeBSD/x86.
611 (thanks to NIIMI Satoshi)
612 * improvement: runtime option --dynamic-space-size can be used
613 to set the size of the dynamic space reserved on startup.
614 * improvement: floating point modes in effect are now saved in
615 core, and restored on startup.
616 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
617 startup, not time since first call to GET-INTERNAL-REAL-TIME.
618 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
619 threads are not running after *SAVE-HOOKS* have run.
620 * improvement: writes to CLOS instance slots are type-checked in code
621 compiled with (SAFETY 3)
622 * improvement: floating-point exception handling on FreeBSD (thanks to
624 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
625 (thanks to Zach Beane)
626 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
628 * improvement: added support for the Shift-JIS external format.
629 (contributed by NIIMI Satoshi)
630 * improvement: callbacks are supported on Linux/PPC. (thanks to
632 * bug fix: compiler bug triggered by a (non-standard) VALUES
633 declaration in a LET* was fixed. (reported by Kaersten Poeck)
634 * bug fix: file compiler no longer confuses validated and already
635 dumped structurres. (reported by Kaersten Poeck)
636 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
637 fill-pointer signals a type-error as required. (thanks to
639 * bug fix: disassemly of funcallable instances works.
640 * bug fix: single stepping on PPC.
641 * bug fix: fix thread-safety problems in the type system (generally
642 manifesting as nonsensical errors like "STRING is a bad type specifier
643 for sequences" or "The value 1 is not of type FIXNUM")
644 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
645 (reported by Josip Gracin)
646 * bug fix: an error is signaled for attempts to displace arrays with
647 incompatible element types (thanks to Mario Mommer)
648 * bug fix: more correct handling of wide characters in debug info
649 (bug reported by Attila Lendvai and fixed by Juho Snellman)
650 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
651 and don't cause extra consing
652 * optimization: MAP and MAP-INTO are significantly faster on vectors
653 whose elements types have been declared.
654 * Improvements to SB-SPROF:
655 ** Support for allocation profiling
656 ** Reduced profiling overhead, especially for long profiling runs
657 * Improvements to the Windows port:
658 ** floating point exceptions are now reported correctly.
659 ** stack exhaustion detection works partially.
660 ** more accurate GET-INTERNAL-REAL-TIME.
661 ** better RUN-PROGRAM behaviour in case of errors in the child
663 ** PROBE-FILE now simplifies pathnames correctly.
664 ** DIRECTORY now works correctly with :WILD-INFERIORS.
665 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
666 1st 1970 (although time timezone data will be unreliable).
667 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
669 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
670 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
671 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
673 * bug fix: two potential GC deadlocks affecting threaded builds.
674 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
675 returns T (reported by Anton Kazennikov)
676 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
677 with non-variable places
678 * bug fix: remove a race condition in the setting of
679 funcallable-instance functions, this should make threaded CLOS
680 code more stable against memory faults.
681 * bug fix: corruption of specials when unbinding is interrupted by an
682 asynchronous unwind (reported by Hannu Koivisto)
683 * improvement: the debugger will now also display local variables that
684 are only used once, for code compiled with a DEBUG optimization quality
687 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
688 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
689 * incompatible change: External-format support for FFI calls. The
690 SB-ALIEN:C-STRING no longer implies an ASCII
691 external-format. Instead the string is subject to external-format
692 conversion. Additionally return values of type SB-ALIEN:C-STRING
693 are by default of type (SIMPLE-ARRAY CHARACTER), not
694 SIMPLE-BASE-STRING. For an alien type that behaves like the old
695 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
696 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
697 * incompatible change: SB-EXT package no longer contains the
698 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
699 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
701 * incompatible change: the single-stepper is no longer available
702 on Alpha, Mips and Sparc platforms.
703 * minor incompatible change: the direct superclasses of
704 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
705 not (STANDARD-OBJECT FUNCTION). This makes the
706 class-precedence-lists of GENERIC-FUNCTION and
707 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
709 * minor incompatible change: the default stream external format on
710 non-unicode SBCL is no longer detected from the locale, but is always
712 * new feature: Add a version of evaluator that uses an interpreter instead
713 of the compiler. EVAL still uses the compiler by default, to switch it
714 to use the interpreter, set the value of the variable
715 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
716 * minor incompatible change: the single-stepper REPL has been merged
717 with the normal debugger (see the "Stepping" heading of the debugger help
718 for more details). The debugger command STEP will no longer switch
719 to the single-stepper REPL.
720 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
721 for a type now works.
722 * bug fix: loading of default sysinit file works. (thanks to Leonid
724 * bug fix: better detection of circularities in the file-compiler.
725 (reported by Marco Monteiro)
726 * bug fix: the CL pathname functions now work with files that have
727 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
728 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
729 systems with Unix98 pty semantics.
730 * bug fix: ASDF-INSTALL will now work with bsd tar.
731 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
733 * bug fix: timers expiring in dead threads no longer cause a
734 type-error (reported by Paul "Nonny Mouse").
735 * bug fix: thanks to more lightweight single-stepper instrumentation,
736 code compiled with (DEBUG 3) will compile and execute significantly faster,
737 and will have more accurate type-inferencing than before
738 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
739 whose bindings are modified
740 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
741 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
742 ** replace dummy implementations of CL:MACHINE-INSTANCE and
743 CL:SOFTWARE-VERSION with proper version
745 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
746 * feature: implemented the READER-METHOD-CLASS and
747 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
748 as specified by AMOP.
749 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
751 * optimization: faster LOGCOUNT implementation on x86 and x86-64
752 (thanks to Lutz Euler)
753 * optimization: hashing of general arrays and vectors has been
754 improved. (reported by Any Fingerhut)
755 * enhancement: SB-INTROSPECT is now able to find definitions of
756 profiled functions. (thanks to Troels Henriksen)
757 * enhancement: compiler-macro expansion applies now to FUNCALL forms
758 as well, allowing compiler-macros for SETF-functions to expand.
759 * enhancement: step-instrumentation no longer wraps known
760 single-value functions with multiple-value context, allowing
761 better type inference.
762 * fixed bug #337: use of MAKE-METHOD in method combination now works
763 even in the presence of user-defined method classes. (reported by
764 Bruno Haible and Pascal Costanza)
765 * fixed bug #339(c): if there are applicable methods not part of any
766 long-form method-combination group, call INVALID-METHOD-ERROR.
767 (reported by Bruno Haible)
768 * fixed bug #361: the :FUNCTION initarg in the protocol for
769 initialization of methods can now be used to override
770 internally-produced optimized functions. (reported by Bruno
772 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
773 system-provided lambda expression no longer cause warnings about
774 unbound #:|pv-table| symbols.
775 * bug fix: improved the handling of type declarations and the
776 detection of violations for keyword arguments with non-constant
778 * bug fix: potentially erronous calls to PATHNAME and
779 MERGE-PATHNAMES were being flushed in some cases.
780 (reported by Richard Kreuter)
781 * bug fix: compiled calls to TYPEP were mishandling obsolete
782 instances. (reported by James Bielman and Attila Lendvai)
783 * bug fix: format strings with ~> without matching ~< no longer
784 trigger an AVER, but signal an understandable error instead.
785 (reported by Antonio Martinez)
786 * bug fix: specifying an output-file without a pathname-type for
787 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
788 of input-file instead of "fasl". (reported by Robert Dodier)
789 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
790 (reported by James Y Knight).
791 * bug fix: compiler-macros-function did not consider the environment
792 argument for shadowing by local functions.
793 * bug fix: compiler-macros expansion was inhibited by local INLINE
795 * bug fix: inline expansions of known functions were subject to
796 step-instrumentation in high debug policies, leading to problems
798 * bug fix: compiler failed to differentiate between different CONS
800 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
801 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
802 * bug fix: #368: incorrect use of expressed vs. upgraded array
804 * bug fix: #306a: more precise unions of array types.
805 * thread-safety improvements:
806 ** CONDITION-WAIT could return early on Linux, if the thread was
807 interrupted and subsequently continued with SIGCONT.
808 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
810 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
811 * added support for the ucs-2 external format. (contributed by Ivan
813 * minor incompatible change: pretty printing of objects of type
814 (cons symbol) is, in the default pprint-dispatch-table, now
815 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
817 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
818 called later in a class's lifetime, possibly as late as when the
819 first instance of the class is created. Previously,
820 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
821 class became finalizeable.
822 * fixed bug: FILE-POSITION sometimes returned inconsistent results
823 for multibyte external-format streams. (thanks to "vbzoli")
824 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
825 with :ALLOCATION :CLASS inherited from superclasses of the
827 * fixed bug: anonymous classes can now be created using the :NAME
828 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
829 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
830 * fixed bug: core-files saved with :EXECUTABLE T can again be
831 executed when SBCL_HOME isn't set. (reported by James Knight)
832 * fixed bug: toplevel LOCALLY forms with declarations could
833 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
834 * fixed bug: printing from several different threads using different
835 values of *print-case* could cause invalid output, due to
836 some internal special variables of the printer not being bound
837 thread-locally (reported by Max Mikhanosha)
838 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
839 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
840 (reported by Pascal Costanza)
841 * fixed bug: FINALIZE-INHERITANCE is called from
842 REINITIALIZE-INSTANCE on classes when the class has previously
843 been finalized, as required by AMOP.
844 * minor code generation optimizations:
845 ** better register allocation in CLOS dispatching functions
846 ** overflow detection when coercing signed bytes to fixnums on x86-64
847 ** is now implemented with one IMUL instruction instead of three shifts
848 ** more efficient bit-vector access on x86 and x86-64
849 ** more efficient access to raw structure slots on x86 and x86-64
850 * fixed some bugs revealed by Paul Dietz' test suite:
851 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
854 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
855 * feature: thread support on Solaris/x86, and experimental thread support
857 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
859 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
860 default method for SLOT-UNBOUND.
861 * minor incompatible change: prevent the user from specializing the
862 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
863 somewhat counter to the intent of the protocol, I (CSR) think, and
864 additionally it just doesn't work in SBCL as currently
865 implemented, thanks to optimizations (that are always valid for
866 the other three SLOT-VALUEish functions, but not for the setter).
867 * minor incompatible change: the :SB-LDB feature is now enabled by
868 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
869 the low-level debugger.
870 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
871 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
872 down to an SB-SYS:FD-STREAM, allowing direct communication in
873 more cases. Temporary files and pipes are still used as a fallback
875 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
876 on the non-reentrant C functions localtime(3) and gmtime(3).
877 * bug fix: global optimization policy was not visible in LOCALLY and
879 * bug fix: class objects can be used as specializers in methods.
880 (reported by Pascal Costanza)
881 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
882 with a type error. (reported by blitz_ on #lisp)
883 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
884 failed with a type error. (reported by Pascal Bourguignon)
885 * bug fix: merging pathnames against defaults with :DIRECTORY
886 starting with '(:RELATIVE :BACK) should preserve the :BACK.
887 (reported by James Y Knight)
888 * bug fix: saving large (>2GB) cores on x86-64 now works
889 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
890 constant index and a (simple-array (signed-byte 32)) array
891 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
892 error (patch by Robert J. Macomber)
893 * feature: TIME now displays the amount of run-time spent in GC
894 * bug fix: The debugger now does a better job of respecting
895 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
896 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
897 workaround for bug 403.)
898 * bug fix: floating point exception handling now works on FreeBSD
899 (thanks to NIIMI Satoshi)
900 * fixed some bugs revealed by Paul Dietz' test suite:
901 ** MISC.641: LET-conversion were not supposed to work in late
903 * improvements to the Win32/x86 port:
904 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
905 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
907 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
908 * new feature: source path information is generated for macro-expansion
909 errors for use in IDE's like Slime (thanks to Helmut Eller)
910 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
911 * bug fix: compiler does not loop forever on an invalid type in
913 * improvement: compilation of most CLOS applications is significantly
915 * optimization: added a limited bytecode compiler for simple toplevel
916 forms, speeding up compilation and FASL loading
917 * bug fix: the statistical profiler now properly distinguishes anonymous
920 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
921 * minor incompatible change: in sbcl-0.9.11 (but not earlier
922 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
923 into subclasses of the finalized class. Now user calls to
924 FINALIZE-INHERITANCE finalize just the one class, and calls by the
925 system return before any subclasses are finalized.
926 * minor incompatible change: The reader no longer ignores errors
927 regarding non-existent packages in #+ and #- feature tests.
928 * new feature: command line options --no-sysinit, --no-userinit to
929 inhibit loading the corresponding init files
930 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
931 for sending data through UDP sockets (thanks to François-René Rideau)
932 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
933 error is signalled instead (thanks to François-René Rideau)
934 * minor incompatible change: Error signalling behaviour of lexical
935 operations violating package locks has changed slightly. Refer to
936 documentation on package locks for details.
937 * enhancement: EVAL can process IF-expressions without resorting to the
939 * bug fix: Some operations involving SIN, COS, or TAN weren't being
940 constant-folded properly due to missing out-of-line functions.
941 (reported by Mika Pihlajamäki)
942 * bug fix: LISTEN sometimes returned T even in cases where no data was
943 immediately available from the stream
944 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
945 were reversed. (reported by Levente Mészáros)
946 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
947 when given a vector displaced to an adjustable array. (reported
949 * bug fix: derivation of float boundaries from numbers outside the
950 appropriate float range (reported by John Wiseman)
951 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
953 * bug fix: precomputing cache entries for generic functions with
954 some subclasses of specializers as yet invalid does not attempt to
955 fill a cache line with a negative offset. (reported by Levente
957 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
958 allow condition class objects as arguments to DOCUMENTATION and
959 (SETF DOCUMENTATION); only find and set documentation for
960 structure names for the STRUCTURE doc-type. (suggested by Gary
962 * improvements to the Win32/x86 port:
963 ** (user-homedir-pathname) and default initialization file
964 locations now know about the user's "Documents and Settings"
965 directory (thanks to Yaroslav Kavenchuk)
966 ** run-program is implemented (thanks to Mike Thomas)
967 ** sockets support (thanks to Timothy Ritchey)
968 ** better backtrace support (thanks to Alastair Bridgewater)
969 ** sb-grovel supported
970 ** asdf-install and sb-posix work somewhat
971 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
972 * improvements to the Solaris/x86 port:
973 ** works on Solaris 11/Solaris Express
974 ** floating-point exception handling support
975 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
976 * fixed some bugs revealed by Paul Dietz' test suite:
977 ** REMOVE-METHOD returns its generic function argument even when
978 the method is not one of the generic functions' methods.
979 ** objects declared as MEMBER types can be admissible arguments to
981 ** printing characters should simply be printed by the FORMAT ~:C
983 ** compiler failure when compiling functions with hairy constant
984 defaults for optional parameters.
985 ** compiler produces wrong code when MAYBE-INLINE-expanding a
986 function, which is already optimized.
988 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
989 * new platform: experimental support for SBCL x86/Darwin, including
990 MacOS X 10.4.5 on Intel.
991 * new feature: Unicode character names are now known to the system
992 (through CHAR-NAME and NAME-CHAR).
993 * new feature: the filesystem location of SBCL's core file is
994 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
995 * minor incompatible change: the contrib modules SB-POSIX and
996 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
997 this change is to make it easier to distribute
998 location-independent binaries.
999 * bug fix: as implied by AMOP, standardized classes no longer have
1000 slots named by external symbols of public packages. (reported by
1002 * optimization: calling functions via constant symbols -- as in
1003 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1004 via the function object as in (FUNCALL #'FOO).
1005 * optimization: on x86 and x86-64, the calling convention for the
1006 case of dealing with an unknown number of values has been altered
1007 to be friendlier to the prediction heuristics implemented,
1008 particularly with reference to CALL and RET pairing. (thanks to
1009 Alastair Bridgewater)
1010 * optimization: on x86, the code for access to thread-local symbol
1011 values and binding/unbinding of thread-local symbols is smaller.
1012 (thanks to Alastair Bridgewater)
1013 * enhancement: CONSTANTP is now able to determine constantness of
1014 more complex forms, including calls to constant-foldable standardized
1015 functions and some special forms beyond QUOTE.
1016 * fixed bug: occasional GC crashes on Solaris/x86
1017 * optimization: x86-64 supports stack allocation of results of simple
1018 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1019 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1020 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1021 (thanks to James Knight)
1022 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1023 by the standard (thanks to Stephen Compall)
1025 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1026 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1027 be used for bundling the runtime and the core file into one
1028 executable binary. This feature is not currently supported on all SBCL
1029 platforms. (thanks to James Bielman and NIIMI Satoshi)
1030 * new feature: a generational or ephemeral garbage collector is now
1031 the default on the PowerPC platform (both Linux and Darwin). The
1032 old Cheney (stop and copy) collector is a build-time option.
1033 (thanks to Cyrus Harmon, after Raymond Toy)
1034 * minor incompatible change: the method by which SBCL finds its
1035 contributed modules has changed; it no longer relies on symbolic
1036 links from an $SBCL_HOME/systems directory, but searches directly
1037 in the subdirectories of $SBCL_HOME.
1038 * enhancement: the dynamic heap size on the Linux/PPC platform is
1039 markedly larger, even using the older Cheney garbage collector.
1040 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1041 certain complicated string types. (reported by Gary King)
1042 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1043 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1044 * fixed bug: DOCUMENTATION on structures no longer signals an
1045 error if no structure type of the right name exists. (reported by
1047 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1048 applicable non-standard primary or :AROUND method on
1049 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1050 no longer cause unbound variable errors. (reported by Kevin Reid)
1051 * fixed bug: metacircle resolution in cases where methods have slots
1052 added before the slots from STANDARD-METHOD. (reported by Jean
1054 * fixed bug: the Power PC assembly code for calling into lisp
1055 sometimes computed its return address wrongly (depending on the
1056 15th bit of the address, fixed at link time). This used to
1057 manifest itself in a segmentation violation while building PCL.
1058 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1059 many others over the years)
1060 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1061 the stack pointer (treated as a signed integer) is negative.
1062 (thanks to Peter van Eynde)
1064 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1065 * new platform: experimental support for the Windows operating
1066 system has been added. (thanks to Alastair Bridgewater)
1067 * fixed several bugs in and robustified the PPC FFI (including
1068 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1069 * bug fix: allow non-simple string symbol names (reported by Paul
1071 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1073 * bug fix: saving a core corrupted callbacks on x86/x86-64
1074 * bug fix: closed a loophole in metacircularity detection and
1075 grounding in the PCL implementation of CLOS.
1076 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1077 (reported by tomppa on #lisp)
1078 * bug fix: clear the direction flag on Lisp -> C transitions, as
1079 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1080 (reported by Andrej Grozin and Hendrik Maryns)
1081 * optimization: major improvements to GC efficiency on GENCGC platforms
1082 * optimization: faster implementation of EQUAL
1083 * optimization: emit more efficient opcodes for some common
1084 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1085 * optimization: several other minor code-generation improvements on x86-64
1086 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1088 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1089 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1090 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1091 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1092 by ANSI as a generic function, and for consistency (SETF
1093 GENERIC-FUNCTION-NAME) is treated likewise.
1094 * fixed bug #233.b: make constraint propagation notice when a variable
1095 value is changed after it is referenced but before it is used
1096 * fixed bug #296: no more arbitrary behaviour from filenames with
1097 odd characters as --load arguments. (reported by Adam Warner)
1098 * fixed bug #390: :CHARACTER-SET pathname components now work as
1099 expected. (reported by Tim Daly Jr)
1100 * fixed bug #391: complicated :TYPE intersections in slot
1101 definitions no longer cause an error in PCL internals.
1102 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1103 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1104 returning the number of octets which would be written to the
1105 file-stream. (thanks to Robert J. Macomber)
1106 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1107 arguments to the internal stream functions. (thanks to David
1109 * bug fix: stack exhaustion is now handled on the x86/SunOS
1110 platform. (thanks to Daisuke Homma)
1111 * bug fix: more accurate ROOM results on GENCGC platforms
1112 * optimization: improved type inference for arithmetic-for
1113 index variables in LOOP
1114 * optimization: faster floating-point SQRT on x86-64
1115 * fixed some bugs revealed by Paul Dietz' test suite:
1116 ** DOCUMENTATION returns NIL instead of "" for method combinations
1117 that don't have a docstring
1119 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1120 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1121 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1122 therefore are not customizeable by user code (as seems to be at
1123 least permitted and maybe required by AMOP). As a consolation,
1124 however, the SBCL implementation of these functions now calls
1125 REINITIALIZE-INSTANCE as specified by AMOP.
1126 * bug fix: slot-definition documentation is propagated as per ANSI
1127 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1128 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1129 argument, as do eslotd initialization methods. (from Pascal
1130 Costanza's "Closer" project)
1131 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1132 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1134 * bug fix: it is now possible to have more than one subclass of
1135 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1136 (reported by Bruno Haible, Pascal Costanza and others)
1137 * bug fix: the dependent update protocol now works for generic
1138 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1139 and Pascal Costanza)
1140 * bug fix: condition-class instances corresponding to
1141 DEFINE-CONDITION forms are now created eagerly. (reported by
1142 Kalle Olavi Niemitalo on comp.lang.lisp)
1143 * bug fix: floating point printing is more accurate in some
1144 circumstances. (thanks to Simon Alexander)
1145 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1146 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1147 * bug fix: callbacks on OS X now preserve stack-alignment.
1148 (thanks to Cyrus Harmon)
1149 * enhancement: the x86-64 disassembler is much better at
1150 disassembling SSE instructions. (thanks to Lutz Euler)
1151 * enhancement: CHECK-TYPE on a variable now allows type-inference
1152 (thanks to Vincent Arkesteijn)
1153 * optimization: improved performance of EUC-JP external format.
1154 (thanks to NIIMI Satoshi)
1155 * optimization: performance improvements to IO on file streams of
1156 :ELEMENT-TYPE CHARACTER
1157 * optimization: much faster memory allocation on x86-64
1158 * optimization: faster garbage collections (latency and throughput)
1160 * optimization: faster bignum subtraction and division on x86 and x86-64
1161 (thanks to Lutz Euler)
1162 * optimization: more accurate type-inference for loops that use a
1163 floating point index variable or a negative step.
1165 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1166 * bug fix: add a workaround to SBCL looping infinitely at startup on
1167 Linux kernels with apparently buggy implementations of personality().
1168 (thanks to Svein Ove Aas)
1169 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1170 (thanks to David Lichteblau)
1171 * bug fix: local bindings shadow global symbol macros for
1172 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1173 * optimization: non-open coded uses of numeric comparison operators
1174 (e.g. >) no longer cons when called with more than one parameter
1175 on platforms supporting dynamic-extent allocation.
1176 * enhancement: saving cores with foreign code loaded is now supported
1177 on MIPS/Linux in addition to the previously supported platforms.
1178 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1179 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1180 * bug fix: division by zero in sb-sprof when no samples were collected
1181 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1182 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1183 (thanks to Alexander Kjeldaas)
1184 * new feature: ignoring the return values of destructive functions
1185 like NREVERSE signals a compile-time style-warning.
1186 (thanks to Kevin Reid)
1187 * new feature: the alignment of alien structure fields can be
1188 explicitly specified. (thanks to Cyrus Harmon)
1189 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1190 and dump core on SIGQUIT
1192 ** incompatible change: threads do not inherit values of specials
1193 from their parents (see manual)
1194 ** bug fix: threads stacks belonging to dead threads are freed by the
1195 next exiting thread, no need to gc to collect thread stacks anymore
1196 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1197 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1198 ** bug fix: dynamic variable and thread start related gc lossage
1199 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1201 * fixed some bugs revealed by Paul Dietz' test suite:
1202 ** SUBTYPEP is slightly more accurate on heinously complicated
1203 CONS types where some of the members have uncertain (in the
1204 NIL, NIL sense) type relationships to each other.
1206 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1207 no longer purified unless :PURIFY T is explicitly specified.
1208 ** Non-purified cores are significantly smaller than before
1210 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1211 * new feature: timers based on Zach Beane's excellent timer package
1212 * added support for the following external formats: koi8-u,
1213 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1214 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1215 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1216 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1217 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1218 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1219 * incompatible change: a threaded SBCL will no longer revert to
1220 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1221 * bug fix: interrupts are disabled until startup is complete; no
1222 more sigsegvs when receiving a signal to soon
1223 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1225 * bug fix: add a workaround for the memory randomization features in
1226 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1227 workaround will only be in effect on systems with the proc filesystem
1229 * bug fix: printing objects of type HASH-TABLE signals a
1230 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1232 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1233 have been running for over 50 days. (reported by Gilbert Baumann)
1234 * bug fix: the logic for getting names of functions gets less
1235 confused when confronded with alternate-metaclass
1236 funcallable-instances. (reported by Cyrus Harmon)
1237 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1238 consistent, even on internal alternate-metaclass objects.
1239 * bug fix: flush closure information collected by physical
1240 environment analysis prepass before the main pass. (bug reported
1242 * bug fix: compiler pack phase does not modify a hashtable, which is
1243 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1244 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1245 may be expanded. (reported by Alexander <malishev> on c.l.l)
1246 * bug fix: no more style-warnings when compiling code using LAST
1247 under high-SPEED low-SPACE settings. (reported by David Wallin
1249 * fixed some incompatibilities between SBCL's MOP and the MOP
1251 ** the METAOBJECT class is now implemented;
1252 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1253 STANDARD-OBJECT, as required;
1254 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1255 now compatible (as required by VALIDATE-SUPERCLASS); there
1256 remains a consistency requirement over the presence or absence
1257 of the FUNCTION class in the superclass list of finalized
1258 classes; see the manual for more details;
1259 ** the :AROUND method for COMPUTE-SLOTS on
1260 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1261 requested slot ordering.
1263 ** bug fix: parent thread now can be gc'ed even with a live
1265 ** bug fix: binding a special with PROGV to no value is not
1266 the same as that symbol not having been bound (thanks to
1268 ** bug fix: binding specials is thread safe (thanks to
1270 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1271 and SB-SPROF do not die with 'no handler for signal XX in
1272 interrupt_handle_now(..)' anymore
1273 ** bug fix: WITH-TIMEOUT works with multiple running threads
1274 * fixed some bugs revealed by Paul Dietz' test suite:
1275 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1276 the :method-class keyword argument.
1278 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1279 * new port: the Solaris operating system on x86 processors is now
1280 mostly supported, though some rough edges in the environment
1281 remain. (thanks to Daisuke Homma)
1282 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1283 space for its heap. (thanks to Thiemo Seufer)
1284 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1285 closures. (thanks to Thiemo Seufer)
1286 * minor incompatible change: eof selects abort in the debugger.
1287 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1288 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1290 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1291 deprecated, and will go away in a future revision of SBCL.
1292 * minor incompatible change: GC-ON and GC-OFF are no longer
1293 implemented with a counter, it does not matter how many times gc
1294 is switched on or off
1295 * bug fix: discriminating functions for generic function classes
1296 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1297 make invalid assumptions about method precedence order. (reported
1299 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1300 platforms. (thanks to Thiemo Seufer)
1301 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1302 * bug fix: correct pathname printing with printer escaping is on.
1303 (thanks to Kevin Reid)
1304 * bug fix: complex VOP definitions in "user-space" no longer trigger
1305 package locks. (reported by Zach Beane)
1306 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1307 for funcallable-instance functions, and is no different from
1309 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1310 characters in the current readtable. (reported by Nicholas Neuss)
1311 * bug fix: SUBTYPEP on various CONS types returns more a more
1312 accurate acknowledgment of its certainty.
1313 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1314 lists in some cases. This partially fixes bug 384.
1315 * flush all standard streams before prompting in the REPL and the
1317 * bug fix: signal handling and triggering gc do not conflict
1318 directly anymore, in particular a high frequency sb-sprof does
1319 not prevent gc from running
1320 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1321 approximation for timezone and DST information between the
1322 universal time epoch and the smallest negative 32-bit time_t.
1323 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1324 year 1899 so long as the encoded time is non-negative.
1325 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1326 the correct answer for non-constant rotate amounts when performing
1327 an inline 32-bit rotation.
1329 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1330 there is only one thread in the session
1331 ** bug fix: memory leak for streams created in one thread and
1332 written to in another
1333 ** bug fix: lockup when compiled with gcc4
1334 ** bug fix: race that allows the gc to be triggered when gc is
1336 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1337 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1339 ** bug fix: finished threads are gc'ed properly
1340 * fixed some bugs revealed by Paul Dietz' test suite:
1341 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1342 has arguments with a different set of applicable methods from
1343 the orignal arguments.
1344 ** The type error thrown by MAP now has a correct expected-type
1346 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1347 name a compiled function.
1348 ** Three MISC tests where a large bignum was improperly coerced to
1349 a float that couldn't represent that bignum during type
1350 derivation were fixed.
1351 ** SUBTYPEP can now handle types involving the negation of a
1352 list-form FUNCTION type.
1353 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1354 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1355 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1357 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1358 * New feature: Experimental support for bivalent streams: streams
1359 opened with :element-type :default now allow character and binary
1360 (unsigned-byte 8) I/O
1361 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1362 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1364 * bug fix: no more highly sporadic "couldn't check whether ~S is
1365 readable" when reading a stream and an interrupt hits in the middle
1366 of a select system call
1367 * compiler better recognizes complex arrays (reported by Eduardo
1369 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1370 for variables on SBCL built with linkage-tables. (reported by Luis
1372 * various error reporting improvements.
1373 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1374 (Thanks to Thiemo Seufer)
1375 * bug fix: nested reader invokations work correctly
1376 * bug fix: it is possible to have simultaneous references to foreign
1377 code and foreign data with the same name.
1379 ** added x86-64 support
1380 ** incompatible change: the threading api now works with thread
1381 objects instead of thread ids
1382 ** bug fix: threads are protected from signals and interruption when
1383 starting up or going down
1384 ** bug fix: a race where an exiting thread could lose its stack to gc
1385 ** bug fix: don't halt on infinite error in threads if possible
1386 ** fixed numerous gc deadlocks introduced in the pthread merge
1387 ** bug fix: fixed thread safety issues in read and print
1388 ** bug fix: debugger doesn't hang on session lock if interrupted at
1389 an inappropriate moment
1390 ** bug fix: run-program is now thread safe(r)
1391 ** bug fix: inner with-recursive-lock no longer releases the mutex
1392 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1393 * fixed some bugs revealed by Paul Dietz' test suite:
1394 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1395 EXPECTED-TYPE slots filled.
1396 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1398 ** MAKE-CONDITION accepts classes as type-designators.
1399 ** COMPILE may never return NIL.
1400 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1401 range before calling Unix time functions
1403 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1404 * numerous signal handling fixes to increase stability
1405 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1406 * minor incompatible change: we now correctly canonize default
1407 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1408 as per the MOP, rather than the historical (INITARG INITFUNCTION
1409 INITFORM). (reported by Bruno Haible)
1410 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1411 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1412 for more information.
1413 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1414 pathname is a directory pathname.
1415 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1416 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1418 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1419 :METACLASS argument in addition to a class name. (reported by
1420 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1421 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1422 (reported by Bruno Haible; patch from Gerd Moellmann)
1423 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1425 * bug fix: more cleanups to the floating point exception handling on
1426 x86-64 (thanks to James Knight)
1427 * bug fix: the compiler does not try to constant fold calls of
1428 COERCE and BIT-* functions when they return freshly constructed
1429 objects. (reported by Nikodemus Siivola and Paul Dietz)
1430 * optimization: improved the MIPS versions of generic (in the
1431 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1432 * optimization: direct conversion of (unsigned-byte 32) to floats on
1433 the PowerPC platform.
1434 * optimization: structure instances with raw slots now use less
1435 memory, and probably show better memory locality. (thanks to
1437 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1438 are treated as requests for stack allocation on the x86-64,
1439 Alpha-32, PPC and SPARC platforms.
1440 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1441 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1442 (thanks to Tony Martinez)
1444 ** gcing a dead thread can no longer lead to lockups
1445 ** threads block signals until they are set up properly
1446 ** errno is no longer shared by threads
1447 ** interrupt-thread restores the eflags register on x86
1448 ** fixed some lockups due to gc/thread interaction
1449 * fixed some bugs revealed by Paul Dietz' test suite:
1450 ** invalid dotted lists no longer raise a read error when
1451 *READ-SUPPRESS* is T
1452 ** signal an error if a symbol that names a declaration is used
1453 as the name of a type, or vice versa
1454 ** allow using the (declare (typespec var*)) abbreviation for
1455 (declare (type typespec var*)) with all type specifiers
1456 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1457 ** always return NIL from PPRINT-POP when OBJECT is NIL
1458 ** don't signal errors when pretty-printing malformed LABELS,
1459 FLET or MACROLET forms
1460 ** declarations in a DOLIST body are also in scope for the
1462 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1464 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1467 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1468 * fixed cross-compiler leakages that prevented building a 32-bit
1469 target with a 64-bit host compiler.
1470 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1471 opened with :IF-EXISTS :OVERWRITE.
1472 * fixed bug 281: error for an invalid qualifier in a short-form method
1473 combination method is not signalled until the faulty method is called.
1474 * bug fix: iteration variable type inferrer failed to deal with open
1475 intervals. (reported by Alan Shields)
1476 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1477 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1478 or more runtime options were provided to the sbcl binary.
1479 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1480 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1481 * optimization: there's now a fast-path for fixnum arguments in the
1482 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1483 * optimization: the code generated on x86-64 is more compact thanks
1484 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1485 * bug fix: floating-point exception handling now partly works on
1486 x86-64. (Thanks to James Knight)
1487 * improvement to the MIPS backend from Thiemo Seufer: C-style
1488 64-bit long long arguments and return values to alien functions
1490 * fixed some bugs revealed by Paul Dietz' test suite:
1491 ** the type-error signalled from WARN has a filled-in DATUM slot.
1492 ** the type-error required when a stream is not associated with
1493 a file has the stream as its datum.
1494 ** type-errors on single-floats on x86-64 no longer have
1495 :INVALID-OBJECT as the datum
1496 ** the type-errors signalled for invalid function names now have
1497 a correct expected type
1498 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1499 for structures defined with a :TYPE. Documentation strings for
1500 typed structures are no longer immediately discarded
1501 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1502 broadcast streams and synonym streams. FILE-LENGTH now also works
1503 on broadcast streams.
1505 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1506 * incompatible change: the --noprogrammer option, deprecated since
1507 version 0.7.5, has been removed. Please use the equivalent
1508 --disable-debugger option instead.
1509 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1511 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1512 inoperational for a while now) has been completely removed.
1513 * null lexical environments are now printed as #<NULL-LEXENV>,
1514 significantly reducing the amount of clutter in typical backtraces.
1515 * documentation on weak pointers, finalization, and after GC hooks
1516 has been added to the manual.
1517 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1518 as other specialized array types, is much faster. SUBSEQ and
1519 COPY-SEQ on such arrays have also been sped up.
1520 * optimization: EQL is now more efficient when at least other argument
1521 is known to be of type (OR FIXNUM (NOT NUMBER)).
1522 * fixed bug: the runtime is now less vulnerable to changes in the
1523 size of the SBCL object on OS X, and virtual memory is reserved for
1524 all spaces that need to be at a fixed address.
1525 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1526 * fixed bug: finalizers and after GC hooks that cause consing are now
1528 * fixed bug: compiler error messages and summaries are now printed to
1529 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1530 * fixed inference of the upper bound of an iteration variable.
1531 (reported by Rajat Datta).
1532 * fixed bug 211e: calling local functions with duplicated constant
1533 keyword argument no longer causes a bogus style warning about an
1535 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1536 ftype declaration to be disregarded. (reported by Dave Roberts)
1537 * fixed bug 373: caused by erronous compilation of references to alien
1538 variables in the runtime on ppc/darwin.
1539 * fixed bug 376: CONJUGATE type deriver.
1540 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1541 (reported by Baughn on #lisp)
1542 * a cross-compiler bug on non-x86 platforms has been identified and
1543 fixed. (thanks to Bruno Haible)
1544 * improvements to the MIPS runtime code for increased stability.
1545 (thanks to Thiemo Seufer)
1546 * increased the maximimum compact environment size to allow
1547 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1548 * partial workaround for bug 135: don't name the function we're
1549 calling for hairy cases of EVAL, so as not to accumulate one environment
1550 entry per call to EVAL. (reported by Kevin Reid)
1551 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1552 * optimization: type testing for non-vector arrays should be faster.
1553 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1555 * bug fix: setting 31st element of a bit vector to zero did not work
1557 * bug fix: redefining a class definition which failed due to a
1558 previous accessor / function clash now works (but see BUGS entry
1559 #380 for more problems in this area). (thanks to Zach Beane)
1560 * the long form of DEFINE-METHOD-COMBINATION disables method group
1561 checking when given a single method group with pattern *. (thanks
1563 * on x86 compiler supports stack allocation of results of simple
1564 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1565 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1566 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1567 * fixed some bugs related to Unicode integration:
1568 ** the restarts for recovering from input and output encoding
1569 errors only appear when there is in fact such an error to
1571 * fixed some bugs revealed by Paul Dietz' test suite:
1572 ** MISC.549 and similar: late transformation of unsafe type
1573 assertions into derived types caused unexpected code
1575 ** SCALE-FLOAT type deriver is less wrong.
1576 ** type derivers for EXP, LOG and similar functions compute result
1577 types for complex arguments better.
1578 ** (MISC.563) CONJUGATE type deriver works for very restricted
1580 ** out-of-line type testers for character strings are available.
1581 ** EQUAL compiler transform understands specialness of objects
1583 ** accessing double-floats stored on the stack now works on x86-64.
1584 ** debugger internals could sometimes create invalid lispobjs,
1585 resulting in GC crashes.
1586 ** MISC.548: type check weakening can convert required type into
1588 ** initialization forms for bindings are not in scope of free special
1591 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1592 * incompatible change: thread support for non-NPTL systems has
1593 been removed - locking is buggy and unreliable. A threaded
1594 SBCL build will now warn at startup and refuse to create
1595 new threads, unless futex support is detected in the kernel
1596 * incompatible change: the top level REPL now has only an ABORT
1597 restart associated with it, not TOPLEVEL and ABORT as it used to.
1598 TOP and TOPLEVEL are now available as debugger commands for
1599 returning to the top level.
1600 * incompatible change: forms evaluated in the REPL now use the
1601 global optimization policy.
1602 * incompatible change: user- and system-initialization files are
1603 no longer processed with LOAD, but by READ and EVAL; hence the
1604 global optimization policy, startup package, readtable, etc,
1606 * The .fasl file format number has been incremented because of
1607 various incompatible changes.
1608 * internal entry point details and argument counts no longer appear
1609 in backtraces unless explicitly requested by setting
1610 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1611 * built-in and standard functions no longer have names like "top
1612 level local call to FOO".
1613 * fixed bug 32: functions defined in non-null lexical environments
1614 now have more legible printed representation
1615 * fixed bug 33: functions defined in non-null lexical environemnts
1616 are now more amenable to inspection by INSPECT.
1617 * workaround for bug 354: XEPs no longer appear in backtraces unless
1618 explicitly requested.
1619 * fixed bug: receiving the signal which results from attempting to
1620 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1621 SIGBUS on other *BSDs) on architectures where the C stack is also
1622 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1623 to debugger confusion or wild instability. (reported by Cheuksan
1625 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1626 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1627 (reported by Lutz Euler)
1628 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1629 compiled code. (reported by Paul Dietz)
1630 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1631 specialized parameter to a method to something that is not TYPEP
1632 the specializer is now possible.
1633 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1634 face of package deletion.
1635 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1636 pathname. (reported by David Tolpin on c.l.l)
1637 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1638 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1639 STRING-STREAM and FILE-STREAM, respectively.
1640 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1641 than previously; in addition, its test suite is now run on build.
1642 (thanks to Luke Gorrie)
1643 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1644 * optimization: added a immediate representation for single-floats
1646 * fixed some bugs revealed by Paul Dietz' test suite:
1647 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1648 ** attempting to create a package with a colliding nickname causes
1649 correctable errors to be signalled.
1650 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1651 ** several x86-64 backend bugs related to sign-extension of immediate
1654 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1655 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1656 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1658 * fixed loading of multiply forward-referenced layouts.
1659 (thanks to Cheuksan Wang)
1660 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1661 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1662 (see COMPILE-FILE documentation for details of :PRINT option)
1663 * fixed bugs 19 and 317: fixed-format floating point printing is
1664 more accurate. This also fixes a bug reported by Adam Warner
1665 related to the ~@F format directive.
1666 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1668 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1669 dispatch table if the from-char is a dispatch macro character.
1670 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1671 by the wordsize no longer ignore the last word. (reported by Lutz
1673 * fixed bug in type checking of dynamic-extent variables. (reported
1675 * optimization: sequence traversal functions use their freedom to
1676 coerce function designators to functions.
1677 * optimization: code with many calls to CLOS methods specialized on
1678 CLOS classes has had redundant type checks removed.
1679 * optimization: added declarations to speed up operations that access
1680 the internal character database (for example STRING-UPCASE)
1681 * optimization: comparison operations between floats and sufficiently small
1682 fixnums no longer create extra rationals
1683 * fixed some bugs related to Unicode integration:
1684 ** portions of multibyte characters at the end of buffers for
1685 character-based file input are correctly transferred to the
1686 start of the buffer at the next read.
1687 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1688 passing it through to OPEN.
1689 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1690 argument given to internal calls to OPEN.
1691 ** the built-in comment readers (introduced by character sequences
1692 ";" and "#|") are more forgiving to encoding errors; they will
1693 STYLE-WARN and then attempt to resync the stream at a character
1694 boundary. (thanks to Teemu Kalvas)
1695 * fixed some bugs in the x86-64 port:
1696 ** Negative short int return values from c-calls are sign-extended
1698 ** The stack is aligned to 16-bytes for c-calls, as required by
1699 the ABI. (thanks to Cheuksan Wang)
1700 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1701 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1702 for enums. (thanks to Vincent Arkesteijn)
1703 ** Multiple small optimizations and bugfixes for floating point
1705 * fixed some bugs revealed by Paul Dietz' test suite:
1706 ** Space, Tab, Linefeed, Return and Page have the invalid
1707 secondary constituent character trait.
1708 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1710 ** WITH-INPUT-FROM-STRING should only update the index place on
1712 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1713 works more reliably.
1714 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1715 with one-dimensional arrays with fill pointers.
1716 ** TYPE-OF failed on a complex with an integer realpart and a
1718 ** compiler failure during type inference for the code of form
1719 (IF (EQL X (THE ...)) ...) (MISC.535).
1721 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1722 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1723 platform. The port supports 61-bit fixnums, large memory spaces
1724 and reloading shared object files.
1725 * enhancement: saving cores with foreign code loaded is now
1726 supported on ppc/Darwin in addition to the previously supported
1728 * enhancement: the statistical profiler now walks deeper into the
1729 call stack for more meaningful call-graphs and accrued time
1730 reports (x86/x86-64 only). It also now reports time spent in
1732 * enhancement: it is now possible to trace most individual methods
1733 of a generic function in addition to tracing the generic function
1735 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1737 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1738 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1739 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1740 produces an error. (thanks to Vincent Arkesteijn)
1741 * bug fix: NAMESTRING on pathnames with :WILD components in their
1742 directories works correctly. (thanks to Artem V. Andreev)
1743 * fixed bug 125: compiler preserves identity of closures. (reported
1745 * bug fixed: functions with &REST arguments sometimes failed with
1746 "Undefined function" when compiled with (DEBUG 3). (reported by
1748 * bug fix: overflow during compiling of setting element of a bit
1749 vector with constant index and value. (reported by Timmy Douglas)
1750 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1751 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1752 * on x86 compiler partially supports stack allocation of dynamic-extent
1754 * GO and RETURN-FROM do not check the extent of their exit points
1755 when compiled with SAFETY 0.
1756 * fixed some bugs related to Unicode integration:
1757 ** encoding and decoding errors are now much more robustly
1758 handled; it should now be possible to recover even from invalid
1759 input or output to the terminal. (thanks to Teemu Kalvas)
1760 ** provided a first cut at implementing STRING-TO-OCTETS and
1761 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1762 ** altered the SB-MD5 contributed module slightly, changing the
1763 interface just enough for it to be supportable for builds where
1764 lisp characters are not eight bits.
1765 * fixed some bugs revealed by Paul Dietz' test suite:
1766 ** the FORMATTER-generated functions for ~V[ conditionals require
1767 the correct number of arguments.
1768 ** READ-FROM-STRING returns the mandated second value when applied
1769 to displaced strings.
1770 ** the #\Rubout and #\Backspace characters are treated as invalid
1771 constituent characters by the tokenizer.
1773 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1774 * new feature: reloading changed shared object files with
1775 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1776 * new feature: references to foreign variables and functions
1777 can now be compiled and loaded before the corresponding shared
1778 object file is loaded, as long as the foreign definitions are
1779 available at runtime.
1780 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1781 supported. (thanks to Dan Debertin)
1782 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1783 just keywords. (thanks to Vincent Arkesteijn)
1784 * fixed bug #331: structure-class instances corresponding to
1785 DEFSTRUCT forms are now created eagerly.
1786 * fixed bug #345: backtraces from calls to undefined functions work
1787 on x86 as well. Related bug #61 is now also partially fixed on x86
1788 (backtraces from throws to unknown catch tags.)
1789 * bug fix: lambda-list parsing is now stricter vrt. order and number
1790 of lambda-list keywords.
1791 * bug fix: as specified by AMOP, an error is signalled if a
1792 class-option appears multiple times in a DEFCLASS form. (reported
1794 * bug fix: RANDOM can be compiled when the compiler derives the type
1795 of its numeric argument as a disjoint set of small integers.
1796 (reported by Paul Dietz)
1797 * bug fix: starting a core saved with shared objects loaded when
1798 those objects are not available no longer causes threaded SBCL to
1799 hang. (reported by Sean Ross)
1800 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1802 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1803 argument, which must always be NIL. (reported by Kalle Niemitalo)
1804 * bug fix: printing 1.0d+23 no longer results in an error.
1805 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1806 * bug fix: structure slot setters preserve evaluation order. (thanks
1808 * fixed some bugs related to Unicode integration:
1809 ** RUN-PROGRAM can allow its child to take input from a Lisp
1810 stream. (reported by Stefan Scholl)
1811 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1812 locales. (reported by Ken Causey)
1813 * fixed some bugs revealed by Paul Dietz' test suite:
1814 ** INCF, DECF and REMF evaluate their place form as specified in
1816 ** FORMATTER expands ~{ ~} iteration directives with V or #
1817 parameters correctly.
1818 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1819 consequent uses no arguments correctly.
1820 ** the system has a partial understanding of the (COMPLEX RATIO)
1822 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1825 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1826 * new feature: a build-time option (controlled by the :SB-UNICODE
1827 keyword feature, enabled by default) for building the system with
1828 support for the entire 21-bit character space defined by the
1830 * new feature: the system now has rudimentary external-format
1831 support; the primary user-visible change at this time is that
1832 characters with the high bit set (such as lower-case-e-acute) will
1833 print correctly to a terminal in a UTF-8 environment.
1834 * minor incompatible change: BASE-CHAR no longer names a class;
1835 however, CHARACTER continues to do so, as required by ANSI.
1836 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1837 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1838 has been moved to the SB-EXT package (temporarily re-exported from
1840 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1841 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1843 * fixed bug #351: better error-handlind and reporting for malformed
1845 * fixed bug #350: bignum-printing is now more memory-efficient,
1846 allowing printing of very large bignums, eg. (expt 2 10000000).
1847 (reported by Bruno Haible)
1848 * fixed bug #302: better primitive-type selection for intersection
1850 * fixed bug #308: non-graphic characters now all have names, as
1851 required. (reported by Bruno Haible)
1852 * bug fix: redefining a class with different superclasses now correctly
1853 removes it from the direct-subclasses of its previous superclasses.
1854 (reported by David Morse)
1855 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1856 new value now works. (reported by Bruno Haible)
1857 * bug fix: correct canonicalization of multiple non-standard slot
1858 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1859 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1860 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1861 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1862 now exists, an signals an error.
1863 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1864 during macroexpansion for non-iteration variables as well. (reported
1865 by Bruno Haible for CMUCL)
1866 * bug fix: Cyclic structures and unprintable objects in compiler
1867 messages no longer cause errors. (reported by Bruno Haible)
1868 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1869 and READ-FROM-STRING all now return a primary value of NIL if
1870 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1871 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1872 (reported by Bruno Haible for CMUCL)
1873 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1874 do not designate a condition. (reported by Bruno Haible for
1876 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1877 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1878 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1879 permitting resolution in favour of any of the conflicting symbols.
1880 (reported by Bruno Haible for CMUCL)
1881 * bug fix: EQUAL compiler optimizations is less aggressive on
1882 strings which can potentially compare true despite having distinct
1883 specialized array element types.
1884 * bug fix: unit enumerations can be defined without dividing by
1885 zero. (thanks to Vincent Arkesteijn)
1886 * FORMAT compile-time argument count checking has been enhanced.
1887 (report from Bruno Haible for CMUCL)
1888 * a partial workaround for the bug 262: the compiler does not try to
1889 inline-expand a local function doing RETURN-FROM from a deleted
1890 BLOCK. (thanks to Peter Denno for the bug report and to David
1891 Wragg for the simple test case)
1892 * fixed some bugs revealed by Paul Dietz' test suite:
1893 ** RENAME-PACKAGE allows all package designators as new package
1895 ** constraint propagation and lambda variable substitution are
1896 more cautious in dealing with partially deleted code.
1897 ** compiler sometimes lost reoptimization passes.
1898 ** CERROR, given a condition as condition designator, passes its
1899 remaining arguments to the continue format control without
1901 ** Case-altering FORMAT directives work correctly on non-ASCII
1903 ** The REAL type specifier handles bounds outside the floating
1904 point ranges without signalling FLOATING-POINT-OVERFLOW.
1905 ** Functions with IR1-transformations can create intercomponent
1906 references to global functions.
1907 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1909 ** FORMAT ~R treats a nil value for its first parameter correctly.
1911 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1912 * enhancement: saving cores with foreign code loaded is now
1913 supported on x86/NetBSD and sparc/Linux in addition to the previously
1914 supported platforms.
1915 * bug fix: on some platforms repeated installations caused multiple
1916 copies of HTML documentation to be installed -- should not happen
1917 any more. (reported by Stefan Scholl)
1918 * bug fix: parsing self-recursive alien record types multiple times
1919 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1920 original patch by Helmut Eller for CMUCL)
1921 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1922 (thanks to Richard Kreuter)
1923 * bug fix: defining classes whose accessors are methods on existing
1924 generic functions in other (locked) packages no longer signals
1925 bogus package lock violations. (reported by François-René Rideau)
1926 * bug fix: special variables as DEFMETHOD parameters no longer have
1927 associated bogus type declarations. (reported by David Wragg and
1929 * bug fix: read-write consistency on streams of element-type
1930 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1931 * bug fix: redefiniton of the only method of a generic function with
1932 no DEFGENERIC no longer emits a full WARNING. In addition,
1933 redefinition of generic functions with no DEFGENERIC to an
1934 incompatible lambda list now signals an error. (thanks to Zach
1936 * bug fix: DEFGENERIC now works even when there's a function of the
1937 same name in an enclosing lexical environment. (thanks to Zach
1939 * fixed compiler failure, caused by instrumenting code during
1940 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1941 * optimization: added loop analysis and improved register allocation
1942 * fixed some bugs revealed by Paul Dietz' test suite:
1943 ** POSITION on displaced vectors with non-zero displacement
1944 returns the right answer.
1945 ** (SIMPLE-STRING) is a valid type specifier for sequence
1947 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1949 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1950 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1952 ** The justification version of the FORMAT ~< directive treats
1953 non-zero minpad parameter correctly.
1955 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1956 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1957 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1958 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1959 the supported interface.
1960 * new feature: Single-stepping of code compiled with DEBUG 2 or
1961 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1962 * new feature: saving cores with foreign code loaded is now
1963 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1964 Timothy Moore's work for CMUCL)
1965 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1966 parameters to * instead of NIL if no initform is supplied.
1967 (reported by Johan Bockgaard)
1968 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1969 correctly when FUNCALL appears as the car of the form. Note:
1970 despite this FUNCALL forms are not currently subject to
1971 compiler-macro expansion. (port of Raymond Toy's fix for the
1972 same from CMUCL, reported by Johan Bockgaard)
1973 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1974 (thanks for Teemu Kalvas)
1975 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1976 case instead of terminating the iteration (thanks for Julian
1977 Squires, Sean Champ and Raymond Toy)
1978 * bug fix: incorrect expansion of defgeneric that caused a style
1979 warning. (thanks for Zach Beane)
1980 * bug fix: slot accessor effective method computation works properly
1981 for classes with multiple non-standard applicable methods on
1982 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1983 * on x86 compiler supports stack allocation of results of LIST and
1984 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1985 CMUCL implementation by Gerd Moellmann)
1986 * fixed some bugs revealed by Paul Dietz' test suite:
1987 ** FORMAT strings with both the ~<~:;~> form of the justification
1988 directive and pretty-printing directives cause an error.
1990 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1991 * incompatible change: the internal functions
1992 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1993 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1994 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1995 instead of the old functions.
1996 * new feature: on platforms where "dladdr" is available foreign
1997 function names now appear in backtraces. (based on Helmut Eller's
1999 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2000 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2002 * documentation: ASDF manual is now distributed with SBCL.
2003 * bug fix: SBCL can now load its contributed modules with REQUIRE
2004 even if the system-provided entries have been removed from
2005 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2007 * bug fix: backtraces involving undefined functions or assembly
2008 routines are more informative. (thanks to Brian Downing)
2009 * bug fix: mutually referent alien structures now work correctly.
2010 (reported by Rick Taube)
2011 * bug fix: structures defined by WITH-ALIEN can be referred to
2012 within other definitions in the same WITH-ALIEN.
2013 * bug fix: division operators (MOD, TRUNCATE and the like) with
2014 constant zero divisors and integer dividends no longer generate
2016 * bug fix: provide default methods for INPUT-STREAM-P and
2017 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2018 * bug fix: improve the detection and resolution of MOP metacycles
2019 (where effective method computation depends on the generic function
2020 having its effective method computed). (reported by Bruno Haible)
2021 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2022 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2023 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2024 * bug fix: pathnames with relative directory components can be
2025 represented relative to default pathnames.
2026 * optimization: in taking the GCD of bignums, reduce the two bignums
2027 to approximately the same size (using Euclid's algorithm) before
2028 applying the more sophisticated binary GCD. (thanks to Juho
2030 * optimization: COUNT on bitvectors now operates word-at-a-time.
2031 * optimization: ASH with a positive, but not necessarily constant,
2032 (leftwards) shift, when in a modular context, is compiled to a
2034 * fixed some bugs revealed by Paul Dietz' test suite:
2035 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2036 if the corresponding argument is NIL.
2037 ** FORMAT directives accepting a minpad parameter treat negative
2039 ** literal commas as character parameters to FORMAT directives are
2041 ** literal spaces directly after ~<Newline> directives within a
2042 format-logical-block (~:< ~@:>) do not induce :FILL-style
2043 conditional newlines.
2044 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2045 as specified: it no longer includes conditional newlines).
2046 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2048 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2049 :PER-LINE-PREFIX argument does not evaluate to a string.
2051 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2052 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2053 the manual for details; for now, package locks can be disabled by
2054 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2055 if no major problems are found then it is likely that they will be
2056 compiled in unconditionally.
2057 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2058 unsupported operators on all platforms. To load a shared library
2059 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2060 object file, link it into a shared library outside of SBCL and
2061 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2063 * bug workaround: ROOM T and the GC were not working together
2064 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2065 reporting facility aren't preserved. That reporting has been
2066 disabled, so now until and unless someone figures out how to make it
2067 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2068 * minor incompatible change: as threatened around sbcl-0.8.0, the
2069 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2070 The symbols are also exported from SB-PCL for backwards
2071 compatibility, but more so than before SB-PCL should be treated as
2072 an implementation-internal package.
2073 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2075 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2076 It is known as of this release not to work on the Alpha, however.
2077 * fixed bug #167: errors signalled due to illegal syntax in method
2078 bodies are now more legible.
2079 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2080 designators and can hence be used with TYPEP.
2081 * fixed bug #333: CHECK-TYPE now ensures that the type error
2082 signalled, if any, has the right object to be accessed by
2083 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2084 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2085 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2087 * fixed bug #334: programmatic addition of slots using specialized
2088 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2089 and :ALLOCATION :CLASS slots.
2090 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2091 float exponents. (rereported by Peter Seibel)
2092 * fixed a bug: #\Space (and other whitespace characters) are no
2093 longer considered to be macro characters in standard syntax by
2094 GET-MACRO-CHARACTER.
2095 * fixed bug: initialization of condition class metaobjects no longer
2096 causes an instance of the condition to be created. (reported by
2098 * fixed bug: it is now possible to have slots such that
2099 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2100 description is neither :INSTANCE nor :CLASS.
2101 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2102 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2103 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2104 non-standard SLOT-DEFINITION-ALLOCATION values.
2105 * fixed bugs: various race conditions handling exiting threads.
2106 CL-PPCRE's thread torture test now passes.
2107 * fixed arguably-a-bug: GC time proportional to number of threads in
2108 system even when most of them are idle
2109 * optimization: improved performance of BIT and SBIT on bit-vectors.
2110 * .fasl file incompatibility: The fasl file version number has
2111 been incremented because of changes associated with package locks.
2113 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2114 * minor incompatible change: the system no longer provides
2115 optimization hints (conditions which are TYPEP
2116 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2117 mode; these hints are still emitted when higher SPEED optimization
2119 * new contrib module: a sampling profiler (profiling by statistical
2120 sampling, rather than by instrumenting functions) is available as
2121 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2122 * the behaviour of the standard function ED is now customizeable by
2123 third parties through a hook variable: see ED's documentation
2124 string for information on the protocol.
2125 * the compiler no longer emits efficiency notes for (FUNCALL X)
2126 when the type of X is uncertain under default optimization
2128 * fixed bug 276: mutating a binding of a specialized parameter to a
2129 method to something that is not TYPEP the specializer is now
2131 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2132 exhibit the required behaviour on the x86 platform. (thanks to
2133 Peter van Eynde, Eric Marsden and Bruno Haible)
2134 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2135 even for difficult arguments. (reported by Peter Graves)
2136 * fixed bug 141a: the backquote printer now descends quoted
2138 * fixed another bug in backquote printing: no more destructive
2139 modification of the form's list structure. (reported by Brian
2141 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2142 move between its address being taken and the call to
2143 interrupt_thread, fixing a crashing race condition.
2144 * the SB-POSIX contrib implementation has been adjusted so that it
2145 no longer exhibits ridiculously poor performance when constructing
2146 instances corresponding to C structs.
2148 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2149 * minor incompatible change: the sb-grovel contrib now treats C
2150 structures as alien (in the sense of SB-ALIEN) objects rather than
2151 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2152 has implications for memory management of client code
2153 (sb-grovel-returned objects must now be manually managed) and for
2154 type safety (alien objects now have full types).
2155 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2156 used to control emission of compiler diagnostics, rather than the
2157 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2158 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2159 quality should be considered deprecated.
2160 * install.sh now installs the user manual as well
2161 * (not quite a new documentable feature, but worth considering in
2162 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2163 beginnings of a semantically meaningful condition hierarchy is
2164 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2166 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2167 designator as the defaults argument.
2168 * fixed bug: Displaced arrays whose displaced-to array has become
2169 too small now cause ARRAY-DIMENSION to signal an error, providing
2170 for safer bounds-checking. (reported by Bruno Haible)
2171 * fixed bug: DEFCLASS slot definitions with identical :READER and
2172 :WRITER names now signal a reasonable error. (reported by Thomas
2174 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2176 * fixed bug: Invalid slot specification errors now print correctly.
2177 (thanks to Zach Beane)
2178 * fixed bug 320: Shared to local slot value transfers in class
2179 redefinitions now happen corrently from superclasses as
2180 well. (reported by Bruno Haible)
2181 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2183 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2184 improper lists correctly. (reported by Bruno Haible)
2185 * fixed bug 313: source-transform for <fun-name> was erroneously
2186 applied to a call of a value of a variable with name <fun-name>.
2187 (reported by Antonio Menezes Leitao)
2188 * fixed bug 307: The obsolete instance protocol ensures that
2189 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2190 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2191 remove dead unknown-values globs from the middle of the stack.
2192 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2193 easier to resume long computations after using *BREAK-ON-SIGNALS*
2194 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2195 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2196 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2197 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2198 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2199 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2201 * fixed a (fairly theoretical) bug in string printing: if
2202 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2203 does not have array-element-type equal to the most general string
2205 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2206 function name. (thanks to Nikodemus Siivola)
2207 * fixed bug: compile-time format string checker failed on
2208 non-closed ~{. (reported by Thomas F Burdick)
2209 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2210 don't warn on legal loop code involving more than one aggregate
2211 boolean. (thanks to Nikodemus Siivola)
2212 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2213 in keyword position to MAKE-PACKAGE was not being reported as
2215 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2216 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2218 * fixed bugs 280 and 312: the checking for multiple definitions in a
2219 file is less likely to become confused by uses of inline
2221 * fixed bug: the #S reader macro performs the keyword coercion
2222 specified for slot names. (reported by Kalle Niemitalo)
2223 * fixed bug: lambda lists may contain symbols whose names start with
2224 & but are not lambda-list-keywords; their occurrence triggers a
2226 * fixed bug 321: define-method-combination argument lambda lists do
2227 not cause invalid code to be generated when &optional or &aux
2228 variables have default values. (reported by Bruno Haible)
2229 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2230 have CLOS classes; this appears also to have fixed reported
2231 instability in stack exhaustion detection.
2232 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2233 MUFFLE-WARNING when no associated restart is present is now
2235 * optimization: rearranged the expansion of various defining macros
2236 so that each expands into only one top-level form in a
2237 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2239 * optimization: used a previously little-used slot in symbols to
2240 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2242 * fixed some bugs revealed by Paul Dietz' test suite:
2243 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2245 ** Fixed an optimization bug related to inheritance of initforms
2246 from local to shared slots.
2247 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2248 ** CAST optimizer forgot to flush argument derived type.
2249 ** print/read consistency on floats is now orders of magnitude
2250 more likely. (thanks also to Bruno Haible for a similar report
2252 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2253 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2254 for objects of type REAL. Make it so.
2255 ** SXHASH is specified (infelicitously) to respect similarity,
2256 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2257 it so. (thanks to Markus Ziegler)
2258 ** on the Alpha, the compiler succeeds in compiling functions
2259 returning a known number of arguments greater than 63.
2260 ** fixed handling of invalid NIL arguments in keyword position
2262 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2264 ** provided a readably-printable representation for RANDOM-STATE
2266 ** ensured that pathnames, where they have a namestring, always
2267 print using #P"..." syntax.
2269 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2270 * Support for the forthcoming 2.0 version of the NetBSD kernel
2271 running on x86 hardware has been added. (thanks to Perry
2272 E. Metzger most immediately, and others for their past work)
2273 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2274 current release version; SBCL's OpenBSD support had been broken
2275 since about the time of OpenBSD's switch to ELF binary format.)
2276 * [placeholder for DX summary]
2277 ** user code with &REST lists declared dynamic-extent, under high
2278 speed or space and low safety and debug optimization policy.
2279 * The manual has been converted to Texinfo format and the debugger
2280 chapter from the cmucl manual has been added.
2281 * A facility has been added to extract documentation strings from
2282 sbcl and store them as Texinfo-formatted snippets for inclusion in
2283 the manual (via Texinfo's @include directive)
2284 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2285 the test case to Dave Roberts)
2286 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2287 pointers. (reported by Sean Ross)
2288 * bug fix: PROFILE output is printed nicely even for large numerical
2289 values. (thanks to Zach Beane)
2290 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2291 greater than 32 handle EOF correctly.
2292 * bug fix: on X86 an immediate argument of the IMUL instruction is
2293 correctly printed in disassembly. (thanks to Lutz Euler)
2294 * bug fix: class slots in redefined classes preserve their old
2295 values. (thanks to Bruno Haible and Nikodemus Siivola)
2296 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2297 succeeds. (reported by Marco Baringer)
2298 * fixed some bugs revealed by Paul Dietz' test suite:
2299 ** READ-SEQUENCE now works on ECHO-STREAMs.
2300 ** RATIONALIZE works more according to its specification. (thanks
2303 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2304 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2305 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2306 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2307 should matter to you only if you rebind the printer control
2308 variables and then find you want different bindings in the
2309 debugger than in the ordinary execution of your program.)
2310 * The runtime build system has been tweaked to support building
2311 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2312 assembler and linker. (thanks to Nikodemus Siivola)
2313 * Unbound, undefined, undeclared variables now trigger full
2314 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2315 more often programmer error than deliberate exploitation of undefined
2317 * optimization: the hash algorithm for strings has changed to one
2318 that is less vulnerable to spurious collisions. (thanks to Juho
2320 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2321 needless bounds checking. (thanks to Juho Snellman)
2322 * optimization: implemented multiplication as a modular
2323 (UNSIGNED-BYTE 32) operation on the PPC backend.
2324 * fixed some bugs revealed by Paul Dietz' test suite:
2325 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2327 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2328 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2329 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2331 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2333 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2335 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2336 their output stream on EOF from read.
2337 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2338 have been read to end-of-file.
2339 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2341 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2342 description of determination of which consecutive characters
2344 ** Printing the "Space" character with escaping on now yields "#\\ ",
2345 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2346 ** Reading floating-point numbers with *READ-BASE* set to a number
2347 less than 10 works correctly.
2348 ** Reading floating-point numbers with *READ-BASE* set to a number
2349 more than 10 works correctly.
2350 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2351 the readtable currently in effect.
2353 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2354 * minor incompatible change: parsing of namestrings on a physical
2355 (Unix) host has changed; numbers after the final #\. in a
2356 namestring are no longer interpreted as a version field. This is
2357 intented to be largely invisible to the user, except that the
2358 meaning of the namestring "*.*.*" has changed: it now refers to a
2359 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2360 should usually be replaced by
2361 (make-pathname :name :wild :type :wild :version :wild)
2362 with the added benefit that this is more likely to be portable.
2363 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2364 option to OPEN now signals an error if the file being opened
2365 exists; this may have an impact on existing code.
2366 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2367 signals received, on the PowerPC platforms (both Linux and
2368 Darwin). (thanks to Pierre Mai for pointing out the location of
2370 * several fixes on OS X: The system now builds and runs cleanly on
2371 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2372 (thanks to Brian Mastenbrook)
2373 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2374 second-resolution: integer multiples of 1/3600 between -24 and 24.
2375 (thanks to Vincent Arkesteijn)
2376 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2377 type when called with 1 argument; PEEK-CHAR checked type of
2378 PEEK-TYPE only after having read first character from a
2379 stream. (reported by Peter Graves)
2380 * bug fix: the garbage collector now has much better locality
2381 behaviour, and in particular no longer treats all memory as being
2382 exhausted when many small objects point to each other in a deeply
2384 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2385 recognized as being TYPEP their class.
2386 * bug fix: the PUSHNEW documentation string has been corrected.
2387 (thanks to Vincent Arkesteijn)
2388 * bug fix: defaulting of the value for the last of an atypically
2389 large number of multiple values being bound was not being
2390 performed correctly on the Alpha or PPC platforms
2391 * optimization: implemented multiplication as a modular
2392 (UNSIGNED-BYTE 32) operation on the x86 backend.
2393 * optimization: SEARCH on simple-base-strings can now be open-coded.
2394 (see also contrib/compiler-extras.lisp for inspiration for
2395 teaching the compiler about the Boyer-Moore algorithm).
2396 * value, returned by MAX (and MIN) called with several EQUALP, but
2397 not EQL, arguments now does not depend on compiler settings.
2398 * fixed some bugs revealed by Paul Dietz' test suite:
2399 ** in stack analysis liveness information is propagated from
2400 non-local entry points.
2401 ** pathwise CAST removing failed when the CAST node did not start
2403 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2404 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2406 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2407 host is already defined.
2408 ** RENAME-FILE works on streams instead of signalling an internal
2410 ** PEEK-CHAR uses the current readtable when determining whether
2411 or not a character is whitespace.
2412 ** MERGE-PATHNAMES handles the case when the pathname does not
2413 specify a name while the default-pathname specifies a version.
2414 ** Pathnames now stand a better chance of respecting print/read
2416 ** Attempting to use standardized file system operators with a
2417 pathname with invalid :DIRECTORY components signals a
2419 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2421 ** DIRECTORY on logical pathnames is more correct.
2422 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2423 signal a TYPE-ERROR if their argument is not a stream.
2424 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2425 designator argument does not designate a stream.
2426 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2427 examining the synonym.
2428 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2430 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2431 element-type larger than ([UN]SIGNED-BYTE 32).
2433 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2434 * When built with the :SB-FUTEX feature, threaded builds now take
2435 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2436 for faster/more reliable mutex and condition variable support.
2437 * Incompatible change (but one you probably shouldn't have been using
2438 anyway): the interface and code for arbitrating between multiple
2439 threads in the same user session has been redesigned.
2440 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2441 error when called without an explicit environment argument.
2442 (thanks to Nikodemus Siivola)
2443 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2444 confused about their position. (thanks to Adam Warner and Gerd
2446 * bug fix: Pretty printing backquoted forms with unquotations in the
2447 argument list position of various code constructs such as LAMBDA
2448 now works correctly. (reported by Paul Dietz)
2449 * bug fix: Pretty printing unquotations no longer loses all
2450 stream position information.
2451 * optimization: performance of string output streams is now less
2452 poor for multiple small sequence writes.
2453 * optimization: performance of CSUBTYPEP in the presence of complex
2454 expressions involving CONS and NOT many times has been improved.
2455 (reported by Paul Dietz)
2456 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2458 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2459 streams for the REPL, for the debugger, and so on. (thanks to
2461 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2462 * fixed some bugs revealed by Paul Dietz' test suite:
2463 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2464 is ignored if it is too small, rather than propagating through
2465 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2466 ** extremely complex negations of CONS types were not being
2467 sufficiently canonicalized, leading to inconsistencies in
2469 ** VALUES tranformer lost derived type.
2471 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2472 * fixed a bootstrapping bug: the build process no longer assumes
2473 that the various BOOLE-related constants have the same value in
2474 host and target lisps. (noted by Paul Dietz' test suite on an
2475 SBCL binary built from CLISP)
2476 * The system can now be dynamically linked on the MIPS platform,
2477 which enables dynamic loading of foreign code from Lisp. (thanks
2478 to Ralf Baechle for discussions on the MIPS ABI)
2479 * The system now records debugging information for its own source
2480 files in a filesystem-position-independent manner, relative to
2481 the "SYS" logical host.
2482 * fixed a compiler bug: MV-LET convertion did not check references
2483 to the "max args" entry point. (reported by Brian Downing)
2484 * tweaked disassembly notes to be less confident about proclaiming
2485 some instruction as an LRA. (thanks to Brian Downing)
2486 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2487 now each have their own history, command character, and other
2488 characteristics. (thanks to David Lichteblau)
2489 * fixed some bugs revealed by Paul Dietz' test suite:
2490 ** compiler failure in compiling LOGAND expressions including a
2492 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2493 shift greater than 32.
2494 ** FLUSH-DEST did not mark blocks for type check regeneration.
2495 ** HANDLER-CASE failed to accept declarations in handler clauses
2496 in some circumstances.
2498 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2499 * New code in contrib/sb-introspect (still probably not entirely
2500 stable yet) provides some support for smart Lisp development
2501 environments like SLIME.
2502 * The conditions signalled for errors occurring when loading .fasl
2503 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2504 in a way which should help ASDF recover gracefully.
2505 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2506 stuff has been cleaned up. If you code contrib/ stuff, this might
2507 affect you, and you can look at contrib/README, contrib/STANDARDS,
2508 and/or the 0.8.4.27 diff to check.
2509 * In full calls the compiler now does not generate checks for declared
2510 argument types for all arguments.
2511 * various threading fixes
2512 ** and some experimental patches which didn't make it into
2513 the main tree for this release, but which are shipped in
2514 contrib/experimental-thread.patch as a possible fix for some
2515 failures (deadlock, spinning...) in GC-intensive multithreaded
2517 * fixed PPC build problem (source code incompatibility of different
2518 library versions): added offsetof() hackery which attempts to divine
2519 where glibc maintainers put uc_mcontext today
2520 * fixed bug 282: compiler does not trust type assertions while passing
2521 arguments to a full call.
2522 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2523 &OPTIONAL VALUES type parameter.
2524 * fix bug 214: algorithm for noting rejected templates is now more
2525 similar to that of template seletion. (also reported by rydis on
2527 * fixed bug 141b: printing backquoted information readably and prettily
2528 inserts a space where necessary.
2529 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2530 (CHARACTER 10) are now reported as errors, rather than propagated
2531 as unknown types. (reported by piso on #lisp)
2532 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2533 with values NIL and :ERROR. (thanks to Milan Zamazal)
2534 * fixed bug 191c: CLOS now does proper keyword argument checking as
2535 described in CLHS 7.6.5 and 7.6.5.1.
2536 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2537 longer raise an error; further, using a list as a for-as-arithmetic
2538 counter now raises a meaningful error.
2539 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2540 now understood by sequence creation functions such as MAKE-SEQUENCE
2542 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2543 a STRING-INPUT-STREAM.
2544 * compiler enhancement: SIGNUM is now better able to derive the type
2546 * type declarations inside WITH-SLOTS are checked. (reported by
2548 * fixed some bugs revealed by Paul Dietz' test suite:
2549 ** incorrect optimization of TRUNCATE for a positive first
2550 argument and negative second.
2551 ** compiler failure in let-convertion during flushing dead code.
2552 ** compiler failure while deriving type of TRUNCATE on an
2553 interval, containing 0.
2554 ** ASH of a negative bignum by a negative bignum count now returns
2556 ** intersection of CONS types now canonicalizes properly, fixing
2557 inconsistencies in SUBTYPEP.
2559 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2560 * incompatible change: The --disable-debugger command line
2561 option now clobbers the debugger at a more fundamental
2562 level, by redefining #'INVOKE-DEBUGGER instead of by
2563 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2564 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2565 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2566 enter the debugger REPL and then suffer recursive errors
2567 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2568 while under the new system, BREAK in --disable-debugger mode
2569 terminates the system just as an unhandled error would.
2570 * fixed compiler performance when processing loops with a step >1;
2571 * bug fix: DOCUMENTATION now retrieves generic function
2572 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2573 support has been systematized, and now supports the methods
2574 specified by ANSI, along with a default method and a method for
2575 slot documentation. (reported by Nathan Froyd)
2576 * bug fix: effective methods associated with a generic function are
2577 no longer cached over a change of that generic function's method
2578 combination. (reported by Andreas Fuchs)
2579 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2580 names a non-existent directory. (thanks to Andreas Fuchs)
2581 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2582 the PPC platform, lead to stack corruption; this has been fixed.
2583 (reported by Rainer Joswig)
2584 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2585 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2586 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2587 arithmetic in general, is now much more reliable on the PPC
2589 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2590 platform now returns the right answer.
2591 * optimization: restored some effective method precomputation in
2592 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2593 precomputation is now tunable.
2594 * optimization: compiler-internal data structure use has been
2595 reviewed, and changes have been made that should improve the
2596 performance of the compiler by about 20%.
2597 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2598 simple-base-strings and simple-bit-vectors is improved.
2599 * optimization: the optimization of 32-bit logical and arithmetic
2600 functions introduced in version 0.8.3 on the x86 has been
2601 implemented on the mips, ppc and sparc platforms; an
2602 implementation of the same facility, but for 64-bit arithmetic,
2603 has been added for the alpha.
2604 * microoptimization: the compiler is better able to make use of the
2605 x86 LEA instruction for multiplication by constants.
2606 * bug fix: in some situations compiler did not report usage of
2607 generic arithmetic in (SPEED 3) policy.
2608 * bug 145b fix: compiler used wrong type specifier while converting
2609 MEMBER-types to numeric.
2610 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2612 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2613 index argument now works properly on the Alpha platform.
2614 * bug fix: floating point exception treatment on the Alpha platform
2616 * bug fix: FILE-POSITION works much better on string input and
2617 output streams. (thanks to Nikodemus Siivola)
2618 * bug fix: many threading/garbage collection symptoms sorted.
2619 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2620 might be pseudo-atomic.
2621 * internal change: Stopping for GC is now done with signals not
2622 ptrace. GC is now done in whichever thread wanted it, instead of
2624 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2626 * bug fix: result form in DO is not contained in the implicit
2628 * incompatible change: ICR structure is changed; the value part of
2629 CONTINUATION is now called LVAR; corresponding functions are
2630 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2631 * added type deriver for ISQRT (thanks to Robert E. Brown).
2632 * bug fix: better support for loading from the command line when an
2633 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2635 * fixed some bugs revealed by Paul Dietz' test suite:
2636 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2637 ** ROUND and FROUND now give the right answer when given very
2638 small float arguments.
2639 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2641 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2642 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2643 ** LCM should return a non-negative integer.
2644 ** PARSE-INTEGER returned the index of a terminator instead of the
2645 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2646 ** PARSE-INTEGER returned an incorrect index being applied to a
2648 ** LCM with two arguments of 0 returns 0 rather than signalling
2650 ** unsigned addition of a 32-bit constant with the high bit set no
2651 longer causes an internal compiler error on the x86.
2652 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2653 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2654 with negative last argument.
2655 ** byte specifiers with zero size and position no longer cause
2656 an error during type derivation.
2657 ** bignum multiplication on the Alpha platform now returns the
2659 * porting: The system now builds on SuSE AMD64, although it still
2660 generates a 32-bit binary.
2661 * .fasl file incompatibility: The fasl file version number has
2662 been incremented (because of the changes to internal compiler
2663 data structures referred to above).
2665 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2666 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2667 more accurately, on the Darwin kernel running on PowerPC hardware.
2668 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2669 * Compiler code deletion notes now signal a condition of type
2670 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2671 an associated MUFFLE-WARNING restart.
2672 * The compiler now performs limited argument count validation of
2673 constant format strings in FORMAT, and where appropriate in ERROR,
2674 CERROR and WARN. (thanks to Gerd Moellmann)
2675 * New ASDF-INSTALL contrib can be used for automatic download and
2676 installation of third-party Lisp code from CCLAN or other sites
2678 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2679 which forces another thread to execute a function supplied by the
2681 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2682 now accept and act upon their :ELEMENT-TYPE keyword argument.
2683 (reported by Martin Atzmueller, Edi Weitz)
2684 * bug fix: FILE-POSITION now accepts position designators up to
2685 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2686 is the greater. (thanks to Patrik Nordebo)
2687 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2688 Kalle Olavi Niemitalo)
2689 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2690 printable. (reported by Eric Marsden)
2691 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2692 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2693 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2694 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2695 (thanks to Stig Sandoe)
2696 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2697 UNREAD-CHAR now correctly includes the unread character in the
2698 target sequence. (thanks to Gerd Moellmann)
2699 * bug fix in threaded builds: the system can now be suspended and
2700 resumed by shell job control with minimal disruption.
2701 * bug fixes in times and timezones >2038AD
2702 * better handling of "where is GNU make?" problem in build scripts
2703 (thanks to Nikodemus Siivola)
2704 * new optimization: inside a named function any reference to a
2705 function with the same name is considered to be a self-reference;
2706 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2707 optimization quality.
2708 * new optimization on x86: logical functions and + now have
2709 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2710 used when the result is truncated to 32 bits.
2711 * VALUES declaration is partially enabled.
2712 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2713 * fixed some bugs revealed by Paul Dietz' test suite:
2714 ** The system now obeys the constraint imposed by
2715 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2716 types form a lattice under type intersection.
2717 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2718 ** ASSOC now ignores NIL elements in an alist.
2719 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2720 and (1+ MOST-POSITIVE-FIXNUM) answers.
2721 ** The addition of a method with invalid qualifiers to a generic
2722 function does not cause an error to be signalled immediately;
2723 a warning is signalled, and the error is generated only on
2724 calling the generic function.
2725 * changed .fasl file version number, in part to add type codes for
2726 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2727 obscure ANSI requirements
2729 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2730 * fixed bug 148: failure to inline-expand a local function left
2731 garbage, confusing the compiler.
2732 * fixed bugs 3cd: structure slot readers perform type check if the
2733 slot can have an invalid value (i.e. it is either not initialized
2734 or can be written with a less specific slot writer).
2735 * bug fix: the compiler now traps array references to elements off
2736 the end of an array; previously, the bounds checking in some
2737 circumstances could go off-by-one.
2738 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2740 * type declarations for array element types now obey the description
2741 on the CLHS page "Declaration TYPE", as per discussions on
2742 sbcl-help around 2003-05-08. This means that a declaration
2743 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2744 declaration, all references to BAR will be asserted or assumed
2745 (with THE, so dependent on compiler policy) to involve objects of
2746 type FOO. Note that no such declaration is implied in
2747 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2748 * declared types of functions from the "Conditions"
2749 chapter. (reported by Paul Dietz)
2750 * bug fix: CERROR accepts a function as its first argument.
2751 * bug fix: NTH an NTHCDR accept a bignum as index
2752 arguments. (reported by Adam Warner)
2753 * optimization: character compare routines now optimize comparing
2754 against a constant character. (reported by Gilbert Baumann)
2755 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2756 argument now works properly on the MIPS platform.
2757 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2758 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2759 declared to have a functional type, does not cause an error
2761 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2762 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2763 * fixed bug: if last continuation of a deleted block has a
2764 destination, this destination should be deleted too. (reported by
2766 * fixed a bug in the bootstrap process: the host compiler's values
2767 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2768 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2769 test case from Patrik Nordebo)
2770 * improved the ability of the disassembler on the PPC platform to
2771 provide helpful disassembly notes.
2772 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2773 the class in more cases than previously.
2774 * bug fix: FILE-POSITION now understands :START and :END for
2775 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2776 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2777 8) data. (thanks to David Lichteblau)
2778 * bug fix: it is possible to add a method to a generic function
2779 without lambda list.
2780 * bug fix: reader failed to signal END-OF-FILE inside an
2781 object representation. (reported by Nikodemus Siivola)
2782 * fixed some bugs revealed by Paul Dietz' test suite:
2783 ** LAST and [N]BUTLAST should accept a bignum.
2784 ** condition slot accessors are methods.
2785 ** (VECTOR NIL) is a subtype of STRING.
2787 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2788 * minor incompatible change: some nonsensical specialized lambda
2789 lists (used in DEFMETHOD) which were previously ignored now signal
2791 * minor incompatible change: the system is now aware of the types of
2792 variables in the COMMON-LISP package, and will signal errors for
2793 most violations of these type constraints (where previously they
2794 were silently accepted).
2795 * minor incompatible change: COMPILE-FILE now uses the freedom
2796 afforded (ANSI 3.2.2.3) to use derived function types for
2797 functions defined in the same file. This also permits the system
2798 to warn on static type mismatches and function
2799 redefinition. (Currently it does not work with high DEBUG level.)
2800 * minor incompatible change: VALUES declaration is disabled.
2801 * When issuing notes, the compiler now signals a condition of type
2802 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2803 restart for use in user handlers. It is expected that the
2804 COMPILER-NOTE condition will eventually become a condition
2805 supertype to a hierarchy of note types, which will then be
2806 handleable in a similar fashion. However, at the moment, no such
2807 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2808 but it's an implementation detail, not a classification for the
2810 * Changes in type checking closed the following bugs:
2811 ** type checking of unused values (192b, 194d, 203);
2812 ** template selection based on unsafe type assertions (192c, 236);
2813 ** type checking in branches (194bc).
2814 * A short form of VALUES type specifier has ANSI meaning (it has
2815 increased the number of situations when SBCL cannot perform type
2817 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2818 keywords or constants is permissible.
2819 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2820 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2821 classes. (thanks to Antonio Martinez)
2822 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2823 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2824 argument to a pretty-print stream. Instead, it leaves any such
2825 operations to DESCRIBE-OBJECT methods.
2826 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2828 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2829 lambda lists. (thanks to David Lichteblau)
2830 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2831 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2832 (thanks to Gerd Moellmann)
2833 * fixed bug 246: increased compilation speed of long
2834 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2836 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2838 * DEFCONSTANT now throws a condition of type
2839 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2840 constant to a non-EQL value; CONTINUE and ABORT restarts
2841 respectively change and preserve the value.
2842 * fixed bug 63: The code walker, part of the implementation of CLOS,
2843 is now better at handling symbol macros.
2844 * bug fix: There is no longer an internal implementation type named
2845 CL:LENGTH. (reported by Raymond Toy)
2846 * bug fix: In macro-like defining macros/special operators the
2847 implicit block does not enclose the lambda list.
2848 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2849 enclosed in parentheses) are not suitable as type specifiers, and
2850 their use properly signals an error now.
2851 * bug fix: An argument count mismatch for a type specifier in code
2852 being compiled no longer causes an unhandled error at compile
2853 time, but signals a compile-time warning.
2854 * fixed simple vector readable printing
2855 * bug fix: DESCRIBE takes more care over whether the class
2856 precedence list slot of a class is bound before accessing it.
2857 (reported by Markus Krummenacker)
2858 * bug fix: FORMATTER can successfully compile pretty-printer format
2859 strings which use variants of the ~* directive inside.
2860 * bug fix: SEARCH now applies its TEST predicate to the elements of
2861 the arguments in the correct order. (thanks to Wolfhard Buss)
2862 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2863 in type inference. (thanks to Robert E. Brown)
2864 * bug fix: Reading in symbols with an explicit package name of ""
2865 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2866 * fixed some bugs revealed by Paul Dietz' test suite:
2867 ** NIL is now allowed as a structure slot name.
2868 ** Arbitrary numbers, not just REALs, are allowed in certain
2869 circumstances in LOOP for-as-arithmetic clauses.
2870 ** Multiple class redefinitions before slot access no longer
2871 causes a type error.
2872 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2873 association between the name and a class.
2874 ** Generic functions with non-standard method-combination and over
2875 five methods all of which return constants no longer return NIL
2876 after the first few invocations. (thanks to Gerd Moellmann)
2877 ** CALL-NEXT-METHOD with no arguments now passes the original
2878 values of the arguments, even in the presence of assignment.
2879 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2881 ** DISASSEMBLE works with closures and funcallable instances.
2882 ** ADD-METHOD now returns the generic function, not the new method.
2883 ** FIND-METHOD signals an error if the lengths of the specializers
2884 is incompatible with the generic function, even if the ERRORP
2886 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2887 which its argument is a member.
2888 ** DEFCLASS only redefines the class named by its class-name
2889 argument if that name is the proper name of the class;
2890 otherwise, it creates a new class.
2891 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2892 of the UNBOUND-SLOT condition to the name of the slot.
2893 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2894 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2895 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2896 SLOT-MAKUNBOUND in the specified fashion.
2898 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2899 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2900 cross-compilation host. As a consequence, we can now bootstrap our
2901 way up to SBCL starting with a bare gcc toolchain and human-readable
2902 source code (first the source to CLISP, then the source to SBCL).
2903 * A contributed module containing a partial implementation of the
2904 simple-streams interface has been included. (thanks to Rudi
2906 * A contributed module implementing the RFC1321 Message Digest
2907 Algorithm, known as MD5, has been included.
2908 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2909 no longer has any effect, as the code controlled by this feature
2910 has been deleted. (As far as we know, no-one has ever built using
2911 this feature, and its semantics were confused in any case).
2912 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2913 work on conditions (as required by the ANSI specification),
2914 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2915 expected behaviour on conditions. Users should note, however,
2916 that such behaviour is not required by the ANSI specification,
2917 and so use of this behaviour may render their code unportable.
2918 * fixed some bugs revealed by Paul Dietz' test suite:
2919 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2921 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2922 ** SXHASH on condition objects no longer returns NIL.
2923 ** :ALLOCATION :CLASS slots are better treated; their values are
2924 updated on class redefinition, and initforms inherited from
2925 superclasses are applied.
2926 ** REMOVE-METHOD returns its generic function argument even when
2927 no method was removed.
2928 ** SHARED-INITIALIZE now initializes the values of the requested
2929 slots, including those with :ALLOCATION :CLASS.
2930 ** ALLOCATE-INSTANCE now works on structure classes defined via
2931 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2933 ** SLOT-EXISTS-P now works on conditions, as well as structures
2935 ** MAKE-LOAD-FORM now has the required methods on
2936 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2937 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2938 keyword, and now implements the SLOT-NAMES argument.
2939 ** methods with &OPTIONAL arguments no longer allow too many
2940 arguments to be passed in the call without error.
2941 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2942 option is consistent with the required arguments of the generic
2943 function lambda list.
2944 * bug fix: REQUIRE accepts a string designator. (Thanks to
2946 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2947 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2948 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2949 * bug fix: The compiler now checks for duplicated variables in macro
2951 * bug fix: SETQ on globals returns the correct value.
2952 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2953 not a COMPILER-ERROR (followed by some other strange error on
2954 choosing the CONTINUE restart).
2955 * bug fix: make.sh and friends are now more consistent in the way that
2956 they look for GNU "make".
2958 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2959 * experimental native threads support (on x86 Linux >=2.4 only).
2960 This is not compiled in by default: you need to add :SB-THREAD to
2961 the target features. See the "Beyond ANSI" chapter of the manual
2963 * fix for longstanding nonANSIism: The old distinction between
2964 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2965 The return value from CL:FIND-CLASS is now a CLOS class, and
2966 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2967 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2968 * An interface to the MetaObject Protocol, as described in Kiczales,
2969 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2970 MIT Press, 1991, is available from the SB-MOP package.
2971 * incompatible change: the SB-PCL package should now be considered
2972 a private implementation detail, and no longer a semi-private MOP
2974 * minor incompatible change: due to rearrangement for threads, the
2975 control stack and binding stack are now allocated at arbitrary
2976 addresses instead of being hardcoded per-port. Users affected by
2977 this probably have to be doing advanced things with shared
2978 libraries, and will know who they are.
2979 * minor incompatible change: Previously, all --eval forms used were
2980 processed with READ before any of them were processed with EVAL.
2981 Now each --eval form is processed with both READ and EVAL before
2982 the next --eval form is processed. (Thus package operations like
2983 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2984 work as the user might reasonably expect.)
2985 * minor incompatible change: *STANDARD-INPUT* is now only an
2986 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2988 * minor incompatible change: Y-OR-N-P is now character-oriented,
2989 not line oriented. Also, YES-OR-NO-P now works without errors.
2990 (thanks to Antonio Martinez)
2991 * sb-aclrepl module improvements: an integrated inspector, added
2992 repl features, and a bug fix to :trace command.
2993 * Known functions, which cannot be open coded by the backend, are
2994 considered to be able to check types of their arguments. (fixing
2995 a bug report by Nathan J. Froyd)
2996 * fixed a bug in computing method discriminating functions: It is
2997 now possible to define methods specialized on classes which have
2998 forward-referenced superclasses. (thanks to Gerd Moellmann)
2999 * fixed evaluation order in optional entries (reported by Gilbert
3001 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3002 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3003 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3004 required-by-AMOP NAME argument, as well as CLASS and
3005 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3006 * fixed bug 20: DEFMETHOD can define methods using names that are
3007 not the proper names of classes to designate class specializers.
3008 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3009 Unix file descriptors, instead of blowing up. (thanks to Antonio
3011 * Garbage collection refactoring: user-visible change is that a
3012 call to the GC function during WITHOUT-GCING will not do garbage
3013 collection until the end of the WITHOUT-GCING. If you were doing
3014 this you were probably losing anyway.
3015 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3016 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3017 TYPEP the latter but not the former.
3018 * The compiler issues a full WARNING for calls to undefined functions
3019 with names from the CL package.
3020 * MAP-INTO for a vector destination is open coded. (reported by
3021 Brian Downing on c.l.l)
3022 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3023 documentation string.
3024 * fixed some bugs revealed by Paul Dietz' test suite:
3025 ** COPY-ALIST now signals an error if its argument is a dotted
3027 ** Condition slots are now accessed more correctly in the presence
3028 of multiple initargs for a given slot.
3029 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3030 exclude restarts of the same name associated with a different
3032 ** DEFCLASS of forward-referenced classes with another
3033 forward-referenced class in the superclasses list no longer
3035 ** Condition slots are now initialized once each, not multiple
3036 times. (thanks to Gerd Moellmann)
3037 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3039 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3040 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3041 arguments contain duplicated elements.
3042 ** RESTART-CASE understands local macros.
3043 ** RESTART-CASE associates exactly its own restarts with a condition.
3044 ** ENDP in safe mode checks its argument to be of type LIST.
3045 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3046 in question is unbound.
3047 ** Optimized MAKE-INSTANCE functions no longer cause internal
3048 assertion failures in the presence of duplicate initargs.
3049 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3050 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3051 propagated even in the ctor optimized implementation of
3053 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3055 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3056 * a better implementation of SXHASH on (simple) bit vectors,
3057 measured both in execution speed and in distribution of results
3058 over the positive fixnums, has been installed. Likewise, a better
3059 implementation of EQUAL for simple bit vectors is now available.
3060 * fixed CEILING optimization for a divisor of form 2^k.
3061 * fixed bug 240 (emitting extra style warnings "using the lexical
3062 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3063 by Antonio Martinez)
3064 * fixed SXHASH, giving different results for NIL depending on type
3065 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3066 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3067 variable. (found by Rolf Wester)
3068 * SBCL does not ignore type declarations for special
3069 variables. (reported by rif on c.l.l 2003-03-05)
3070 * some bug fixes in contrib/sb-aclrepl/
3071 * fixed some bugs revealed by Paul Dietz' test suite:
3072 ** a bug in the CONS type specifier, whereby the CAR and CDR
3073 types got intertwined, has been fixed;
3074 ** the type system is now able to reason about the interaction
3075 between INTEGER and RATIO types more completely;
3076 ** APPEND, [N]REVERSE and NRECONC check that those their
3077 arguments, which must be proper lists, are really so;
3078 ** An array specialized to be unable to hold elements has been
3079 implemented, as required -- yes, really -- by ANSI;
3080 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3081 on malformed property lists;
3083 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3084 * incompatible packaging change: in line with Unix convention,
3085 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3086 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3087 * REQUIRE and PROVIDE are now optionally capable of doing something
3088 useful. See the documentation string for REQUIRE.
3089 * infrastructure for a managed SBCL contrib system: contributed
3090 modules in this release include:
3091 ** the ASDF system definition facility;
3092 ** an interface to the BSD Sockets API;
3093 ** an ACL-like convenience interface to the repl;
3094 (thanks to Kevin Rosenberg)
3095 ** an implementation of ROTATE-BYTE, with efficient implementation
3097 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3098 gives a closer approximation to the right answer than previously.
3099 (thanks to Raymond Toy)
3100 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3101 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3102 optional environment arguments, as required by ANSI.
3103 * fixed bugs in other functions taking environment objects, allowing
3104 calls with an explicit NIL environment argument to be compiled
3106 * fixed bug 228: primary return values from
3107 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3108 COMPILE or FUNCTION.
3109 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3110 :TYPE will no longer signal errors on innocuous objects.
3111 * fixed bug 231b: SETQ is better at respecting type declarations in
3112 the lexical environment.
3113 * fixed a bug in DEFCLASS: classes named by symbols with no or
3114 unprintable packages can now be defined.
3115 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3116 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3117 * fixed compiler failure related to checking types of functions.
3118 (reported by Robert E. Brown)
3119 * the compiler is now much more consistent in its error-checking
3120 treatment of bounding index arguments to sequence functions: in
3121 (SAFETY 3) code, errors will be signalled in almost all cases if
3122 invalid sequence bounding indices are passed to functions defined
3123 by ANSI to operate on sequences.
3124 * fixed a bug in the build procedure: documentation of SBCL-specific
3125 packages is now preserved and available in the final Lisp image.
3126 * lifted FDEFINITION lookup out of loops in the implementation of
3127 many list operations. (thanks to Robert E. Brown)
3128 * fixed a bug in the reader: the #n# reader macro now works for
3129 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3130 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3131 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3132 E. Brown and Tony Martinez respectively)
3133 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3134 SLOT-BOUNDP: in particular, it is now able to optimize them much
3135 better, and is now not vulnerable to having packages renamed.
3136 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3137 always called when a slot is not present in an instance. (thanks
3139 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3140 longer signal an error on unreading a character following EOF on
3141 the previous constituent stream. (thanks to Tony Martinez)
3142 * fixed some bugs revealed by Paul Dietz' test suite:
3143 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3144 not just nonnegative fixnums;
3145 ** the logical bit-array operators such as BIT-AND now accept an
3146 explicit NIL for their "opt-arg" argument (to indicate a
3147 freshly-consed result bit-array);
3148 ** ELT now signals an error on an invalid sequence index in safe
3150 ** the type system is now cleverer about negations of numeric
3151 types, and consequently understands the BIGNUM and RATIO types
3153 ** the type system is now cleverer about the interaction between
3154 INTEGER and RATIO types: while bugs still remain, many more
3155 cases are accurately computed;
3156 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3157 if it is in the last clause;
3158 ** CONSTANTLY now correctly returns a side-effect-free function in
3160 ** DECLARE is no longer treated as a special-operator; in
3161 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3162 * incremented fasl file version number due to the change in the
3163 DEFSTRUCT-SLOT-DESCRIPTION structure.
3165 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3166 * minor incompatible change: code processed by the "interpreter" or
3167 EVAL now has a compilation optimization policy of (DEBUG 2)
3168 (changed from (DEBUG 1)) to improve debuggability of interactive
3169 development, and to allow the use of the debug RETURN command in
3171 * an experimental implementation of the RETURN command for the
3172 debugger has been included. (thanks to Frederik Kuivinen)
3173 * fixed bug 62: constraints were not propagated into a loop.
3174 * fixed bug in embedded calls of SORT (reported and investigated by
3176 * fixed some bugs revealed by Paul Dietz' test suite:
3177 ** printing and reading of arrays with some dimensions having
3178 length 0 (thanks to Gerd Moellmann);
3179 ** BOA constructor with &AUX argument without a default value does
3180 not cause a type error;
3181 ** CONSTANTP now returns true for all self-evaluating objects.
3183 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3184 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3185 accessors that are related by inheritance, as specified in the
3186 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3187 Valtteri Vuorikoski)
3188 * The compiler is now able to inline functions that were defined in
3189 a complex lexical environment (e.g. inside a MACROLET).
3190 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3192 * Support for the upcoming FreeBSD-5.0 release has been included.
3193 (thanks to Dag-Erling Smorgrav)
3194 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3195 effect when it is not in a toplevel context.
3196 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3197 stand a better chance of being correct. (thanks to Gerd
3199 * fixed bug in COERCE, which now signals an error on coercing a
3200 rational to a bounded real type which excludes the expected
3202 * The compiler is now able to derive types more accurately from the
3203 COERCE and COMPILE functions.
3204 * fixed bug 223: functional binding is considered to be constant
3205 only for symbols in the CL package.
3206 * fixed bug 231: SETQ did not check the type of a variable being set
3207 (reported by Robert E. Brown)
3208 * A new optimization for MAKE-INSTANCE has been included, fixing
3209 various bugs (including relating to :ALLOCATION :CLASS slots and
3210 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3212 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3213 ** As required by ANSI, LOOP now disallows anonymous collection
3214 clauses such as COLLECT I in conjunction with aggregate boolean
3215 clauses such as THEREIS (= I 1);
3216 ** LOOP now signals an error when any variable is reused in the
3217 same loop (including the potentially useful construct analogous
3218 to WITH A = 1 WITH A = (1+ A);
3219 ** IT is only a special loop symbol within the first clause of a
3220 conditional loop clause;
3221 ** LOOP with a typed iteration variable over a hashtable now
3222 signals a type error iff it should.
3223 * fixed some other bugs revealed by Paul Dietz' test suite:
3224 ** FILE-STREAM now names the class previously known as FD-STREAM;
3225 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3226 argument) no longer signals an error;
3227 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3228 of the slot symbol, rather than using the current package
3229 ((:CONC-NAME "") continues to intern the slot's name in the
3231 * incremented fasl file version number, because of the incompatible
3232 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3233 of the new implementation of DEFINE-COMPILER-MACRO.
3235 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3236 * Support for building SBCL for MIPS platforms running in
3237 little-endian mode has now been checked in, and basic
3238 functionality on said platforms verified.
3239 * minor incompatible change: PCL now records the pathname of a file
3240 in which methods and the like are defined, rather than its
3242 * minor incompatible change: TRUENAME now considers the truename of
3243 a file naming a directory to be the pathname with :DIRECTORY
3244 component indicating that directory.
3245 * minor incompatible change: a NAMED clause in the extended form of
3246 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3247 reason for the previous behaviour is unclear.
3248 * more systematization and improvement of CLOS and MOP conformance
3249 in PCL (thanks to Gerd Moellman and Pierre Mai):
3250 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3252 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3253 primary methods with no specializers;
3254 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3256 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3257 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3258 FORWARD-REFERENCED-CLASSes; error reporting on
3259 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3261 ** SXHASH on CLOS instances now uses a slot internal to the
3262 instance to return different numbers on distinct instances,
3263 while preserving the same return value through invocations of
3265 ** DEFMETHOD signals errors when methods with longer incongruent
3266 lambda lists are added to generic functions;
3267 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3268 CLASS, as specified in AMOP;
3269 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3270 on the order returned by the primary method for classes of
3271 class STANDARD-CLASS;
3272 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3273 * fixed some bugs shown by Paul Dietz' test suite:
3274 ** DOLIST puts its body in TAGBODY;
3275 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3277 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3278 value producing form;
3279 ** if more variables are given to PROGV than values, extra
3280 variables are bound and made to have no value;
3281 ** NSUBSTITUTE on list arguments gets the right answer with
3283 ** ELT signals an error of type TYPE-ERROR when the index argument
3284 is not a valid sequence index;
3285 ** LOOP signals (at macroexpansion time) an error of type
3286 PROGRAM-ERROR when duplicate variable names are found;
3287 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3288 ** FUNCALL of special-operators now cause an error of type
3290 ** PSETQ now works as required in the presence of side-effecting
3291 symbol-macro places;
3292 ** NCONC accepts any object as its last argument;
3293 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3295 ** loop-for-as-package does not require a package to be explicitely
3297 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3299 * fixed bug 166: compiler preserves "there is a way to go"
3300 invariant when deleting code.
3301 * fixed bug 172: macro lambda lists with required arguments after
3302 &REST arguments now cause an error to be signalled. (thanks to
3304 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3305 bug. (thanks to Matthew Danish)
3306 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3308 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3309 arguments are lexically rebound. (thanks to Gerd Moellmann and
3311 * fixed bug 194: error messages are now more informative when there
3312 is no primary method applicable in a call to a generic
3313 function. (thanks to Gerd Moellmann)
3314 * fixed bug in command line argument checking (thanks to Julian
3316 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3317 (thanks to Matthew Danish)
3318 * incremented fasl file version number, because of the
3319 SXHASH-related changes in the layout of CLOS data structures
3321 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3322 * minor incompatible change: The runtime (the Unix executable named
3323 "sbcl") is now much pickier about the .core files it will load.
3324 Essentially it now requires .core files to descend from the same
3325 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3326 as the runtime does. (The intent is to prevent the crashes which
3327 can occur, and which can even be reported as mysterious failures,
3328 when people patch the sources or change the build parameters
3329 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3330 sbcl and .core files.)
3331 * fixed bug: VALUES-LIST is no longer optimized away.
3332 * fixed bug 142: The FFI conversion of C string values to Lisp
3333 string values no longer conses excessively. (thanks to Nathan
3334 Froyd porting Raymond Toy's fix to CMU CL)
3335 * began to systematize and improve MOP conformance in PCL (thanks to
3336 Nathan Froyd, Gerd Moellman and Pierre Mai):
3337 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3339 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3340 ** FINALIZE-INHERITANCE is now called on class finalization;
3341 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3342 argument precedence order.
3343 * fixed bug 202: The compiler no longer fails on functions whose
3344 derived types contradict their declared type.
3345 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3346 so it can be non-toplevel.
3347 * The fasl file version number has changed (because of the new
3348 implementation of DEFMACRO).
3349 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3350 safe code, that any length requirement by their type-specifier
3351 argument is valid. The exceptions to this are described in bug
3353 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3354 functions now check the types of their inputs as required by ANSI.
3355 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3356 introduced symbol is DECLAREd to be SPECIAL.
3357 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3358 * fixed bug: :COUNT argument to sequence functions may be negative
3359 * fixed bug: body of DO-SYMBOLS may contain declarations
3360 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3361 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3362 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3363 symbol macro only once
3364 * fixed printing of call frame when argument list is unavailable
3365 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3366 * compiler no longer signals WARNING on unknown keyword
3369 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3370 * A beta-quality port to the mips architecture running Linux,
3371 based on the old CMUCL backend, has been made. It has been tested
3372 on a big-endian kernel, and works sufficiently well to be able to
3373 rebuild itself; it has not been tested in little-endian mode.
3374 * fixed an inconsistency between gencgc.c and purify.c which made
3375 dumping/loading .core files unreliable
3376 * fixed bug 120a: The compiler now deals correctly with IFs where
3377 the consequent is the same as the alternative, instead of
3378 misderiving the return type. (thanks to Alexey Dejneka)
3379 * fixed bug 113: Logical pathnames are now dumpable (the logical
3380 host is resolved at load-time, throwing an error if it is not
3382 * fixed bug 174: FORMAT's error message is slightly clearer when a
3383 non-printing character is used in a format directive.
3384 * fixed several bugs in compiler checking of type declarations, i.e.
3385 violations of the Python "declarations are assertions" principle
3386 (thanks to Alexey Dejneka)
3387 * fixed several bugs in PCL's error checking (thanks to Gerd
3389 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3391 * fixed bug in compilation of functions as first class values
3392 (thanks to Antonio Martinez-Shotton)
3393 * The compiler's handling TYPE-ERRORs which it can prove will
3394 inevitably happen at runtime has been cleaned up and corrected
3395 in several ways. (thanks to Alexey Dejneka)
3396 * improved argument type checking for various basic arithmetic
3397 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3398 lavished on them in the past that they can be compiled in many
3399 ways in different special cases
3400 * fixed bug 181: compiler checks validity of user supplied type
3402 * cleaned up code flushing in optimization: Function calls which
3403 should signal errors for safety purposes (e.g. which ANSI says
3404 should signal errors when their arguments are of incorrect type)
3405 are no longer optimized away.
3406 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3407 * incremented fasl file version number, because changes in the
3408 implementation of sequence functions like COERCE caused
3409 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3410 in old inline expansions) to become undefined. (Actually these
3411 changes were later undone, so we might very well be binary
3412 compatible with 0.7.7 after all, but leaving the version number
3413 incremented seemed like the simplest and most conservative
3416 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3417 * An alpha-quality port to the parisc architecture running Linux,
3418 based on the old CMUCL backend, has been made. This, even more so
3419 than the other backends, should be considered still a work in
3420 progress; known problems include that the Linux kernel in 64-bit
3421 mode does not propagate the correct sigcontext structure to
3422 userspace, and consequently SBCL on a parisc64 kernel will not
3424 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3425 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3426 suppressed.) Also now that the compiler is looking at declarations
3427 in the environment, it checks optimization declarations as well,
3428 and suppresses inlining when (> DEBUG SPEED).
3429 * More fixes have been made to treatment of floating point exception
3430 treatment and other Unix signals. In particular, floating point
3431 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3432 * The detection and handling of control stack exhaustion (infinite
3433 or very deeply nested recursion) has changed. Stack exhaustion
3434 detection is now done by write-protecting pages at the OS level
3435 and applies at all optimization settings; when found, a
3436 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3437 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3438 be caught using IGNORE-ERRORS.
3439 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3440 correctly when transforming recursive LABELS functions to LETs.
3441 (thanks to Alexey Dejneka)
3442 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3443 that are names of constants or global variables.
3444 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3445 alien routines with docstrings.
3446 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3447 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3449 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3450 (thanks to Eric Marsden)
3451 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3452 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3453 * Bug fix: The SPARC backend can now compile functions involving
3454 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3455 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3456 to a functions expecting a PATHNAME-DESIGNATOR.
3457 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3458 lambda lists. (thanks to Alexey Dejneka)
3459 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3460 functions so that EXPORTing the name of the class doesn't cause
3461 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3462 to Antonio Martinez for reporting this)
3463 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3464 constant-folded without failing an assertion. (thanks to Einar
3465 Floystad Dorum for reporting this)
3466 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3467 types are now dealt with more correctly by the compiler.
3468 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3469 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3470 behaviour analogous to RENAME-FILE. This puts its behaviour more
3471 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3473 * The fasl file version number has changed again. (because of the
3474 bug fix involving the names of PCL MAKE-INSTANCE functions)
3476 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3477 * bug fix: Floating point exceptions are treated much more
3478 consistently on the x86/Linux and PPC/Linux platforms.
3479 * Array initialization with :INITIAL-ELEMENT is now much faster for
3480 cases when the compiler cannot open code the array creation, but
3481 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3482 array accesses have also seen a speed increase.
3483 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3484 specified. (This was at the root of some bad interactions between
3485 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3487 * bug fix: Internal error arguments for undefined functions are now
3488 computed correctly on the PPC/Linux platform.
3489 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3490 Raymond Toy's patch for CMU CL)
3491 * Support for the Solaris 9 operating environment has been included
3492 (thanks to Daniel Merritt)
3493 * A very ugly but hopefully complete draft of the missing FFI chapter
3494 of the manual has been created by reformatting the corresponding
3495 CMU CL manual chapter into (currently very ugly and incoherent)
3496 DocBook and bringing it up to date for SBCL behavior. Thus, the
3497 manual is now essentially complete, at least by my extreme
3498 once-and-only-once standards, whereby it's acceptable to refer to
3499 the doc strings of SB-EXT functions as the primary documentation.
3500 * The fasl file version number has changed again, due to cleanup of
3501 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3502 is no longer a static symbol.)
3504 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3505 * SBCL now builds with OpenMCL (version 0.12) as the
3506 cross-compilation host; also, more progress has been made toward
3507 bootstrapping under CLISP.
3508 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3510 * bug 158 fixed: The compiler can now deal with integer loop
3511 increments different from 1; fixing this turned out also to fix
3513 * bug 169 fixed: no more bogus warnings about using lexical bindings
3514 despite the presence of perfectly good SPECIAL declarations (thanks
3515 to David Lichteblau)
3516 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3517 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3518 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3520 * bug 180 fixed: Method combination specifications no longer ignore
3521 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3522 * bug fix: Structure type predicate functions now check their argument
3523 count as they should.
3524 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3525 correctly. (thanks to Pierre Mai)
3526 * minor incompatible change: The --noprogrammer option is deprecated
3527 in favor of the new --disable-debugger option, which is very similar.
3528 (The major difference is that it takes effect at a slightly different
3529 time at startup, causing handling of errors in --sysinit and
3530 --userinit files will be affected differently.) The
3531 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3532 been added to allow this functionality to be controlled from ordinary
3533 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3534 maintainers, who might want to run non-interactive scripts to
3535 build SBCL cores which will later be used interactively.)
3536 * minor incompatible change: The LOAD function no longer, when given
3537 a wild pathname to load, loads all files matching that pathname.
3538 Instead, an error of type FILE-ERROR is signalled.
3540 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3541 * bug 147 fixed: The compiler preserves its block link/count
3542 invariants more correctly now so that it doesn't crash. (thanks
3544 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3546 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3547 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3548 no-value-for-this-character as NIL (as specified by ANSI).
3549 * HOST-NAMESTRING on physical pathnames now returns a string that is
3550 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3551 (thanks to Christophe Rhodes)
3552 * The Alpha port handles icache flushing more correctly. (thanks to
3554 * More progress has been made toward bootstrapping under CLISP. (thanks
3555 to Christophe Rhodes)
3556 * The fasl file format has changed again, because dynamic loading
3557 on OpenBSD (which has non-ELF object files) motivated some cleanups
3558 in the way that foreign symbols are transformed and passed around.
3559 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3560 is no longer treated as whitespace by the reader, but instead as
3561 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3562 |A
\x7fB|, instead of A as it used to.
3564 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3565 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3566 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3567 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3568 this as of 0.7.1.45, but was left out of the previous news section
3569 (thanks to Dan Barlow)
3570 * SBCL now runs on the Solaris operating system on SPARC architectures
3571 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3572 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3573 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3574 * SPARC backend cleanups, allowing builds of cores optimized for V8
3575 and V9 SPARCS, and also emission of code targeted to a particular
3576 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3578 * SBCL is closer to bootstrapping under CLISP, thanks to various
3579 fixes by Christophe Rhodes.
3580 * The fasl file format has changed again, to allow the compiler's
3581 INFO database to support symbol macros.
3582 * The user manual (in doc/) is formatted into HTML more nicely.
3583 (thanks to coreythomas)
3584 * The system is smarter about SUBTYPEP relationships, especially
3585 those involving NOT types (including types such as ATOM which are
3586 represented internally using NOT types). Thus SUBTYPEP is less
3587 likely to return (VALUES NIL NIL) in general, and in particular
3588 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3590 * The fasl file format has changed again, because the internal
3591 representation of types now includes a new slot to support the new
3592 SUBTYPEP-of-NOT-types logic.
3593 * (not a change in the main branch of SBCL, but a related prototype
3594 which can hopefully be merged into the main branch of SBCL in the
3595 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3596 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3597 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3598 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3599 dispatch macro character combinations. (thanks to Alexey Dejneka)
3600 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3601 that superficially look like logical namestrings correctly.
3602 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3604 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3605 default supplied-p) for &optional and &key arguments. (thanks to
3608 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3609 * incompatible change: The compiler is now less aggressive about
3610 tail call optimization, doing it only when (> SPACE DEBUG) or
3611 (> SPEED DEBUG). (This is an incompatible change because there are
3612 programs which relied on the old CMU-CL-style behavior to optimize
3613 away their unbounded recursion which will now die of stack overflow.)
3614 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3615 for non-GENCGC systems has been increased to 20M (since that
3616 seems much closer to the likely performance optimum for modern
3617 systems than the old 4M value was)
3618 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3619 and *DEBUG-PRINT-LEVEL*
3620 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3621 of CMU CL's support for SPARC, and various endianness and other
3622 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3623 * new syntactic sugar for the Unix command line: --load foo.bar is now
3624 an alternate notation for --eval '(load "foo.bar")'.
3626 ** The system now detects stack overflow and handles it gracefully,
3627 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3628 optimization settings. (This is a good thing in general, and
3629 its introduction in this version should be particularly timely
3630 for anyone whose code fails because of suppression of tail
3632 ** The system now hunts for the C variable "environ" in a more
3633 devious way, to avoid segfaults when the C library version differs
3634 between compile time and run time. (thanks to Christophe Rhodes)
3635 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3636 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3637 ** The compiler no longer issues bogus style warnings for undefined
3638 classes in the same source file as the DEFCLASSes which defined
3639 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3641 ** fixes in CONDITION class precedence list for undefined function
3642 errors (thanks to Alexei Dejneka)
3643 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3644 correctly. (thanks to Dan Barlow)
3645 ** portability fixes aiming at bootstrapping under CLISP (thanks
3646 to Dave McDonald and Christophe Rhodes)
3647 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3648 ** fixes in type translation and and type inference (thanks to
3650 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3651 ** various fixes in the new ports (thanks to Dan Barlow)
3652 * several changes related to debugging:
3653 ** suppression of tail recursion, as noted above
3654 ** stack overflow detection, as noted above
3655 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3656 is now the default. (For some time encapsulation has been more
3657 reliable than the breakpoint-based :ENCAPSULATE NIL
3658 implementation, at least on X86 systems; and I just noticed that
3659 encapsulation also seems closer to the spirit of the ANSI
3662 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3664 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3665 up properly again. (There was a packaging bug in 0.7.0 which
3666 left their definitions in SB-SYS::LOAD-FOREIGN and
3667 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3668 vital for most things which interface to C-level interfaces,
3669 like extensions working with sockets or databases or
3670 Perl-compatible regexes or whatever, and the need to fix
3671 this bug is the main reason that 0.7.1 was released so
3673 ** DEFGENERIC is now choosier about the methods it redefines, so that
3674 reLOADing a previously-LOADed file containing DEFGENERICs does
3675 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3676 cycle now works as it should. (thanks to Alexey Dejneka)
3677 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3678 fixed by Christophe Rhodes. (He actually submitted this patch
3679 months ago, and I delayed until after 0.7.0.)
3680 ** Bug 111 (internal compiler confusion about runtime checks on
3681 FUNCTION types) was fixed by Alexey Dejneka.
3682 * Some internal cleanups (getting rid of variables which aren't
3683 needed now that the byte interpreter is gone) caused the fasl
3684 file format number to change again.
3686 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3687 * major incompatible change: The default fasl file extension, i.e. the
3688 default extension for files produced by COMPILE-FILE, has changed
3689 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3691 ** There are many changes in the implementation of the compiler.
3692 SBCL is now essentially a compiler-only implementation of ANSI
3693 Common Lisp. EVAL still "interprets" a few special cases, but
3694 almost all the interesting cases are handled by creating
3695 a LAMBDA expression, calling COMPILE on it, then calling
3696 FUNCALL on the result.
3697 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3698 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3699 Since the code is newer, there might still be some new bugs
3700 (though not as many as before Martin Atzmueller's fixes:-). But
3701 the new code is substantially simpler and clearer, and hopefully
3702 any remaining bugs will be simpler, less fundamental, and more
3703 fixable then the bugs in the old code.
3704 ** The revised compiler is still a little unsteady on its feet.
3706 *** The debugging information it produces (particularly the names
3707 of FUNCTION objects) is sometimes much less useful than what
3708 the old compiler produced.
3709 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3710 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3711 within a MACROLET) has been temporarily weakened.
3712 ** There are new compiler optimizations for various functions:
3713 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3714 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3715 *** the math functions TRUNCATE, FLOOR, and CEILING
3716 *** the function-of-all-trades COERCE
3717 Mostly these should be transparent, but there's one
3718 potentially-annoying problem (bug 117): when the compiler
3719 inline-expands a function and does type analysis on the result,
3720 it can create control paths which have type mismatches, and
3721 when it can't prove that those control paths aren't taken,
3722 it will issue WARNINGs about the type mismatches. This is
3723 a particular problem in practice for the new sequence functions.
3724 It's not clear how this should be fixed, and for now, a
3725 workaround is given in the entry for 117 in the BUGS file.
3726 ** (Because of the interaction between the two previous items --
3727 occasional inlining problems and new inline expansions -- some
3728 of the new sequence function optimizations won't really kick in
3729 completely until debugging information, and then inlining, are
3730 straightened out in some future version.)
3731 * minor incompatible changes:
3732 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3733 DIRECTORY no longer implicitly promotes NIL slots of its
3734 pathname argument to :WILD. In particular, when you ask for the
3735 contents of a directory (which you used to be able to do without
3736 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3737 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3738 ** changes in behavior that ANSI explicitly defines to be
3739 implementation dependent:
3740 *** The new compiler-only implementation still conforms with ANSI,
3741 but acts a little different than before. Besides the obvious
3742 changes in performance tradeoffs (that the cost per form passed
3743 to EVAL has gone up, and the cost per form executed by EVAL
3744 has gone down), the behavior of the system changes a little
3745 because there are no longer any interpreted function objects.
3746 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3747 e.g. doing COMPILE on the output of interactive DEFUN is
3749 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3751 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3752 has changed from (:CL) to NIL.
3753 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3754 e.g. #\Newline and #\Tab, don't have names specified in the
3755 ANSI Common Lisp standard, is now based on their ASCII symbolic
3756 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3757 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3758 are no longer used for output.
3759 ** changes in internal implementation constants:
3760 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3761 4 million. (If your application spends a lot of time GCing and
3762 you have a lot of RAM, you might want to experiment with
3763 increasing it even more.)
3764 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3765 However, almost all old code should still continue to work without
3766 immediate update, as SB-C-CALL is now a (deprecated) nickname
3768 ** Old operator names in the style DEF-FOO are now deprecated in
3769 favor of new corresponding names DEFINE-FOO, for consistency with
3770 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3771 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3772 internal symbols, but a few supported extensions like
3773 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3774 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3775 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3776 etc. as you get deeper into recursive calls to the debugger
3777 command loop, instead of the old "5]", "5]]", "5]]]"
3778 sequence. (I was motivated to do this when squabbles between
3779 ILISP and SBCL left me very deeply nested in the debugger. In the
3780 short term, this change will probably provoke more ILISP/SBCL
3781 squabbles, but hopefully it will be an improvement in the long run.)
3782 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3783 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3784 function, since declaiming return types involving aliens is
3785 (1) annoyingly messy to do by hand and (2) vital to efficient
3786 compilation of code which calls such functions.
3787 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3788 longer reexported by the SB-EXT package. They're solely useful
3789 for alien code, so it seems more logical that you should get
3790 them from the SB-ALIEN package, not in SB-EXT.
3791 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3792 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3793 features. Instead, the code that they used to control is always
3794 built into the system.
3795 * many other bug fixes
3796 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3797 advantage of the new EVAL-WHEN stuff and to clean them up in
3798 general, and they are now more ANSI-compliant in a number of
3799 ways. Martin Atzmueller is responsible for a lot of this.
3800 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3802 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3803 *** correct ERROR type for various file operations
3804 *** some fixes for Lisp streams
3805 *** DEFMETHOD syntax checking
3806 *** changing old weird representation of debug information as
3807 strings (which, among their other deficiencies, don't transform
3808 correctly when you rename packages, and don't change their
3809 print representation when you change things like *PACKAGE*
3810 and *PRINT-LENGTH*) to symbols and lists of symbols
3811 He also made several improvements and fixed several bugs in DESCRIBE.
3812 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3814 *** misbehavior of WRITE-STRING/WRITE-LINE
3815 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3816 and several other LOOP problems as well
3817 *** DIRECTORY when similar filenames are present
3818 *** DEFGENERIC with :METHOD options
3819 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3820 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3821 *** argument ordering in FIND with :TEST option
3822 *** mishandled package designator argument in APROPOS-LIST
3823 *** various problems in the backquote readmacro
3824 *** a bug in APROPOS
3825 *** probably some others that I'm not describing very well here,
3826 since the CVS log documents them by reference to sbcl-devel
3827 messages, and the SourceForge archives aren't working well.:-(
3828 ** Dan Barlow improved the Alpha port (and is making progress on the
3829 PPC port, for those of you who think different).
3830 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3831 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3832 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3833 continues to work on the SPARC port (for those of you in a position
3834 to look down upon our little PC-compatible boxes from a great height).
3835 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3836 and uses that copy, rather than the current dynamic value, when
3837 it's trying to decide whether to truncate output. Thus e.g.
3838 (let ((*print-lines* 50))
3839 (pprint-logical-block (stream nil)
3841 (let ((*print-lines* 8))
3842 (print (aref possiblybigthings i) stream)))))
3843 should now truncate the logical block only at 50 lines, instead of
3844 often truncating it at 8 lines, as it did before.
3845 * The doc/cmucl/ directory, containing old CMU CL documentation
3846 from the time of the fork, is no longer part of the base system.
3847 SourceForge has shut down its anonymous FTP service, and with it
3848 my original plan for distributing the old CMU CL documentation
3849 there. For now, if you need these files you can download an old
3850 SBCL source release and extract them from it.
3851 * The fasl file version number changed again, for dozens of reasons,
3852 some of which are apparent above.
3854 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3855 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3856 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3857 patch, so that DISASSEMBLE output is much nicer.
3858 * The code in the SB-PROFILE package now seems reasonably stable.
3859 I still haven't decided what the final interface should look like
3860 (I'd like PROFILE to interact cleanly with TRACE, since both
3861 facilities use function encapsulation) but if you have a need
3862 for profiling now, you can probably use it successfully with
3863 the current CMU-CL-style interface.
3864 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3865 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3866 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3867 link now returns the dangling link itself, and for similar
3868 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3869 link itself. (In these cases the old code signalled an error and
3870 looped endlessly, respectively.) Thus, DIRECTORY now works even
3871 in the presence of dangling and cyclic symbolic links.
3872 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3873 is now a supported extension again, since the consensus on
3874 sbcl-devel was that it can be useful for ordinary development
3875 work, not just for debugging SBCL itself.
3876 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3877 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3878 that currently-defined functions might be redefined later with
3879 different return types.
3880 * Hash tables can be printed readably, as inspired by CMU CL code
3881 of Eric Marsden and SBCL code of Martin Atzmueller.
3882 * better error handling in CLOS method combination, thanks to
3883 Martin Atzmueller porting Pierre Mai's CMU CL patches
3884 * more overflow fixes for >16Mbyte I/O buffers
3885 * A bug in READ has been fixed, so that now a single Ctrl-D
3886 character suffices to cause end-of-file on character streams.
3887 In particular, now you only need one Ctrl-D at the command
3888 line (not two) to exit SBCL.
3889 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3891 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3892 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3893 does the right thing.
3894 * The implementation of some type tests, especially for CONDITION
3895 types, is now tidier and maybe faster, due to CMU CL code
3896 originally by Douglas Crosher, ported by Martin Atzmueller.
3897 * Some math functions have been fixed, and there are new
3898 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3899 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3900 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3901 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3902 likely to be built into sbcl-0.7.0.)
3903 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3904 CL, ported to SBCL by Martin Atzmueller)
3905 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3906 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3907 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3908 * There's a new slam.sh hack to shorten the edit/compile/debug
3909 cycle for low-level changes to SBCL itself, and a new
3910 :SB-AFTER-XC-CORE target feature to control the generation of
3911 the after-xc.core file needed by slam.sh.
3912 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3913 COMPILE-FILE is no longer supported, so that now every function
3914 gets an entry point, so that block compilation looks a little
3915 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3916 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3917 returns the number of bytes consed since the system started,
3918 rather than the number consed since the first time the function
3919 was called. (The new definition parallels ANSI functions like
3920 CL:GET-INTERNAL-RUN-TIME.)
3921 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3922 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3923 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3924 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3925 * Fasl file version numbers are now independent of the target CPU,
3926 since historically most system changes which required version
3927 number changes have affected all CPUs equally. Similarly,
3928 the byte fasl file version is now equal to the ordinary
3931 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3932 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3933 is no longer recognized. I apologize for this, because it was
3934 listed in SB-EXT as a supported extension, but I found that
3935 its existing behavior was poorly specified, as well as incorrectly
3936 specified, and it looked like too much of a mess to straighten it
3937 out. I have enough on my hands trying to get ANSI stuff to work..
3938 * many patches ported from CMU CL by Martin Atzmueller, with
3939 half a dozen bug fixes in pretty-printing and the debugger, and
3940 half a dozen others elsewhere
3941 * fixed bug 13: Floating point infinities are now supported again.
3942 They might still be a little bit flaky, but thanks to bug reports
3943 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3944 as flaky as they were.
3945 * The --noprogrammer command line option is now supported. (Its
3946 behavior is slightly different in detail from what the old man
3947 page claimed it would do, but it's still appropriate under the
3948 same circumstances that the man page talks about.)
3949 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3950 are now supported, and enabled by default. Thus, the compiler can
3951 handle many floating point and complex operations much less
3952 inefficiently. (Thus e.g. you can implement a complex FFT
3954 * The compiler now detects type mismatches between DECLAIM FTYPE
3955 and DEFUN better, and implements CHECK-TYPE more correctly, and
3956 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3957 thanks to patches from Martin Atzmueller.
3958 * various fixes to make the cross-compiler more portable to
3959 ANSI-conforming-but-different cross-compilation hosts (notably
3960 Lispworks for Windows, following bug reports from Arthur Lemmens)
3961 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3962 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3963 fixed thanks to Pierre Mai's CMU CL patches.
3964 * fixes to keep the system from overflowing internal counters when
3965 it tries to use i/o buffers larger than 16M bytes
3966 * fixed bug 45a: Various internal functions required to support
3967 complex special functions have been merged from CMU CL sources.
3968 (When I was first setting up SBCL, I misunderstood a compile-time
3969 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3970 * improved support for type intersection and union, fixing bug 12
3971 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3972 more obscure bugs as well
3973 * some steps toward byte-compiling non-performance-critical
3974 parts of the system, courtesy of patches from Martin Atzmueller
3975 * Christophe Rhodes has made some debian packages of sbcl at
3976 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3977 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3978 stable, but are nonetheless usable. When he's ready, I'd be happy
3979 to add them to the SourceForge "File Releases" section. (And if
3980 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3981 * new fasl file format version number (because of changes in
3982 internal representation of (OR ..) types to accommodate the new
3983 support for (AND ..) types, among other things)
3985 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3986 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3988 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3990 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3991 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3992 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3993 * DESCRIBE now works on structure objects again.
3994 * Most function call argument type mismatches are now handled as
3995 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3996 know whether the function will be redefined before the call is
3997 executed. (The compiler could flag local calls with full WARNINGs,
3998 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3999 it doesn't keep track of enough information to know whether calls
4000 are local in this sense.)
4001 * Compiler output is now more verbose, with messages truncated
4002 later than before. (There should be some supported way for users
4003 to override the default verbosity, but I haven't decided how to
4004 provide it yet, so this behavior is still controlled by the internal
4005 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4006 src/compiler/ir1util.lisp.)
4007 * Fasl file format version numbers have increased again, because
4008 support for the Gray streams extension changes the layout of the
4009 system's STREAM objects.
4010 * The Gray subclassable streams extension now works, thanks to a
4011 patch from Martin Atzmueller.
4012 * The full LOAD-FOREIGN extension (not just the primitive
4013 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4014 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4015 but like most other programs, it defaults to copying the Unix
4016 environment from the original process instead of starting the
4017 new process in an empty environment.
4018 * Extensions which manipulate the Unix environment now support
4019 an :ENVIRONMENT keyword option which doesn't smash case or
4020 do other bad things. The CMU-CL-style :ENV option is retained
4021 for porting convenience.
4022 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4023 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4025 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4027 * A patch from Martin Atzmueller seems to have solved the SIGINT
4028 problem, and as far as we know, signal-handling now works cleanly.
4029 (If you find any new bugs, please report them!)
4030 * The system no longer defaults Lisp source file names to types
4031 ".l", ".cl", or ".lsp", but only to ".lisp".
4032 * The compiler no longer uses special default file extensions for
4033 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4034 seems to expect a single default extension for all compiled code,
4035 and there's no compelling reason to try to stretch the standard
4036 to allow two different extensions.) Instead, byte-compiled files
4037 default to the same extension as native-compiled files.
4038 * Fasl file format version numbers have increased again, because
4039 a rearrangement of internal implementation packages made some
4040 dumped symbols in old fasl files unreadable in new cores.
4041 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4042 many fewer weird special cases.
4043 * Bug #17 (differing COMPILE-FILE behavior between logical and
4044 physical pathnames) has been fixed, and some related misbehavior too,
4045 thanks to a patch from Martin Atzmueller.
4046 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4047 by Tim Moore, ported to SBCL by Martin Atzmueller.
4048 * Martin Atzmueller fixed several filesystem-related problems,
4049 including bug #36, in part by porting CMU CL patches, which were
4050 written in part by Paul Werkowski.
4051 * More compiler warnings in src/runtime/ are gone, thanks to
4052 more patches from Martin Atzmueller.
4053 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4056 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4058 * DESCRIBE now works on CONDITION objects.
4059 * The debugger now handles errors which arise when trying to print
4060 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4062 * The build system now uses an additional file, customize-target-features.lisp,
4063 to allow local modifications to the target *FEATURES* list. (The point of
4064 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4065 debugging features enabled, without having to worry about propagating it
4066 into everyone's system when I do a "cvs update".) When no
4067 customize-target-features.lisp file exists, the target *FEATURES* list
4068 should be constructed the same way as before.
4069 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4070 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4071 warn about behavior which is undefined under the ANSI spec. Note: This
4072 is specified by ANSI, but it's not very popular with programmers.
4073 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4074 macro in the SBCL sources for an example of a workaround which you
4075 might use to make portable ANSI-standard code which does what you want.
4076 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4077 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4078 when it's used as a non-toplevel form. (This is required in order
4079 to implement the DEFCONSTANT-EQX macro.)
4080 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4081 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4082 funny things, due to bugs in the implementation of EVAL-WHEN
4083 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4084 non-toplevel DEFCONSTANTs are uncommon.)
4085 * The core file version number and fasl file version number have been
4086 incremented, because the old noncompliant DEFCONSTANT behavior involved
4087 calling functions which no longer exist, and because I also took the
4088 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4089 * fixed bug 1 (error handling before read-eval-print loop starts), and
4090 redid debugger restarts and related debugger commands somewhat while
4092 ** The QUIT debugger command is gone, since it did something
4093 rather different than the SB-EXT:QUIT command, and since it never
4094 worked properly outside the main toplevel read/eval/print loop.
4095 Invoking the new TOPLEVEL restart provides the same functionality.
4096 ** The GO debugger command is also gone, since you can just invoke
4097 the CONTINUE restart directly instead.
4098 ** The TOP debugger command is also gone, since it's redundant with the
4099 FRAME 0 command, and since it interfered with abbreviations for the
4101 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4103 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4104 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4105 some porting work by Martin Atzmueller)
4106 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4107 with a boatload of patches which he ported to SBCL. Now that those
4109 ** The system tries to make sure that its low-priority messages
4110 are prefixed by semicolons, to help people who like to use
4111 syntax highlighting in their ILISP buffer. (This patch
4112 was originally due to Raymond Toy.)
4113 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4114 patches originally written by Raymond Toy.
4115 ** The compiler understands coercion between single-value and
4116 multiple-VALUES type expressions better, getting rid of some very
4117 weird behavior, thanks to patches originally by Robert MacLachlan
4118 and Douglas Crosher.
4119 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4120 lambda lists, thanks to a patch originally by Pierre Mai.
4121 ** The system no longer bogusly warns about "abbreviated type
4123 ** The compiler gets less confused by inlining and RETURN-FROM,
4124 thanks to some patches originally by Tim Moore.
4125 ** The system no longer hangs when dumping circular lists to fasl
4126 files, thanks to a patch originally from Douglas Crosher.
4127 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4128 undefined function error.
4129 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4130 in structure slot initforms) for now, documented workaround instead:-|
4131 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4132 * fixed bug 5: added stubs for various Gray stream functions called
4133 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4134 installed, at least appropriate type errors are generated
4135 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4136 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4137 consistently in DEFMETHOD forms.
4138 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4139 it doesn't seem to affect SBCL after all
4140 * The C runtime system now builds with better optimization and many
4141 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4143 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4145 * The system is now under CVS at SourceForge (instead of the
4146 CVS repository on my home machine).
4147 * The new signal handling code has been tweaked to treat register
4148 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4149 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4150 downstream, e.g. in debug-int.lisp, has implicit dependencies
4151 on the unsignedness of integer representation of machine words,
4152 and that caused the system to bomb out with infinite regress
4153 when trying to recover from type errors involving signed values,
4154 e.g. (BUTLAST '(1 2 3) -1).)
4155 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4156 (This was one of the bugs Peter Van Eynde reported back in July.)
4157 * The system now uses code inspired by Colin Walters' O(N)
4158 implementation of MAP (from the cmucl-imp@cons.org mailing
4159 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4160 inline the MAP operation, and there is more than one
4161 sequence argument to the MAP call (so that it can't just
4162 do ETYPECASE once and for all based on the type of the
4163 single sequence argument). (The old non-inline implementation
4164 of the general M-argument sequence-of-length-N case required
4165 O(M*N*N) time when any of the sequence arguments were LISTs.)
4166 * The QUIT :UNIX-CODE keyword argument has been renamed to
4167 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4168 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4170 * Raymond Wiker's patches to port dynamic loading from Linux to
4171 FreeBSD have been added.
4172 * The BUGS file is now more nearly up to date, thanks in large part
4173 to Martin Atzmueller's review of it.
4174 * The debugger now flushes standard output streams before it begins
4175 its output ("debugger invoked" and so forth).
4176 * The core version number and fasl file version number have both
4177 been incremented, because of incompatible changes in the layout
4179 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4180 used to not be called for a saved Lisp image.)
4181 * Martin Atzmueller's version of a patch to fix a compiler crash,
4182 as posted on sbcl-devel 13 September 2000, has been installed.
4183 * Instead of installing Martin Atzmueller's patch for the
4184 compiler transform for SUBSEQ, I deleted the compiler transform,
4185 and transforms for some similar consing operations.
4186 * A bug in signal handling which kept TRACE from working on OpenBSD
4188 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4189 away by constant folding
4190 * The system now defines its address space constants in one place
4191 (in the Lisp sources), and propagates them automatically elsewhere
4192 (through GENESIS and the sbcl.h file). Therefore, patching the
4193 address map is less unnecessarily tedious and error-prone. The
4194 Lisp names of address space constants have also been systematized.
4195 * CVS tags like dollar-Header-dollar have been removed from
4196 the sources, because they have never saved me trouble and
4197 they've been source of trouble working with patches and other
4198 diff-related operations.
4199 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4200 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4202 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4204 * The system has been ported to OpenBSD.
4205 * The system now compiles with a simple "sh make.sh" on the systems
4206 that it's supported on. I.e., now you no longer need to tweak
4207 text in the target-features.lisp-expr and symlinks in src/runtime/
4208 by hand, the make.sh takes care of it for you.
4209 * The system is no longer so grossly inefficient when compiling code
4210 involving vectors implemented as general (not simple) vectors (VECTOR T),
4211 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4212 which dares to use the various sequence functions on non-simple
4213 vectors, takes less of a performance hit.
4214 * There is now a primitive type predicate VECTOR-T-P
4215 to test for the (VECTOR T) type, so that e.g.
4216 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4217 can now be compiled with some semblance of efficiency. (The old code
4218 turned the type declaration into a full call to %TYPEP at runtime!)
4219 * AREF on (VECTOR T) is still not fast, since it's still compiled
4220 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4221 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4222 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4223 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4224 without making full calls to SUBTYPEP at runtime.
4225 (Some analogous efficiency issues for non-simple vectors specialized to
4226 element types other than T, or for non-simple multidimensional arrays,
4227 have not been addressed. They could almost certainly be handled the
4228 same way if anyone is motivated to do so.)
4229 * The changes in array handling break binary compatibility, so
4230 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4231 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4232 instead of (VALUES T T).
4233 * By following the instructions that Dan Barlow posted to sbcl-devel
4234 on 2 July 2000, I was able to enable primitive dynamic object
4235 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4236 functionality is not implemented (since it calls ld to resolve
4237 library references automatically, requiring RUN-PROGRAM for its
4238 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4239 try to resolve library references) is now supported.
4240 * The system now flushes the standard output streams when it terminates,
4241 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4242 them at several other probably-convenient times, e.g. in each pass of
4243 the toplevel read-eval-print loop, and after evaluating a form given
4244 as an "--eval" command-line option. (These changes were motivated by a
4245 discussion of stream flushing issues on cmucl-imp in August 2000.)
4246 * The source transform for TYPEP of array types no longer assumes
4247 that an array whose element type is a not-yet-defined type
4248 is implemented as an array of T, but instead punts, so that the
4249 type will be interpreted at runtime.
4250 * There is now some support for cross-compiling in make.sh: each of
4251 the phases of make.sh has its own script. (This should be transparent
4252 to people doing ordinary, non-cross-compile builds.)
4253 * Since my laptop doesn't have hundreds of megabytes of memory like
4254 my desktop machine, I became more motivated to do some items on
4255 my to-do list in order to reduce the size of the system a little:
4256 ** Arrange for various needed-only-at-cold-init things to be
4257 uninterned after cold init. To support this, those things have
4258 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4259 symbols with such names are now uninterned after cold init).
4260 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4261 fasl files for cold load.
4262 ** Remove the old compiler structure pooling code (which used to
4263 be conditional on the target feature :SB-ALLOC) completely.
4264 ** Redo the representation of some data in cold init to be more compact.
4265 (I also looked into supporting byte compiled code at bootstrap time,
4266 which would probably reduce the size of the system a lot, but that
4267 looked too complicated, so I punted for now.)
4268 * The maximum signal nesting depth in the src/runtime/ support code has
4269 been reduced from 4096 to 256. (I don't know any reason for the very
4270 large old value. If the new smaller value turns out to break something,
4271 I'll probably just bump it back up.)
4272 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4274 * Many, many bugs reported by Peter Van Eynde have been added to
4275 the BUGS list; some have even been fixed.
4276 * While enabling dynamic object file loading, I tried to make the
4277 code easier to understand, renaming various functions and variables
4278 with less ambiguous names, and changing some function calling
4279 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4280 * While trying to figure out how to do the OpenBSD port, I tried to
4281 clean up some of the code in src/runtime/. In particular, I dropped
4282 support for non-POSIX signal handling, added various comments,
4283 tweaked the code to reduce the number of compilation warnings, and
4284 renamed some files to increase consistency.
4285 * To support the new automatic configuration functionality in make.sh,
4286 the source file target-features.lisp-expr has been replaced with the
4287 source file base-target-features.lisp-expr and the machine-generated
4288 file local-target-features.lisp-expr.
4289 * fixed a stupid quoting error in make.sh so that using CMU CL
4290 "lisp -batch" as cross-compilation host works again
4292 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4294 * DESCRIBE no longer tries to call itself recursively to describe
4295 bound/fbound values, so that it no longer fails on symbols which are
4296 bound to themselves (like keywords, T, and NIL).
4297 * DESCRIBE now works on generic functions.
4298 * The printer now prints less-screwed-up representations of closures
4299 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4300 * A private symbol is used instead of the :EMPTY keyword previously
4301 used to mark empty slots in hash tables. Thus
4302 (DEFVAR *HT* (MAKE-HASH-TABLE))
4303 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4304 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4305 now does what ANSI says that it should. (You can still get
4306 similar noncompliant behavior if bang on the hash table
4307 implementation with all the symbols you get back from
4308 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4309 This breaks binary compatibility, since tests for equality to
4310 :EMPTY are wired into things like the macroexpansion of
4311 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4313 * There's now a minimal placeholder implementation for CL:STEP,
4314 as required by ANSI.
4315 * An obscure bug in the interaction of the normal compiler, the byte
4316 compiler, inlining, and structure predicates has been patched
4317 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4318 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4320 * Missing ordinary arguments in a macro call are now detected even
4321 when the macro lambda list contains &KEY or &REST.
4322 * The debugger no longer complains about encountering the top of the
4323 stack when you type "FRAME 0" to explicitly instruct it to go to
4324 the top of the stack. And it now prints the frame you request even
4325 if it's the current frame (instead of saying "You are here.").
4326 * As specified by ANSI, the system now always prints keywords
4327 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4329 * The default initial SIZE of HASH-TABLEs is now smaller.
4330 * Type information from CLOS class dispatch is now propagated
4331 into DEFMETHOD bodies, so that e.g.
4332 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4334 is now basically equivalent to
4335 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4336 (DECLARE (TYPE SINGLE-FLOAT X))
4338 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4339 operation, without having to do run-time type dispatch.
4340 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4341 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4342 * Since I don't seem to be making big file reorganizations very often
4343 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4344 is overflowing my ability to conveniently back them up), I've finally
4345 checked the system into CVS. (The CVS repository is on my home system,
4346 not at SourceForge -- putting it on SourceForge might come later.)
4347 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4348 high-level GC-NOTIFY-FOO functions send their output. (There's
4349 still very little control of where low-level verbose GC functions
4350 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4351 less than it used to -- the GC-NOTIFY-FOO functions are now under
4352 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4353 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4354 in only one place in the source code, and propagates it automatically
4355 everywhere that it's needed. Thus e.g. when I bump the version from
4356 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4357 * The C source files now include boilerplate legalese and documentation
4358 at the head of each file (just as the Lisp source files already did).
4359 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4360 to his page will be replaced with a link to his new CLiki service.
4362 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4364 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4365 * The build process now looks for GNU make under the default name "gmake",
4366 instead of "make" as it used to. If GNU make is not available as "gmake"
4367 on your system, you can change this default behavior by setting the
4368 GNUMAKE environment variable.
4369 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4370 can build without error under CMU CL.
4372 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4374 * There is now a partial SBCL user manual (with some new text and some
4375 text cribbed from the CMU CL manual).
4376 * The beginnings of a profiler have been added (starting with the
4377 CMU CL profiler and simplifying and cleaning up). Eventually the
4378 main interface should be through the TRACE macro, but for now,
4379 it's still accessed through vaguely CMU-CL-style functions and macros
4380 exported from the package SB-PROFILE.
4381 * Some problems left over from porting CMU CL to the new
4382 cross-compilation bootstrap process have been cleaned up:
4383 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4384 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4385 part when getting rid of the compiler *BACKEND* stuff.)
4386 ** The value of *NULL-TYPE* was screwed up, because it was
4387 being initialized before the type system knew the final
4388 definition of the 'NULL type. This screwed up several key
4389 optimizations in the compiler, causing inefficiency in all sorts
4390 of places. (I found it because I wanted to understand why
4391 GET-INTERNAL-RUN-TIME was consing.)
4392 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4393 PROCLAIM FTYPE information. Unfortunately this broke binary
4394 compatibility again, since now the forms output by DEFGENERIC
4395 to refer to functions which didn't exist in 0.6.3.
4396 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4397 can use the new (as of 0.6.3) transform for SOME into MAP into
4399 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4400 values to (INTEGER 0 1000000), so that the time code will no longer
4401 occasionally get blown up by Linux returning 1000000 microseconds
4402 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4403 its output conform to the ANSI spec. (Alas, this makes its output
4404 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4405 * A full call to MAP NIL with a single sequence argument no longer conses.
4406 * fixes to problems pointed out by Martin Atzmueller:
4407 * The manual page no longer talks about multiprocessing as though
4408 it were currently supported.
4409 * The ILISP support patches have been removed from the distribution,
4410 because as of version 5.10.1, ILISP now supports SBCL without us
4411 having to maintain patches.
4412 * added a modified version of Raymond Toy's recent CMU CL patch for
4413 EQUALP comparison of HASH-TABLE
4415 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4417 * The system still can't cross-compile itself with
4418 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4419 entails), but at least it can compile more of itself that way
4420 than it used to be able to, and various buglets which were uncovered
4421 by trying to cross-compile itself that way have now been fixed.
4422 * This release breaks binary compatibility again. This time
4423 at least I've incremented the FASL file format version to 2, so that the
4424 problem can be detected reliably instead of just causing weird errors.
4425 * various new style warnings:
4426 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4427 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4428 ** using the lexical binding of a variable named in the *FOO* style
4429 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4430 as specified by ANSI.
4431 * *RANDOM-STATE* is no longer automatically initialized from
4432 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4433 default behavior of the system is to repeat its behavior every time
4434 it's run. If you'd like to change this behavior, you can always
4435 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4436 old convention there was no comparably easy way to get the system to
4437 repeat its behavior every time it was run.
4438 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4439 FTYPE declarations has been removed, in favor of their ANSI
4440 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4441 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4442 the types of their sequence arguments can be determined at compile time.
4443 This is done through a new open code expansion for MAP which eliminates
4444 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4445 argument types can be determined at compile time.
4446 * The optimizer now transforms COERCE into an identity operation when it
4447 can prove that the coerced object is already of the correct type. (This
4448 can be a win for machine generated code, including the output of other
4449 optimization transforms, such as the MAP transform above.)
4450 * Credit information has been moved from source file headers into CREDITS.
4451 * Source file headers have been made more standard.
4452 * The CASE macro now compiles without complaining even when it has
4455 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4457 * (Note that the way that the PCL macroexpansions were rewritten
4458 to accommodate the change in DEFGENERIC below breaks binary
4459 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4460 not run under sbcl-0.6.2. Once we get out of alpha releases,
4461 i.e. hit release 1.0.0, we'll probably try to maintain binary
4462 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4463 and sbcl-1.4.4. Until then, however, it might be fairly common
4464 for maintenance releases to break binary compatibility.)
4465 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4467 * The reporting of unhandled signals has been changed to print some
4468 explanatory text as well as the report form. (Previously only
4469 the report form was printed.)
4470 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4471 it defines, so that the compiler no longer issues undefined function
4472 warnings for compiled-but-not-yet-loaded generic functions.
4473 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4474 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4475 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4476 section of the ANSI spec are supported.
4477 * The "" nickname for the "KEYWORD" package has been removed.
4478 The reader still handles symbol tokens which begin with a package marker
4479 as keywords, but it doesn't expose its mechanism for doing so in the
4480 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4481 * The system now issues STYLE-WARNINGs for contradictory TYPE
4482 proclamations. (Warnings for contradictory FTYPE proclamations would
4483 be nice too, but those can't be done usefully unless the type system
4484 is made smarter about FUNCTION types.)
4485 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4486 systematized, so that "*target-*.lisp is supposed to exist only on the
4487 target and imply that there's a related file which exists on the
4488 host, and *host-*.lisp is supposed to exist only on the host and imply
4489 that there's a related file which exists on the target. This involves a
4490 lot of renaming. Hopefully the acute confusion caused by the renaming
4491 will be justified by the reduction in chronic confusion..
4492 ** runtime-type.lisp -> early-target-type.lisp
4493 ** target-type.lisp -> late-target-type.lisp
4494 ** early-host-format.lisp -> early-format.lisp
4495 ** late-host-format.lisp -> late-format.lisp
4496 ** host-error.lisp -> misc-error.lisp
4497 ** early-error.lisp -> early-target-error.lisp
4498 ** late-error.lisp -> late-target-error.lisp
4499 ** host-defboot.lisp -> early-defboot.lisp
4500 ** code/misc.lisp -> code/target-misc.lisp
4501 ** code/host-misc.lisp -> code/misc.lisp
4502 ** code/numbers.lisp -> code/target-numbers.lisp
4503 ** code/early-numbers.lisp -> numbers.lisp
4504 ** early-host-type.lisp -> early-type.lisp
4505 ** late-host-type.lisp -> late-type.lisp
4506 ** host-typep.lisp -> typep.lisp
4507 ** load.lisp -> target-load.lisp
4508 ** host-load.lisp -> load.lisp
4509 ** host-disassem.lisp -> disassem.lisp
4510 ** host-insts.lisp -> insts.lisp
4511 ** byte-comp.lisp -> target-byte-comp.lisp
4512 ** host-byte-comp.lisp -> byte-comp.lisp
4513 ** host-signal.lisp -> signal.lisp
4514 ** host-defstruct.lisp -> defstruct.lisp
4515 ** late-target-type.lisp -> deftypes-for-target.lisp
4516 Furthermore, several other previously target-only files foo.lisp (e.g.
4517 hash-table.lisp and random.lisp) have been split into a target-and-host
4518 foo.lisp file and a target-only target-foo.lisp file, with their key type
4519 definitions in the target-and-host part, so that the cross-compiler will
4520 know more about target types.
4521 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4522 gone away. In their place are various *BACKEND-FOO* variables
4523 corresponding to the slots of the old structure.
4524 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4525 into the target SBCL has been fixed.
4526 * The chill.lisp system for loading cold code into a running SBCL
4528 * Support for the CMU CL "scavenger hook" extension has been removed.
4529 (It was undocumented and unused in the CMU CL sources that SBCL was
4530 derived from, and stale in sbcl-0.6.1.)
4531 * Various errors in the cross-compiler type system were detected
4532 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4533 (enabling various consistency checks). Many of them were fixed,
4534 but some hard problems remain, so the compiler is back to
4535 running without *TYPE-SYSTEM-INITIALIZED* for now.
4536 * As part of the cross-compiler type system cleanup, I implemented
4537 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4538 * I have started adding UNCROSS calls throughout the type system
4539 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4540 on cross-compiler input files will be able to go away, and various
4542 * CONSTANTP now returns true for quoted forms (as explicitly required
4545 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4547 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4548 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4549 (MAKE-LIST -1). In the longer term, it ought to become true
4550 that declarations are assertions even at SAFETY 1. For now, it's not
4551 quite true even at SAFETY 3, but it's at least more nearly true..
4552 (Note that this change seems to increases the size of the system by
4553 O(5%) and to decrease the speed of the compiler by 20% or more.)
4554 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4555 for the problem of printing dozens of lines of distracting information
4556 about low-level system machinery as part of the top stack frame
4557 on entry to the debugger when an undefined function was called.
4558 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4559 is not reset to COMMON-LISP-USER.
4560 * Compilation of stuff related to dyncount.lisp has been made conditional
4561 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4562 smaller. The various dyncount-related symbols have been moved into
4563 a new "SB-DYNCOUNT" package.
4564 * tty-inspect.lisp has been renamed to inspect.lisp.
4565 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4566 feature has gone away. (When we eventually port to other flavors of
4567 libc and/or Unix, we'll try to make the differences between flavors
4568 invisible at the user level.)
4569 * Various other *FEATURES* tags, and/or their associated conditionals,
4570 have been removed if obsolescent, or given better documentation, or
4571 sometimes given more-mnemonic names.
4573 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4575 * tidied up "make.sh" script
4576 * tidied up system directory structure
4577 * better "clean.sh" behavior
4578 * added doc/FOR-CMUCL-DEVELOPERS
4579 * many many small tweaks to output format, e.g. removing possibly-confusing
4580 trailing #\. character in DESCRIBE-INSTANCE
4581 * (EQUALP #\A 'A) no longer signals an error.
4582 * new hashing code, including EQUALP hashing
4583 * tidied up Lisp initialization and toplevel
4584 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4585 * command line argument processing
4586 * added POSIX-GETENV function to deal with Unix-ish environment variables
4587 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4588 terminating SBCL on EOF
4589 * non-verbose GC by default
4590 * There is no more "sbcl" shell script; the sbcl file is now the C
4591 runtime executable (just like CMU CL).
4592 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4594 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4595 symbol and package information as Lisp native symbol and package objects
4596 instead of strings naming symbols and strings naming packages. This way,
4597 whenever packages are renamed (as in warm init), debug information is
4598 transformed along with everything else.
4599 * tweaked the optimization policy declarations which control the building
4600 of SBCL itself. Now, among other things, the system no longer saves
4601 source location debugging information. (This helps two problems at once
4602 by reducing SBCL size and by keeping SBCL from trying to look for its
4603 sources -- which may not exist -- when reporting errors.)
4604 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4605 debugging and testing purposes
4606 * cleaned up printing, making the printer call PRINT-OBJECT for
4607 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4608 methods, giving nearly-ANSI behavior
4609 * converted almost all special variables to use *FOO* naming convention
4610 * deleted PARSE-TIME functionality, since it can be done portably
4611 * moved some files out of cold init into warm init
4612 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4614 * regularized formatting of source files
4615 * added an install.sh script
4616 * fixed ridiculous memory usage of cross-compiler by making
4617 compiler/alloc.lisp not try to do pooling unless it can hook
4618 itself into the GC of the cross-compilation host. Now the system
4619 builds nicely on my old laptop.
4620 * added :SB-ALLOC in target-features.lisp-expr
4621 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4622 was not implemented)
4623 * re-did condition handling and note reporting in the compiler. Notes
4624 are no longer handled by signalling conditions. Style warnings
4625 and warnings are handled more correctly and reported in such a way
4626 that it's easy to find one or the other in your output (so that you
4627 can e.g. figure out which of many problems caused COMPILE-FILE to
4629 * changed the severity of several compiler warnings from full WARNING
4630 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4631 compiler note reporting so that it doesn't use the condition system
4632 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4633 COMPILE-FILE command)
4634 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4635 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4636 no longer supported, since their implementation was deeply intertwingled
4637 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4638 * removed bogus "support" for compiler macros named (SETF FOO), and
4639 removed the compiler macro for SETF INFO (but only after making a fool
4640 of myself on the cmucl-imp mailing list by posting a bogus patch for
4641 DEFINE-COMPILER-MACRO..)
4642 * Compiled files containing forms which have side effects on the Lisp
4643 reader (such as DEFPACKAGE forms) are now handled more correctly.
4644 (Compiler queuing of top level lambdas has been suppressed by setting
4645 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4646 * deleted various currently-unused source files, e.g. gengc.lisp. They
4647 may be added back at some point e.g. when porting to other architectures,
4648 but until they are it's distracting to distribute them and to try to
4650 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4651 there were so many of them they're just distractions, and UNCROSS is
4652 known to be able to handle the current sources
4653 * moved PROFILE functionality into TRACE, so that it will be clear
4654 how the wrapping and unwrapping of functions when you profile them
4655 interacts with the wrapping and unwrapping of functions when you
4656 trace them. (Actually, the functionality isn't there yet, but at least
4657 the interface specification is there. Hopefully, the functionality will
4658 arrive with some future maintenance release.)
4659 * removed host-oops.lisp
4660 * changed signature of QUIT function to allow UNIX-CODE argument
4661 * fixed READ-SEQUENCE bug
4662 * tweaked verbose GC output so that it looks more like the progress
4663 output that ANSI specifies for functions like LOAD
4664 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4665 * added <http://sbcl.sourceforge.com> to the banner information printed by