1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.18 relative to 1.0.17:
3 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
4 profiles only the current thread.
5 * enhancement: SB-SPROF now has support for wallclock profiling,
6 and is also able to profile specific threads. REPORT output
7 has also additional sorting options.
8 * optimization: structure allocation has been improved
9 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
10 ** out of line constructors are ~10% faster.
11 ** inline constructors are ~15% faster.
12 ** inline constructors are capable of dynamic extent allocation
13 (generally on x86 and x86-64, in some cases on other platforms
15 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
17 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
19 * fixed some bugs revealed by Paul Dietz' test suite:
20 ** NIL is a valid function name (regression at 1.0.13.38)
21 ** FILL on lists was missing its return value (regression at 1.0.12.27)
23 changes in sbcl-1.0.17 relative to 1.0.16:
24 * temporary regression: user code can no longer allocate closure
25 variable storage on stack, due to bug 419 without explicitly
26 requesting it. Please consult sbcl-devel for advice if you need to
27 use this feature in the meanwhile.
28 * new feature: runtime argument --control-stack-size can be used to
29 adjust thread default control stack size.
30 * enhancement: improved TIME output
31 ** all times are reported using the measured accuracy (milliseconds
32 for real and GC times, microseconds for everything else.)
33 ** processor cycle counts on x86 and x86-64.
34 ** interpreted forms are counted for both evaluator modes.
35 ** number of lambdas converted by the compiler is reported.
36 ** CPU percentage report (computed from real and total run time.)
37 ** more comprehensive run time reporting, using a condenced format
38 ** interperted form, lambda, and page fault counts are omitted
40 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
42 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
43 * optimization: two argument forms of LAST are upto ~10% faster
44 in normal SPEED policies.
45 * optimization: NCONC no longer needs to heap cons its &REST list
46 in normal SPEED policies.
47 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
49 * bug fix: UNION and NUNION work with :TEST-NOT once more,
50 regression since 1.0.9.1. (thanks to Eric Marsden)
51 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
52 since 1.0.15.36. (thanks to Paul Khuong)
53 * bug fix: LAST when always returned the whole list when given a bignum
54 as the second argument.
55 * bug fix: dynamic extent allocation of nested lists and vectors
56 could leak to otherwise accessible parts.
57 * bug fix: invalid optimization of heap-allocated alien variable
59 * bug fix: fasl header checking is less vulnerable to different
60 platform word lengths.
61 * bug fix: more correct assembler syntax for GNU binutils
62 2.18.50.0.4 support. (thanks to Marijn Schouten)
63 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
65 * bug fix: the fopcompiler can handle LOCALLY forms (with no
66 declarations) successfully. (reported by Attila Lendvai)
68 changes in sbcl-1.0.16 relative to 1.0.15:
69 * minor incompatible change: revert the changes to sb-posix's error
70 signaling added in 1.0.14.
71 * minor incompatible change: change PROBE-FILE back to returning
72 NIL whenever we can't get a truename, as was the case before 1.0.14.
73 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
74 inherits from ERROR instead of just CONDITION.
75 * new feature: SB-INTROSPECT can provide source locations for instances
76 as well. (thanks to Tobian Ritterweiler)
77 * optimization: binding special variables now generates smaller code
78 on threaded platforms.
79 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
80 and cases where no :TEST is given but the compiler can infer that
81 the element to search is of type (OR FIXNUM (NOT NUMBER)).
82 * optimization: better LOGNOT on fixnums.
83 * optimization: modular arithmetic for a particular requested width
84 is implemented using a tagged representation unless a better
85 representation is available.
86 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
87 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
88 for wellformedness and returns a specialized vector. (reported by
90 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
91 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
92 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
93 unscheduling timer at the same time as another timer fires.
94 * bug fix: don't reschedule timers for dead threads.
95 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
96 * bug fix: copying output from RUN-PROGRAM to a stream signalled
97 bogus errors if select() was interrupted.
98 * enhancement: add support for fcntl's struct flock to SB-POSIX.
100 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
101 * enhancement: cleaner backtraces for interactive interrupts, as
102 well as other cases where the interesting frames used to be
103 obscured by interrupt handling frames.
104 * enhancement: untracing a whole package using (UNTRACE "FOO") is
105 now supported, and tracing a whole package using (TRACE "FOO") now
106 traces SETF-functions as well.
107 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
108 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
109 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
110 * unreadably printed representation of hash-tables now includes
112 * bug fix: partially fixed #188: type propagation from assignments
113 is now more efficient.
114 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
115 'DOUBLE-FLOAT) are not flushable.
116 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
117 full address of the object, and none of the tag bits.
118 * bug fix: readably printing hash-tables now respects other printer
119 control variables. (reported by Cedric St-Jean)
120 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
121 keyword with MAKE-HASH-TABLE.
122 * bug fix: export SB-POSIX:MKSTEMP.
123 * bug fix: SORT was not interrupt safe.
124 * bug fix: XREF accounts for the last node of each basic-block as
126 * bug fix: MAKE-INSTANCE optimizations interacted badly with
127 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
128 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
129 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
130 create obsolete instances in the case of redefinition or
131 obsoletion of a superclass. (thanks to Andy Hefner)
132 * bug fix: Support for the Alpha architecture has been revived; it had
133 suffered somewhat from lack of maintenance since sbcl-1.0.
134 * improvements to the Windows port:
135 ** The system detects the codepage to use at startup, instead of
136 using the value from the saved core. (thanks to Kei Suzuki)
138 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
139 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
140 (see documentation for details.)
141 * revived support for OpenBSD (contributed by Josh Elsasser)
142 * partially fixed bug #108: ROOM no longer suffers from occasional
143 (AVER (SAP= CURRENT END)) failures.
144 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
145 bodies no longer cause a WARNING to be signalled. (reported by
147 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
148 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
149 the bug in TRUNCATE.) (reported by Michael Weber)
150 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
151 no samples. (reported by Andy Hefner)
152 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
153 no longer appear as (NIL ...) frames in backtraces.
154 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
155 is now actually exported from SB-EXT as documented. (reported by
157 * bug fix: sb-aclrepl now correctly understands how to inspect
158 single-floats on 64-bit platforms where single-floats are not boxed.
159 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
160 yet been finalized. (reported by Levente Meszaros)
161 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
162 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
163 SBCL-specific optimize qualities.
165 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
166 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
167 an executable in the search path, and does so in the child
168 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
169 been removed; it can be found in the CVS history, for somebody who
170 needs that search behavior (see the manual).
171 * minor incompatible change: compiler policy re. weakening type
172 checks has changed: now type checks are weakened only if SAFETY < 2
174 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
175 unparsing of directory pathnames as files. Analogously,
176 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
177 filename to parse into a directory pathname.
178 * enhancement: implicit generic function creation now signals a
179 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
180 users can bind handlers for and muffle around calls to LOAD.
181 * enhancement: RUN-PROGRAM allows unicode arguments and environments
182 to be used (using the default stream external format), and allows
183 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
184 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
185 strings and vectors whose element-type or simplicity is not fully
186 known at compile-time.
187 * optimization: STRING-TRIM and related functions no longer allocate
188 a new string when no trimming needs to be performed. These functions
189 are also faster than before when the input string has been declared
191 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
192 * optimization: READ-LINE is significantly faster for files containing
194 * optimization: non-open coded uses of character comparison operators
195 (e.g. char=) no longer cons when called with more than one parameter
196 on platforms supporting dynamic-extent allocation.
197 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
198 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
199 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
201 * bug fix: some sequence functions elided bounds checking when
203 * bug fix: too liberal weakening of union-type checks when SPEED >
205 * bug fix: more bogus fixnum declarations in ROOM implementation
208 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
209 * new feature: MAKE-HASH-TABLE now experimentally accepts a
210 :SYNCHRONIZED argument, which makes the hash-table safe for
211 concurrent accesses (but not iteration.) See also:
212 SB-EXT:WITH-LOCKED-HASH-TABLE, and
213 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
214 * optimization: CONCATENATE on strings is an order of magnitude faster
215 in code compiled with (> SPEED SPACE).
216 * optimization: SUBSEQ is ~50% faster on lists.
217 * bug fix: bug 417 fixed -- source location reporting is now more robust.
218 * bug fix: SUBSEQ on a list will now correctly signal an error if if
219 END is smaller then START.
220 * bug fix: SB-PROFILE will no longer report extra consing for nested
221 calls to profiled functions.
222 * bug fix: ROOM implementation had bogus fixnum declarations which
223 could cause type-errors when calling ROOM on large images.
224 * bug fix: if file compilation is aborted, the partial fasl is now
225 deleted, and COMPILE-FILE returns NIL as the primary value.
226 * bug fix: number of thread safety issues relating to SBCL's internal
227 hash-table usage have been fixed.
228 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
229 be returned from its body when the values were being returned
230 using unknown-values return convection and the W-P-O was wrapped
231 inside an UNWIND-PROTECT.
232 * bug fix: sb-posix should now compile again under Windows, enabling
235 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
236 * incompatible change: hash-table accessor functions are no longer
237 automatically protected by locks. Concurrent accesses on the same hash-table
238 from multiple threads can give inconsistent results or even corrupt the
239 hash-table completely. Multi-threaded applications should do their own
240 locking at the correct granularity. In the current implementation it is
241 still safe to have multiple readers access the same table, but it's not
242 guaranteed that this property will be maintained in future releases.
243 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
244 and will signal an error at runtime.
245 * enhancement: SB-THREAD package now exports a semaphore interface.
246 * enhancement: CONS can now stack-allocate on x86 and
247 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
249 * enhancement: nested lists can now be stack allocated on
250 platforms providing stack allocation support.
251 * enhancement: dynamic-extent support has been extended to support
252 cases where there are multiple possible sources for the stack
254 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
255 if the mutex is uncontested on Linux.
256 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
257 as specified by the CLHS. (thanks to Tobias Rittweiler)
258 * bug fix: NaN comparison now works on x86-64.
259 * bug fix: CLOSE :ABORT T on a stream with pending output now
261 * bug fix: instances of non-standard metaclasses using standard
262 instance structure protocol sometimes missed the slot type checks
264 * bug fix: known functions can be inlined in high-debug code.
266 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
267 * minor incompatible change: the MSI installer on Windows no longer
268 associates .lisp and .fasl files with the installed SBCL.
269 * minor incompatible change: :UNIX is no longer present in *FEATURES*
270 on Windows. (thanks to Luis Oliviera)
271 * new feature: SB-CLTL2 contrib module now implements
272 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
273 * optimization: scavenging weak pointers is now more efficient,
274 requiring O(1) instead of O(N) per weak pointer to identify
275 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
276 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
277 method bodies using constant slot-names when the first argument is
278 a specializer parameter for the method.
279 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
280 names, when the class of the instance is a direct instance
281 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
282 as slow as the constant slot-name case.)
283 * optimization: member type construction is now O(N) instead
285 * optimization: UNION and NUNION are now O(N+M) for large
286 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
288 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
289 now more readable in environments like Slime which display it.
290 (thanks to Tobias C. Rittweiler)
291 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
292 was able to infer the correct argument type for the object on which
293 the CAS operation was being performed.
294 * bug fix: copy propagation interfered with parallel assignment
295 semantics in local calls. (reported by Paul Khuong)
296 * bug fix: the signed modular fixnum shift compiled to wrong code on
297 x86 and x86-64. (spotted by a slight modification to some of
300 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
301 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
302 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
303 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
305 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
306 and generic function now print the object identity for anonymous
307 instances. (thanks to Attila Lendvai)
308 * enhancement: as an extension to MOP, SBCL now supports SETF of
309 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
310 (thanks to Attila Lendvai)
311 * enhancement: sb-cover has an alternate annotation mode which puts
312 the form annotations at the car of the form, instead of annotating
313 it completely. The benefit of this mode is that it shows better
314 which forms were instrumented by the compiler.
315 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
316 by printer control variables. (reported by Dan Corkill)
317 * bug fix: system leaked memory when delayed output was performed by
318 the OS in smaller chunks then expected. (thanks to David Smith)
319 * bug fix: system leaked memory when file streams were not closed
321 * bug fix: large objects written to slow streams that were modified
322 after the write could end up with the modified state written to
323 the underlying file descriptor.
324 * bug fix: multiple threads operating in parallel on the same stream
325 could cause buffer-overflows.
326 * bug fix: source location information is stored correctly
327 (broken since 1.0.6). This bug would generally show up as the
328 Slime debugger highlighting the wrong form when the sldb-show-source
330 * bug fix: Branch forms are again annotated as branches in the sb-cover
332 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
333 on x86-64. (reported by Gregory Vanuxem)
334 * bug fix: EQUALP could return wrong results for structures with raw
335 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
336 word). (reported by Vjacheslav Fyodorov)
337 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
340 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
341 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
342 atomic compare-and-swap operations on threaded platforms.
343 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
344 allows assigning a global minimum value to optimization qualities
345 (overriding proclamations and declarations).
346 * enhancement: closed over variables can be stack-allocated on x86
348 * performance bug fix: GETHASH and (SETF GETHASH) are once again
350 * optimization: slot definition lookup is now O(1). This speeds up
351 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
352 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
353 * optimization: ASSOC and MEMBER can now be open-coded for all
354 combinations of keyword arguments when second argument is constant
355 and SPEED >= SPACE. In other cases a specialized version is
357 * bug fix: using obsoleted structure instances with TYPEP and
358 generic functions now signals a sensible error.
359 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
360 (reported by Kristoffer Kvello)
361 * bug fix: backtrace construction is now more careful when making
362 lisp-objects from pointers on the stack, to avoid creating bogus
363 objects that can be seen by the GC.
364 * bug fix: defaulting of values in contexts expecting more than 7
365 variables now works on x86-64. (reported by Christopher Laux)
366 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
368 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
369 as the property-list of a symbol.
370 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
371 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
372 situations requested, are once again file-compileable. (reported
375 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
376 * MOP improvement: support for user-defined subclasses of
377 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
378 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
379 the expansion of DEFMETHOD, is responsible for generating a form
380 which creates a list of specializers when evaluated. Additional
381 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
382 debugging and introspective support.
383 * minor incompatible change: the (unsupported) spinlock interface
384 has changed: free spinlock now has the value NIL, and a held spinlock
385 has the owning thread as its value.
386 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
387 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
389 * enhancement: name of a socket-stream is now "a socket" instead of
391 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
392 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
393 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
394 for generic functions. (Thanks to Tobias C. Rittweiler)
395 * optimization: bignum printing speed has been improved by 20-40%
396 (depending on the bignum size.)
397 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
399 * bug fix: the cache used by the CLOS to store precomputed effective
400 methods, slot offsets, and constant return values is now thread and
402 * bug fix: generic function dispatch function updating is now thread
403 and interrupt safe (in the sense that the known issues have been
405 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
406 * bug fix: interning EQL-specializers is now thread and interrupt safe.
407 * bug fix: asdf systems with dependencies to the SB-POSIX or
408 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
409 * bug fix: interrupt safety of applicable method computation has been
412 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
413 * new contrib: sb-cover, an experimental code coverage tool, is included
415 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
416 significantly faster.
417 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
418 produce more efficient code on x86-64 if the compiler can determine
419 that (+ INDEX <constant>) does not require a bounds check and FOO
420 has an element type at least 8 bits wide.
421 * enhancement: a new, experimental synchronous timeout facility is
422 provided. Refer to SB-SYS:WITH-DEADLINE for details.
423 * enhancement: when a symbol name conflict error arises, the
424 conflicting symbols are always printed with a package prefix.
425 (thanks to Kevin Reid)
426 * enhancement: stepping is now once again supported on the SPARC and
427 MIPS platforms. (It is also now more likely to work on CheneyGC
429 * enhancement: sb-sprof can now also track and report accurate call
431 * bug fixes: the treatment of non-standard subclasses of
432 SB-MOP:SPECIALIZER is more correct.
433 * incompatible change: PURIFY no longer copies the data from the
434 dynamic space into the static and read-only spaces on platforms
435 that use the generational garbage collector
436 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
438 * bug fix: GC race condition occasionally resulting in crashes with
439 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
441 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
443 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
444 a STYLE-WARNING to be signalled (regression from 1.0.4.)
445 * bug fix: an asynchronous interrupt could previously leave the
446 system running with GC inhibited.
447 * bug fix: a DECLARE form evaluated at top-level now causes an error
448 rather than silently (or verbosely) returning NIL.
449 * bug fix: trying to dissassemble functions compiled at high DEBUG could
450 result in a error being signalled due to source form lookup errors.
451 (reported by Peter Graves)
453 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
454 * incompatible change: removed writer methods for host-ent-name,
455 host-ent-addresses -- changing the values did not update the DNS
457 * minor incompatible change: changed experimental JOIN-THREAD interface
458 * documentation: the manual now lists reader and writer methods
459 in class slot documentation sections. (thanks to Richard M Kreuter)
460 * documentation: unwinding from asyncronous events has been
461 documented as unsafe.
462 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
463 in multithreaded application code.
464 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
465 platforms. (thanks to James Anderson for the optimization hint)
466 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
468 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
469 variants no longer cons.
470 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
471 their NOT- variants no longer cons.
472 * optimization: EQUAL hash tables no longer use SXHASH for objects
473 of all data types, but instead use an EQL hash for types for which
474 EQUAL is the same as EQL.
475 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
476 are significantly faster.
477 * optimization: new STRING-OUTPUT-STREAM implementation is much
478 faster for large outputs and conses 30% less on average.
479 * enhancement: XREF information is now collected for references made
480 to global variables using SYMBOL-VALUE and a constant argument.
481 * enhancement: SIGINT now causes a specific condition
482 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
483 * bug fix: FILE-POSITION used to signal an error in some cases where
484 ANSI requires it to return NIL.
485 * bug fix: ADJUST-ARRAY is now interrupt-safe.
486 * bug fix: adding and removing fd-handlers is now interrupt-safe.
487 * bug fix: inlined calls to C now ensure 16byte stack alignment on
489 * bug fix: bad type declaration in the CLOS implementation has
490 been fixed. (reported by James Anderson)
491 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
492 (thanks to Lutz Euler)
493 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
494 required by the CLHS. (reported by Samium Gromoff)
495 * bug fix: dead unbound variable references now signal an error.
496 (reported by Marco Monteiro)
497 * bug fix: / with an unused value was being deleted in safe code.
498 (thanks to Marco Monteiro and Kevin Reid)
499 * bug fix: number of characters that can be written onto a single
500 line in a file is unlimited.
501 * bug fix: some GC deadlocks caused by asynchronous interrupts have
502 been fixed by inhibiting interrupts for when GC is disabled.
503 * bug fix: some interrupt safety issues with GETHASH, (SETF
504 GETHASH), CLRHASH and REMHASH have been fixed.
505 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
506 specifier no longer causes infinite recursion.
507 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
508 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
509 * bug fix: an &environment argument with the correct variable information
510 is passed to macros that are expanded during byte compilation.
511 (reported by Samium Gromoff)
512 * bug fix: a logic error was causing the hash function for bignums to
513 have excessive amounts of collisions. (reported by Faré Rideau)
514 * bug fix: modifying the contents of an array could change the return
515 value of SXHASH on that array, which is only allowed for strings
516 and bit vectors. (bug introduced in 0.9.16)
517 * improvement: the x86-64/darwin port now passes all tests (except
518 for the debugger tests) but should still be considered
519 experimental until this is fixed.
520 * improvement: a style-warning is signaled for CASE (etc) clauses with
521 duplicate keys. (thanks to Kevin Reid)
522 * improvement: macos/x86 and macos/x86-64 now use mach exception
523 handlers for dealing with illegal instructions (for trapping and
524 error handling) and memory protection violations (for GC).
526 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
527 * new platform: experimental support for x86-64/darwin (MacOS).
528 * incompatible change: the thread-safe (on most platforms) getaddrinfo
529 and getnameinfo sockets functions are used instead of gethostbyaddr
530 and gethostbyname, on platforms where the newer functions are available.
531 As a result, the ALIASES field of HOST-ENT will always be NIL on these
533 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
534 don't print the startup banner, but behave as if --noinform was passed
535 as a command line argument. (thanks to Kevin Reid)
536 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
537 * optimization: code using alien values with undeclared types is much faster.
538 * optimization: the compiler is now able to open code SEARCH in more cases.
539 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
540 * bug fix: using standardized COMMON-LISP special variables as loop
541 variables no longer signals bogus package lock violations. (reported
543 * bug fix: declaring local loop variables to be of a range-limited type
544 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
545 (reported by Andras Simon)
546 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
547 bugs remain on x86-64.)
548 * bug fix: the #= and ## reader macros now interact reasonably with
549 funcallable instances.
550 * bug fix: type-checks for function arguments were compiled using the
551 compiler policy settings of the wrong lexical environment (in 1.0.2
553 * bug fix: SHADOW accepts characters as string designators, as required
554 by the spec (thanks to Eric Marsden)
555 * bug fix: fixed GC safety issues when foreign functions are called with
556 non-base strings as arguments
557 * bug fix: more consistent error output for fatal-compiler-warnings, like
559 * bug fix: functions with proper names were showing up as NIL in
560 backtraces (reported by Edi Weitz, regression in 1.0.2)
562 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
563 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
564 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
565 produce more efficient code on the x86 if the compiler can determine
566 that (+ INDEX <constant>) does not require a bounds check and FOO
567 has an element type at least 8 bits wide.
568 * bug fix: references to undefined variables are handled the same way
569 in toplevel forms as in the normal compiler
570 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
572 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
573 (thanks to Jon Buller)
574 * improvement: faster compilation times for complex functions
575 * improvement: added readlink support to SB-POSIX (thanks to Richard
578 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
579 * improvement: experimental support for mach exception handling on
580 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
582 * improvement: support for GBK external format.
583 (thanks to Chun Tian (binghe))
584 * improvement: the debugger now displays variables that have been closed
585 over, in code compiled with (DEBUG 2) or higher
586 * improvement: support for executable cores on NetBSD (thanks to
588 * new feature: added a RESTART-FRAME debugger command
589 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
590 be used to provide an implementation for FILE-POSITION on Gray streams
591 (thanks to Eric Marsden)
592 * new feature: add syslog, openlog and closelog support to SB-POSIX
593 (thanks to Richard Kreuter)
594 * optimization: the function call overhead in code compiled with
595 a high DEBUG optimization setting is significantly
596 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
597 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
598 (thanks to Tony Martinez)
599 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
600 evaluated once (reported by Kevin Reid)
601 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
602 works again (thanks to Kilian Sprotte)
603 * bug fix: an error is signaled for tagbodies with duplicate tags
604 (thanks to Stephen Wilson)
605 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
606 * bug fix: Win32 port can now handle foreign code unwinding Lisp
607 stack frames from alien callbacks.
608 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
609 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
610 65k pages (thanks to David Woodhouse)
611 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
613 changes in sbcl-1.0.1 relative to sbcl-1.0:
614 * new platform: FreeBSD/x86-64, including support for threading.
615 * new feature: the compiler stores cross-referencing information
616 about function calls (who-calls), macroexpansion (who-macroexpands)
617 and special variables (who-binds, who-sets, who-references) for code
618 compiled with (< SPACE 3). This information is available through the
619 sb-introspect contrib.
620 * new feature: users may subclass SEQUENCE, and have instances of
621 these classes interoperate with standard Common Lisp functions if
622 a number of methods are defined. (This feature is experimental
623 and the interface subject to change based on feedback from SBCL
624 users and the general community)
625 * improvement: sb-sprof traces call stacks to an arbitrary depth on
626 x86/x86-64, rather than the previous fixed depth of 8
627 * improvement: another pthread back-end of mutex "pthread-futex".
628 * bug fix: non-ascii command-line arguments are processed correctly
629 (thanks to Yaroslav Kavenchuk)
630 * bug fix: non-required arguments were not passed correctly when a method
631 defined using DEFMETHOD was called from a mop-generated method using
632 CALL-NEXT-METHOD (reported by Pascal Costanza)
633 * bug fix: recursion is now permitted in accessors through
634 SLOT-UNBOUND. (reported by Pascal Costanza)
635 * bug fix: an error was signaled at startup if the HOME environment
636 variable was defined, but had an empty value (reported by Peter Van Eynde)
637 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
638 signal a PROGRAM-ERROR, not a BUG.
639 * bug fix: SB-POSIX:READDIR works when built with large file support.
640 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
642 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
643 for better automatic indentation support. (thanks to Matt Pillsbury)
644 * optimization: loading generic functions no longer takes O(n^2) time,
645 proportional to the amount of methods in the generic function
646 (reported by Todd Sabin and Jeremy Brown)
647 * optimization: the FIND and POSITION family of sequence functions
648 are significantly faster on arrays whose element types have been
650 * improvements to the Windows port:
651 ** Intermittent heap corruption problems have been fixed. (thanks
652 to Alastair Bridgewater)
653 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
655 ** Lisp is able to unwind foreign exception frames from alien
656 callbacks. (thanks to Alastair Bridgewater)
658 changes in sbcl-1.0 relative to sbcl-0.9.18:
659 * improvement: experimental support for threading on FreeBSD/x86.
660 (thanks to NIIMI Satoshi)
661 * improvement: runtime option --dynamic-space-size can be used
662 to set the size of the dynamic space reserved on startup.
663 * improvement: floating point modes in effect are now saved in
664 core, and restored on startup.
665 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
666 startup, not time since first call to GET-INTERNAL-REAL-TIME.
667 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
668 threads are not running after *SAVE-HOOKS* have run.
669 * improvement: writes to CLOS instance slots are type-checked in code
670 compiled with (SAFETY 3)
671 * improvement: floating-point exception handling on FreeBSD (thanks to
673 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
674 (thanks to Zach Beane)
675 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
677 * improvement: added support for the Shift-JIS external format.
678 (contributed by NIIMI Satoshi)
679 * improvement: callbacks are supported on Linux/PPC. (thanks to
681 * bug fix: compiler bug triggered by a (non-standard) VALUES
682 declaration in a LET* was fixed. (reported by Kaersten Poeck)
683 * bug fix: file compiler no longer confuses validated and already
684 dumped structurres. (reported by Kaersten Poeck)
685 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
686 fill-pointer signals a type-error as required. (thanks to
688 * bug fix: disassemly of funcallable instances works.
689 * bug fix: single stepping on PPC.
690 * bug fix: fix thread-safety problems in the type system (generally
691 manifesting as nonsensical errors like "STRING is a bad type specifier
692 for sequences" or "The value 1 is not of type FIXNUM")
693 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
694 (reported by Josip Gracin)
695 * bug fix: an error is signaled for attempts to displace arrays with
696 incompatible element types (thanks to Mario Mommer)
697 * bug fix: more correct handling of wide characters in debug info
698 (bug reported by Attila Lendvai and fixed by Juho Snellman)
699 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
700 and don't cause extra consing
701 * optimization: MAP and MAP-INTO are significantly faster on vectors
702 whose elements types have been declared.
703 * Improvements to SB-SPROF:
704 ** Support for allocation profiling
705 ** Reduced profiling overhead, especially for long profiling runs
706 * Improvements to the Windows port:
707 ** floating point exceptions are now reported correctly.
708 ** stack exhaustion detection works partially.
709 ** more accurate GET-INTERNAL-REAL-TIME.
710 ** better RUN-PROGRAM behaviour in case of errors in the child
712 ** PROBE-FILE now simplifies pathnames correctly.
713 ** DIRECTORY now works correctly with :WILD-INFERIORS.
714 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
715 1st 1970 (although time timezone data will be unreliable).
716 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
718 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
719 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
720 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
722 * bug fix: two potential GC deadlocks affecting threaded builds.
723 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
724 returns T (reported by Anton Kazennikov)
725 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
726 with non-variable places
727 * bug fix: remove a race condition in the setting of
728 funcallable-instance functions, this should make threaded CLOS
729 code more stable against memory faults.
730 * bug fix: corruption of specials when unbinding is interrupted by an
731 asynchronous unwind (reported by Hannu Koivisto)
732 * improvement: the debugger will now also display local variables that
733 are only used once, for code compiled with a DEBUG optimization quality
736 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
737 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
738 * incompatible change: External-format support for FFI calls. The
739 SB-ALIEN:C-STRING no longer implies an ASCII
740 external-format. Instead the string is subject to external-format
741 conversion. Additionally return values of type SB-ALIEN:C-STRING
742 are by default of type (SIMPLE-ARRAY CHARACTER), not
743 SIMPLE-BASE-STRING. For an alien type that behaves like the old
744 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
745 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
746 * incompatible change: SB-EXT package no longer contains the
747 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
748 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
750 * incompatible change: the single-stepper is no longer available
751 on Alpha, Mips and Sparc platforms.
752 * minor incompatible change: the direct superclasses of
753 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
754 not (STANDARD-OBJECT FUNCTION). This makes the
755 class-precedence-lists of GENERIC-FUNCTION and
756 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
758 * minor incompatible change: the default stream external format on
759 non-unicode SBCL is no longer detected from the locale, but is always
761 * new feature: Add a version of evaluator that uses an interpreter instead
762 of the compiler. EVAL still uses the compiler by default, to switch it
763 to use the interpreter, set the value of the variable
764 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
765 * minor incompatible change: the single-stepper REPL has been merged
766 with the normal debugger (see the "Stepping" heading of the debugger help
767 for more details). The debugger command STEP will no longer switch
768 to the single-stepper REPL.
769 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
770 for a type now works.
771 * bug fix: loading of default sysinit file works. (thanks to Leonid
773 * bug fix: better detection of circularities in the file-compiler.
774 (reported by Marco Monteiro)
775 * bug fix: the CL pathname functions now work with files that have
776 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
777 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
778 systems with Unix98 pty semantics.
779 * bug fix: ASDF-INSTALL will now work with bsd tar.
780 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
782 * bug fix: timers expiring in dead threads no longer cause a
783 type-error (reported by Paul "Nonny Mouse").
784 * bug fix: thanks to more lightweight single-stepper instrumentation,
785 code compiled with (DEBUG 3) will compile and execute significantly faster,
786 and will have more accurate type-inferencing than before
787 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
788 whose bindings are modified
789 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
790 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
791 ** replace dummy implementations of CL:MACHINE-INSTANCE and
792 CL:SOFTWARE-VERSION with proper version
794 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
795 * feature: implemented the READER-METHOD-CLASS and
796 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
797 as specified by AMOP.
798 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
800 * optimization: faster LOGCOUNT implementation on x86 and x86-64
801 (thanks to Lutz Euler)
802 * optimization: hashing of general arrays and vectors has been
803 improved. (reported by Any Fingerhut)
804 * enhancement: SB-INTROSPECT is now able to find definitions of
805 profiled functions. (thanks to Troels Henriksen)
806 * enhancement: compiler-macro expansion applies now to FUNCALL forms
807 as well, allowing compiler-macros for SETF-functions to expand.
808 * enhancement: step-instrumentation no longer wraps known
809 single-value functions with multiple-value context, allowing
810 better type inference.
811 * fixed bug #337: use of MAKE-METHOD in method combination now works
812 even in the presence of user-defined method classes. (reported by
813 Bruno Haible and Pascal Costanza)
814 * fixed bug #339(c): if there are applicable methods not part of any
815 long-form method-combination group, call INVALID-METHOD-ERROR.
816 (reported by Bruno Haible)
817 * fixed bug #361: the :FUNCTION initarg in the protocol for
818 initialization of methods can now be used to override
819 internally-produced optimized functions. (reported by Bruno
821 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
822 system-provided lambda expression no longer cause warnings about
823 unbound #:|pv-table| symbols.
824 * bug fix: improved the handling of type declarations and the
825 detection of violations for keyword arguments with non-constant
827 * bug fix: potentially erronous calls to PATHNAME and
828 MERGE-PATHNAMES were being flushed in some cases.
829 (reported by Richard Kreuter)
830 * bug fix: compiled calls to TYPEP were mishandling obsolete
831 instances. (reported by James Bielman and Attila Lendvai)
832 * bug fix: format strings with ~> without matching ~< no longer
833 trigger an AVER, but signal an understandable error instead.
834 (reported by Antonio Martinez)
835 * bug fix: specifying an output-file without a pathname-type for
836 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
837 of input-file instead of "fasl". (reported by Robert Dodier)
838 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
839 (reported by James Y Knight).
840 * bug fix: compiler-macros-function did not consider the environment
841 argument for shadowing by local functions.
842 * bug fix: compiler-macros expansion was inhibited by local INLINE
844 * bug fix: inline expansions of known functions were subject to
845 step-instrumentation in high debug policies, leading to problems
847 * bug fix: compiler failed to differentiate between different CONS
849 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
850 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
851 * bug fix: #368: incorrect use of expressed vs. upgraded array
853 * bug fix: #306a: more precise unions of array types.
854 * thread-safety improvements:
855 ** CONDITION-WAIT could return early on Linux, if the thread was
856 interrupted and subsequently continued with SIGCONT.
857 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
859 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
860 * added support for the ucs-2 external format. (contributed by Ivan
862 * minor incompatible change: pretty printing of objects of type
863 (cons symbol) is, in the default pprint-dispatch-table, now
864 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
866 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
867 called later in a class's lifetime, possibly as late as when the
868 first instance of the class is created. Previously,
869 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
870 class became finalizeable.
871 * fixed bug: FILE-POSITION sometimes returned inconsistent results
872 for multibyte external-format streams. (thanks to "vbzoli")
873 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
874 with :ALLOCATION :CLASS inherited from superclasses of the
876 * fixed bug: anonymous classes can now be created using the :NAME
877 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
878 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
879 * fixed bug: core-files saved with :EXECUTABLE T can again be
880 executed when SBCL_HOME isn't set. (reported by James Knight)
881 * fixed bug: toplevel LOCALLY forms with declarations could
882 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
883 * fixed bug: printing from several different threads using different
884 values of *print-case* could cause invalid output, due to
885 some internal special variables of the printer not being bound
886 thread-locally (reported by Max Mikhanosha)
887 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
888 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
889 (reported by Pascal Costanza)
890 * fixed bug: FINALIZE-INHERITANCE is called from
891 REINITIALIZE-INSTANCE on classes when the class has previously
892 been finalized, as required by AMOP.
893 * minor code generation optimizations:
894 ** better register allocation in CLOS dispatching functions
895 ** overflow detection when coercing signed bytes to fixnums on x86-64
896 ** is now implemented with one IMUL instruction instead of three shifts
897 ** more efficient bit-vector access on x86 and x86-64
898 ** more efficient access to raw structure slots on x86 and x86-64
899 * fixed some bugs revealed by Paul Dietz' test suite:
900 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
903 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
904 * feature: thread support on Solaris/x86, and experimental thread support
906 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
908 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
909 default method for SLOT-UNBOUND.
910 * minor incompatible change: prevent the user from specializing the
911 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
912 somewhat counter to the intent of the protocol, I (CSR) think, and
913 additionally it just doesn't work in SBCL as currently
914 implemented, thanks to optimizations (that are always valid for
915 the other three SLOT-VALUEish functions, but not for the setter).
916 * minor incompatible change: the :SB-LDB feature is now enabled by
917 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
918 the low-level debugger.
919 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
920 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
921 down to an SB-SYS:FD-STREAM, allowing direct communication in
922 more cases. Temporary files and pipes are still used as a fallback
924 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
925 on the non-reentrant C functions localtime(3) and gmtime(3).
926 * bug fix: global optimization policy was not visible in LOCALLY and
928 * bug fix: class objects can be used as specializers in methods.
929 (reported by Pascal Costanza)
930 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
931 with a type error. (reported by blitz_ on #lisp)
932 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
933 failed with a type error. (reported by Pascal Bourguignon)
934 * bug fix: merging pathnames against defaults with :DIRECTORY
935 starting with '(:RELATIVE :BACK) should preserve the :BACK.
936 (reported by James Y Knight)
937 * bug fix: saving large (>2GB) cores on x86-64 now works
938 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
939 constant index and a (simple-array (signed-byte 32)) array
940 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
941 error (patch by Robert J. Macomber)
942 * feature: TIME now displays the amount of run-time spent in GC
943 * bug fix: The debugger now does a better job of respecting
944 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
945 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
946 workaround for bug 403.)
947 * bug fix: floating point exception handling now works on FreeBSD
948 (thanks to NIIMI Satoshi)
949 * fixed some bugs revealed by Paul Dietz' test suite:
950 ** MISC.641: LET-conversion were not supposed to work in late
952 * improvements to the Win32/x86 port:
953 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
954 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
956 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
957 * new feature: source path information is generated for macro-expansion
958 errors for use in IDE's like Slime (thanks to Helmut Eller)
959 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
960 * bug fix: compiler does not loop forever on an invalid type in
962 * improvement: compilation of most CLOS applications is significantly
964 * optimization: added a limited bytecode compiler for simple toplevel
965 forms, speeding up compilation and FASL loading
966 * bug fix: the statistical profiler now properly distinguishes anonymous
969 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
970 * minor incompatible change: in sbcl-0.9.11 (but not earlier
971 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
972 into subclasses of the finalized class. Now user calls to
973 FINALIZE-INHERITANCE finalize just the one class, and calls by the
974 system return before any subclasses are finalized.
975 * minor incompatible change: The reader no longer ignores errors
976 regarding non-existent packages in #+ and #- feature tests.
977 * new feature: command line options --no-sysinit, --no-userinit to
978 inhibit loading the corresponding init files
979 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
980 for sending data through UDP sockets (thanks to François-René Rideau)
981 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
982 error is signalled instead (thanks to François-René Rideau)
983 * minor incompatible change: Error signalling behaviour of lexical
984 operations violating package locks has changed slightly. Refer to
985 documentation on package locks for details.
986 * enhancement: EVAL can process IF-expressions without resorting to the
988 * bug fix: Some operations involving SIN, COS, or TAN weren't being
989 constant-folded properly due to missing out-of-line functions.
990 (reported by Mika Pihlajamäki)
991 * bug fix: LISTEN sometimes returned T even in cases where no data was
992 immediately available from the stream
993 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
994 were reversed. (reported by Levente Mészáros)
995 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
996 when given a vector displaced to an adjustable array. (reported
998 * bug fix: derivation of float boundaries from numbers outside the
999 appropriate float range (reported by John Wiseman)
1000 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1002 * bug fix: precomputing cache entries for generic functions with
1003 some subclasses of specializers as yet invalid does not attempt to
1004 fill a cache line with a negative offset. (reported by Levente
1006 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1007 allow condition class objects as arguments to DOCUMENTATION and
1008 (SETF DOCUMENTATION); only find and set documentation for
1009 structure names for the STRUCTURE doc-type. (suggested by Gary
1011 * improvements to the Win32/x86 port:
1012 ** (user-homedir-pathname) and default initialization file
1013 locations now know about the user's "Documents and Settings"
1014 directory (thanks to Yaroslav Kavenchuk)
1015 ** run-program is implemented (thanks to Mike Thomas)
1016 ** sockets support (thanks to Timothy Ritchey)
1017 ** better backtrace support (thanks to Alastair Bridgewater)
1018 ** sb-grovel supported
1019 ** asdf-install and sb-posix work somewhat
1020 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1021 * improvements to the Solaris/x86 port:
1022 ** works on Solaris 11/Solaris Express
1023 ** floating-point exception handling support
1024 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1025 * fixed some bugs revealed by Paul Dietz' test suite:
1026 ** REMOVE-METHOD returns its generic function argument even when
1027 the method is not one of the generic functions' methods.
1028 ** objects declared as MEMBER types can be admissible arguments to
1029 structure accessors.
1030 ** printing characters should simply be printed by the FORMAT ~:C
1032 ** compiler failure when compiling functions with hairy constant
1033 defaults for optional parameters.
1034 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1035 function, which is already optimized.
1037 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1038 * new platform: experimental support for SBCL x86/Darwin, including
1039 MacOS X 10.4.5 on Intel.
1040 * new feature: Unicode character names are now known to the system
1041 (through CHAR-NAME and NAME-CHAR).
1042 * new feature: the filesystem location of SBCL's core file is
1043 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1044 * minor incompatible change: the contrib modules SB-POSIX and
1045 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1046 this change is to make it easier to distribute
1047 location-independent binaries.
1048 * bug fix: as implied by AMOP, standardized classes no longer have
1049 slots named by external symbols of public packages. (reported by
1051 * optimization: calling functions via constant symbols -- as in
1052 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1053 via the function object as in (FUNCALL #'FOO).
1054 * optimization: on x86 and x86-64, the calling convention for the
1055 case of dealing with an unknown number of values has been altered
1056 to be friendlier to the prediction heuristics implemented,
1057 particularly with reference to CALL and RET pairing. (thanks to
1058 Alastair Bridgewater)
1059 * optimization: on x86, the code for access to thread-local symbol
1060 values and binding/unbinding of thread-local symbols is smaller.
1061 (thanks to Alastair Bridgewater)
1062 * enhancement: CONSTANTP is now able to determine constantness of
1063 more complex forms, including calls to constant-foldable standardized
1064 functions and some special forms beyond QUOTE.
1065 * fixed bug: occasional GC crashes on Solaris/x86
1066 * optimization: x86-64 supports stack allocation of results of simple
1067 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1068 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1069 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1070 (thanks to James Knight)
1071 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1072 by the standard (thanks to Stephen Compall)
1074 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1075 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1076 be used for bundling the runtime and the core file into one
1077 executable binary. This feature is not currently supported on all SBCL
1078 platforms. (thanks to James Bielman and NIIMI Satoshi)
1079 * new feature: a generational or ephemeral garbage collector is now
1080 the default on the PowerPC platform (both Linux and Darwin). The
1081 old Cheney (stop and copy) collector is a build-time option.
1082 (thanks to Cyrus Harmon, after Raymond Toy)
1083 * minor incompatible change: the method by which SBCL finds its
1084 contributed modules has changed; it no longer relies on symbolic
1085 links from an $SBCL_HOME/systems directory, but searches directly
1086 in the subdirectories of $SBCL_HOME.
1087 * enhancement: the dynamic heap size on the Linux/PPC platform is
1088 markedly larger, even using the older Cheney garbage collector.
1089 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1090 certain complicated string types. (reported by Gary King)
1091 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1092 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1093 * fixed bug: DOCUMENTATION on structures no longer signals an
1094 error if no structure type of the right name exists. (reported by
1096 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1097 applicable non-standard primary or :AROUND method on
1098 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1099 no longer cause unbound variable errors. (reported by Kevin Reid)
1100 * fixed bug: metacircle resolution in cases where methods have slots
1101 added before the slots from STANDARD-METHOD. (reported by Jean
1103 * fixed bug: the Power PC assembly code for calling into lisp
1104 sometimes computed its return address wrongly (depending on the
1105 15th bit of the address, fixed at link time). This used to
1106 manifest itself in a segmentation violation while building PCL.
1107 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1108 many others over the years)
1109 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1110 the stack pointer (treated as a signed integer) is negative.
1111 (thanks to Peter van Eynde)
1113 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1114 * new platform: experimental support for the Windows operating
1115 system has been added. (thanks to Alastair Bridgewater)
1116 * fixed several bugs in and robustified the PPC FFI (including
1117 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1118 * bug fix: allow non-simple string symbol names (reported by Paul
1120 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1122 * bug fix: saving a core corrupted callbacks on x86/x86-64
1123 * bug fix: closed a loophole in metacircularity detection and
1124 grounding in the PCL implementation of CLOS.
1125 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1126 (reported by tomppa on #lisp)
1127 * bug fix: clear the direction flag on Lisp -> C transitions, as
1128 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1129 (reported by Andrej Grozin and Hendrik Maryns)
1130 * optimization: major improvements to GC efficiency on GENCGC platforms
1131 * optimization: faster implementation of EQUAL
1132 * optimization: emit more efficient opcodes for some common
1133 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1134 * optimization: several other minor code-generation improvements on x86-64
1135 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1137 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1138 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1139 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1140 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1141 by ANSI as a generic function, and for consistency (SETF
1142 GENERIC-FUNCTION-NAME) is treated likewise.
1143 * fixed bug #233.b: make constraint propagation notice when a variable
1144 value is changed after it is referenced but before it is used
1145 * fixed bug #296: no more arbitrary behaviour from filenames with
1146 odd characters as --load arguments. (reported by Adam Warner)
1147 * fixed bug #390: :CHARACTER-SET pathname components now work as
1148 expected. (reported by Tim Daly Jr)
1149 * fixed bug #391: complicated :TYPE intersections in slot
1150 definitions no longer cause an error in PCL internals.
1151 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1152 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1153 returning the number of octets which would be written to the
1154 file-stream. (thanks to Robert J. Macomber)
1155 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1156 arguments to the internal stream functions. (thanks to David
1158 * bug fix: stack exhaustion is now handled on the x86/SunOS
1159 platform. (thanks to Daisuke Homma)
1160 * bug fix: more accurate ROOM results on GENCGC platforms
1161 * optimization: improved type inference for arithmetic-for
1162 index variables in LOOP
1163 * optimization: faster floating-point SQRT on x86-64
1164 * fixed some bugs revealed by Paul Dietz' test suite:
1165 ** DOCUMENTATION returns NIL instead of "" for method combinations
1166 that don't have a docstring
1168 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1169 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1170 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1171 therefore are not customizeable by user code (as seems to be at
1172 least permitted and maybe required by AMOP). As a consolation,
1173 however, the SBCL implementation of these functions now calls
1174 REINITIALIZE-INSTANCE as specified by AMOP.
1175 * bug fix: slot-definition documentation is propagated as per ANSI
1176 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1177 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1178 argument, as do eslotd initialization methods. (from Pascal
1179 Costanza's "Closer" project)
1180 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1181 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1183 * bug fix: it is now possible to have more than one subclass of
1184 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1185 (reported by Bruno Haible, Pascal Costanza and others)
1186 * bug fix: the dependent update protocol now works for generic
1187 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1188 and Pascal Costanza)
1189 * bug fix: condition-class instances corresponding to
1190 DEFINE-CONDITION forms are now created eagerly. (reported by
1191 Kalle Olavi Niemitalo on comp.lang.lisp)
1192 * bug fix: floating point printing is more accurate in some
1193 circumstances. (thanks to Simon Alexander)
1194 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1195 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1196 * bug fix: callbacks on OS X now preserve stack-alignment.
1197 (thanks to Cyrus Harmon)
1198 * enhancement: the x86-64 disassembler is much better at
1199 disassembling SSE instructions. (thanks to Lutz Euler)
1200 * enhancement: CHECK-TYPE on a variable now allows type-inference
1201 (thanks to Vincent Arkesteijn)
1202 * optimization: improved performance of EUC-JP external format.
1203 (thanks to NIIMI Satoshi)
1204 * optimization: performance improvements to IO on file streams of
1205 :ELEMENT-TYPE CHARACTER
1206 * optimization: much faster memory allocation on x86-64
1207 * optimization: faster garbage collections (latency and throughput)
1209 * optimization: faster bignum subtraction and division on x86 and x86-64
1210 (thanks to Lutz Euler)
1211 * optimization: more accurate type-inference for loops that use a
1212 floating point index variable or a negative step.
1214 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1215 * bug fix: add a workaround to SBCL looping infinitely at startup on
1216 Linux kernels with apparently buggy implementations of personality().
1217 (thanks to Svein Ove Aas)
1218 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1219 (thanks to David Lichteblau)
1220 * bug fix: local bindings shadow global symbol macros for
1221 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1222 * optimization: non-open coded uses of numeric comparison operators
1223 (e.g. >) no longer cons when called with more than one parameter
1224 on platforms supporting dynamic-extent allocation.
1225 * enhancement: saving cores with foreign code loaded is now supported
1226 on MIPS/Linux in addition to the previously supported platforms.
1227 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1228 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1229 * bug fix: division by zero in sb-sprof when no samples were collected
1230 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1231 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1232 (thanks to Alexander Kjeldaas)
1233 * new feature: ignoring the return values of destructive functions
1234 like NREVERSE signals a compile-time style-warning.
1235 (thanks to Kevin Reid)
1236 * new feature: the alignment of alien structure fields can be
1237 explicitly specified. (thanks to Cyrus Harmon)
1238 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1239 and dump core on SIGQUIT
1241 ** incompatible change: threads do not inherit values of specials
1242 from their parents (see manual)
1243 ** bug fix: threads stacks belonging to dead threads are freed by the
1244 next exiting thread, no need to gc to collect thread stacks anymore
1245 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1246 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1247 ** bug fix: dynamic variable and thread start related gc lossage
1248 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1250 * fixed some bugs revealed by Paul Dietz' test suite:
1251 ** SUBTYPEP is slightly more accurate on heinously complicated
1252 CONS types where some of the members have uncertain (in the
1253 NIL, NIL sense) type relationships to each other.
1255 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1256 no longer purified unless :PURIFY T is explicitly specified.
1257 ** Non-purified cores are significantly smaller than before
1259 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1260 * new feature: timers based on Zach Beane's excellent timer package
1261 * added support for the following external formats: koi8-u,
1262 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1263 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1264 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1265 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1266 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1267 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1268 * incompatible change: a threaded SBCL will no longer revert to
1269 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1270 * bug fix: interrupts are disabled until startup is complete; no
1271 more sigsegvs when receiving a signal to soon
1272 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1274 * bug fix: add a workaround for the memory randomization features in
1275 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1276 workaround will only be in effect on systems with the proc filesystem
1278 * bug fix: printing objects of type HASH-TABLE signals a
1279 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1281 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1282 have been running for over 50 days. (reported by Gilbert Baumann)
1283 * bug fix: the logic for getting names of functions gets less
1284 confused when confronded with alternate-metaclass
1285 funcallable-instances. (reported by Cyrus Harmon)
1286 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1287 consistent, even on internal alternate-metaclass objects.
1288 * bug fix: flush closure information collected by physical
1289 environment analysis prepass before the main pass. (bug reported
1291 * bug fix: compiler pack phase does not modify a hashtable, which is
1292 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1293 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1294 may be expanded. (reported by Alexander <malishev> on c.l.l)
1295 * bug fix: no more style-warnings when compiling code using LAST
1296 under high-SPEED low-SPACE settings. (reported by David Wallin
1298 * fixed some incompatibilities between SBCL's MOP and the MOP
1300 ** the METAOBJECT class is now implemented;
1301 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1302 STANDARD-OBJECT, as required;
1303 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1304 now compatible (as required by VALIDATE-SUPERCLASS); there
1305 remains a consistency requirement over the presence or absence
1306 of the FUNCTION class in the superclass list of finalized
1307 classes; see the manual for more details;
1308 ** the :AROUND method for COMPUTE-SLOTS on
1309 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1310 requested slot ordering.
1312 ** bug fix: parent thread now can be gc'ed even with a live
1314 ** bug fix: binding a special with PROGV to no value is not
1315 the same as that symbol not having been bound (thanks to
1317 ** bug fix: binding specials is thread safe (thanks to
1319 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1320 and SB-SPROF do not die with 'no handler for signal XX in
1321 interrupt_handle_now(..)' anymore
1322 ** bug fix: WITH-TIMEOUT works with multiple running threads
1323 * fixed some bugs revealed by Paul Dietz' test suite:
1324 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1325 the :method-class keyword argument.
1327 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1328 * new port: the Solaris operating system on x86 processors is now
1329 mostly supported, though some rough edges in the environment
1330 remain. (thanks to Daisuke Homma)
1331 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1332 space for its heap. (thanks to Thiemo Seufer)
1333 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1334 closures. (thanks to Thiemo Seufer)
1335 * minor incompatible change: eof selects abort in the debugger.
1336 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1337 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1339 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1340 deprecated, and will go away in a future revision of SBCL.
1341 * minor incompatible change: GC-ON and GC-OFF are no longer
1342 implemented with a counter, it does not matter how many times gc
1343 is switched on or off
1344 * bug fix: discriminating functions for generic function classes
1345 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1346 make invalid assumptions about method precedence order. (reported
1348 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1349 platforms. (thanks to Thiemo Seufer)
1350 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1351 * bug fix: correct pathname printing with printer escaping is on.
1352 (thanks to Kevin Reid)
1353 * bug fix: complex VOP definitions in "user-space" no longer trigger
1354 package locks. (reported by Zach Beane)
1355 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1356 for funcallable-instance functions, and is no different from
1358 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1359 characters in the current readtable. (reported by Nicholas Neuss)
1360 * bug fix: SUBTYPEP on various CONS types returns more a more
1361 accurate acknowledgment of its certainty.
1362 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1363 lists in some cases. This partially fixes bug 384.
1364 * flush all standard streams before prompting in the REPL and the
1366 * bug fix: signal handling and triggering gc do not conflict
1367 directly anymore, in particular a high frequency sb-sprof does
1368 not prevent gc from running
1369 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1370 approximation for timezone and DST information between the
1371 universal time epoch and the smallest negative 32-bit time_t.
1372 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1373 year 1899 so long as the encoded time is non-negative.
1374 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1375 the correct answer for non-constant rotate amounts when performing
1376 an inline 32-bit rotation.
1378 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1379 there is only one thread in the session
1380 ** bug fix: memory leak for streams created in one thread and
1381 written to in another
1382 ** bug fix: lockup when compiled with gcc4
1383 ** bug fix: race that allows the gc to be triggered when gc is
1385 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1386 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1388 ** bug fix: finished threads are gc'ed properly
1389 * fixed some bugs revealed by Paul Dietz' test suite:
1390 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1391 has arguments with a different set of applicable methods from
1392 the orignal arguments.
1393 ** The type error thrown by MAP now has a correct expected-type
1395 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1396 name a compiled function.
1397 ** Three MISC tests where a large bignum was improperly coerced to
1398 a float that couldn't represent that bignum during type
1399 derivation were fixed.
1400 ** SUBTYPEP can now handle types involving the negation of a
1401 list-form FUNCTION type.
1402 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1403 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1404 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1406 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1407 * New feature: Experimental support for bivalent streams: streams
1408 opened with :element-type :default now allow character and binary
1409 (unsigned-byte 8) I/O
1410 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1411 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1413 * bug fix: no more highly sporadic "couldn't check whether ~S is
1414 readable" when reading a stream and an interrupt hits in the middle
1415 of a select system call
1416 * compiler better recognizes complex arrays (reported by Eduardo
1418 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1419 for variables on SBCL built with linkage-tables. (reported by Luis
1421 * various error reporting improvements.
1422 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1423 (Thanks to Thiemo Seufer)
1424 * bug fix: nested reader invokations work correctly
1425 * bug fix: it is possible to have simultaneous references to foreign
1426 code and foreign data with the same name.
1428 ** added x86-64 support
1429 ** incompatible change: the threading api now works with thread
1430 objects instead of thread ids
1431 ** bug fix: threads are protected from signals and interruption when
1432 starting up or going down
1433 ** bug fix: a race where an exiting thread could lose its stack to gc
1434 ** bug fix: don't halt on infinite error in threads if possible
1435 ** fixed numerous gc deadlocks introduced in the pthread merge
1436 ** bug fix: fixed thread safety issues in read and print
1437 ** bug fix: debugger doesn't hang on session lock if interrupted at
1438 an inappropriate moment
1439 ** bug fix: run-program is now thread safe(r)
1440 ** bug fix: inner with-recursive-lock no longer releases the mutex
1441 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1442 * fixed some bugs revealed by Paul Dietz' test suite:
1443 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1444 EXPECTED-TYPE slots filled.
1445 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1447 ** MAKE-CONDITION accepts classes as type-designators.
1448 ** COMPILE may never return NIL.
1449 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1450 range before calling Unix time functions
1452 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1453 * numerous signal handling fixes to increase stability
1454 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1455 * minor incompatible change: we now correctly canonize default
1456 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1457 as per the MOP, rather than the historical (INITARG INITFUNCTION
1458 INITFORM). (reported by Bruno Haible)
1459 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1460 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1461 for more information.
1462 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1463 pathname is a directory pathname.
1464 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1465 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1467 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1468 :METACLASS argument in addition to a class name. (reported by
1469 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1470 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1471 (reported by Bruno Haible; patch from Gerd Moellmann)
1472 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1474 * bug fix: more cleanups to the floating point exception handling on
1475 x86-64 (thanks to James Knight)
1476 * bug fix: the compiler does not try to constant fold calls of
1477 COERCE and BIT-* functions when they return freshly constructed
1478 objects. (reported by Nikodemus Siivola and Paul Dietz)
1479 * optimization: improved the MIPS versions of generic (in the
1480 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1481 * optimization: direct conversion of (unsigned-byte 32) to floats on
1482 the PowerPC platform.
1483 * optimization: structure instances with raw slots now use less
1484 memory, and probably show better memory locality. (thanks to
1486 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1487 are treated as requests for stack allocation on the x86-64,
1488 Alpha-32, PPC and SPARC platforms.
1489 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1490 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1491 (thanks to Tony Martinez)
1493 ** gcing a dead thread can no longer lead to lockups
1494 ** threads block signals until they are set up properly
1495 ** errno is no longer shared by threads
1496 ** interrupt-thread restores the eflags register on x86
1497 ** fixed some lockups due to gc/thread interaction
1498 * fixed some bugs revealed by Paul Dietz' test suite:
1499 ** invalid dotted lists no longer raise a read error when
1500 *READ-SUPPRESS* is T
1501 ** signal an error if a symbol that names a declaration is used
1502 as the name of a type, or vice versa
1503 ** allow using the (declare (typespec var*)) abbreviation for
1504 (declare (type typespec var*)) with all type specifiers
1505 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1506 ** always return NIL from PPRINT-POP when OBJECT is NIL
1507 ** don't signal errors when pretty-printing malformed LABELS,
1508 FLET or MACROLET forms
1509 ** declarations in a DOLIST body are also in scope for the
1511 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1513 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1516 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1517 * fixed cross-compiler leakages that prevented building a 32-bit
1518 target with a 64-bit host compiler.
1519 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1520 opened with :IF-EXISTS :OVERWRITE.
1521 * fixed bug 281: error for an invalid qualifier in a short-form method
1522 combination method is not signalled until the faulty method is called.
1523 * bug fix: iteration variable type inferrer failed to deal with open
1524 intervals. (reported by Alan Shields)
1525 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1526 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1527 or more runtime options were provided to the sbcl binary.
1528 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1529 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1530 * optimization: there's now a fast-path for fixnum arguments in the
1531 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1532 * optimization: the code generated on x86-64 is more compact thanks
1533 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1534 * bug fix: floating-point exception handling now partly works on
1535 x86-64. (Thanks to James Knight)
1536 * improvement to the MIPS backend from Thiemo Seufer: C-style
1537 64-bit long long arguments and return values to alien functions
1539 * fixed some bugs revealed by Paul Dietz' test suite:
1540 ** the type-error signalled from WARN has a filled-in DATUM slot.
1541 ** the type-error required when a stream is not associated with
1542 a file has the stream as its datum.
1543 ** type-errors on single-floats on x86-64 no longer have
1544 :INVALID-OBJECT as the datum
1545 ** the type-errors signalled for invalid function names now have
1546 a correct expected type
1547 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1548 for structures defined with a :TYPE. Documentation strings for
1549 typed structures are no longer immediately discarded
1550 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1551 broadcast streams and synonym streams. FILE-LENGTH now also works
1552 on broadcast streams.
1554 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1555 * incompatible change: the --noprogrammer option, deprecated since
1556 version 0.7.5, has been removed. Please use the equivalent
1557 --disable-debugger option instead.
1558 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1560 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1561 inoperational for a while now) has been completely removed.
1562 * null lexical environments are now printed as #<NULL-LEXENV>,
1563 significantly reducing the amount of clutter in typical backtraces.
1564 * documentation on weak pointers, finalization, and after GC hooks
1565 has been added to the manual.
1566 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1567 as other specialized array types, is much faster. SUBSEQ and
1568 COPY-SEQ on such arrays have also been sped up.
1569 * optimization: EQL is now more efficient when at least other argument
1570 is known to be of type (OR FIXNUM (NOT NUMBER)).
1571 * fixed bug: the runtime is now less vulnerable to changes in the
1572 size of the SBCL object on OS X, and virtual memory is reserved for
1573 all spaces that need to be at a fixed address.
1574 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1575 * fixed bug: finalizers and after GC hooks that cause consing are now
1577 * fixed bug: compiler error messages and summaries are now printed to
1578 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1579 * fixed inference of the upper bound of an iteration variable.
1580 (reported by Rajat Datta).
1581 * fixed bug 211e: calling local functions with duplicated constant
1582 keyword argument no longer causes a bogus style warning about an
1584 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1585 ftype declaration to be disregarded. (reported by Dave Roberts)
1586 * fixed bug 373: caused by erronous compilation of references to alien
1587 variables in the runtime on ppc/darwin.
1588 * fixed bug 376: CONJUGATE type deriver.
1589 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1590 (reported by Baughn on #lisp)
1591 * a cross-compiler bug on non-x86 platforms has been identified and
1592 fixed. (thanks to Bruno Haible)
1593 * improvements to the MIPS runtime code for increased stability.
1594 (thanks to Thiemo Seufer)
1595 * increased the maximimum compact environment size to allow
1596 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1597 * partial workaround for bug 135: don't name the function we're
1598 calling for hairy cases of EVAL, so as not to accumulate one environment
1599 entry per call to EVAL. (reported by Kevin Reid)
1600 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1601 * optimization: type testing for non-vector arrays should be faster.
1602 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1604 * bug fix: setting 31st element of a bit vector to zero did not work
1606 * bug fix: redefining a class definition which failed due to a
1607 previous accessor / function clash now works (but see BUGS entry
1608 #380 for more problems in this area). (thanks to Zach Beane)
1609 * the long form of DEFINE-METHOD-COMBINATION disables method group
1610 checking when given a single method group with pattern *. (thanks
1612 * on x86 compiler supports stack allocation of results of simple
1613 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1614 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1615 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1616 * fixed some bugs related to Unicode integration:
1617 ** the restarts for recovering from input and output encoding
1618 errors only appear when there is in fact such an error to
1620 * fixed some bugs revealed by Paul Dietz' test suite:
1621 ** MISC.549 and similar: late transformation of unsafe type
1622 assertions into derived types caused unexpected code
1624 ** SCALE-FLOAT type deriver is less wrong.
1625 ** type derivers for EXP, LOG and similar functions compute result
1626 types for complex arguments better.
1627 ** (MISC.563) CONJUGATE type deriver works for very restricted
1629 ** out-of-line type testers for character strings are available.
1630 ** EQUAL compiler transform understands specialness of objects
1632 ** accessing double-floats stored on the stack now works on x86-64.
1633 ** debugger internals could sometimes create invalid lispobjs,
1634 resulting in GC crashes.
1635 ** MISC.548: type check weakening can convert required type into
1637 ** initialization forms for bindings are not in scope of free special
1640 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1641 * incompatible change: thread support for non-NPTL systems has
1642 been removed - locking is buggy and unreliable. A threaded
1643 SBCL build will now warn at startup and refuse to create
1644 new threads, unless futex support is detected in the kernel
1645 * incompatible change: the top level REPL now has only an ABORT
1646 restart associated with it, not TOPLEVEL and ABORT as it used to.
1647 TOP and TOPLEVEL are now available as debugger commands for
1648 returning to the top level.
1649 * incompatible change: forms evaluated in the REPL now use the
1650 global optimization policy.
1651 * incompatible change: user- and system-initialization files are
1652 no longer processed with LOAD, but by READ and EVAL; hence the
1653 global optimization policy, startup package, readtable, etc,
1655 * The .fasl file format number has been incremented because of
1656 various incompatible changes.
1657 * internal entry point details and argument counts no longer appear
1658 in backtraces unless explicitly requested by setting
1659 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1660 * built-in and standard functions no longer have names like "top
1661 level local call to FOO".
1662 * fixed bug 32: functions defined in non-null lexical environments
1663 now have more legible printed representation
1664 * fixed bug 33: functions defined in non-null lexical environemnts
1665 are now more amenable to inspection by INSPECT.
1666 * workaround for bug 354: XEPs no longer appear in backtraces unless
1667 explicitly requested.
1668 * fixed bug: receiving the signal which results from attempting to
1669 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1670 SIGBUS on other *BSDs) on architectures where the C stack is also
1671 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1672 to debugger confusion or wild instability. (reported by Cheuksan
1674 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1675 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1676 (reported by Lutz Euler)
1677 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1678 compiled code. (reported by Paul Dietz)
1679 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1680 specialized parameter to a method to something that is not TYPEP
1681 the specializer is now possible.
1682 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1683 face of package deletion.
1684 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1685 pathname. (reported by David Tolpin on c.l.l)
1686 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1687 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1688 STRING-STREAM and FILE-STREAM, respectively.
1689 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1690 than previously; in addition, its test suite is now run on build.
1691 (thanks to Luke Gorrie)
1692 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1693 * optimization: added a immediate representation for single-floats
1695 * fixed some bugs revealed by Paul Dietz' test suite:
1696 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1697 ** attempting to create a package with a colliding nickname causes
1698 correctable errors to be signalled.
1699 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1700 ** several x86-64 backend bugs related to sign-extension of immediate
1703 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1704 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1705 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1707 * fixed loading of multiply forward-referenced layouts.
1708 (thanks to Cheuksan Wang)
1709 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1710 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1711 (see COMPILE-FILE documentation for details of :PRINT option)
1712 * fixed bugs 19 and 317: fixed-format floating point printing is
1713 more accurate. This also fixes a bug reported by Adam Warner
1714 related to the ~@F format directive.
1715 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1717 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1718 dispatch table if the from-char is a dispatch macro character.
1719 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1720 by the wordsize no longer ignore the last word. (reported by Lutz
1722 * fixed bug in type checking of dynamic-extent variables. (reported
1724 * optimization: sequence traversal functions use their freedom to
1725 coerce function designators to functions.
1726 * optimization: code with many calls to CLOS methods specialized on
1727 CLOS classes has had redundant type checks removed.
1728 * optimization: added declarations to speed up operations that access
1729 the internal character database (for example STRING-UPCASE)
1730 * optimization: comparison operations between floats and sufficiently small
1731 fixnums no longer create extra rationals
1732 * fixed some bugs related to Unicode integration:
1733 ** portions of multibyte characters at the end of buffers for
1734 character-based file input are correctly transferred to the
1735 start of the buffer at the next read.
1736 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1737 passing it through to OPEN.
1738 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1739 argument given to internal calls to OPEN.
1740 ** the built-in comment readers (introduced by character sequences
1741 ";" and "#|") are more forgiving to encoding errors; they will
1742 STYLE-WARN and then attempt to resync the stream at a character
1743 boundary. (thanks to Teemu Kalvas)
1744 * fixed some bugs in the x86-64 port:
1745 ** Negative short int return values from c-calls are sign-extended
1747 ** The stack is aligned to 16-bytes for c-calls, as required by
1748 the ABI. (thanks to Cheuksan Wang)
1749 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1750 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1751 for enums. (thanks to Vincent Arkesteijn)
1752 ** Multiple small optimizations and bugfixes for floating point
1754 * fixed some bugs revealed by Paul Dietz' test suite:
1755 ** Space, Tab, Linefeed, Return and Page have the invalid
1756 secondary constituent character trait.
1757 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1759 ** WITH-INPUT-FROM-STRING should only update the index place on
1761 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1762 works more reliably.
1763 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1764 with one-dimensional arrays with fill pointers.
1765 ** TYPE-OF failed on a complex with an integer realpart and a
1767 ** compiler failure during type inference for the code of form
1768 (IF (EQL X (THE ...)) ...) (MISC.535).
1770 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1771 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1772 platform. The port supports 61-bit fixnums, large memory spaces
1773 and reloading shared object files.
1774 * enhancement: saving cores with foreign code loaded is now
1775 supported on ppc/Darwin in addition to the previously supported
1777 * enhancement: the statistical profiler now walks deeper into the
1778 call stack for more meaningful call-graphs and accrued time
1779 reports (x86/x86-64 only). It also now reports time spent in
1781 * enhancement: it is now possible to trace most individual methods
1782 of a generic function in addition to tracing the generic function
1784 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1786 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1787 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1788 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1789 produces an error. (thanks to Vincent Arkesteijn)
1790 * bug fix: NAMESTRING on pathnames with :WILD components in their
1791 directories works correctly. (thanks to Artem V. Andreev)
1792 * fixed bug 125: compiler preserves identity of closures. (reported
1794 * bug fixed: functions with &REST arguments sometimes failed with
1795 "Undefined function" when compiled with (DEBUG 3). (reported by
1797 * bug fix: overflow during compiling of setting element of a bit
1798 vector with constant index and value. (reported by Timmy Douglas)
1799 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1800 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1801 * on x86 compiler partially supports stack allocation of dynamic-extent
1803 * GO and RETURN-FROM do not check the extent of their exit points
1804 when compiled with SAFETY 0.
1805 * fixed some bugs related to Unicode integration:
1806 ** encoding and decoding errors are now much more robustly
1807 handled; it should now be possible to recover even from invalid
1808 input or output to the terminal. (thanks to Teemu Kalvas)
1809 ** provided a first cut at implementing STRING-TO-OCTETS and
1810 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1811 ** altered the SB-MD5 contributed module slightly, changing the
1812 interface just enough for it to be supportable for builds where
1813 lisp characters are not eight bits.
1814 * fixed some bugs revealed by Paul Dietz' test suite:
1815 ** the FORMATTER-generated functions for ~V[ conditionals require
1816 the correct number of arguments.
1817 ** READ-FROM-STRING returns the mandated second value when applied
1818 to displaced strings.
1819 ** the #\Rubout and #\Backspace characters are treated as invalid
1820 constituent characters by the tokenizer.
1822 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1823 * new feature: reloading changed shared object files with
1824 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1825 * new feature: references to foreign variables and functions
1826 can now be compiled and loaded before the corresponding shared
1827 object file is loaded, as long as the foreign definitions are
1828 available at runtime.
1829 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1830 supported. (thanks to Dan Debertin)
1831 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1832 just keywords. (thanks to Vincent Arkesteijn)
1833 * fixed bug #331: structure-class instances corresponding to
1834 DEFSTRUCT forms are now created eagerly.
1835 * fixed bug #345: backtraces from calls to undefined functions work
1836 on x86 as well. Related bug #61 is now also partially fixed on x86
1837 (backtraces from throws to unknown catch tags.)
1838 * bug fix: lambda-list parsing is now stricter vrt. order and number
1839 of lambda-list keywords.
1840 * bug fix: as specified by AMOP, an error is signalled if a
1841 class-option appears multiple times in a DEFCLASS form. (reported
1843 * bug fix: RANDOM can be compiled when the compiler derives the type
1844 of its numeric argument as a disjoint set of small integers.
1845 (reported by Paul Dietz)
1846 * bug fix: starting a core saved with shared objects loaded when
1847 those objects are not available no longer causes threaded SBCL to
1848 hang. (reported by Sean Ross)
1849 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1851 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1852 argument, which must always be NIL. (reported by Kalle Niemitalo)
1853 * bug fix: printing 1.0d+23 no longer results in an error.
1854 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1855 * bug fix: structure slot setters preserve evaluation order. (thanks
1857 * fixed some bugs related to Unicode integration:
1858 ** RUN-PROGRAM can allow its child to take input from a Lisp
1859 stream. (reported by Stefan Scholl)
1860 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1861 locales. (reported by Ken Causey)
1862 * fixed some bugs revealed by Paul Dietz' test suite:
1863 ** INCF, DECF and REMF evaluate their place form as specified in
1865 ** FORMATTER expands ~{ ~} iteration directives with V or #
1866 parameters correctly.
1867 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1868 consequent uses no arguments correctly.
1869 ** the system has a partial understanding of the (COMPLEX RATIO)
1871 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1874 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1875 * new feature: a build-time option (controlled by the :SB-UNICODE
1876 keyword feature, enabled by default) for building the system with
1877 support for the entire 21-bit character space defined by the
1879 * new feature: the system now has rudimentary external-format
1880 support; the primary user-visible change at this time is that
1881 characters with the high bit set (such as lower-case-e-acute) will
1882 print correctly to a terminal in a UTF-8 environment.
1883 * minor incompatible change: BASE-CHAR no longer names a class;
1884 however, CHARACTER continues to do so, as required by ANSI.
1885 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1886 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1887 has been moved to the SB-EXT package (temporarily re-exported from
1889 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1890 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1892 * fixed bug #351: better error-handlind and reporting for malformed
1894 * fixed bug #350: bignum-printing is now more memory-efficient,
1895 allowing printing of very large bignums, eg. (expt 2 10000000).
1896 (reported by Bruno Haible)
1897 * fixed bug #302: better primitive-type selection for intersection
1899 * fixed bug #308: non-graphic characters now all have names, as
1900 required. (reported by Bruno Haible)
1901 * bug fix: redefining a class with different superclasses now correctly
1902 removes it from the direct-subclasses of its previous superclasses.
1903 (reported by David Morse)
1904 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1905 new value now works. (reported by Bruno Haible)
1906 * bug fix: correct canonicalization of multiple non-standard slot
1907 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1908 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1909 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1910 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1911 now exists, an signals an error.
1912 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1913 during macroexpansion for non-iteration variables as well. (reported
1914 by Bruno Haible for CMUCL)
1915 * bug fix: Cyclic structures and unprintable objects in compiler
1916 messages no longer cause errors. (reported by Bruno Haible)
1917 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1918 and READ-FROM-STRING all now return a primary value of NIL if
1919 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1920 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1921 (reported by Bruno Haible for CMUCL)
1922 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1923 do not designate a condition. (reported by Bruno Haible for
1925 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1926 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1927 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1928 permitting resolution in favour of any of the conflicting symbols.
1929 (reported by Bruno Haible for CMUCL)
1930 * bug fix: EQUAL compiler optimizations is less aggressive on
1931 strings which can potentially compare true despite having distinct
1932 specialized array element types.
1933 * bug fix: unit enumerations can be defined without dividing by
1934 zero. (thanks to Vincent Arkesteijn)
1935 * FORMAT compile-time argument count checking has been enhanced.
1936 (report from Bruno Haible for CMUCL)
1937 * a partial workaround for the bug 262: the compiler does not try to
1938 inline-expand a local function doing RETURN-FROM from a deleted
1939 BLOCK. (thanks to Peter Denno for the bug report and to David
1940 Wragg for the simple test case)
1941 * fixed some bugs revealed by Paul Dietz' test suite:
1942 ** RENAME-PACKAGE allows all package designators as new package
1944 ** constraint propagation and lambda variable substitution are
1945 more cautious in dealing with partially deleted code.
1946 ** compiler sometimes lost reoptimization passes.
1947 ** CERROR, given a condition as condition designator, passes its
1948 remaining arguments to the continue format control without
1950 ** Case-altering FORMAT directives work correctly on non-ASCII
1952 ** The REAL type specifier handles bounds outside the floating
1953 point ranges without signalling FLOATING-POINT-OVERFLOW.
1954 ** Functions with IR1-transformations can create intercomponent
1955 references to global functions.
1956 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1958 ** FORMAT ~R treats a nil value for its first parameter correctly.
1960 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1961 * enhancement: saving cores with foreign code loaded is now
1962 supported on x86/NetBSD and sparc/Linux in addition to the previously
1963 supported platforms.
1964 * bug fix: on some platforms repeated installations caused multiple
1965 copies of HTML documentation to be installed -- should not happen
1966 any more. (reported by Stefan Scholl)
1967 * bug fix: parsing self-recursive alien record types multiple times
1968 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1969 original patch by Helmut Eller for CMUCL)
1970 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1971 (thanks to Richard Kreuter)
1972 * bug fix: defining classes whose accessors are methods on existing
1973 generic functions in other (locked) packages no longer signals
1974 bogus package lock violations. (reported by François-René Rideau)
1975 * bug fix: special variables as DEFMETHOD parameters no longer have
1976 associated bogus type declarations. (reported by David Wragg and
1978 * bug fix: read-write consistency on streams of element-type
1979 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1980 * bug fix: redefiniton of the only method of a generic function with
1981 no DEFGENERIC no longer emits a full WARNING. In addition,
1982 redefinition of generic functions with no DEFGENERIC to an
1983 incompatible lambda list now signals an error. (thanks to Zach
1985 * bug fix: DEFGENERIC now works even when there's a function of the
1986 same name in an enclosing lexical environment. (thanks to Zach
1988 * fixed compiler failure, caused by instrumenting code during
1989 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1990 * optimization: added loop analysis and improved register allocation
1991 * fixed some bugs revealed by Paul Dietz' test suite:
1992 ** POSITION on displaced vectors with non-zero displacement
1993 returns the right answer.
1994 ** (SIMPLE-STRING) is a valid type specifier for sequence
1996 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1998 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1999 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2001 ** The justification version of the FORMAT ~< directive treats
2002 non-zero minpad parameter correctly.
2004 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2005 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2006 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2007 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2008 the supported interface.
2009 * new feature: Single-stepping of code compiled with DEBUG 2 or
2010 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2011 * new feature: saving cores with foreign code loaded is now
2012 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2013 Timothy Moore's work for CMUCL)
2014 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2015 parameters to * instead of NIL if no initform is supplied.
2016 (reported by Johan Bockgaard)
2017 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2018 correctly when FUNCALL appears as the car of the form. Note:
2019 despite this FUNCALL forms are not currently subject to
2020 compiler-macro expansion. (port of Raymond Toy's fix for the
2021 same from CMUCL, reported by Johan Bockgaard)
2022 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2023 (thanks for Teemu Kalvas)
2024 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2025 case instead of terminating the iteration (thanks for Julian
2026 Squires, Sean Champ and Raymond Toy)
2027 * bug fix: incorrect expansion of defgeneric that caused a style
2028 warning. (thanks for Zach Beane)
2029 * bug fix: slot accessor effective method computation works properly
2030 for classes with multiple non-standard applicable methods on
2031 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2032 * on x86 compiler supports stack allocation of results of LIST and
2033 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2034 CMUCL implementation by Gerd Moellmann)
2035 * fixed some bugs revealed by Paul Dietz' test suite:
2036 ** FORMAT strings with both the ~<~:;~> form of the justification
2037 directive and pretty-printing directives cause an error.
2039 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2040 * incompatible change: the internal functions
2041 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2042 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2043 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2044 instead of the old functions.
2045 * new feature: on platforms where "dladdr" is available foreign
2046 function names now appear in backtraces. (based on Helmut Eller's
2048 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2049 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2051 * documentation: ASDF manual is now distributed with SBCL.
2052 * bug fix: SBCL can now load its contributed modules with REQUIRE
2053 even if the system-provided entries have been removed from
2054 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2056 * bug fix: backtraces involving undefined functions or assembly
2057 routines are more informative. (thanks to Brian Downing)
2058 * bug fix: mutually referent alien structures now work correctly.
2059 (reported by Rick Taube)
2060 * bug fix: structures defined by WITH-ALIEN can be referred to
2061 within other definitions in the same WITH-ALIEN.
2062 * bug fix: division operators (MOD, TRUNCATE and the like) with
2063 constant zero divisors and integer dividends no longer generate
2065 * bug fix: provide default methods for INPUT-STREAM-P and
2066 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2067 * bug fix: improve the detection and resolution of MOP metacycles
2068 (where effective method computation depends on the generic function
2069 having its effective method computed). (reported by Bruno Haible)
2070 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2071 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2072 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2073 * bug fix: pathnames with relative directory components can be
2074 represented relative to default pathnames.
2075 * optimization: in taking the GCD of bignums, reduce the two bignums
2076 to approximately the same size (using Euclid's algorithm) before
2077 applying the more sophisticated binary GCD. (thanks to Juho
2079 * optimization: COUNT on bitvectors now operates word-at-a-time.
2080 * optimization: ASH with a positive, but not necessarily constant,
2081 (leftwards) shift, when in a modular context, is compiled to a
2083 * fixed some bugs revealed by Paul Dietz' test suite:
2084 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2085 if the corresponding argument is NIL.
2086 ** FORMAT directives accepting a minpad parameter treat negative
2088 ** literal commas as character parameters to FORMAT directives are
2090 ** literal spaces directly after ~<Newline> directives within a
2091 format-logical-block (~:< ~@:>) do not induce :FILL-style
2092 conditional newlines.
2093 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2094 as specified: it no longer includes conditional newlines).
2095 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2097 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2098 :PER-LINE-PREFIX argument does not evaluate to a string.
2100 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2101 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2102 the manual for details; for now, package locks can be disabled by
2103 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2104 if no major problems are found then it is likely that they will be
2105 compiled in unconditionally.
2106 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2107 unsupported operators on all platforms. To load a shared library
2108 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2109 object file, link it into a shared library outside of SBCL and
2110 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2112 * bug workaround: ROOM T and the GC were not working together
2113 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2114 reporting facility aren't preserved. That reporting has been
2115 disabled, so now until and unless someone figures out how to make it
2116 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2117 * minor incompatible change: as threatened around sbcl-0.8.0, the
2118 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2119 The symbols are also exported from SB-PCL for backwards
2120 compatibility, but more so than before SB-PCL should be treated as
2121 an implementation-internal package.
2122 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2124 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2125 It is known as of this release not to work on the Alpha, however.
2126 * fixed bug #167: errors signalled due to illegal syntax in method
2127 bodies are now more legible.
2128 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2129 designators and can hence be used with TYPEP.
2130 * fixed bug #333: CHECK-TYPE now ensures that the type error
2131 signalled, if any, has the right object to be accessed by
2132 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2133 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2134 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2136 * fixed bug #334: programmatic addition of slots using specialized
2137 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2138 and :ALLOCATION :CLASS slots.
2139 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2140 float exponents. (rereported by Peter Seibel)
2141 * fixed a bug: #\Space (and other whitespace characters) are no
2142 longer considered to be macro characters in standard syntax by
2143 GET-MACRO-CHARACTER.
2144 * fixed bug: initialization of condition class metaobjects no longer
2145 causes an instance of the condition to be created. (reported by
2147 * fixed bug: it is now possible to have slots such that
2148 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2149 description is neither :INSTANCE nor :CLASS.
2150 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2151 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2152 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2153 non-standard SLOT-DEFINITION-ALLOCATION values.
2154 * fixed bugs: various race conditions handling exiting threads.
2155 CL-PPCRE's thread torture test now passes.
2156 * fixed arguably-a-bug: GC time proportional to number of threads in
2157 system even when most of them are idle
2158 * optimization: improved performance of BIT and SBIT on bit-vectors.
2159 * .fasl file incompatibility: The fasl file version number has
2160 been incremented because of changes associated with package locks.
2162 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2163 * minor incompatible change: the system no longer provides
2164 optimization hints (conditions which are TYPEP
2165 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2166 mode; these hints are still emitted when higher SPEED optimization
2168 * new contrib module: a sampling profiler (profiling by statistical
2169 sampling, rather than by instrumenting functions) is available as
2170 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2171 * the behaviour of the standard function ED is now customizeable by
2172 third parties through a hook variable: see ED's documentation
2173 string for information on the protocol.
2174 * the compiler no longer emits efficiency notes for (FUNCALL X)
2175 when the type of X is uncertain under default optimization
2177 * fixed bug 276: mutating a binding of a specialized parameter to a
2178 method to something that is not TYPEP the specializer is now
2180 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2181 exhibit the required behaviour on the x86 platform. (thanks to
2182 Peter van Eynde, Eric Marsden and Bruno Haible)
2183 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2184 even for difficult arguments. (reported by Peter Graves)
2185 * fixed bug 141a: the backquote printer now descends quoted
2187 * fixed another bug in backquote printing: no more destructive
2188 modification of the form's list structure. (reported by Brian
2190 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2191 move between its address being taken and the call to
2192 interrupt_thread, fixing a crashing race condition.
2193 * the SB-POSIX contrib implementation has been adjusted so that it
2194 no longer exhibits ridiculously poor performance when constructing
2195 instances corresponding to C structs.
2197 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2198 * minor incompatible change: the sb-grovel contrib now treats C
2199 structures as alien (in the sense of SB-ALIEN) objects rather than
2200 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2201 has implications for memory management of client code
2202 (sb-grovel-returned objects must now be manually managed) and for
2203 type safety (alien objects now have full types).
2204 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2205 used to control emission of compiler diagnostics, rather than the
2206 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2207 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2208 quality should be considered deprecated.
2209 * install.sh now installs the user manual as well
2210 * (not quite a new documentable feature, but worth considering in
2211 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2212 beginnings of a semantically meaningful condition hierarchy is
2213 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2215 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2216 designator as the defaults argument.
2217 * fixed bug: Displaced arrays whose displaced-to array has become
2218 too small now cause ARRAY-DIMENSION to signal an error, providing
2219 for safer bounds-checking. (reported by Bruno Haible)
2220 * fixed bug: DEFCLASS slot definitions with identical :READER and
2221 :WRITER names now signal a reasonable error. (reported by Thomas
2223 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2225 * fixed bug: Invalid slot specification errors now print correctly.
2226 (thanks to Zach Beane)
2227 * fixed bug 320: Shared to local slot value transfers in class
2228 redefinitions now happen corrently from superclasses as
2229 well. (reported by Bruno Haible)
2230 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2232 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2233 improper lists correctly. (reported by Bruno Haible)
2234 * fixed bug 313: source-transform for <fun-name> was erroneously
2235 applied to a call of a value of a variable with name <fun-name>.
2236 (reported by Antonio Menezes Leitao)
2237 * fixed bug 307: The obsolete instance protocol ensures that
2238 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2239 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2240 remove dead unknown-values globs from the middle of the stack.
2241 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2242 easier to resume long computations after using *BREAK-ON-SIGNALS*
2243 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2244 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2245 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2246 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2247 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2248 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2250 * fixed a (fairly theoretical) bug in string printing: if
2251 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2252 does not have array-element-type equal to the most general string
2254 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2255 function name. (thanks to Nikodemus Siivola)
2256 * fixed bug: compile-time format string checker failed on
2257 non-closed ~{. (reported by Thomas F Burdick)
2258 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2259 don't warn on legal loop code involving more than one aggregate
2260 boolean. (thanks to Nikodemus Siivola)
2261 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2262 in keyword position to MAKE-PACKAGE was not being reported as
2264 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2265 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2267 * fixed bugs 280 and 312: the checking for multiple definitions in a
2268 file is less likely to become confused by uses of inline
2270 * fixed bug: the #S reader macro performs the keyword coercion
2271 specified for slot names. (reported by Kalle Niemitalo)
2272 * fixed bug: lambda lists may contain symbols whose names start with
2273 & but are not lambda-list-keywords; their occurrence triggers a
2275 * fixed bug 321: define-method-combination argument lambda lists do
2276 not cause invalid code to be generated when &optional or &aux
2277 variables have default values. (reported by Bruno Haible)
2278 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2279 have CLOS classes; this appears also to have fixed reported
2280 instability in stack exhaustion detection.
2281 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2282 MUFFLE-WARNING when no associated restart is present is now
2284 * optimization: rearranged the expansion of various defining macros
2285 so that each expands into only one top-level form in a
2286 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2288 * optimization: used a previously little-used slot in symbols to
2289 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2291 * fixed some bugs revealed by Paul Dietz' test suite:
2292 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2294 ** Fixed an optimization bug related to inheritance of initforms
2295 from local to shared slots.
2296 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2297 ** CAST optimizer forgot to flush argument derived type.
2298 ** print/read consistency on floats is now orders of magnitude
2299 more likely. (thanks also to Bruno Haible for a similar report
2301 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2302 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2303 for objects of type REAL. Make it so.
2304 ** SXHASH is specified (infelicitously) to respect similarity,
2305 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2306 it so. (thanks to Markus Ziegler)
2307 ** on the Alpha, the compiler succeeds in compiling functions
2308 returning a known number of arguments greater than 63.
2309 ** fixed handling of invalid NIL arguments in keyword position
2311 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2313 ** provided a readably-printable representation for RANDOM-STATE
2315 ** ensured that pathnames, where they have a namestring, always
2316 print using #P"..." syntax.
2318 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2319 * Support for the forthcoming 2.0 version of the NetBSD kernel
2320 running on x86 hardware has been added. (thanks to Perry
2321 E. Metzger most immediately, and others for their past work)
2322 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2323 current release version; SBCL's OpenBSD support had been broken
2324 since about the time of OpenBSD's switch to ELF binary format.)
2325 * [placeholder for DX summary]
2326 ** user code with &REST lists declared dynamic-extent, under high
2327 speed or space and low safety and debug optimization policy.
2328 * The manual has been converted to Texinfo format and the debugger
2329 chapter from the cmucl manual has been added.
2330 * A facility has been added to extract documentation strings from
2331 sbcl and store them as Texinfo-formatted snippets for inclusion in
2332 the manual (via Texinfo's @include directive)
2333 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2334 the test case to Dave Roberts)
2335 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2336 pointers. (reported by Sean Ross)
2337 * bug fix: PROFILE output is printed nicely even for large numerical
2338 values. (thanks to Zach Beane)
2339 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2340 greater than 32 handle EOF correctly.
2341 * bug fix: on X86 an immediate argument of the IMUL instruction is
2342 correctly printed in disassembly. (thanks to Lutz Euler)
2343 * bug fix: class slots in redefined classes preserve their old
2344 values. (thanks to Bruno Haible and Nikodemus Siivola)
2345 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2346 succeeds. (reported by Marco Baringer)
2347 * fixed some bugs revealed by Paul Dietz' test suite:
2348 ** READ-SEQUENCE now works on ECHO-STREAMs.
2349 ** RATIONALIZE works more according to its specification. (thanks
2352 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2353 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2354 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2355 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2356 should matter to you only if you rebind the printer control
2357 variables and then find you want different bindings in the
2358 debugger than in the ordinary execution of your program.)
2359 * The runtime build system has been tweaked to support building
2360 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2361 assembler and linker. (thanks to Nikodemus Siivola)
2362 * Unbound, undefined, undeclared variables now trigger full
2363 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2364 more often programmer error than deliberate exploitation of undefined
2366 * optimization: the hash algorithm for strings has changed to one
2367 that is less vulnerable to spurious collisions. (thanks to Juho
2369 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2370 needless bounds checking. (thanks to Juho Snellman)
2371 * optimization: implemented multiplication as a modular
2372 (UNSIGNED-BYTE 32) operation on the PPC backend.
2373 * fixed some bugs revealed by Paul Dietz' test suite:
2374 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2376 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2377 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2378 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2380 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2382 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2384 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2385 their output stream on EOF from read.
2386 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2387 have been read to end-of-file.
2388 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2390 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2391 description of determination of which consecutive characters
2393 ** Printing the "Space" character with escaping on now yields "#\\ ",
2394 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2395 ** Reading floating-point numbers with *READ-BASE* set to a number
2396 less than 10 works correctly.
2397 ** Reading floating-point numbers with *READ-BASE* set to a number
2398 more than 10 works correctly.
2399 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2400 the readtable currently in effect.
2402 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2403 * minor incompatible change: parsing of namestrings on a physical
2404 (Unix) host has changed; numbers after the final #\. in a
2405 namestring are no longer interpreted as a version field. This is
2406 intented to be largely invisible to the user, except that the
2407 meaning of the namestring "*.*.*" has changed: it now refers to a
2408 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2409 should usually be replaced by
2410 (make-pathname :name :wild :type :wild :version :wild)
2411 with the added benefit that this is more likely to be portable.
2412 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2413 option to OPEN now signals an error if the file being opened
2414 exists; this may have an impact on existing code.
2415 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2416 signals received, on the PowerPC platforms (both Linux and
2417 Darwin). (thanks to Pierre Mai for pointing out the location of
2419 * several fixes on OS X: The system now builds and runs cleanly on
2420 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2421 (thanks to Brian Mastenbrook)
2422 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2423 second-resolution: integer multiples of 1/3600 between -24 and 24.
2424 (thanks to Vincent Arkesteijn)
2425 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2426 type when called with 1 argument; PEEK-CHAR checked type of
2427 PEEK-TYPE only after having read first character from a
2428 stream. (reported by Peter Graves)
2429 * bug fix: the garbage collector now has much better locality
2430 behaviour, and in particular no longer treats all memory as being
2431 exhausted when many small objects point to each other in a deeply
2433 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2434 recognized as being TYPEP their class.
2435 * bug fix: the PUSHNEW documentation string has been corrected.
2436 (thanks to Vincent Arkesteijn)
2437 * bug fix: defaulting of the value for the last of an atypically
2438 large number of multiple values being bound was not being
2439 performed correctly on the Alpha or PPC platforms
2440 * optimization: implemented multiplication as a modular
2441 (UNSIGNED-BYTE 32) operation on the x86 backend.
2442 * optimization: SEARCH on simple-base-strings can now be open-coded.
2443 (see also contrib/compiler-extras.lisp for inspiration for
2444 teaching the compiler about the Boyer-Moore algorithm).
2445 * value, returned by MAX (and MIN) called with several EQUALP, but
2446 not EQL, arguments now does not depend on compiler settings.
2447 * fixed some bugs revealed by Paul Dietz' test suite:
2448 ** in stack analysis liveness information is propagated from
2449 non-local entry points.
2450 ** pathwise CAST removing failed when the CAST node did not start
2452 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2453 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2455 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2456 host is already defined.
2457 ** RENAME-FILE works on streams instead of signalling an internal
2459 ** PEEK-CHAR uses the current readtable when determining whether
2460 or not a character is whitespace.
2461 ** MERGE-PATHNAMES handles the case when the pathname does not
2462 specify a name while the default-pathname specifies a version.
2463 ** Pathnames now stand a better chance of respecting print/read
2465 ** Attempting to use standardized file system operators with a
2466 pathname with invalid :DIRECTORY components signals a
2468 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2470 ** DIRECTORY on logical pathnames is more correct.
2471 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2472 signal a TYPE-ERROR if their argument is not a stream.
2473 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2474 designator argument does not designate a stream.
2475 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2476 examining the synonym.
2477 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2479 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2480 element-type larger than ([UN]SIGNED-BYTE 32).
2482 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2483 * When built with the :SB-FUTEX feature, threaded builds now take
2484 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2485 for faster/more reliable mutex and condition variable support.
2486 * Incompatible change (but one you probably shouldn't have been using
2487 anyway): the interface and code for arbitrating between multiple
2488 threads in the same user session has been redesigned.
2489 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2490 error when called without an explicit environment argument.
2491 (thanks to Nikodemus Siivola)
2492 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2493 confused about their position. (thanks to Adam Warner and Gerd
2495 * bug fix: Pretty printing backquoted forms with unquotations in the
2496 argument list position of various code constructs such as LAMBDA
2497 now works correctly. (reported by Paul Dietz)
2498 * bug fix: Pretty printing unquotations no longer loses all
2499 stream position information.
2500 * optimization: performance of string output streams is now less
2501 poor for multiple small sequence writes.
2502 * optimization: performance of CSUBTYPEP in the presence of complex
2503 expressions involving CONS and NOT many times has been improved.
2504 (reported by Paul Dietz)
2505 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2507 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2508 streams for the REPL, for the debugger, and so on. (thanks to
2510 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2511 * fixed some bugs revealed by Paul Dietz' test suite:
2512 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2513 is ignored if it is too small, rather than propagating through
2514 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2515 ** extremely complex negations of CONS types were not being
2516 sufficiently canonicalized, leading to inconsistencies in
2518 ** VALUES tranformer lost derived type.
2520 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2521 * fixed a bootstrapping bug: the build process no longer assumes
2522 that the various BOOLE-related constants have the same value in
2523 host and target lisps. (noted by Paul Dietz' test suite on an
2524 SBCL binary built from CLISP)
2525 * The system can now be dynamically linked on the MIPS platform,
2526 which enables dynamic loading of foreign code from Lisp. (thanks
2527 to Ralf Baechle for discussions on the MIPS ABI)
2528 * The system now records debugging information for its own source
2529 files in a filesystem-position-independent manner, relative to
2530 the "SYS" logical host.
2531 * fixed a compiler bug: MV-LET convertion did not check references
2532 to the "max args" entry point. (reported by Brian Downing)
2533 * tweaked disassembly notes to be less confident about proclaiming
2534 some instruction as an LRA. (thanks to Brian Downing)
2535 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2536 now each have their own history, command character, and other
2537 characteristics. (thanks to David Lichteblau)
2538 * fixed some bugs revealed by Paul Dietz' test suite:
2539 ** compiler failure in compiling LOGAND expressions including a
2541 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2542 shift greater than 32.
2543 ** FLUSH-DEST did not mark blocks for type check regeneration.
2544 ** HANDLER-CASE failed to accept declarations in handler clauses
2545 in some circumstances.
2547 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2548 * New code in contrib/sb-introspect (still probably not entirely
2549 stable yet) provides some support for smart Lisp development
2550 environments like SLIME.
2551 * The conditions signalled for errors occurring when loading .fasl
2552 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2553 in a way which should help ASDF recover gracefully.
2554 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2555 stuff has been cleaned up. If you code contrib/ stuff, this might
2556 affect you, and you can look at contrib/README, contrib/STANDARDS,
2557 and/or the 0.8.4.27 diff to check.
2558 * In full calls the compiler now does not generate checks for declared
2559 argument types for all arguments.
2560 * various threading fixes
2561 ** and some experimental patches which didn't make it into
2562 the main tree for this release, but which are shipped in
2563 contrib/experimental-thread.patch as a possible fix for some
2564 failures (deadlock, spinning...) in GC-intensive multithreaded
2566 * fixed PPC build problem (source code incompatibility of different
2567 library versions): added offsetof() hackery which attempts to divine
2568 where glibc maintainers put uc_mcontext today
2569 * fixed bug 282: compiler does not trust type assertions while passing
2570 arguments to a full call.
2571 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2572 &OPTIONAL VALUES type parameter.
2573 * fix bug 214: algorithm for noting rejected templates is now more
2574 similar to that of template seletion. (also reported by rydis on
2576 * fixed bug 141b: printing backquoted information readably and prettily
2577 inserts a space where necessary.
2578 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2579 (CHARACTER 10) are now reported as errors, rather than propagated
2580 as unknown types. (reported by piso on #lisp)
2581 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2582 with values NIL and :ERROR. (thanks to Milan Zamazal)
2583 * fixed bug 191c: CLOS now does proper keyword argument checking as
2584 described in CLHS 7.6.5 and 7.6.5.1.
2585 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2586 longer raise an error; further, using a list as a for-as-arithmetic
2587 counter now raises a meaningful error.
2588 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2589 now understood by sequence creation functions such as MAKE-SEQUENCE
2591 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2592 a STRING-INPUT-STREAM.
2593 * compiler enhancement: SIGNUM is now better able to derive the type
2595 * type declarations inside WITH-SLOTS are checked. (reported by
2597 * fixed some bugs revealed by Paul Dietz' test suite:
2598 ** incorrect optimization of TRUNCATE for a positive first
2599 argument and negative second.
2600 ** compiler failure in let-convertion during flushing dead code.
2601 ** compiler failure while deriving type of TRUNCATE on an
2602 interval, containing 0.
2603 ** ASH of a negative bignum by a negative bignum count now returns
2605 ** intersection of CONS types now canonicalizes properly, fixing
2606 inconsistencies in SUBTYPEP.
2608 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2609 * incompatible change: The --disable-debugger command line
2610 option now clobbers the debugger at a more fundamental
2611 level, by redefining #'INVOKE-DEBUGGER instead of by
2612 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2613 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2614 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2615 enter the debugger REPL and then suffer recursive errors
2616 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2617 while under the new system, BREAK in --disable-debugger mode
2618 terminates the system just as an unhandled error would.
2619 * fixed compiler performance when processing loops with a step >1;
2620 * bug fix: DOCUMENTATION now retrieves generic function
2621 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2622 support has been systematized, and now supports the methods
2623 specified by ANSI, along with a default method and a method for
2624 slot documentation. (reported by Nathan Froyd)
2625 * bug fix: effective methods associated with a generic function are
2626 no longer cached over a change of that generic function's method
2627 combination. (reported by Andreas Fuchs)
2628 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2629 names a non-existent directory. (thanks to Andreas Fuchs)
2630 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2631 the PPC platform, lead to stack corruption; this has been fixed.
2632 (reported by Rainer Joswig)
2633 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2634 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2635 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2636 arithmetic in general, is now much more reliable on the PPC
2638 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2639 platform now returns the right answer.
2640 * optimization: restored some effective method precomputation in
2641 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2642 precomputation is now tunable.
2643 * optimization: compiler-internal data structure use has been
2644 reviewed, and changes have been made that should improve the
2645 performance of the compiler by about 20%.
2646 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2647 simple-base-strings and simple-bit-vectors is improved.
2648 * optimization: the optimization of 32-bit logical and arithmetic
2649 functions introduced in version 0.8.3 on the x86 has been
2650 implemented on the mips, ppc and sparc platforms; an
2651 implementation of the same facility, but for 64-bit arithmetic,
2652 has been added for the alpha.
2653 * microoptimization: the compiler is better able to make use of the
2654 x86 LEA instruction for multiplication by constants.
2655 * bug fix: in some situations compiler did not report usage of
2656 generic arithmetic in (SPEED 3) policy.
2657 * bug 145b fix: compiler used wrong type specifier while converting
2658 MEMBER-types to numeric.
2659 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2661 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2662 index argument now works properly on the Alpha platform.
2663 * bug fix: floating point exception treatment on the Alpha platform
2665 * bug fix: FILE-POSITION works much better on string input and
2666 output streams. (thanks to Nikodemus Siivola)
2667 * bug fix: many threading/garbage collection symptoms sorted.
2668 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2669 might be pseudo-atomic.
2670 * internal change: Stopping for GC is now done with signals not
2671 ptrace. GC is now done in whichever thread wanted it, instead of
2673 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2675 * bug fix: result form in DO is not contained in the implicit
2677 * incompatible change: ICR structure is changed; the value part of
2678 CONTINUATION is now called LVAR; corresponding functions are
2679 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2680 * added type deriver for ISQRT (thanks to Robert E. Brown).
2681 * bug fix: better support for loading from the command line when an
2682 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2684 * fixed some bugs revealed by Paul Dietz' test suite:
2685 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2686 ** ROUND and FROUND now give the right answer when given very
2687 small float arguments.
2688 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2690 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2691 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2692 ** LCM should return a non-negative integer.
2693 ** PARSE-INTEGER returned the index of a terminator instead of the
2694 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2695 ** PARSE-INTEGER returned an incorrect index being applied to a
2697 ** LCM with two arguments of 0 returns 0 rather than signalling
2699 ** unsigned addition of a 32-bit constant with the high bit set no
2700 longer causes an internal compiler error on the x86.
2701 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2702 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2703 with negative last argument.
2704 ** byte specifiers with zero size and position no longer cause
2705 an error during type derivation.
2706 ** bignum multiplication on the Alpha platform now returns the
2708 * porting: The system now builds on SuSE AMD64, although it still
2709 generates a 32-bit binary.
2710 * .fasl file incompatibility: The fasl file version number has
2711 been incremented (because of the changes to internal compiler
2712 data structures referred to above).
2714 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2715 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2716 more accurately, on the Darwin kernel running on PowerPC hardware.
2717 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2718 * Compiler code deletion notes now signal a condition of type
2719 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2720 an associated MUFFLE-WARNING restart.
2721 * The compiler now performs limited argument count validation of
2722 constant format strings in FORMAT, and where appropriate in ERROR,
2723 CERROR and WARN. (thanks to Gerd Moellmann)
2724 * New ASDF-INSTALL contrib can be used for automatic download and
2725 installation of third-party Lisp code from CCLAN or other sites
2727 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2728 which forces another thread to execute a function supplied by the
2730 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2731 now accept and act upon their :ELEMENT-TYPE keyword argument.
2732 (reported by Martin Atzmueller, Edi Weitz)
2733 * bug fix: FILE-POSITION now accepts position designators up to
2734 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2735 is the greater. (thanks to Patrik Nordebo)
2736 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2737 Kalle Olavi Niemitalo)
2738 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2739 printable. (reported by Eric Marsden)
2740 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2741 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2742 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2743 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2744 (thanks to Stig Sandoe)
2745 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2746 UNREAD-CHAR now correctly includes the unread character in the
2747 target sequence. (thanks to Gerd Moellmann)
2748 * bug fix in threaded builds: the system can now be suspended and
2749 resumed by shell job control with minimal disruption.
2750 * bug fixes in times and timezones >2038AD
2751 * better handling of "where is GNU make?" problem in build scripts
2752 (thanks to Nikodemus Siivola)
2753 * new optimization: inside a named function any reference to a
2754 function with the same name is considered to be a self-reference;
2755 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2756 optimization quality.
2757 * new optimization on x86: logical functions and + now have
2758 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2759 used when the result is truncated to 32 bits.
2760 * VALUES declaration is partially enabled.
2761 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2762 * fixed some bugs revealed by Paul Dietz' test suite:
2763 ** The system now obeys the constraint imposed by
2764 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2765 types form a lattice under type intersection.
2766 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2767 ** ASSOC now ignores NIL elements in an alist.
2768 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2769 and (1+ MOST-POSITIVE-FIXNUM) answers.
2770 ** The addition of a method with invalid qualifiers to a generic
2771 function does not cause an error to be signalled immediately;
2772 a warning is signalled, and the error is generated only on
2773 calling the generic function.
2774 * changed .fasl file version number, in part to add type codes for
2775 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2776 obscure ANSI requirements
2778 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2779 * fixed bug 148: failure to inline-expand a local function left
2780 garbage, confusing the compiler.
2781 * fixed bugs 3cd: structure slot readers perform type check if the
2782 slot can have an invalid value (i.e. it is either not initialized
2783 or can be written with a less specific slot writer).
2784 * bug fix: the compiler now traps array references to elements off
2785 the end of an array; previously, the bounds checking in some
2786 circumstances could go off-by-one.
2787 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2789 * type declarations for array element types now obey the description
2790 on the CLHS page "Declaration TYPE", as per discussions on
2791 sbcl-help around 2003-05-08. This means that a declaration
2792 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2793 declaration, all references to BAR will be asserted or assumed
2794 (with THE, so dependent on compiler policy) to involve objects of
2795 type FOO. Note that no such declaration is implied in
2796 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2797 * declared types of functions from the "Conditions"
2798 chapter. (reported by Paul Dietz)
2799 * bug fix: CERROR accepts a function as its first argument.
2800 * bug fix: NTH an NTHCDR accept a bignum as index
2801 arguments. (reported by Adam Warner)
2802 * optimization: character compare routines now optimize comparing
2803 against a constant character. (reported by Gilbert Baumann)
2804 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2805 argument now works properly on the MIPS platform.
2806 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2807 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2808 declared to have a functional type, does not cause an error
2810 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2811 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2812 * fixed bug: if last continuation of a deleted block has a
2813 destination, this destination should be deleted too. (reported by
2815 * fixed a bug in the bootstrap process: the host compiler's values
2816 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2817 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2818 test case from Patrik Nordebo)
2819 * improved the ability of the disassembler on the PPC platform to
2820 provide helpful disassembly notes.
2821 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2822 the class in more cases than previously.
2823 * bug fix: FILE-POSITION now understands :START and :END for
2824 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2825 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2826 8) data. (thanks to David Lichteblau)
2827 * bug fix: it is possible to add a method to a generic function
2828 without lambda list.
2829 * bug fix: reader failed to signal END-OF-FILE inside an
2830 object representation. (reported by Nikodemus Siivola)
2831 * fixed some bugs revealed by Paul Dietz' test suite:
2832 ** LAST and [N]BUTLAST should accept a bignum.
2833 ** condition slot accessors are methods.
2834 ** (VECTOR NIL) is a subtype of STRING.
2836 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2837 * minor incompatible change: some nonsensical specialized lambda
2838 lists (used in DEFMETHOD) which were previously ignored now signal
2840 * minor incompatible change: the system is now aware of the types of
2841 variables in the COMMON-LISP package, and will signal errors for
2842 most violations of these type constraints (where previously they
2843 were silently accepted).
2844 * minor incompatible change: COMPILE-FILE now uses the freedom
2845 afforded (ANSI 3.2.2.3) to use derived function types for
2846 functions defined in the same file. This also permits the system
2847 to warn on static type mismatches and function
2848 redefinition. (Currently it does not work with high DEBUG level.)
2849 * minor incompatible change: VALUES declaration is disabled.
2850 * When issuing notes, the compiler now signals a condition of type
2851 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2852 restart for use in user handlers. It is expected that the
2853 COMPILER-NOTE condition will eventually become a condition
2854 supertype to a hierarchy of note types, which will then be
2855 handleable in a similar fashion. However, at the moment, no such
2856 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2857 but it's an implementation detail, not a classification for the
2859 * Changes in type checking closed the following bugs:
2860 ** type checking of unused values (192b, 194d, 203);
2861 ** template selection based on unsafe type assertions (192c, 236);
2862 ** type checking in branches (194bc).
2863 * A short form of VALUES type specifier has ANSI meaning (it has
2864 increased the number of situations when SBCL cannot perform type
2866 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2867 keywords or constants is permissible.
2868 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2869 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2870 classes. (thanks to Antonio Martinez)
2871 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2872 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2873 argument to a pretty-print stream. Instead, it leaves any such
2874 operations to DESCRIBE-OBJECT methods.
2875 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2877 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2878 lambda lists. (thanks to David Lichteblau)
2879 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2880 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2881 (thanks to Gerd Moellmann)
2882 * fixed bug 246: increased compilation speed of long
2883 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2885 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2887 * DEFCONSTANT now throws a condition of type
2888 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2889 constant to a non-EQL value; CONTINUE and ABORT restarts
2890 respectively change and preserve the value.
2891 * fixed bug 63: The code walker, part of the implementation of CLOS,
2892 is now better at handling symbol macros.
2893 * bug fix: There is no longer an internal implementation type named
2894 CL:LENGTH. (reported by Raymond Toy)
2895 * bug fix: In macro-like defining macros/special operators the
2896 implicit block does not enclose the lambda list.
2897 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2898 enclosed in parentheses) are not suitable as type specifiers, and
2899 their use properly signals an error now.
2900 * bug fix: An argument count mismatch for a type specifier in code
2901 being compiled no longer causes an unhandled error at compile
2902 time, but signals a compile-time warning.
2903 * fixed simple vector readable printing
2904 * bug fix: DESCRIBE takes more care over whether the class
2905 precedence list slot of a class is bound before accessing it.
2906 (reported by Markus Krummenacker)
2907 * bug fix: FORMATTER can successfully compile pretty-printer format
2908 strings which use variants of the ~* directive inside.
2909 * bug fix: SEARCH now applies its TEST predicate to the elements of
2910 the arguments in the correct order. (thanks to Wolfhard Buss)
2911 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2912 in type inference. (thanks to Robert E. Brown)
2913 * bug fix: Reading in symbols with an explicit package name of ""
2914 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2915 * fixed some bugs revealed by Paul Dietz' test suite:
2916 ** NIL is now allowed as a structure slot name.
2917 ** Arbitrary numbers, not just REALs, are allowed in certain
2918 circumstances in LOOP for-as-arithmetic clauses.
2919 ** Multiple class redefinitions before slot access no longer
2920 causes a type error.
2921 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2922 association between the name and a class.
2923 ** Generic functions with non-standard method-combination and over
2924 five methods all of which return constants no longer return NIL
2925 after the first few invocations. (thanks to Gerd Moellmann)
2926 ** CALL-NEXT-METHOD with no arguments now passes the original
2927 values of the arguments, even in the presence of assignment.
2928 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2930 ** DISASSEMBLE works with closures and funcallable instances.
2931 ** ADD-METHOD now returns the generic function, not the new method.
2932 ** FIND-METHOD signals an error if the lengths of the specializers
2933 is incompatible with the generic function, even if the ERRORP
2935 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2936 which its argument is a member.
2937 ** DEFCLASS only redefines the class named by its class-name
2938 argument if that name is the proper name of the class;
2939 otherwise, it creates a new class.
2940 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2941 of the UNBOUND-SLOT condition to the name of the slot.
2942 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2943 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2944 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2945 SLOT-MAKUNBOUND in the specified fashion.
2947 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2948 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2949 cross-compilation host. As a consequence, we can now bootstrap our
2950 way up to SBCL starting with a bare gcc toolchain and human-readable
2951 source code (first the source to CLISP, then the source to SBCL).
2952 * A contributed module containing a partial implementation of the
2953 simple-streams interface has been included. (thanks to Rudi
2955 * A contributed module implementing the RFC1321 Message Digest
2956 Algorithm, known as MD5, has been included.
2957 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2958 no longer has any effect, as the code controlled by this feature
2959 has been deleted. (As far as we know, no-one has ever built using
2960 this feature, and its semantics were confused in any case).
2961 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2962 work on conditions (as required by the ANSI specification),
2963 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2964 expected behaviour on conditions. Users should note, however,
2965 that such behaviour is not required by the ANSI specification,
2966 and so use of this behaviour may render their code unportable.
2967 * fixed some bugs revealed by Paul Dietz' test suite:
2968 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2970 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2971 ** SXHASH on condition objects no longer returns NIL.
2972 ** :ALLOCATION :CLASS slots are better treated; their values are
2973 updated on class redefinition, and initforms inherited from
2974 superclasses are applied.
2975 ** REMOVE-METHOD returns its generic function argument even when
2976 no method was removed.
2977 ** SHARED-INITIALIZE now initializes the values of the requested
2978 slots, including those with :ALLOCATION :CLASS.
2979 ** ALLOCATE-INSTANCE now works on structure classes defined via
2980 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2982 ** SLOT-EXISTS-P now works on conditions, as well as structures
2984 ** MAKE-LOAD-FORM now has the required methods on
2985 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2986 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2987 keyword, and now implements the SLOT-NAMES argument.
2988 ** methods with &OPTIONAL arguments no longer allow too many
2989 arguments to be passed in the call without error.
2990 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2991 option is consistent with the required arguments of the generic
2992 function lambda list.
2993 * bug fix: REQUIRE accepts a string designator. (Thanks to
2995 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2996 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2997 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2998 * bug fix: The compiler now checks for duplicated variables in macro
3000 * bug fix: SETQ on globals returns the correct value.
3001 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3002 not a COMPILER-ERROR (followed by some other strange error on
3003 choosing the CONTINUE restart).
3004 * bug fix: make.sh and friends are now more consistent in the way that
3005 they look for GNU "make".
3007 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3008 * experimental native threads support (on x86 Linux >=2.4 only).
3009 This is not compiled in by default: you need to add :SB-THREAD to
3010 the target features. See the "Beyond ANSI" chapter of the manual
3012 * fix for longstanding nonANSIism: The old distinction between
3013 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3014 The return value from CL:FIND-CLASS is now a CLOS class, and
3015 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3016 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3017 * An interface to the MetaObject Protocol, as described in Kiczales,
3018 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3019 MIT Press, 1991, is available from the SB-MOP package.
3020 * incompatible change: the SB-PCL package should now be considered
3021 a private implementation detail, and no longer a semi-private MOP
3023 * minor incompatible change: due to rearrangement for threads, the
3024 control stack and binding stack are now allocated at arbitrary
3025 addresses instead of being hardcoded per-port. Users affected by
3026 this probably have to be doing advanced things with shared
3027 libraries, and will know who they are.
3028 * minor incompatible change: Previously, all --eval forms used were
3029 processed with READ before any of them were processed with EVAL.
3030 Now each --eval form is processed with both READ and EVAL before
3031 the next --eval form is processed. (Thus package operations like
3032 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3033 work as the user might reasonably expect.)
3034 * minor incompatible change: *STANDARD-INPUT* is now only an
3035 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3037 * minor incompatible change: Y-OR-N-P is now character-oriented,
3038 not line oriented. Also, YES-OR-NO-P now works without errors.
3039 (thanks to Antonio Martinez)
3040 * sb-aclrepl module improvements: an integrated inspector, added
3041 repl features, and a bug fix to :trace command.
3042 * Known functions, which cannot be open coded by the backend, are
3043 considered to be able to check types of their arguments. (fixing
3044 a bug report by Nathan J. Froyd)
3045 * fixed a bug in computing method discriminating functions: It is
3046 now possible to define methods specialized on classes which have
3047 forward-referenced superclasses. (thanks to Gerd Moellmann)
3048 * fixed evaluation order in optional entries (reported by Gilbert
3050 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3051 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3052 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3053 required-by-AMOP NAME argument, as well as CLASS and
3054 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3055 * fixed bug 20: DEFMETHOD can define methods using names that are
3056 not the proper names of classes to designate class specializers.
3057 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3058 Unix file descriptors, instead of blowing up. (thanks to Antonio
3060 * Garbage collection refactoring: user-visible change is that a
3061 call to the GC function during WITHOUT-GCING will not do garbage
3062 collection until the end of the WITHOUT-GCING. If you were doing
3063 this you were probably losing anyway.
3064 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3065 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3066 TYPEP the latter but not the former.
3067 * The compiler issues a full WARNING for calls to undefined functions
3068 with names from the CL package.
3069 * MAP-INTO for a vector destination is open coded. (reported by
3070 Brian Downing on c.l.l)
3071 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3072 documentation string.
3073 * fixed some bugs revealed by Paul Dietz' test suite:
3074 ** COPY-ALIST now signals an error if its argument is a dotted
3076 ** Condition slots are now accessed more correctly in the presence
3077 of multiple initargs for a given slot.
3078 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3079 exclude restarts of the same name associated with a different
3081 ** DEFCLASS of forward-referenced classes with another
3082 forward-referenced class in the superclasses list no longer
3084 ** Condition slots are now initialized once each, not multiple
3085 times. (thanks to Gerd Moellmann)
3086 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3088 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3089 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3090 arguments contain duplicated elements.
3091 ** RESTART-CASE understands local macros.
3092 ** RESTART-CASE associates exactly its own restarts with a condition.
3093 ** ENDP in safe mode checks its argument to be of type LIST.
3094 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3095 in question is unbound.
3096 ** Optimized MAKE-INSTANCE functions no longer cause internal
3097 assertion failures in the presence of duplicate initargs.
3098 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3099 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3100 propagated even in the ctor optimized implementation of
3102 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3104 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3105 * a better implementation of SXHASH on (simple) bit vectors,
3106 measured both in execution speed and in distribution of results
3107 over the positive fixnums, has been installed. Likewise, a better
3108 implementation of EQUAL for simple bit vectors is now available.
3109 * fixed CEILING optimization for a divisor of form 2^k.
3110 * fixed bug 240 (emitting extra style warnings "using the lexical
3111 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3112 by Antonio Martinez)
3113 * fixed SXHASH, giving different results for NIL depending on type
3114 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3115 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3116 variable. (found by Rolf Wester)
3117 * SBCL does not ignore type declarations for special
3118 variables. (reported by rif on c.l.l 2003-03-05)
3119 * some bug fixes in contrib/sb-aclrepl/
3120 * fixed some bugs revealed by Paul Dietz' test suite:
3121 ** a bug in the CONS type specifier, whereby the CAR and CDR
3122 types got intertwined, has been fixed;
3123 ** the type system is now able to reason about the interaction
3124 between INTEGER and RATIO types more completely;
3125 ** APPEND, [N]REVERSE and NRECONC check that those their
3126 arguments, which must be proper lists, are really so;
3127 ** An array specialized to be unable to hold elements has been
3128 implemented, as required -- yes, really -- by ANSI;
3129 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3130 on malformed property lists;
3132 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3133 * incompatible packaging change: in line with Unix convention,
3134 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3135 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3136 * REQUIRE and PROVIDE are now optionally capable of doing something
3137 useful. See the documentation string for REQUIRE.
3138 * infrastructure for a managed SBCL contrib system: contributed
3139 modules in this release include:
3140 ** the ASDF system definition facility;
3141 ** an interface to the BSD Sockets API;
3142 ** an ACL-like convenience interface to the repl;
3143 (thanks to Kevin Rosenberg)
3144 ** an implementation of ROTATE-BYTE, with efficient implementation
3146 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3147 gives a closer approximation to the right answer than previously.
3148 (thanks to Raymond Toy)
3149 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3150 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3151 optional environment arguments, as required by ANSI.
3152 * fixed bugs in other functions taking environment objects, allowing
3153 calls with an explicit NIL environment argument to be compiled
3155 * fixed bug 228: primary return values from
3156 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3157 COMPILE or FUNCTION.
3158 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3159 :TYPE will no longer signal errors on innocuous objects.
3160 * fixed bug 231b: SETQ is better at respecting type declarations in
3161 the lexical environment.
3162 * fixed a bug in DEFCLASS: classes named by symbols with no or
3163 unprintable packages can now be defined.
3164 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3165 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3166 * fixed compiler failure related to checking types of functions.
3167 (reported by Robert E. Brown)
3168 * the compiler is now much more consistent in its error-checking
3169 treatment of bounding index arguments to sequence functions: in
3170 (SAFETY 3) code, errors will be signalled in almost all cases if
3171 invalid sequence bounding indices are passed to functions defined
3172 by ANSI to operate on sequences.
3173 * fixed a bug in the build procedure: documentation of SBCL-specific
3174 packages is now preserved and available in the final Lisp image.
3175 * lifted FDEFINITION lookup out of loops in the implementation of
3176 many list operations. (thanks to Robert E. Brown)
3177 * fixed a bug in the reader: the #n# reader macro now works for
3178 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3179 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3180 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3181 E. Brown and Tony Martinez respectively)
3182 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3183 SLOT-BOUNDP: in particular, it is now able to optimize them much
3184 better, and is now not vulnerable to having packages renamed.
3185 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3186 always called when a slot is not present in an instance. (thanks
3188 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3189 longer signal an error on unreading a character following EOF on
3190 the previous constituent stream. (thanks to Tony Martinez)
3191 * fixed some bugs revealed by Paul Dietz' test suite:
3192 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3193 not just nonnegative fixnums;
3194 ** the logical bit-array operators such as BIT-AND now accept an
3195 explicit NIL for their "opt-arg" argument (to indicate a
3196 freshly-consed result bit-array);
3197 ** ELT now signals an error on an invalid sequence index in safe
3199 ** the type system is now cleverer about negations of numeric
3200 types, and consequently understands the BIGNUM and RATIO types
3202 ** the type system is now cleverer about the interaction between
3203 INTEGER and RATIO types: while bugs still remain, many more
3204 cases are accurately computed;
3205 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3206 if it is in the last clause;
3207 ** CONSTANTLY now correctly returns a side-effect-free function in
3209 ** DECLARE is no longer treated as a special-operator; in
3210 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3211 * incremented fasl file version number due to the change in the
3212 DEFSTRUCT-SLOT-DESCRIPTION structure.
3214 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3215 * minor incompatible change: code processed by the "interpreter" or
3216 EVAL now has a compilation optimization policy of (DEBUG 2)
3217 (changed from (DEBUG 1)) to improve debuggability of interactive
3218 development, and to allow the use of the debug RETURN command in
3220 * an experimental implementation of the RETURN command for the
3221 debugger has been included. (thanks to Frederik Kuivinen)
3222 * fixed bug 62: constraints were not propagated into a loop.
3223 * fixed bug in embedded calls of SORT (reported and investigated by
3225 * fixed some bugs revealed by Paul Dietz' test suite:
3226 ** printing and reading of arrays with some dimensions having
3227 length 0 (thanks to Gerd Moellmann);
3228 ** BOA constructor with &AUX argument without a default value does
3229 not cause a type error;
3230 ** CONSTANTP now returns true for all self-evaluating objects.
3232 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3233 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3234 accessors that are related by inheritance, as specified in the
3235 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3236 Valtteri Vuorikoski)
3237 * The compiler is now able to inline functions that were defined in
3238 a complex lexical environment (e.g. inside a MACROLET).
3239 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3241 * Support for the upcoming FreeBSD-5.0 release has been included.
3242 (thanks to Dag-Erling Smorgrav)
3243 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3244 effect when it is not in a toplevel context.
3245 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3246 stand a better chance of being correct. (thanks to Gerd
3248 * fixed bug in COERCE, which now signals an error on coercing a
3249 rational to a bounded real type which excludes the expected
3251 * The compiler is now able to derive types more accurately from the
3252 COERCE and COMPILE functions.
3253 * fixed bug 223: functional binding is considered to be constant
3254 only for symbols in the CL package.
3255 * fixed bug 231: SETQ did not check the type of a variable being set
3256 (reported by Robert E. Brown)
3257 * A new optimization for MAKE-INSTANCE has been included, fixing
3258 various bugs (including relating to :ALLOCATION :CLASS slots and
3259 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3261 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3262 ** As required by ANSI, LOOP now disallows anonymous collection
3263 clauses such as COLLECT I in conjunction with aggregate boolean
3264 clauses such as THEREIS (= I 1);
3265 ** LOOP now signals an error when any variable is reused in the
3266 same loop (including the potentially useful construct analogous
3267 to WITH A = 1 WITH A = (1+ A);
3268 ** IT is only a special loop symbol within the first clause of a
3269 conditional loop clause;
3270 ** LOOP with a typed iteration variable over a hashtable now
3271 signals a type error iff it should.
3272 * fixed some other bugs revealed by Paul Dietz' test suite:
3273 ** FILE-STREAM now names the class previously known as FD-STREAM;
3274 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3275 argument) no longer signals an error;
3276 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3277 of the slot symbol, rather than using the current package
3278 ((:CONC-NAME "") continues to intern the slot's name in the
3280 * incremented fasl file version number, because of the incompatible
3281 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3282 of the new implementation of DEFINE-COMPILER-MACRO.
3284 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3285 * Support for building SBCL for MIPS platforms running in
3286 little-endian mode has now been checked in, and basic
3287 functionality on said platforms verified.
3288 * minor incompatible change: PCL now records the pathname of a file
3289 in which methods and the like are defined, rather than its
3291 * minor incompatible change: TRUENAME now considers the truename of
3292 a file naming a directory to be the pathname with :DIRECTORY
3293 component indicating that directory.
3294 * minor incompatible change: a NAMED clause in the extended form of
3295 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3296 reason for the previous behaviour is unclear.
3297 * more systematization and improvement of CLOS and MOP conformance
3298 in PCL (thanks to Gerd Moellman and Pierre Mai):
3299 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3301 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3302 primary methods with no specializers;
3303 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3305 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3306 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3307 FORWARD-REFERENCED-CLASSes; error reporting on
3308 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3310 ** SXHASH on CLOS instances now uses a slot internal to the
3311 instance to return different numbers on distinct instances,
3312 while preserving the same return value through invocations of
3314 ** DEFMETHOD signals errors when methods with longer incongruent
3315 lambda lists are added to generic functions;
3316 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3317 CLASS, as specified in AMOP;
3318 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3319 on the order returned by the primary method for classes of
3320 class STANDARD-CLASS;
3321 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3322 * fixed some bugs shown by Paul Dietz' test suite:
3323 ** DOLIST puts its body in TAGBODY;
3324 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3326 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3327 value producing form;
3328 ** if more variables are given to PROGV than values, extra
3329 variables are bound and made to have no value;
3330 ** NSUBSTITUTE on list arguments gets the right answer with
3332 ** ELT signals an error of type TYPE-ERROR when the index argument
3333 is not a valid sequence index;
3334 ** LOOP signals (at macroexpansion time) an error of type
3335 PROGRAM-ERROR when duplicate variable names are found;
3336 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3337 ** FUNCALL of special-operators now cause an error of type
3339 ** PSETQ now works as required in the presence of side-effecting
3340 symbol-macro places;
3341 ** NCONC accepts any object as its last argument;
3342 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3344 ** loop-for-as-package does not require a package to be explicitely
3346 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3348 * fixed bug 166: compiler preserves "there is a way to go"
3349 invariant when deleting code.
3350 * fixed bug 172: macro lambda lists with required arguments after
3351 &REST arguments now cause an error to be signalled. (thanks to
3353 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3354 bug. (thanks to Matthew Danish)
3355 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3357 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3358 arguments are lexically rebound. (thanks to Gerd Moellmann and
3360 * fixed bug 194: error messages are now more informative when there
3361 is no primary method applicable in a call to a generic
3362 function. (thanks to Gerd Moellmann)
3363 * fixed bug in command line argument checking (thanks to Julian
3365 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3366 (thanks to Matthew Danish)
3367 * incremented fasl file version number, because of the
3368 SXHASH-related changes in the layout of CLOS data structures
3370 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3371 * minor incompatible change: The runtime (the Unix executable named
3372 "sbcl") is now much pickier about the .core files it will load.
3373 Essentially it now requires .core files to descend from the same
3374 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3375 as the runtime does. (The intent is to prevent the crashes which
3376 can occur, and which can even be reported as mysterious failures,
3377 when people patch the sources or change the build parameters
3378 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3379 sbcl and .core files.)
3380 * fixed bug: VALUES-LIST is no longer optimized away.
3381 * fixed bug 142: The FFI conversion of C string values to Lisp
3382 string values no longer conses excessively. (thanks to Nathan
3383 Froyd porting Raymond Toy's fix to CMU CL)
3384 * began to systematize and improve MOP conformance in PCL (thanks to
3385 Nathan Froyd, Gerd Moellman and Pierre Mai):
3386 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3388 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3389 ** FINALIZE-INHERITANCE is now called on class finalization;
3390 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3391 argument precedence order.
3392 * fixed bug 202: The compiler no longer fails on functions whose
3393 derived types contradict their declared type.
3394 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3395 so it can be non-toplevel.
3396 * The fasl file version number has changed (because of the new
3397 implementation of DEFMACRO).
3398 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3399 safe code, that any length requirement by their type-specifier
3400 argument is valid. The exceptions to this are described in bug
3402 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3403 functions now check the types of their inputs as required by ANSI.
3404 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3405 introduced symbol is DECLAREd to be SPECIAL.
3406 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3407 * fixed bug: :COUNT argument to sequence functions may be negative
3408 * fixed bug: body of DO-SYMBOLS may contain declarations
3409 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3410 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3411 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3412 symbol macro only once
3413 * fixed printing of call frame when argument list is unavailable
3414 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3415 * compiler no longer signals WARNING on unknown keyword
3418 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3419 * A beta-quality port to the mips architecture running Linux,
3420 based on the old CMUCL backend, has been made. It has been tested
3421 on a big-endian kernel, and works sufficiently well to be able to
3422 rebuild itself; it has not been tested in little-endian mode.
3423 * fixed an inconsistency between gencgc.c and purify.c which made
3424 dumping/loading .core files unreliable
3425 * fixed bug 120a: The compiler now deals correctly with IFs where
3426 the consequent is the same as the alternative, instead of
3427 misderiving the return type. (thanks to Alexey Dejneka)
3428 * fixed bug 113: Logical pathnames are now dumpable (the logical
3429 host is resolved at load-time, throwing an error if it is not
3431 * fixed bug 174: FORMAT's error message is slightly clearer when a
3432 non-printing character is used in a format directive.
3433 * fixed several bugs in compiler checking of type declarations, i.e.
3434 violations of the Python "declarations are assertions" principle
3435 (thanks to Alexey Dejneka)
3436 * fixed several bugs in PCL's error checking (thanks to Gerd
3438 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3440 * fixed bug in compilation of functions as first class values
3441 (thanks to Antonio Martinez-Shotton)
3442 * The compiler's handling TYPE-ERRORs which it can prove will
3443 inevitably happen at runtime has been cleaned up and corrected
3444 in several ways. (thanks to Alexey Dejneka)
3445 * improved argument type checking for various basic arithmetic
3446 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3447 lavished on them in the past that they can be compiled in many
3448 ways in different special cases
3449 * fixed bug 181: compiler checks validity of user supplied type
3451 * cleaned up code flushing in optimization: Function calls which
3452 should signal errors for safety purposes (e.g. which ANSI says
3453 should signal errors when their arguments are of incorrect type)
3454 are no longer optimized away.
3455 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3456 * incremented fasl file version number, because changes in the
3457 implementation of sequence functions like COERCE caused
3458 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3459 in old inline expansions) to become undefined. (Actually these
3460 changes were later undone, so we might very well be binary
3461 compatible with 0.7.7 after all, but leaving the version number
3462 incremented seemed like the simplest and most conservative
3465 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3466 * An alpha-quality port to the parisc architecture running Linux,
3467 based on the old CMUCL backend, has been made. This, even more so
3468 than the other backends, should be considered still a work in
3469 progress; known problems include that the Linux kernel in 64-bit
3470 mode does not propagate the correct sigcontext structure to
3471 userspace, and consequently SBCL on a parisc64 kernel will not
3473 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3474 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3475 suppressed.) Also now that the compiler is looking at declarations
3476 in the environment, it checks optimization declarations as well,
3477 and suppresses inlining when (> DEBUG SPEED).
3478 * More fixes have been made to treatment of floating point exception
3479 treatment and other Unix signals. In particular, floating point
3480 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3481 * The detection and handling of control stack exhaustion (infinite
3482 or very deeply nested recursion) has changed. Stack exhaustion
3483 detection is now done by write-protecting pages at the OS level
3484 and applies at all optimization settings; when found, a
3485 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3486 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3487 be caught using IGNORE-ERRORS.
3488 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3489 correctly when transforming recursive LABELS functions to LETs.
3490 (thanks to Alexey Dejneka)
3491 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3492 that are names of constants or global variables.
3493 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3494 alien routines with docstrings.
3495 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3496 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3498 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3499 (thanks to Eric Marsden)
3500 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3501 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3502 * Bug fix: The SPARC backend can now compile functions involving
3503 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3504 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3505 to a functions expecting a PATHNAME-DESIGNATOR.
3506 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3507 lambda lists. (thanks to Alexey Dejneka)
3508 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3509 functions so that EXPORTing the name of the class doesn't cause
3510 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3511 to Antonio Martinez for reporting this)
3512 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3513 constant-folded without failing an assertion. (thanks to Einar
3514 Floystad Dorum for reporting this)
3515 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3516 types are now dealt with more correctly by the compiler.
3517 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3518 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3519 behaviour analogous to RENAME-FILE. This puts its behaviour more
3520 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3522 * The fasl file version number has changed again. (because of the
3523 bug fix involving the names of PCL MAKE-INSTANCE functions)
3525 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3526 * bug fix: Floating point exceptions are treated much more
3527 consistently on the x86/Linux and PPC/Linux platforms.
3528 * Array initialization with :INITIAL-ELEMENT is now much faster for
3529 cases when the compiler cannot open code the array creation, but
3530 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3531 array accesses have also seen a speed increase.
3532 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3533 specified. (This was at the root of some bad interactions between
3534 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3536 * bug fix: Internal error arguments for undefined functions are now
3537 computed correctly on the PPC/Linux platform.
3538 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3539 Raymond Toy's patch for CMU CL)
3540 * Support for the Solaris 9 operating environment has been included
3541 (thanks to Daniel Merritt)
3542 * A very ugly but hopefully complete draft of the missing FFI chapter
3543 of the manual has been created by reformatting the corresponding
3544 CMU CL manual chapter into (currently very ugly and incoherent)
3545 DocBook and bringing it up to date for SBCL behavior. Thus, the
3546 manual is now essentially complete, at least by my extreme
3547 once-and-only-once standards, whereby it's acceptable to refer to
3548 the doc strings of SB-EXT functions as the primary documentation.
3549 * The fasl file version number has changed again, due to cleanup of
3550 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3551 is no longer a static symbol.)
3553 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3554 * SBCL now builds with OpenMCL (version 0.12) as the
3555 cross-compilation host; also, more progress has been made toward
3556 bootstrapping under CLISP.
3557 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3559 * bug 158 fixed: The compiler can now deal with integer loop
3560 increments different from 1; fixing this turned out also to fix
3562 * bug 169 fixed: no more bogus warnings about using lexical bindings
3563 despite the presence of perfectly good SPECIAL declarations (thanks
3564 to David Lichteblau)
3565 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3566 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3567 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3569 * bug 180 fixed: Method combination specifications no longer ignore
3570 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3571 * bug fix: Structure type predicate functions now check their argument
3572 count as they should.
3573 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3574 correctly. (thanks to Pierre Mai)
3575 * minor incompatible change: The --noprogrammer option is deprecated
3576 in favor of the new --disable-debugger option, which is very similar.
3577 (The major difference is that it takes effect at a slightly different
3578 time at startup, causing handling of errors in --sysinit and
3579 --userinit files will be affected differently.) The
3580 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3581 been added to allow this functionality to be controlled from ordinary
3582 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3583 maintainers, who might want to run non-interactive scripts to
3584 build SBCL cores which will later be used interactively.)
3585 * minor incompatible change: The LOAD function no longer, when given
3586 a wild pathname to load, loads all files matching that pathname.
3587 Instead, an error of type FILE-ERROR is signalled.
3589 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3590 * bug 147 fixed: The compiler preserves its block link/count
3591 invariants more correctly now so that it doesn't crash. (thanks
3593 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3595 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3596 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3597 no-value-for-this-character as NIL (as specified by ANSI).
3598 * HOST-NAMESTRING on physical pathnames now returns a string that is
3599 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3600 (thanks to Christophe Rhodes)
3601 * The Alpha port handles icache flushing more correctly. (thanks to
3603 * More progress has been made toward bootstrapping under CLISP. (thanks
3604 to Christophe Rhodes)
3605 * The fasl file format has changed again, because dynamic loading
3606 on OpenBSD (which has non-ELF object files) motivated some cleanups
3607 in the way that foreign symbols are transformed and passed around.
3608 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3609 is no longer treated as whitespace by the reader, but instead as
3610 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3611 |A
\x7fB|, instead of A as it used to.
3613 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3614 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3615 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3616 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3617 this as of 0.7.1.45, but was left out of the previous news section
3618 (thanks to Dan Barlow)
3619 * SBCL now runs on the Solaris operating system on SPARC architectures
3620 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3621 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3622 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3623 * SPARC backend cleanups, allowing builds of cores optimized for V8
3624 and V9 SPARCS, and also emission of code targeted to a particular
3625 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3627 * SBCL is closer to bootstrapping under CLISP, thanks to various
3628 fixes by Christophe Rhodes.
3629 * The fasl file format has changed again, to allow the compiler's
3630 INFO database to support symbol macros.
3631 * The user manual (in doc/) is formatted into HTML more nicely.
3632 (thanks to coreythomas)
3633 * The system is smarter about SUBTYPEP relationships, especially
3634 those involving NOT types (including types such as ATOM which are
3635 represented internally using NOT types). Thus SUBTYPEP is less
3636 likely to return (VALUES NIL NIL) in general, and in particular
3637 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3639 * The fasl file format has changed again, because the internal
3640 representation of types now includes a new slot to support the new
3641 SUBTYPEP-of-NOT-types logic.
3642 * (not a change in the main branch of SBCL, but a related prototype
3643 which can hopefully be merged into the main branch of SBCL in the
3644 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3645 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3646 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3647 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3648 dispatch macro character combinations. (thanks to Alexey Dejneka)
3649 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3650 that superficially look like logical namestrings correctly.
3651 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3653 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3654 default supplied-p) for &optional and &key arguments. (thanks to
3657 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3658 * incompatible change: The compiler is now less aggressive about
3659 tail call optimization, doing it only when (> SPACE DEBUG) or
3660 (> SPEED DEBUG). (This is an incompatible change because there are
3661 programs which relied on the old CMU-CL-style behavior to optimize
3662 away their unbounded recursion which will now die of stack overflow.)
3663 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3664 for non-GENCGC systems has been increased to 20M (since that
3665 seems much closer to the likely performance optimum for modern
3666 systems than the old 4M value was)
3667 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3668 and *DEBUG-PRINT-LEVEL*
3669 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3670 of CMU CL's support for SPARC, and various endianness and other
3671 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3672 * new syntactic sugar for the Unix command line: --load foo.bar is now
3673 an alternate notation for --eval '(load "foo.bar")'.
3675 ** The system now detects stack overflow and handles it gracefully,
3676 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3677 optimization settings. (This is a good thing in general, and
3678 its introduction in this version should be particularly timely
3679 for anyone whose code fails because of suppression of tail
3681 ** The system now hunts for the C variable "environ" in a more
3682 devious way, to avoid segfaults when the C library version differs
3683 between compile time and run time. (thanks to Christophe Rhodes)
3684 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3685 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3686 ** The compiler no longer issues bogus style warnings for undefined
3687 classes in the same source file as the DEFCLASSes which defined
3688 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3690 ** fixes in CONDITION class precedence list for undefined function
3691 errors (thanks to Alexei Dejneka)
3692 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3693 correctly. (thanks to Dan Barlow)
3694 ** portability fixes aiming at bootstrapping under CLISP (thanks
3695 to Dave McDonald and Christophe Rhodes)
3696 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3697 ** fixes in type translation and and type inference (thanks to
3699 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3700 ** various fixes in the new ports (thanks to Dan Barlow)
3701 * several changes related to debugging:
3702 ** suppression of tail recursion, as noted above
3703 ** stack overflow detection, as noted above
3704 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3705 is now the default. (For some time encapsulation has been more
3706 reliable than the breakpoint-based :ENCAPSULATE NIL
3707 implementation, at least on X86 systems; and I just noticed that
3708 encapsulation also seems closer to the spirit of the ANSI
3711 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3713 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3714 up properly again. (There was a packaging bug in 0.7.0 which
3715 left their definitions in SB-SYS::LOAD-FOREIGN and
3716 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3717 vital for most things which interface to C-level interfaces,
3718 like extensions working with sockets or databases or
3719 Perl-compatible regexes or whatever, and the need to fix
3720 this bug is the main reason that 0.7.1 was released so
3722 ** DEFGENERIC is now choosier about the methods it redefines, so that
3723 reLOADing a previously-LOADed file containing DEFGENERICs does
3724 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3725 cycle now works as it should. (thanks to Alexey Dejneka)
3726 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3727 fixed by Christophe Rhodes. (He actually submitted this patch
3728 months ago, and I delayed until after 0.7.0.)
3729 ** Bug 111 (internal compiler confusion about runtime checks on
3730 FUNCTION types) was fixed by Alexey Dejneka.
3731 * Some internal cleanups (getting rid of variables which aren't
3732 needed now that the byte interpreter is gone) caused the fasl
3733 file format number to change again.
3735 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3736 * major incompatible change: The default fasl file extension, i.e. the
3737 default extension for files produced by COMPILE-FILE, has changed
3738 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3740 ** There are many changes in the implementation of the compiler.
3741 SBCL is now essentially a compiler-only implementation of ANSI
3742 Common Lisp. EVAL still "interprets" a few special cases, but
3743 almost all the interesting cases are handled by creating
3744 a LAMBDA expression, calling COMPILE on it, then calling
3745 FUNCALL on the result.
3746 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3747 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3748 Since the code is newer, there might still be some new bugs
3749 (though not as many as before Martin Atzmueller's fixes:-). But
3750 the new code is substantially simpler and clearer, and hopefully
3751 any remaining bugs will be simpler, less fundamental, and more
3752 fixable then the bugs in the old code.
3753 ** The revised compiler is still a little unsteady on its feet.
3755 *** The debugging information it produces (particularly the names
3756 of FUNCTION objects) is sometimes much less useful than what
3757 the old compiler produced.
3758 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3759 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3760 within a MACROLET) has been temporarily weakened.
3761 ** There are new compiler optimizations for various functions:
3762 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3763 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3764 *** the math functions TRUNCATE, FLOOR, and CEILING
3765 *** the function-of-all-trades COERCE
3766 Mostly these should be transparent, but there's one
3767 potentially-annoying problem (bug 117): when the compiler
3768 inline-expands a function and does type analysis on the result,
3769 it can create control paths which have type mismatches, and
3770 when it can't prove that those control paths aren't taken,
3771 it will issue WARNINGs about the type mismatches. This is
3772 a particular problem in practice for the new sequence functions.
3773 It's not clear how this should be fixed, and for now, a
3774 workaround is given in the entry for 117 in the BUGS file.
3775 ** (Because of the interaction between the two previous items --
3776 occasional inlining problems and new inline expansions -- some
3777 of the new sequence function optimizations won't really kick in
3778 completely until debugging information, and then inlining, are
3779 straightened out in some future version.)
3780 * minor incompatible changes:
3781 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3782 DIRECTORY no longer implicitly promotes NIL slots of its
3783 pathname argument to :WILD. In particular, when you ask for the
3784 contents of a directory (which you used to be able to do without
3785 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3786 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3787 ** changes in behavior that ANSI explicitly defines to be
3788 implementation dependent:
3789 *** The new compiler-only implementation still conforms with ANSI,
3790 but acts a little different than before. Besides the obvious
3791 changes in performance tradeoffs (that the cost per form passed
3792 to EVAL has gone up, and the cost per form executed by EVAL
3793 has gone down), the behavior of the system changes a little
3794 because there are no longer any interpreted function objects.
3795 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3796 e.g. doing COMPILE on the output of interactive DEFUN is
3798 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3800 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3801 has changed from (:CL) to NIL.
3802 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3803 e.g. #\Newline and #\Tab, don't have names specified in the
3804 ANSI Common Lisp standard, is now based on their ASCII symbolic
3805 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3806 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3807 are no longer used for output.
3808 ** changes in internal implementation constants:
3809 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3810 4 million. (If your application spends a lot of time GCing and
3811 you have a lot of RAM, you might want to experiment with
3812 increasing it even more.)
3813 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3814 However, almost all old code should still continue to work without
3815 immediate update, as SB-C-CALL is now a (deprecated) nickname
3817 ** Old operator names in the style DEF-FOO are now deprecated in
3818 favor of new corresponding names DEFINE-FOO, for consistency with
3819 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3820 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3821 internal symbols, but a few supported extensions like
3822 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3823 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3824 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3825 etc. as you get deeper into recursive calls to the debugger
3826 command loop, instead of the old "5]", "5]]", "5]]]"
3827 sequence. (I was motivated to do this when squabbles between
3828 ILISP and SBCL left me very deeply nested in the debugger. In the
3829 short term, this change will probably provoke more ILISP/SBCL
3830 squabbles, but hopefully it will be an improvement in the long run.)
3831 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3832 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3833 function, since declaiming return types involving aliens is
3834 (1) annoyingly messy to do by hand and (2) vital to efficient
3835 compilation of code which calls such functions.
3836 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3837 longer reexported by the SB-EXT package. They're solely useful
3838 for alien code, so it seems more logical that you should get
3839 them from the SB-ALIEN package, not in SB-EXT.
3840 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3841 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3842 features. Instead, the code that they used to control is always
3843 built into the system.
3844 * many other bug fixes
3845 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3846 advantage of the new EVAL-WHEN stuff and to clean them up in
3847 general, and they are now more ANSI-compliant in a number of
3848 ways. Martin Atzmueller is responsible for a lot of this.
3849 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3851 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3852 *** correct ERROR type for various file operations
3853 *** some fixes for Lisp streams
3854 *** DEFMETHOD syntax checking
3855 *** changing old weird representation of debug information as
3856 strings (which, among their other deficiencies, don't transform
3857 correctly when you rename packages, and don't change their
3858 print representation when you change things like *PACKAGE*
3859 and *PRINT-LENGTH*) to symbols and lists of symbols
3860 He also made several improvements and fixed several bugs in DESCRIBE.
3861 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3863 *** misbehavior of WRITE-STRING/WRITE-LINE
3864 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3865 and several other LOOP problems as well
3866 *** DIRECTORY when similar filenames are present
3867 *** DEFGENERIC with :METHOD options
3868 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3869 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3870 *** argument ordering in FIND with :TEST option
3871 *** mishandled package designator argument in APROPOS-LIST
3872 *** various problems in the backquote readmacro
3873 *** a bug in APROPOS
3874 *** probably some others that I'm not describing very well here,
3875 since the CVS log documents them by reference to sbcl-devel
3876 messages, and the SourceForge archives aren't working well.:-(
3877 ** Dan Barlow improved the Alpha port (and is making progress on the
3878 PPC port, for those of you who think different).
3879 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3880 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3881 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3882 continues to work on the SPARC port (for those of you in a position
3883 to look down upon our little PC-compatible boxes from a great height).
3884 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3885 and uses that copy, rather than the current dynamic value, when
3886 it's trying to decide whether to truncate output. Thus e.g.
3887 (let ((*print-lines* 50))
3888 (pprint-logical-block (stream nil)
3890 (let ((*print-lines* 8))
3891 (print (aref possiblybigthings i) stream)))))
3892 should now truncate the logical block only at 50 lines, instead of
3893 often truncating it at 8 lines, as it did before.
3894 * The doc/cmucl/ directory, containing old CMU CL documentation
3895 from the time of the fork, is no longer part of the base system.
3896 SourceForge has shut down its anonymous FTP service, and with it
3897 my original plan for distributing the old CMU CL documentation
3898 there. For now, if you need these files you can download an old
3899 SBCL source release and extract them from it.
3900 * The fasl file version number changed again, for dozens of reasons,
3901 some of which are apparent above.
3903 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3904 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3905 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3906 patch, so that DISASSEMBLE output is much nicer.
3907 * The code in the SB-PROFILE package now seems reasonably stable.
3908 I still haven't decided what the final interface should look like
3909 (I'd like PROFILE to interact cleanly with TRACE, since both
3910 facilities use function encapsulation) but if you have a need
3911 for profiling now, you can probably use it successfully with
3912 the current CMU-CL-style interface.
3913 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3914 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3915 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3916 link now returns the dangling link itself, and for similar
3917 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3918 link itself. (In these cases the old code signalled an error and
3919 looped endlessly, respectively.) Thus, DIRECTORY now works even
3920 in the presence of dangling and cyclic symbolic links.
3921 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3922 is now a supported extension again, since the consensus on
3923 sbcl-devel was that it can be useful for ordinary development
3924 work, not just for debugging SBCL itself.
3925 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3926 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3927 that currently-defined functions might be redefined later with
3928 different return types.
3929 * Hash tables can be printed readably, as inspired by CMU CL code
3930 of Eric Marsden and SBCL code of Martin Atzmueller.
3931 * better error handling in CLOS method combination, thanks to
3932 Martin Atzmueller porting Pierre Mai's CMU CL patches
3933 * more overflow fixes for >16Mbyte I/O buffers
3934 * A bug in READ has been fixed, so that now a single Ctrl-D
3935 character suffices to cause end-of-file on character streams.
3936 In particular, now you only need one Ctrl-D at the command
3937 line (not two) to exit SBCL.
3938 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3940 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3941 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3942 does the right thing.
3943 * The implementation of some type tests, especially for CONDITION
3944 types, is now tidier and maybe faster, due to CMU CL code
3945 originally by Douglas Crosher, ported by Martin Atzmueller.
3946 * Some math functions have been fixed, and there are new
3947 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3948 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3949 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3950 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3951 likely to be built into sbcl-0.7.0.)
3952 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3953 CL, ported to SBCL by Martin Atzmueller)
3954 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3955 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3956 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3957 * There's a new slam.sh hack to shorten the edit/compile/debug
3958 cycle for low-level changes to SBCL itself, and a new
3959 :SB-AFTER-XC-CORE target feature to control the generation of
3960 the after-xc.core file needed by slam.sh.
3961 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3962 COMPILE-FILE is no longer supported, so that now every function
3963 gets an entry point, so that block compilation looks a little
3964 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3965 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3966 returns the number of bytes consed since the system started,
3967 rather than the number consed since the first time the function
3968 was called. (The new definition parallels ANSI functions like
3969 CL:GET-INTERNAL-RUN-TIME.)
3970 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3971 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3972 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3973 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3974 * Fasl file version numbers are now independent of the target CPU,
3975 since historically most system changes which required version
3976 number changes have affected all CPUs equally. Similarly,
3977 the byte fasl file version is now equal to the ordinary
3980 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3981 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3982 is no longer recognized. I apologize for this, because it was
3983 listed in SB-EXT as a supported extension, but I found that
3984 its existing behavior was poorly specified, as well as incorrectly
3985 specified, and it looked like too much of a mess to straighten it
3986 out. I have enough on my hands trying to get ANSI stuff to work..
3987 * many patches ported from CMU CL by Martin Atzmueller, with
3988 half a dozen bug fixes in pretty-printing and the debugger, and
3989 half a dozen others elsewhere
3990 * fixed bug 13: Floating point infinities are now supported again.
3991 They might still be a little bit flaky, but thanks to bug reports
3992 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3993 as flaky as they were.
3994 * The --noprogrammer command line option is now supported. (Its
3995 behavior is slightly different in detail from what the old man
3996 page claimed it would do, but it's still appropriate under the
3997 same circumstances that the man page talks about.)
3998 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3999 are now supported, and enabled by default. Thus, the compiler can
4000 handle many floating point and complex operations much less
4001 inefficiently. (Thus e.g. you can implement a complex FFT
4003 * The compiler now detects type mismatches between DECLAIM FTYPE
4004 and DEFUN better, and implements CHECK-TYPE more correctly, and
4005 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4006 thanks to patches from Martin Atzmueller.
4007 * various fixes to make the cross-compiler more portable to
4008 ANSI-conforming-but-different cross-compilation hosts (notably
4009 Lispworks for Windows, following bug reports from Arthur Lemmens)
4010 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4011 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4012 fixed thanks to Pierre Mai's CMU CL patches.
4013 * fixes to keep the system from overflowing internal counters when
4014 it tries to use i/o buffers larger than 16M bytes
4015 * fixed bug 45a: Various internal functions required to support
4016 complex special functions have been merged from CMU CL sources.
4017 (When I was first setting up SBCL, I misunderstood a compile-time
4018 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4019 * improved support for type intersection and union, fixing bug 12
4020 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4021 more obscure bugs as well
4022 * some steps toward byte-compiling non-performance-critical
4023 parts of the system, courtesy of patches from Martin Atzmueller
4024 * Christophe Rhodes has made some debian packages of sbcl at
4025 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4026 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4027 stable, but are nonetheless usable. When he's ready, I'd be happy
4028 to add them to the SourceForge "File Releases" section. (And if
4029 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4030 * new fasl file format version number (because of changes in
4031 internal representation of (OR ..) types to accommodate the new
4032 support for (AND ..) types, among other things)
4034 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4035 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4037 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4039 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4040 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4041 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4042 * DESCRIBE now works on structure objects again.
4043 * Most function call argument type mismatches are now handled as
4044 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4045 know whether the function will be redefined before the call is
4046 executed. (The compiler could flag local calls with full WARNINGs,
4047 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4048 it doesn't keep track of enough information to know whether calls
4049 are local in this sense.)
4050 * Compiler output is now more verbose, with messages truncated
4051 later than before. (There should be some supported way for users
4052 to override the default verbosity, but I haven't decided how to
4053 provide it yet, so this behavior is still controlled by the internal
4054 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4055 src/compiler/ir1util.lisp.)
4056 * Fasl file format version numbers have increased again, because
4057 support for the Gray streams extension changes the layout of the
4058 system's STREAM objects.
4059 * The Gray subclassable streams extension now works, thanks to a
4060 patch from Martin Atzmueller.
4061 * The full LOAD-FOREIGN extension (not just the primitive
4062 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4063 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4064 but like most other programs, it defaults to copying the Unix
4065 environment from the original process instead of starting the
4066 new process in an empty environment.
4067 * Extensions which manipulate the Unix environment now support
4068 an :ENVIRONMENT keyword option which doesn't smash case or
4069 do other bad things. The CMU-CL-style :ENV option is retained
4070 for porting convenience.
4071 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4072 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4074 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4076 * A patch from Martin Atzmueller seems to have solved the SIGINT
4077 problem, and as far as we know, signal-handling now works cleanly.
4078 (If you find any new bugs, please report them!)
4079 * The system no longer defaults Lisp source file names to types
4080 ".l", ".cl", or ".lsp", but only to ".lisp".
4081 * The compiler no longer uses special default file extensions for
4082 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4083 seems to expect a single default extension for all compiled code,
4084 and there's no compelling reason to try to stretch the standard
4085 to allow two different extensions.) Instead, byte-compiled files
4086 default to the same extension as native-compiled files.
4087 * Fasl file format version numbers have increased again, because
4088 a rearrangement of internal implementation packages made some
4089 dumped symbols in old fasl files unreadable in new cores.
4090 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4091 many fewer weird special cases.
4092 * Bug #17 (differing COMPILE-FILE behavior between logical and
4093 physical pathnames) has been fixed, and some related misbehavior too,
4094 thanks to a patch from Martin Atzmueller.
4095 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4096 by Tim Moore, ported to SBCL by Martin Atzmueller.
4097 * Martin Atzmueller fixed several filesystem-related problems,
4098 including bug #36, in part by porting CMU CL patches, which were
4099 written in part by Paul Werkowski.
4100 * More compiler warnings in src/runtime/ are gone, thanks to
4101 more patches from Martin Atzmueller.
4102 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4105 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4107 * DESCRIBE now works on CONDITION objects.
4108 * The debugger now handles errors which arise when trying to print
4109 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4111 * The build system now uses an additional file, customize-target-features.lisp,
4112 to allow local modifications to the target *FEATURES* list. (The point of
4113 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4114 debugging features enabled, without having to worry about propagating it
4115 into everyone's system when I do a "cvs update".) When no
4116 customize-target-features.lisp file exists, the target *FEATURES* list
4117 should be constructed the same way as before.
4118 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4119 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4120 warn about behavior which is undefined under the ANSI spec. Note: This
4121 is specified by ANSI, but it's not very popular with programmers.
4122 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4123 macro in the SBCL sources for an example of a workaround which you
4124 might use to make portable ANSI-standard code which does what you want.
4125 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4126 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4127 when it's used as a non-toplevel form. (This is required in order
4128 to implement the DEFCONSTANT-EQX macro.)
4129 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4130 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4131 funny things, due to bugs in the implementation of EVAL-WHEN
4132 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4133 non-toplevel DEFCONSTANTs are uncommon.)
4134 * The core file version number and fasl file version number have been
4135 incremented, because the old noncompliant DEFCONSTANT behavior involved
4136 calling functions which no longer exist, and because I also took the
4137 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4138 * fixed bug 1 (error handling before read-eval-print loop starts), and
4139 redid debugger restarts and related debugger commands somewhat while
4141 ** The QUIT debugger command is gone, since it did something
4142 rather different than the SB-EXT:QUIT command, and since it never
4143 worked properly outside the main toplevel read/eval/print loop.
4144 Invoking the new TOPLEVEL restart provides the same functionality.
4145 ** The GO debugger command is also gone, since you can just invoke
4146 the CONTINUE restart directly instead.
4147 ** The TOP debugger command is also gone, since it's redundant with the
4148 FRAME 0 command, and since it interfered with abbreviations for the
4150 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4152 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4153 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4154 some porting work by Martin Atzmueller)
4155 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4156 with a boatload of patches which he ported to SBCL. Now that those
4158 ** The system tries to make sure that its low-priority messages
4159 are prefixed by semicolons, to help people who like to use
4160 syntax highlighting in their ILISP buffer. (This patch
4161 was originally due to Raymond Toy.)
4162 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4163 patches originally written by Raymond Toy.
4164 ** The compiler understands coercion between single-value and
4165 multiple-VALUES type expressions better, getting rid of some very
4166 weird behavior, thanks to patches originally by Robert MacLachlan
4167 and Douglas Crosher.
4168 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4169 lambda lists, thanks to a patch originally by Pierre Mai.
4170 ** The system no longer bogusly warns about "abbreviated type
4172 ** The compiler gets less confused by inlining and RETURN-FROM,
4173 thanks to some patches originally by Tim Moore.
4174 ** The system no longer hangs when dumping circular lists to fasl
4175 files, thanks to a patch originally from Douglas Crosher.
4176 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4177 undefined function error.
4178 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4179 in structure slot initforms) for now, documented workaround instead:-|
4180 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4181 * fixed bug 5: added stubs for various Gray stream functions called
4182 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4183 installed, at least appropriate type errors are generated
4184 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4185 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4186 consistently in DEFMETHOD forms.
4187 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4188 it doesn't seem to affect SBCL after all
4189 * The C runtime system now builds with better optimization and many
4190 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4192 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4194 * The system is now under CVS at SourceForge (instead of the
4195 CVS repository on my home machine).
4196 * The new signal handling code has been tweaked to treat register
4197 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4198 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4199 downstream, e.g. in debug-int.lisp, has implicit dependencies
4200 on the unsignedness of integer representation of machine words,
4201 and that caused the system to bomb out with infinite regress
4202 when trying to recover from type errors involving signed values,
4203 e.g. (BUTLAST '(1 2 3) -1).)
4204 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4205 (This was one of the bugs Peter Van Eynde reported back in July.)
4206 * The system now uses code inspired by Colin Walters' O(N)
4207 implementation of MAP (from the cmucl-imp@cons.org mailing
4208 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4209 inline the MAP operation, and there is more than one
4210 sequence argument to the MAP call (so that it can't just
4211 do ETYPECASE once and for all based on the type of the
4212 single sequence argument). (The old non-inline implementation
4213 of the general M-argument sequence-of-length-N case required
4214 O(M*N*N) time when any of the sequence arguments were LISTs.)
4215 * The QUIT :UNIX-CODE keyword argument has been renamed to
4216 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4217 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4219 * Raymond Wiker's patches to port dynamic loading from Linux to
4220 FreeBSD have been added.
4221 * The BUGS file is now more nearly up to date, thanks in large part
4222 to Martin Atzmueller's review of it.
4223 * The debugger now flushes standard output streams before it begins
4224 its output ("debugger invoked" and so forth).
4225 * The core version number and fasl file version number have both
4226 been incremented, because of incompatible changes in the layout
4228 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4229 used to not be called for a saved Lisp image.)
4230 * Martin Atzmueller's version of a patch to fix a compiler crash,
4231 as posted on sbcl-devel 13 September 2000, has been installed.
4232 * Instead of installing Martin Atzmueller's patch for the
4233 compiler transform for SUBSEQ, I deleted the compiler transform,
4234 and transforms for some similar consing operations.
4235 * A bug in signal handling which kept TRACE from working on OpenBSD
4237 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4238 away by constant folding
4239 * The system now defines its address space constants in one place
4240 (in the Lisp sources), and propagates them automatically elsewhere
4241 (through GENESIS and the sbcl.h file). Therefore, patching the
4242 address map is less unnecessarily tedious and error-prone. The
4243 Lisp names of address space constants have also been systematized.
4244 * CVS tags like dollar-Header-dollar have been removed from
4245 the sources, because they have never saved me trouble and
4246 they've been source of trouble working with patches and other
4247 diff-related operations.
4248 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4249 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4251 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4253 * The system has been ported to OpenBSD.
4254 * The system now compiles with a simple "sh make.sh" on the systems
4255 that it's supported on. I.e., now you no longer need to tweak
4256 text in the target-features.lisp-expr and symlinks in src/runtime/
4257 by hand, the make.sh takes care of it for you.
4258 * The system is no longer so grossly inefficient when compiling code
4259 involving vectors implemented as general (not simple) vectors (VECTOR T),
4260 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4261 which dares to use the various sequence functions on non-simple
4262 vectors, takes less of a performance hit.
4263 * There is now a primitive type predicate VECTOR-T-P
4264 to test for the (VECTOR T) type, so that e.g.
4265 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4266 can now be compiled with some semblance of efficiency. (The old code
4267 turned the type declaration into a full call to %TYPEP at runtime!)
4268 * AREF on (VECTOR T) is still not fast, since it's still compiled
4269 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4270 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4271 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4272 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4273 without making full calls to SUBTYPEP at runtime.
4274 (Some analogous efficiency issues for non-simple vectors specialized to
4275 element types other than T, or for non-simple multidimensional arrays,
4276 have not been addressed. They could almost certainly be handled the
4277 same way if anyone is motivated to do so.)
4278 * The changes in array handling break binary compatibility, so
4279 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4280 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4281 instead of (VALUES T T).
4282 * By following the instructions that Dan Barlow posted to sbcl-devel
4283 on 2 July 2000, I was able to enable primitive dynamic object
4284 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4285 functionality is not implemented (since it calls ld to resolve
4286 library references automatically, requiring RUN-PROGRAM for its
4287 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4288 try to resolve library references) is now supported.
4289 * The system now flushes the standard output streams when it terminates,
4290 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4291 them at several other probably-convenient times, e.g. in each pass of
4292 the toplevel read-eval-print loop, and after evaluating a form given
4293 as an "--eval" command-line option. (These changes were motivated by a
4294 discussion of stream flushing issues on cmucl-imp in August 2000.)
4295 * The source transform for TYPEP of array types no longer assumes
4296 that an array whose element type is a not-yet-defined type
4297 is implemented as an array of T, but instead punts, so that the
4298 type will be interpreted at runtime.
4299 * There is now some support for cross-compiling in make.sh: each of
4300 the phases of make.sh has its own script. (This should be transparent
4301 to people doing ordinary, non-cross-compile builds.)
4302 * Since my laptop doesn't have hundreds of megabytes of memory like
4303 my desktop machine, I became more motivated to do some items on
4304 my to-do list in order to reduce the size of the system a little:
4305 ** Arrange for various needed-only-at-cold-init things to be
4306 uninterned after cold init. To support this, those things have
4307 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4308 symbols with such names are now uninterned after cold init).
4309 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4310 fasl files for cold load.
4311 ** Remove the old compiler structure pooling code (which used to
4312 be conditional on the target feature :SB-ALLOC) completely.
4313 ** Redo the representation of some data in cold init to be more compact.
4314 (I also looked into supporting byte compiled code at bootstrap time,
4315 which would probably reduce the size of the system a lot, but that
4316 looked too complicated, so I punted for now.)
4317 * The maximum signal nesting depth in the src/runtime/ support code has
4318 been reduced from 4096 to 256. (I don't know any reason for the very
4319 large old value. If the new smaller value turns out to break something,
4320 I'll probably just bump it back up.)
4321 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4323 * Many, many bugs reported by Peter Van Eynde have been added to
4324 the BUGS list; some have even been fixed.
4325 * While enabling dynamic object file loading, I tried to make the
4326 code easier to understand, renaming various functions and variables
4327 with less ambiguous names, and changing some function calling
4328 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4329 * While trying to figure out how to do the OpenBSD port, I tried to
4330 clean up some of the code in src/runtime/. In particular, I dropped
4331 support for non-POSIX signal handling, added various comments,
4332 tweaked the code to reduce the number of compilation warnings, and
4333 renamed some files to increase consistency.
4334 * To support the new automatic configuration functionality in make.sh,
4335 the source file target-features.lisp-expr has been replaced with the
4336 source file base-target-features.lisp-expr and the machine-generated
4337 file local-target-features.lisp-expr.
4338 * fixed a stupid quoting error in make.sh so that using CMU CL
4339 "lisp -batch" as cross-compilation host works again
4341 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4343 * DESCRIBE no longer tries to call itself recursively to describe
4344 bound/fbound values, so that it no longer fails on symbols which are
4345 bound to themselves (like keywords, T, and NIL).
4346 * DESCRIBE now works on generic functions.
4347 * The printer now prints less-screwed-up representations of closures
4348 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4349 * A private symbol is used instead of the :EMPTY keyword previously
4350 used to mark empty slots in hash tables. Thus
4351 (DEFVAR *HT* (MAKE-HASH-TABLE))
4352 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4353 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4354 now does what ANSI says that it should. (You can still get
4355 similar noncompliant behavior if bang on the hash table
4356 implementation with all the symbols you get back from
4357 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4358 This breaks binary compatibility, since tests for equality to
4359 :EMPTY are wired into things like the macroexpansion of
4360 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4362 * There's now a minimal placeholder implementation for CL:STEP,
4363 as required by ANSI.
4364 * An obscure bug in the interaction of the normal compiler, the byte
4365 compiler, inlining, and structure predicates has been patched
4366 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4367 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4369 * Missing ordinary arguments in a macro call are now detected even
4370 when the macro lambda list contains &KEY or &REST.
4371 * The debugger no longer complains about encountering the top of the
4372 stack when you type "FRAME 0" to explicitly instruct it to go to
4373 the top of the stack. And it now prints the frame you request even
4374 if it's the current frame (instead of saying "You are here.").
4375 * As specified by ANSI, the system now always prints keywords
4376 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4378 * The default initial SIZE of HASH-TABLEs is now smaller.
4379 * Type information from CLOS class dispatch is now propagated
4380 into DEFMETHOD bodies, so that e.g.
4381 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4383 is now basically equivalent to
4384 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4385 (DECLARE (TYPE SINGLE-FLOAT X))
4387 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4388 operation, without having to do run-time type dispatch.
4389 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4390 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4391 * Since I don't seem to be making big file reorganizations very often
4392 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4393 is overflowing my ability to conveniently back them up), I've finally
4394 checked the system into CVS. (The CVS repository is on my home system,
4395 not at SourceForge -- putting it on SourceForge might come later.)
4396 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4397 high-level GC-NOTIFY-FOO functions send their output. (There's
4398 still very little control of where low-level verbose GC functions
4399 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4400 less than it used to -- the GC-NOTIFY-FOO functions are now under
4401 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4402 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4403 in only one place in the source code, and propagates it automatically
4404 everywhere that it's needed. Thus e.g. when I bump the version from
4405 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4406 * The C source files now include boilerplate legalese and documentation
4407 at the head of each file (just as the Lisp source files already did).
4408 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4409 to his page will be replaced with a link to his new CLiki service.
4411 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4413 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4414 * The build process now looks for GNU make under the default name "gmake",
4415 instead of "make" as it used to. If GNU make is not available as "gmake"
4416 on your system, you can change this default behavior by setting the
4417 GNUMAKE environment variable.
4418 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4419 can build without error under CMU CL.
4421 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4423 * There is now a partial SBCL user manual (with some new text and some
4424 text cribbed from the CMU CL manual).
4425 * The beginnings of a profiler have been added (starting with the
4426 CMU CL profiler and simplifying and cleaning up). Eventually the
4427 main interface should be through the TRACE macro, but for now,
4428 it's still accessed through vaguely CMU-CL-style functions and macros
4429 exported from the package SB-PROFILE.
4430 * Some problems left over from porting CMU CL to the new
4431 cross-compilation bootstrap process have been cleaned up:
4432 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4433 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4434 part when getting rid of the compiler *BACKEND* stuff.)
4435 ** The value of *NULL-TYPE* was screwed up, because it was
4436 being initialized before the type system knew the final
4437 definition of the 'NULL type. This screwed up several key
4438 optimizations in the compiler, causing inefficiency in all sorts
4439 of places. (I found it because I wanted to understand why
4440 GET-INTERNAL-RUN-TIME was consing.)
4441 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4442 PROCLAIM FTYPE information. Unfortunately this broke binary
4443 compatibility again, since now the forms output by DEFGENERIC
4444 to refer to functions which didn't exist in 0.6.3.
4445 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4446 can use the new (as of 0.6.3) transform for SOME into MAP into
4448 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4449 values to (INTEGER 0 1000000), so that the time code will no longer
4450 occasionally get blown up by Linux returning 1000000 microseconds
4451 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4452 its output conform to the ANSI spec. (Alas, this makes its output
4453 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4454 * A full call to MAP NIL with a single sequence argument no longer conses.
4455 * fixes to problems pointed out by Martin Atzmueller:
4456 * The manual page no longer talks about multiprocessing as though
4457 it were currently supported.
4458 * The ILISP support patches have been removed from the distribution,
4459 because as of version 5.10.1, ILISP now supports SBCL without us
4460 having to maintain patches.
4461 * added a modified version of Raymond Toy's recent CMU CL patch for
4462 EQUALP comparison of HASH-TABLE
4464 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4466 * The system still can't cross-compile itself with
4467 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4468 entails), but at least it can compile more of itself that way
4469 than it used to be able to, and various buglets which were uncovered
4470 by trying to cross-compile itself that way have now been fixed.
4471 * This release breaks binary compatibility again. This time
4472 at least I've incremented the FASL file format version to 2, so that the
4473 problem can be detected reliably instead of just causing weird errors.
4474 * various new style warnings:
4475 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4476 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4477 ** using the lexical binding of a variable named in the *FOO* style
4478 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4479 as specified by ANSI.
4480 * *RANDOM-STATE* is no longer automatically initialized from
4481 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4482 default behavior of the system is to repeat its behavior every time
4483 it's run. If you'd like to change this behavior, you can always
4484 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4485 old convention there was no comparably easy way to get the system to
4486 repeat its behavior every time it was run.
4487 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4488 FTYPE declarations has been removed, in favor of their ANSI
4489 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4490 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4491 the types of their sequence arguments can be determined at compile time.
4492 This is done through a new open code expansion for MAP which eliminates
4493 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4494 argument types can be determined at compile time.
4495 * The optimizer now transforms COERCE into an identity operation when it
4496 can prove that the coerced object is already of the correct type. (This
4497 can be a win for machine generated code, including the output of other
4498 optimization transforms, such as the MAP transform above.)
4499 * Credit information has been moved from source file headers into CREDITS.
4500 * Source file headers have been made more standard.
4501 * The CASE macro now compiles without complaining even when it has
4504 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4506 * (Note that the way that the PCL macroexpansions were rewritten
4507 to accommodate the change in DEFGENERIC below breaks binary
4508 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4509 not run under sbcl-0.6.2. Once we get out of alpha releases,
4510 i.e. hit release 1.0.0, we'll probably try to maintain binary
4511 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4512 and sbcl-1.4.4. Until then, however, it might be fairly common
4513 for maintenance releases to break binary compatibility.)
4514 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4516 * The reporting of unhandled signals has been changed to print some
4517 explanatory text as well as the report form. (Previously only
4518 the report form was printed.)
4519 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4520 it defines, so that the compiler no longer issues undefined function
4521 warnings for compiled-but-not-yet-loaded generic functions.
4522 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4523 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4524 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4525 section of the ANSI spec are supported.
4526 * The "" nickname for the "KEYWORD" package has been removed.
4527 The reader still handles symbol tokens which begin with a package marker
4528 as keywords, but it doesn't expose its mechanism for doing so in the
4529 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4530 * The system now issues STYLE-WARNINGs for contradictory TYPE
4531 proclamations. (Warnings for contradictory FTYPE proclamations would
4532 be nice too, but those can't be done usefully unless the type system
4533 is made smarter about FUNCTION types.)
4534 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4535 systematized, so that "*target-*.lisp is supposed to exist only on the
4536 target and imply that there's a related file which exists on the
4537 host, and *host-*.lisp is supposed to exist only on the host and imply
4538 that there's a related file which exists on the target. This involves a
4539 lot of renaming. Hopefully the acute confusion caused by the renaming
4540 will be justified by the reduction in chronic confusion..
4541 ** runtime-type.lisp -> early-target-type.lisp
4542 ** target-type.lisp -> late-target-type.lisp
4543 ** early-host-format.lisp -> early-format.lisp
4544 ** late-host-format.lisp -> late-format.lisp
4545 ** host-error.lisp -> misc-error.lisp
4546 ** early-error.lisp -> early-target-error.lisp
4547 ** late-error.lisp -> late-target-error.lisp
4548 ** host-defboot.lisp -> early-defboot.lisp
4549 ** code/misc.lisp -> code/target-misc.lisp
4550 ** code/host-misc.lisp -> code/misc.lisp
4551 ** code/numbers.lisp -> code/target-numbers.lisp
4552 ** code/early-numbers.lisp -> numbers.lisp
4553 ** early-host-type.lisp -> early-type.lisp
4554 ** late-host-type.lisp -> late-type.lisp
4555 ** host-typep.lisp -> typep.lisp
4556 ** load.lisp -> target-load.lisp
4557 ** host-load.lisp -> load.lisp
4558 ** host-disassem.lisp -> disassem.lisp
4559 ** host-insts.lisp -> insts.lisp
4560 ** byte-comp.lisp -> target-byte-comp.lisp
4561 ** host-byte-comp.lisp -> byte-comp.lisp
4562 ** host-signal.lisp -> signal.lisp
4563 ** host-defstruct.lisp -> defstruct.lisp
4564 ** late-target-type.lisp -> deftypes-for-target.lisp
4565 Furthermore, several other previously target-only files foo.lisp (e.g.
4566 hash-table.lisp and random.lisp) have been split into a target-and-host
4567 foo.lisp file and a target-only target-foo.lisp file, with their key type
4568 definitions in the target-and-host part, so that the cross-compiler will
4569 know more about target types.
4570 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4571 gone away. In their place are various *BACKEND-FOO* variables
4572 corresponding to the slots of the old structure.
4573 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4574 into the target SBCL has been fixed.
4575 * The chill.lisp system for loading cold code into a running SBCL
4577 * Support for the CMU CL "scavenger hook" extension has been removed.
4578 (It was undocumented and unused in the CMU CL sources that SBCL was
4579 derived from, and stale in sbcl-0.6.1.)
4580 * Various errors in the cross-compiler type system were detected
4581 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4582 (enabling various consistency checks). Many of them were fixed,
4583 but some hard problems remain, so the compiler is back to
4584 running without *TYPE-SYSTEM-INITIALIZED* for now.
4585 * As part of the cross-compiler type system cleanup, I implemented
4586 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4587 * I have started adding UNCROSS calls throughout the type system
4588 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4589 on cross-compiler input files will be able to go away, and various
4591 * CONSTANTP now returns true for quoted forms (as explicitly required
4594 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4596 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4597 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4598 (MAKE-LIST -1). In the longer term, it ought to become true
4599 that declarations are assertions even at SAFETY 1. For now, it's not
4600 quite true even at SAFETY 3, but it's at least more nearly true..
4601 (Note that this change seems to increases the size of the system by
4602 O(5%) and to decrease the speed of the compiler by 20% or more.)
4603 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4604 for the problem of printing dozens of lines of distracting information
4605 about low-level system machinery as part of the top stack frame
4606 on entry to the debugger when an undefined function was called.
4607 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4608 is not reset to COMMON-LISP-USER.
4609 * Compilation of stuff related to dyncount.lisp has been made conditional
4610 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4611 smaller. The various dyncount-related symbols have been moved into
4612 a new "SB-DYNCOUNT" package.
4613 * tty-inspect.lisp has been renamed to inspect.lisp.
4614 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4615 feature has gone away. (When we eventually port to other flavors of
4616 libc and/or Unix, we'll try to make the differences between flavors
4617 invisible at the user level.)
4618 * Various other *FEATURES* tags, and/or their associated conditionals,
4619 have been removed if obsolescent, or given better documentation, or
4620 sometimes given more-mnemonic names.
4622 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4624 * tidied up "make.sh" script
4625 * tidied up system directory structure
4626 * better "clean.sh" behavior
4627 * added doc/FOR-CMUCL-DEVELOPERS
4628 * many many small tweaks to output format, e.g. removing possibly-confusing
4629 trailing #\. character in DESCRIBE-INSTANCE
4630 * (EQUALP #\A 'A) no longer signals an error.
4631 * new hashing code, including EQUALP hashing
4632 * tidied up Lisp initialization and toplevel
4633 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4634 * command line argument processing
4635 * added POSIX-GETENV function to deal with Unix-ish environment variables
4636 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4637 terminating SBCL on EOF
4638 * non-verbose GC by default
4639 * There is no more "sbcl" shell script; the sbcl file is now the C
4640 runtime executable (just like CMU CL).
4641 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4643 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4644 symbol and package information as Lisp native symbol and package objects
4645 instead of strings naming symbols and strings naming packages. This way,
4646 whenever packages are renamed (as in warm init), debug information is
4647 transformed along with everything else.
4648 * tweaked the optimization policy declarations which control the building
4649 of SBCL itself. Now, among other things, the system no longer saves
4650 source location debugging information. (This helps two problems at once
4651 by reducing SBCL size and by keeping SBCL from trying to look for its
4652 sources -- which may not exist -- when reporting errors.)
4653 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4654 debugging and testing purposes
4655 * cleaned up printing, making the printer call PRINT-OBJECT for
4656 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4657 methods, giving nearly-ANSI behavior
4658 * converted almost all special variables to use *FOO* naming convention
4659 * deleted PARSE-TIME functionality, since it can be done portably
4660 * moved some files out of cold init into warm init
4661 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4663 * regularized formatting of source files
4664 * added an install.sh script
4665 * fixed ridiculous memory usage of cross-compiler by making
4666 compiler/alloc.lisp not try to do pooling unless it can hook
4667 itself into the GC of the cross-compilation host. Now the system
4668 builds nicely on my old laptop.
4669 * added :SB-ALLOC in target-features.lisp-expr
4670 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4671 was not implemented)
4672 * re-did condition handling and note reporting in the compiler. Notes
4673 are no longer handled by signalling conditions. Style warnings
4674 and warnings are handled more correctly and reported in such a way
4675 that it's easy to find one or the other in your output (so that you
4676 can e.g. figure out which of many problems caused COMPILE-FILE to
4678 * changed the severity of several compiler warnings from full WARNING
4679 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4680 compiler note reporting so that it doesn't use the condition system
4681 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4682 COMPILE-FILE command)
4683 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4684 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4685 no longer supported, since their implementation was deeply intertwingled
4686 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4687 * removed bogus "support" for compiler macros named (SETF FOO), and
4688 removed the compiler macro for SETF INFO (but only after making a fool
4689 of myself on the cmucl-imp mailing list by posting a bogus patch for
4690 DEFINE-COMPILER-MACRO..)
4691 * Compiled files containing forms which have side effects on the Lisp
4692 reader (such as DEFPACKAGE forms) are now handled more correctly.
4693 (Compiler queuing of top level lambdas has been suppressed by setting
4694 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4695 * deleted various currently-unused source files, e.g. gengc.lisp. They
4696 may be added back at some point e.g. when porting to other architectures,
4697 but until they are it's distracting to distribute them and to try to
4699 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4700 there were so many of them they're just distractions, and UNCROSS is
4701 known to be able to handle the current sources
4702 * moved PROFILE functionality into TRACE, so that it will be clear
4703 how the wrapping and unwrapping of functions when you profile them
4704 interacts with the wrapping and unwrapping of functions when you
4705 trace them. (Actually, the functionality isn't there yet, but at least
4706 the interface specification is there. Hopefully, the functionality will
4707 arrive with some future maintenance release.)
4708 * removed host-oops.lisp
4709 * changed signature of QUIT function to allow UNIX-CODE argument
4710 * fixed READ-SEQUENCE bug
4711 * tweaked verbose GC output so that it looks more like the progress
4712 output that ANSI specifies for functions like LOAD
4713 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4714 * added <http://sbcl.sourceforge.com> to the banner information printed by