1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.20 relative to 1.0.19:
3 * bug fix: fixed #427: unused local aliens no longer cause compiler
4 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
6 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
7 alien-stack leakage. (reported by Andy Hefner)
8 * bug fix: PROGV signals an error when an attempt to violate declared
9 type of a variable or bind a constant is made.
10 * bug fix: SET signals an error when an attempt to violate declared
11 type of a variable is made.
13 changes in sbcl-1.0.19 relative to 1.0.18:
14 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
15 warnings that go otherwise unhandled will be muffled if they are
16 of the type that's the value of this variable.
17 * optimization: stack allocation is slightly more efficient on x86
19 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
20 where a raw slot always is initialized using the initform whose
21 type is not know sufficiently well a compile-time are now compiled
22 correctly. (reported by John Morrison)
23 * bug fix: compiler no longer makes erronous assumptions in the
24 presense of non-foldable SATISFIES types.
25 * bug fix: stack analysis missed cleanups of dynamic-extent
26 arguments in non-let-converted calls to local functions.
27 * improvements to the Windows port:
28 ** adjusted address spaces for building on both Win32 and
29 Win64. (thanks for John Connors)
30 * fixed some bugs revealed by Paul Dietz' test suite:
31 ** interval arithmetic during type derivation used inexact integer
32 to single-float coercions.
33 ** arithmetic operations involving large integers and single
34 floats give the same results in compiled and interpreted code.
35 ** deriving the result type of COERCE no longer signals an error
36 if the derived type of the second argument is a MEMBER type
37 containing invalid type specifiers.
38 ** ADJOIN with constant NIL as second argument works correctly.
39 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
41 changes in sbcl-1.0.18 relative to 1.0.17:
42 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
43 profiles only the current thread.
44 * minor incompatible change: changes to SYMBOL-VALUE of constants
45 defined with DEFCONSTANT now signal an error.
46 * enhancement: SB-SPROF now has support for wallclock profiling,
47 and is also able to profile specific threads. REPORT output
48 has also additional sorting options.
49 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
51 * optimization: structure allocation has been improved
52 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
53 ** out of line constructors are ~10% faster.
54 ** inline constructors are ~15% faster.
55 ** inline constructors are capable of dynamic extent allocation
56 (generally on x86 and x86-64, in some cases on other platforms
58 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
60 * optimization: file compiler is now able to coalesce non-circular
61 lists, non-base strings, and bit-vectors. Additionally, constants
62 are never referenced through SYMBOL-VALUE at runtime.
63 * optimization: code defining methods on PRINT-OBJECT (and other
64 generic functions in the COMMON-LISP package) now loads faster.
65 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
67 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
68 fasl is now deleted. (reported by Attila Lendvai)
69 * bug fix: READ-LINE always returned NIL for the last line in files.
70 (reported by Yoshinori Tahara)
71 * bug fix: more accurate disassembly annotations of foreign function
72 calls. (thanks to Andy Hefner)
73 * bug fix: trimming non-simple strings and non-string string
74 designators when the there is nothing to trim works properly.
75 (thanks to James Knight)
76 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
77 (thanks to Travis Cross)
78 * fixed some bugs revealed by Paul Dietz' test suite:
79 ** NIL is a valid function name (regression at 1.0.13.38)
80 ** FILL on lists was missing its return value (regression at 1.0.12.27)
81 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
82 fill pointers properly (regression at 1.0.12.23)
83 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
84 displacement indices properly (regression at 1.0.12.23)
86 changes in sbcl-1.0.17 relative to 1.0.16:
87 * temporary regression: user code can no longer allocate closure
88 variable storage on stack, due to bug 419 without explicitly
89 requesting it. Please consult sbcl-devel for advice if you need to
90 use this feature in the meanwhile.
91 * new feature: runtime argument --control-stack-size can be used to
92 adjust thread default control stack size.
93 * enhancement: improved TIME output
94 ** all times are reported using the measured accuracy (milliseconds
95 for real and GC times, microseconds for everything else.)
96 ** processor cycle counts on x86 and x86-64.
97 ** interpreted forms are counted for both evaluator modes.
98 ** number of lambdas converted by the compiler is reported.
99 ** CPU percentage report (computed from real and total run time.)
100 ** more comprehensive run time reporting, using a condenced format
101 ** interperted form, lambda, and page fault counts are omitted
103 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
105 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
106 * optimization: two argument forms of LAST are upto ~10% faster
107 in normal SPEED policies.
108 * optimization: NCONC no longer needs to heap cons its &REST list
109 in normal SPEED policies.
110 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
112 * bug fix: UNION and NUNION work with :TEST-NOT once more,
113 regression since 1.0.9.1. (thanks to Eric Marsden)
114 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
115 since 1.0.15.36. (thanks to Paul Khuong)
116 * bug fix: LAST when always returned the whole list when given a bignum
117 as the second argument.
118 * bug fix: dynamic extent allocation of nested lists and vectors
119 could leak to otherwise accessible parts.
120 * bug fix: invalid optimization of heap-allocated alien variable
122 * bug fix: fasl header checking is less vulnerable to different
123 platform word lengths.
124 * bug fix: more correct assembler syntax for GNU binutils
125 2.18.50.0.4 support. (thanks to Marijn Schouten)
126 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
128 * bug fix: the fopcompiler can handle LOCALLY forms (with no
129 declarations) successfully. (reported by Attila Lendvai)
131 changes in sbcl-1.0.16 relative to 1.0.15:
132 * minor incompatible change: revert the changes to sb-posix's error
133 signaling added in 1.0.14.
134 * minor incompatible change: change PROBE-FILE back to returning
135 NIL whenever we can't get a truename, as was the case before 1.0.14.
136 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
137 inherits from ERROR instead of just CONDITION.
138 * new feature: SB-INTROSPECT can provide source locations for instances
139 as well. (thanks to Tobian Ritterweiler)
140 * optimization: binding special variables now generates smaller code
141 on threaded platforms.
142 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
143 and cases where no :TEST is given but the compiler can infer that
144 the element to search is of type (OR FIXNUM (NOT NUMBER)).
145 * optimization: better LOGNOT on fixnums.
146 * optimization: modular arithmetic for a particular requested width
147 is implemented using a tagged representation unless a better
148 representation is available.
149 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
150 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
151 for wellformedness and returns a specialized vector. (reported by
152 Francois-Rene Rideau)
153 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
154 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
155 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
156 unscheduling timer at the same time as another timer fires.
157 * bug fix: don't reschedule timers for dead threads.
158 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
159 * bug fix: copying output from RUN-PROGRAM to a stream signalled
160 bogus errors if select() was interrupted.
161 * enhancement: add support for fcntl's struct flock to SB-POSIX.
163 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
164 * enhancement: cleaner backtraces for interactive interrupts, as
165 well as other cases where the interesting frames used to be
166 obscured by interrupt handling frames.
167 * enhancement: untracing a whole package using (UNTRACE "FOO") is
168 now supported, and tracing a whole package using (TRACE "FOO") now
169 traces SETF-functions as well.
170 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
171 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
172 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
173 * unreadably printed representation of hash-tables now includes
175 * bug fix: partially fixed #188: type propagation from assignments
176 is now more efficient.
177 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
178 'DOUBLE-FLOAT) are not flushable.
179 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
180 full address of the object, and none of the tag bits.
181 * bug fix: readably printing hash-tables now respects other printer
182 control variables. (reported by Cedric St-Jean)
183 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
184 keyword with MAKE-HASH-TABLE.
185 * bug fix: export SB-POSIX:MKSTEMP.
186 * bug fix: SORT was not interrupt safe.
187 * bug fix: XREF accounts for the last node of each basic-block as
189 * bug fix: MAKE-INSTANCE optimizations interacted badly with
190 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
191 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
192 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
193 create obsolete instances in the case of redefinition or
194 obsoletion of a superclass. (thanks to Andy Hefner)
195 * bug fix: Support for the Alpha architecture has been revived; it had
196 suffered somewhat from lack of maintenance since sbcl-1.0.
197 * improvements to the Windows port:
198 ** The system detects the codepage to use at startup, instead of
199 using the value from the saved core. (thanks to Kei Suzuki)
201 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
202 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
203 (see documentation for details.)
204 * revived support for OpenBSD (contributed by Josh Elsasser)
205 * partially fixed bug #108: ROOM no longer suffers from occasional
206 (AVER (SAP= CURRENT END)) failures.
207 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
208 bodies no longer cause a WARNING to be signalled. (reported by
210 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
211 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
212 the bug in TRUNCATE.) (reported by Michael Weber)
213 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
214 no samples. (reported by Andy Hefner)
215 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
216 no longer appear as (NIL ...) frames in backtraces.
217 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
218 is now actually exported from SB-EXT as documented. (reported by
220 * bug fix: sb-aclrepl now correctly understands how to inspect
221 single-floats on 64-bit platforms where single-floats are not boxed.
222 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
223 yet been finalized. (reported by Levente Meszaros)
224 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
225 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
226 SBCL-specific optimize qualities.
228 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
229 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
230 an executable in the search path, and does so in the child
231 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
232 been removed; it can be found in the CVS history, for somebody who
233 needs that search behavior (see the manual).
234 * minor incompatible change: compiler policy re. weakening type
235 checks has changed: now type checks are weakened only if SAFETY < 2
237 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
238 unparsing of directory pathnames as files. Analogously,
239 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
240 filename to parse into a directory pathname.
241 * enhancement: implicit generic function creation now signals a
242 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
243 users can bind handlers for and muffle around calls to LOAD.
244 * enhancement: RUN-PROGRAM allows unicode arguments and environments
245 to be used (using the default stream external format), and allows
246 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
247 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
248 strings and vectors whose element-type or simplicity is not fully
249 known at compile-time.
250 * optimization: STRING-TRIM and related functions no longer allocate
251 a new string when no trimming needs to be performed. These functions
252 are also faster than before when the input string has been declared
254 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
255 * optimization: READ-LINE is significantly faster for files containing
257 * optimization: non-open coded uses of character comparison operators
258 (e.g. char=) no longer cons when called with more than one parameter
259 on platforms supporting dynamic-extent allocation.
260 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
261 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
262 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
264 * bug fix: some sequence functions elided bounds checking when
266 * bug fix: too liberal weakening of union-type checks when SPEED >
268 * bug fix: more bogus fixnum declarations in ROOM implementation
271 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
272 * new feature: MAKE-HASH-TABLE now experimentally accepts a
273 :SYNCHRONIZED argument, which makes the hash-table safe for
274 concurrent accesses (but not iteration.) See also:
275 SB-EXT:WITH-LOCKED-HASH-TABLE, and
276 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
277 * optimization: CONCATENATE on strings is an order of magnitude faster
278 in code compiled with (> SPEED SPACE).
279 * optimization: SUBSEQ is ~50% faster on lists.
280 * bug fix: bug 417 fixed -- source location reporting is now more robust.
281 * bug fix: SUBSEQ on a list will now correctly signal an error if if
282 END is smaller then START.
283 * bug fix: SB-PROFILE will no longer report extra consing for nested
284 calls to profiled functions.
285 * bug fix: ROOM implementation had bogus fixnum declarations which
286 could cause type-errors when calling ROOM on large images.
287 * bug fix: if file compilation is aborted, the partial fasl is now
288 deleted, and COMPILE-FILE returns NIL as the primary value.
289 * bug fix: number of thread safety issues relating to SBCL's internal
290 hash-table usage have been fixed.
291 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
292 be returned from its body when the values were being returned
293 using unknown-values return convection and the W-P-O was wrapped
294 inside an UNWIND-PROTECT.
295 * bug fix: sb-posix should now compile again under Windows, enabling
298 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
299 * incompatible change: hash-table accessor functions are no longer
300 automatically protected by locks. Concurrent accesses on the same hash-table
301 from multiple threads can give inconsistent results or even corrupt the
302 hash-table completely. Multi-threaded applications should do their own
303 locking at the correct granularity. In the current implementation it is
304 still safe to have multiple readers access the same table, but it's not
305 guaranteed that this property will be maintained in future releases.
306 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
307 and will signal an error at runtime.
308 * enhancement: SB-THREAD package now exports a semaphore interface.
309 * enhancement: CONS can now stack-allocate on x86 and
310 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
312 * enhancement: nested lists can now be stack allocated on
313 platforms providing stack allocation support.
314 * enhancement: dynamic-extent support has been extended to support
315 cases where there are multiple possible sources for the stack
317 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
318 if the mutex is uncontested on Linux.
319 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
320 as specified by the CLHS. (thanks to Tobias Rittweiler)
321 * bug fix: NaN comparison now works on x86-64.
322 * bug fix: CLOSE :ABORT T on a stream with pending output now
324 * bug fix: instances of non-standard metaclasses using standard
325 instance structure protocol sometimes missed the slot type checks
327 * bug fix: known functions can be inlined in high-debug code.
329 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
330 * minor incompatible change: the MSI installer on Windows no longer
331 associates .lisp and .fasl files with the installed SBCL.
332 * minor incompatible change: :UNIX is no longer present in *FEATURES*
333 on Windows. (thanks to Luis Oliviera)
334 * new feature: SB-CLTL2 contrib module now implements
335 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
336 * optimization: scavenging weak pointers is now more efficient,
337 requiring O(1) instead of O(N) per weak pointer to identify
338 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
339 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
340 method bodies using constant slot-names when the first argument is
341 a specializer parameter for the method.
342 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
343 names, when the class of the instance is a direct instance
344 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
345 as slow as the constant slot-name case.)
346 * optimization: member type construction is now O(N) instead
348 * optimization: UNION and NUNION are now O(N+M) for large
349 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
351 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
352 now more readable in environments like Slime which display it.
353 (thanks to Tobias C. Rittweiler)
354 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
355 was able to infer the correct argument type for the object on which
356 the CAS operation was being performed.
357 * bug fix: copy propagation interfered with parallel assignment
358 semantics in local calls. (reported by Paul Khuong)
359 * bug fix: the signed modular fixnum shift compiled to wrong code on
360 x86 and x86-64. (spotted by a slight modification to some of
363 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
364 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
365 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
366 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
368 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
369 and generic function now print the object identity for anonymous
370 instances. (thanks to Attila Lendvai)
371 * enhancement: as an extension to MOP, SBCL now supports SETF of
372 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
373 (thanks to Attila Lendvai)
374 * enhancement: sb-cover has an alternate annotation mode which puts
375 the form annotations at the car of the form, instead of annotating
376 it completely. The benefit of this mode is that it shows better
377 which forms were instrumented by the compiler.
378 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
379 by printer control variables. (reported by Dan Corkill)
380 * bug fix: system leaked memory when delayed output was performed by
381 the OS in smaller chunks then expected. (thanks to David Smith)
382 * bug fix: system leaked memory when file streams were not closed
384 * bug fix: large objects written to slow streams that were modified
385 after the write could end up with the modified state written to
386 the underlying file descriptor.
387 * bug fix: multiple threads operating in parallel on the same stream
388 could cause buffer-overflows.
389 * bug fix: source location information is stored correctly
390 (broken since 1.0.6). This bug would generally show up as the
391 Slime debugger highlighting the wrong form when the sldb-show-source
393 * bug fix: Branch forms are again annotated as branches in the sb-cover
395 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
396 on x86-64. (reported by Gregory Vanuxem)
397 * bug fix: EQUALP could return wrong results for structures with raw
398 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
399 word). (reported by Vjacheslav Fyodorov)
400 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
403 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
404 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
405 atomic compare-and-swap operations on threaded platforms.
406 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
407 allows assigning a global minimum value to optimization qualities
408 (overriding proclamations and declarations).
409 * enhancement: closed over variables can be stack-allocated on x86
411 * performance bug fix: GETHASH and (SETF GETHASH) are once again
413 * optimization: slot definition lookup is now O(1). This speeds up
414 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
415 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
416 * optimization: ASSOC and MEMBER can now be open-coded for all
417 combinations of keyword arguments when second argument is constant
418 and SPEED >= SPACE. In other cases a specialized version is
420 * bug fix: using obsoleted structure instances with TYPEP and
421 generic functions now signals a sensible error.
422 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
423 (reported by Kristoffer Kvello)
424 * bug fix: backtrace construction is now more careful when making
425 lisp-objects from pointers on the stack, to avoid creating bogus
426 objects that can be seen by the GC.
427 * bug fix: defaulting of values in contexts expecting more than 7
428 variables now works on x86-64. (reported by Christopher Laux)
429 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
431 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
432 as the property-list of a symbol.
433 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
434 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
435 situations requested, are once again file-compileable. (reported
438 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
439 * MOP improvement: support for user-defined subclasses of
440 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
441 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
442 the expansion of DEFMETHOD, is responsible for generating a form
443 which creates a list of specializers when evaluated. Additional
444 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
445 debugging and introspective support.
446 * minor incompatible change: the (unsupported) spinlock interface
447 has changed: free spinlock now has the value NIL, and a held spinlock
448 has the owning thread as its value.
449 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
450 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
452 * enhancement: name of a socket-stream is now "a socket" instead of
454 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
455 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
456 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
457 for generic functions. (Thanks to Tobias C. Rittweiler)
458 * optimization: bignum printing speed has been improved by 20-40%
459 (depending on the bignum size.)
460 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
462 * bug fix: the cache used by the CLOS to store precomputed effective
463 methods, slot offsets, and constant return values is now thread and
465 * bug fix: generic function dispatch function updating is now thread
466 and interrupt safe (in the sense that the known issues have been
468 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
469 * bug fix: interning EQL-specializers is now thread and interrupt safe.
470 * bug fix: asdf systems with dependencies to the SB-POSIX or
471 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
472 * bug fix: interrupt safety of applicable method computation has been
475 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
476 * new contrib: sb-cover, an experimental code coverage tool, is included
478 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
479 significantly faster.
480 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
481 produce more efficient code on x86-64 if the compiler can determine
482 that (+ INDEX <constant>) does not require a bounds check and FOO
483 has an element type at least 8 bits wide.
484 * enhancement: a new, experimental synchronous timeout facility is
485 provided. Refer to SB-SYS:WITH-DEADLINE for details.
486 * enhancement: when a symbol name conflict error arises, the
487 conflicting symbols are always printed with a package prefix.
488 (thanks to Kevin Reid)
489 * enhancement: stepping is now once again supported on the SPARC and
490 MIPS platforms. (It is also now more likely to work on CheneyGC
492 * enhancement: sb-sprof can now also track and report accurate call
494 * bug fixes: the treatment of non-standard subclasses of
495 SB-MOP:SPECIALIZER is more correct.
496 * incompatible change: PURIFY no longer copies the data from the
497 dynamic space into the static and read-only spaces on platforms
498 that use the generational garbage collector
499 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
501 * bug fix: GC race condition occasionally resulting in crashes with
502 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
504 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
506 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
507 a STYLE-WARNING to be signalled (regression from 1.0.4.)
508 * bug fix: an asynchronous interrupt could previously leave the
509 system running with GC inhibited.
510 * bug fix: a DECLARE form evaluated at top-level now causes an error
511 rather than silently (or verbosely) returning NIL.
512 * bug fix: trying to dissassemble functions compiled at high DEBUG could
513 result in a error being signalled due to source form lookup errors.
514 (reported by Peter Graves)
516 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
517 * incompatible change: removed writer methods for host-ent-name,
518 host-ent-addresses -- changing the values did not update the DNS
520 * minor incompatible change: changed experimental JOIN-THREAD interface
521 * documentation: the manual now lists reader and writer methods
522 in class slot documentation sections. (thanks to Richard M Kreuter)
523 * documentation: unwinding from asyncronous events has been
524 documented as unsafe.
525 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
526 in multithreaded application code.
527 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
528 platforms. (thanks to James Anderson for the optimization hint)
529 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
531 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
532 variants no longer cons.
533 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
534 their NOT- variants no longer cons.
535 * optimization: EQUAL hash tables no longer use SXHASH for objects
536 of all data types, but instead use an EQL hash for types for which
537 EQUAL is the same as EQL.
538 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
539 are significantly faster.
540 * optimization: new STRING-OUTPUT-STREAM implementation is much
541 faster for large outputs and conses 30% less on average.
542 * enhancement: XREF information is now collected for references made
543 to global variables using SYMBOL-VALUE and a constant argument.
544 * enhancement: SIGINT now causes a specific condition
545 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
546 * bug fix: FILE-POSITION used to signal an error in some cases where
547 ANSI requires it to return NIL.
548 * bug fix: ADJUST-ARRAY is now interrupt-safe.
549 * bug fix: adding and removing fd-handlers is now interrupt-safe.
550 * bug fix: inlined calls to C now ensure 16byte stack alignment on
552 * bug fix: bad type declaration in the CLOS implementation has
553 been fixed. (reported by James Anderson)
554 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
555 (thanks to Lutz Euler)
556 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
557 required by the CLHS. (reported by Samium Gromoff)
558 * bug fix: dead unbound variable references now signal an error.
559 (reported by Marco Monteiro)
560 * bug fix: / with an unused value was being deleted in safe code.
561 (thanks to Marco Monteiro and Kevin Reid)
562 * bug fix: number of characters that can be written onto a single
563 line in a file is unlimited.
564 * bug fix: some GC deadlocks caused by asynchronous interrupts have
565 been fixed by inhibiting interrupts for when GC is disabled.
566 * bug fix: some interrupt safety issues with GETHASH, (SETF
567 GETHASH), CLRHASH and REMHASH have been fixed.
568 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
569 specifier no longer causes infinite recursion.
570 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
571 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
572 * bug fix: an &environment argument with the correct variable information
573 is passed to macros that are expanded during byte compilation.
574 (reported by Samium Gromoff)
575 * bug fix: a logic error was causing the hash function for bignums to
576 have excessive amounts of collisions. (reported by Faré Rideau)
577 * bug fix: modifying the contents of an array could change the return
578 value of SXHASH on that array, which is only allowed for strings
579 and bit vectors. (bug introduced in 0.9.16)
580 * improvement: the x86-64/darwin port now passes all tests (except
581 for the debugger tests) but should still be considered
582 experimental until this is fixed.
583 * improvement: a style-warning is signaled for CASE (etc) clauses with
584 duplicate keys. (thanks to Kevin Reid)
585 * improvement: macos/x86 and macos/x86-64 now use mach exception
586 handlers for dealing with illegal instructions (for trapping and
587 error handling) and memory protection violations (for GC).
589 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
590 * new platform: experimental support for x86-64/darwin (MacOS).
591 * incompatible change: the thread-safe (on most platforms) getaddrinfo
592 and getnameinfo sockets functions are used instead of gethostbyaddr
593 and gethostbyname, on platforms where the newer functions are available.
594 As a result, the ALIASES field of HOST-ENT will always be NIL on these
596 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
597 don't print the startup banner, but behave as if --noinform was passed
598 as a command line argument. (thanks to Kevin Reid)
599 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
600 * optimization: code using alien values with undeclared types is much faster.
601 * optimization: the compiler is now able to open code SEARCH in more cases.
602 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
603 * bug fix: using standardized COMMON-LISP special variables as loop
604 variables no longer signals bogus package lock violations. (reported
606 * bug fix: declaring local loop variables to be of a range-limited type
607 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
608 (reported by Andras Simon)
609 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
610 bugs remain on x86-64.)
611 * bug fix: the #= and ## reader macros now interact reasonably with
612 funcallable instances.
613 * bug fix: type-checks for function arguments were compiled using the
614 compiler policy settings of the wrong lexical environment (in 1.0.2
616 * bug fix: SHADOW accepts characters as string designators, as required
617 by the spec (thanks to Eric Marsden)
618 * bug fix: fixed GC safety issues when foreign functions are called with
619 non-base strings as arguments
620 * bug fix: more consistent error output for fatal-compiler-warnings, like
622 * bug fix: functions with proper names were showing up as NIL in
623 backtraces (reported by Edi Weitz, regression in 1.0.2)
625 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
626 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
627 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
628 produce more efficient code on the x86 if the compiler can determine
629 that (+ INDEX <constant>) does not require a bounds check and FOO
630 has an element type at least 8 bits wide.
631 * bug fix: references to undefined variables are handled the same way
632 in toplevel forms as in the normal compiler
633 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
635 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
636 (thanks to Jon Buller)
637 * improvement: faster compilation times for complex functions
638 * improvement: added readlink support to SB-POSIX (thanks to Richard
641 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
642 * improvement: experimental support for mach exception handling on
643 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
645 * improvement: support for GBK external format.
646 (thanks to Chun Tian (binghe))
647 * improvement: the debugger now displays variables that have been closed
648 over, in code compiled with (DEBUG 2) or higher
649 * improvement: support for executable cores on NetBSD (thanks to
651 * new feature: added a RESTART-FRAME debugger command
652 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
653 be used to provide an implementation for FILE-POSITION on Gray streams
654 (thanks to Eric Marsden)
655 * new feature: add syslog, openlog and closelog support to SB-POSIX
656 (thanks to Richard Kreuter)
657 * optimization: the function call overhead in code compiled with
658 a high DEBUG optimization setting is significantly
659 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
660 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
661 (thanks to Tony Martinez)
662 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
663 evaluated once (reported by Kevin Reid)
664 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
665 works again (thanks to Kilian Sprotte)
666 * bug fix: an error is signaled for tagbodies with duplicate tags
667 (thanks to Stephen Wilson)
668 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
669 * bug fix: Win32 port can now handle foreign code unwinding Lisp
670 stack frames from alien callbacks.
671 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
672 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
673 65k pages (thanks to David Woodhouse)
674 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
676 changes in sbcl-1.0.1 relative to sbcl-1.0:
677 * new platform: FreeBSD/x86-64, including support for threading.
678 * new feature: the compiler stores cross-referencing information
679 about function calls (who-calls), macroexpansion (who-macroexpands)
680 and special variables (who-binds, who-sets, who-references) for code
681 compiled with (< SPACE 3). This information is available through the
682 sb-introspect contrib.
683 * new feature: users may subclass SEQUENCE, and have instances of
684 these classes interoperate with standard Common Lisp functions if
685 a number of methods are defined. (This feature is experimental
686 and the interface subject to change based on feedback from SBCL
687 users and the general community)
688 * improvement: sb-sprof traces call stacks to an arbitrary depth on
689 x86/x86-64, rather than the previous fixed depth of 8
690 * improvement: another pthread back-end of mutex "pthread-futex".
691 * bug fix: non-ascii command-line arguments are processed correctly
692 (thanks to Yaroslav Kavenchuk)
693 * bug fix: non-required arguments were not passed correctly when a method
694 defined using DEFMETHOD was called from a mop-generated method using
695 CALL-NEXT-METHOD (reported by Pascal Costanza)
696 * bug fix: recursion is now permitted in accessors through
697 SLOT-UNBOUND. (reported by Pascal Costanza)
698 * bug fix: an error was signaled at startup if the HOME environment
699 variable was defined, but had an empty value (reported by Peter Van Eynde)
700 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
701 signal a PROGRAM-ERROR, not a BUG.
702 * bug fix: SB-POSIX:READDIR works when built with large file support.
703 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
705 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
706 for better automatic indentation support. (thanks to Matt Pillsbury)
707 * optimization: loading generic functions no longer takes O(n^2) time,
708 proportional to the amount of methods in the generic function
709 (reported by Todd Sabin and Jeremy Brown)
710 * optimization: the FIND and POSITION family of sequence functions
711 are significantly faster on arrays whose element types have been
713 * improvements to the Windows port:
714 ** Intermittent heap corruption problems have been fixed. (thanks
715 to Alastair Bridgewater)
716 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
718 ** Lisp is able to unwind foreign exception frames from alien
719 callbacks. (thanks to Alastair Bridgewater)
721 changes in sbcl-1.0 relative to sbcl-0.9.18:
722 * improvement: experimental support for threading on FreeBSD/x86.
723 (thanks to NIIMI Satoshi)
724 * improvement: runtime option --dynamic-space-size can be used
725 to set the size of the dynamic space reserved on startup.
726 * improvement: floating point modes in effect are now saved in
727 core, and restored on startup.
728 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
729 startup, not time since first call to GET-INTERNAL-REAL-TIME.
730 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
731 threads are not running after *SAVE-HOOKS* have run.
732 * improvement: writes to CLOS instance slots are type-checked in code
733 compiled with (SAFETY 3)
734 * improvement: floating-point exception handling on FreeBSD (thanks to
736 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
737 (thanks to Zach Beane)
738 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
740 * improvement: added support for the Shift-JIS external format.
741 (contributed by NIIMI Satoshi)
742 * improvement: callbacks are supported on Linux/PPC. (thanks to
744 * bug fix: compiler bug triggered by a (non-standard) VALUES
745 declaration in a LET* was fixed. (reported by Kaersten Poeck)
746 * bug fix: file compiler no longer confuses validated and already
747 dumped structurres. (reported by Kaersten Poeck)
748 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
749 fill-pointer signals a type-error as required. (thanks to
751 * bug fix: disassemly of funcallable instances works.
752 * bug fix: single stepping on PPC.
753 * bug fix: fix thread-safety problems in the type system (generally
754 manifesting as nonsensical errors like "STRING is a bad type specifier
755 for sequences" or "The value 1 is not of type FIXNUM")
756 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
757 (reported by Josip Gracin)
758 * bug fix: an error is signaled for attempts to displace arrays with
759 incompatible element types (thanks to Mario Mommer)
760 * bug fix: more correct handling of wide characters in debug info
761 (bug reported by Attila Lendvai and fixed by Juho Snellman)
762 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
763 and don't cause extra consing
764 * optimization: MAP and MAP-INTO are significantly faster on vectors
765 whose elements types have been declared.
766 * Improvements to SB-SPROF:
767 ** Support for allocation profiling
768 ** Reduced profiling overhead, especially for long profiling runs
769 * Improvements to the Windows port:
770 ** floating point exceptions are now reported correctly.
771 ** stack exhaustion detection works partially.
772 ** more accurate GET-INTERNAL-REAL-TIME.
773 ** better RUN-PROGRAM behaviour in case of errors in the child
775 ** PROBE-FILE now simplifies pathnames correctly.
776 ** DIRECTORY now works correctly with :WILD-INFERIORS.
777 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
778 1st 1970 (although time timezone data will be unreliable).
779 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
781 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
782 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
783 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
785 * bug fix: two potential GC deadlocks affecting threaded builds.
786 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
787 returns T (reported by Anton Kazennikov)
788 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
789 with non-variable places
790 * bug fix: remove a race condition in the setting of
791 funcallable-instance functions, this should make threaded CLOS
792 code more stable against memory faults.
793 * bug fix: corruption of specials when unbinding is interrupted by an
794 asynchronous unwind (reported by Hannu Koivisto)
795 * improvement: the debugger will now also display local variables that
796 are only used once, for code compiled with a DEBUG optimization quality
799 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
800 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
801 * incompatible change: External-format support for FFI calls. The
802 SB-ALIEN:C-STRING no longer implies an ASCII
803 external-format. Instead the string is subject to external-format
804 conversion. Additionally return values of type SB-ALIEN:C-STRING
805 are by default of type (SIMPLE-ARRAY CHARACTER), not
806 SIMPLE-BASE-STRING. For an alien type that behaves like the old
807 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
808 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
809 * incompatible change: SB-EXT package no longer contains the
810 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
811 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
813 * incompatible change: the single-stepper is no longer available
814 on Alpha, Mips and Sparc platforms.
815 * minor incompatible change: the direct superclasses of
816 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
817 not (STANDARD-OBJECT FUNCTION). This makes the
818 class-precedence-lists of GENERIC-FUNCTION and
819 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
821 * minor incompatible change: the default stream external format on
822 non-unicode SBCL is no longer detected from the locale, but is always
824 * new feature: Add a version of evaluator that uses an interpreter instead
825 of the compiler. EVAL still uses the compiler by default, to switch it
826 to use the interpreter, set the value of the variable
827 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
828 * minor incompatible change: the single-stepper REPL has been merged
829 with the normal debugger (see the "Stepping" heading of the debugger help
830 for more details). The debugger command STEP will no longer switch
831 to the single-stepper REPL.
832 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
833 for a type now works.
834 * bug fix: loading of default sysinit file works. (thanks to Leonid
836 * bug fix: better detection of circularities in the file-compiler.
837 (reported by Marco Monteiro)
838 * bug fix: the CL pathname functions now work with files that have
839 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
840 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
841 systems with Unix98 pty semantics.
842 * bug fix: ASDF-INSTALL will now work with bsd tar.
843 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
845 * bug fix: timers expiring in dead threads no longer cause a
846 type-error (reported by Paul "Nonny Mouse").
847 * bug fix: thanks to more lightweight single-stepper instrumentation,
848 code compiled with (DEBUG 3) will compile and execute significantly faster,
849 and will have more accurate type-inferencing than before
850 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
851 whose bindings are modified
852 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
853 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
854 ** replace dummy implementations of CL:MACHINE-INSTANCE and
855 CL:SOFTWARE-VERSION with proper version
857 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
858 * feature: implemented the READER-METHOD-CLASS and
859 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
860 as specified by AMOP.
861 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
863 * optimization: faster LOGCOUNT implementation on x86 and x86-64
864 (thanks to Lutz Euler)
865 * optimization: hashing of general arrays and vectors has been
866 improved. (reported by Any Fingerhut)
867 * enhancement: SB-INTROSPECT is now able to find definitions of
868 profiled functions. (thanks to Troels Henriksen)
869 * enhancement: compiler-macro expansion applies now to FUNCALL forms
870 as well, allowing compiler-macros for SETF-functions to expand.
871 * enhancement: step-instrumentation no longer wraps known
872 single-value functions with multiple-value context, allowing
873 better type inference.
874 * fixed bug #337: use of MAKE-METHOD in method combination now works
875 even in the presence of user-defined method classes. (reported by
876 Bruno Haible and Pascal Costanza)
877 * fixed bug #339(c): if there are applicable methods not part of any
878 long-form method-combination group, call INVALID-METHOD-ERROR.
879 (reported by Bruno Haible)
880 * fixed bug #361: the :FUNCTION initarg in the protocol for
881 initialization of methods can now be used to override
882 internally-produced optimized functions. (reported by Bruno
884 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
885 system-provided lambda expression no longer cause warnings about
886 unbound #:|pv-table| symbols.
887 * bug fix: improved the handling of type declarations and the
888 detection of violations for keyword arguments with non-constant
890 * bug fix: potentially erronous calls to PATHNAME and
891 MERGE-PATHNAMES were being flushed in some cases.
892 (reported by Richard Kreuter)
893 * bug fix: compiled calls to TYPEP were mishandling obsolete
894 instances. (reported by James Bielman and Attila Lendvai)
895 * bug fix: format strings with ~> without matching ~< no longer
896 trigger an AVER, but signal an understandable error instead.
897 (reported by Antonio Martinez)
898 * bug fix: specifying an output-file without a pathname-type for
899 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
900 of input-file instead of "fasl". (reported by Robert Dodier)
901 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
902 (reported by James Y Knight).
903 * bug fix: compiler-macros-function did not consider the environment
904 argument for shadowing by local functions.
905 * bug fix: compiler-macros expansion was inhibited by local INLINE
907 * bug fix: inline expansions of known functions were subject to
908 step-instrumentation in high debug policies, leading to problems
910 * bug fix: compiler failed to differentiate between different CONS
912 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
913 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
914 * bug fix: #368: incorrect use of expressed vs. upgraded array
916 * bug fix: #306a: more precise unions of array types.
917 * thread-safety improvements:
918 ** CONDITION-WAIT could return early on Linux, if the thread was
919 interrupted and subsequently continued with SIGCONT.
920 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
922 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
923 * added support for the ucs-2 external format. (contributed by Ivan
925 * minor incompatible change: pretty printing of objects of type
926 (cons symbol) is, in the default pprint-dispatch-table, now
927 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
929 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
930 called later in a class's lifetime, possibly as late as when the
931 first instance of the class is created. Previously,
932 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
933 class became finalizeable.
934 * fixed bug: FILE-POSITION sometimes returned inconsistent results
935 for multibyte external-format streams. (thanks to "vbzoli")
936 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
937 with :ALLOCATION :CLASS inherited from superclasses of the
939 * fixed bug: anonymous classes can now be created using the :NAME
940 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
941 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
942 * fixed bug: core-files saved with :EXECUTABLE T can again be
943 executed when SBCL_HOME isn't set. (reported by James Knight)
944 * fixed bug: toplevel LOCALLY forms with declarations could
945 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
946 * fixed bug: printing from several different threads using different
947 values of *print-case* could cause invalid output, due to
948 some internal special variables of the printer not being bound
949 thread-locally (reported by Max Mikhanosha)
950 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
951 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
952 (reported by Pascal Costanza)
953 * fixed bug: FINALIZE-INHERITANCE is called from
954 REINITIALIZE-INSTANCE on classes when the class has previously
955 been finalized, as required by AMOP.
956 * minor code generation optimizations:
957 ** better register allocation in CLOS dispatching functions
958 ** overflow detection when coercing signed bytes to fixnums on x86-64
959 ** is now implemented with one IMUL instruction instead of three shifts
960 ** more efficient bit-vector access on x86 and x86-64
961 ** more efficient access to raw structure slots on x86 and x86-64
962 * fixed some bugs revealed by Paul Dietz' test suite:
963 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
966 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
967 * feature: thread support on Solaris/x86, and experimental thread support
969 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
971 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
972 default method for SLOT-UNBOUND.
973 * minor incompatible change: prevent the user from specializing the
974 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
975 somewhat counter to the intent of the protocol, I (CSR) think, and
976 additionally it just doesn't work in SBCL as currently
977 implemented, thanks to optimizations (that are always valid for
978 the other three SLOT-VALUEish functions, but not for the setter).
979 * minor incompatible change: the :SB-LDB feature is now enabled by
980 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
981 the low-level debugger.
982 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
983 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
984 down to an SB-SYS:FD-STREAM, allowing direct communication in
985 more cases. Temporary files and pipes are still used as a fallback
987 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
988 on the non-reentrant C functions localtime(3) and gmtime(3).
989 * bug fix: global optimization policy was not visible in LOCALLY and
991 * bug fix: class objects can be used as specializers in methods.
992 (reported by Pascal Costanza)
993 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
994 with a type error. (reported by blitz_ on #lisp)
995 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
996 failed with a type error. (reported by Pascal Bourguignon)
997 * bug fix: merging pathnames against defaults with :DIRECTORY
998 starting with '(:RELATIVE :BACK) should preserve the :BACK.
999 (reported by James Y Knight)
1000 * bug fix: saving large (>2GB) cores on x86-64 now works
1001 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1002 constant index and a (simple-array (signed-byte 32)) array
1003 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1004 error (patch by Robert J. Macomber)
1005 * feature: TIME now displays the amount of run-time spent in GC
1006 * bug fix: The debugger now does a better job of respecting
1007 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1008 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1009 workaround for bug 403.)
1010 * bug fix: floating point exception handling now works on FreeBSD
1011 (thanks to NIIMI Satoshi)
1012 * fixed some bugs revealed by Paul Dietz' test suite:
1013 ** MISC.641: LET-conversion were not supposed to work in late
1015 * improvements to the Win32/x86 port:
1016 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1017 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1019 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1020 * new feature: source path information is generated for macro-expansion
1021 errors for use in IDE's like Slime (thanks to Helmut Eller)
1022 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1023 * bug fix: compiler does not loop forever on an invalid type in
1025 * improvement: compilation of most CLOS applications is significantly
1027 * optimization: added a limited bytecode compiler for simple toplevel
1028 forms, speeding up compilation and FASL loading
1029 * bug fix: the statistical profiler now properly distinguishes anonymous
1032 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1033 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1034 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1035 into subclasses of the finalized class. Now user calls to
1036 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1037 system return before any subclasses are finalized.
1038 * minor incompatible change: The reader no longer ignores errors
1039 regarding non-existent packages in #+ and #- feature tests.
1040 * new feature: command line options --no-sysinit, --no-userinit to
1041 inhibit loading the corresponding init files
1042 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1043 for sending data through UDP sockets (thanks to François-René Rideau)
1044 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1045 error is signalled instead (thanks to François-René Rideau)
1046 * minor incompatible change: Error signalling behaviour of lexical
1047 operations violating package locks has changed slightly. Refer to
1048 documentation on package locks for details.
1049 * enhancement: EVAL can process IF-expressions without resorting to the
1051 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1052 constant-folded properly due to missing out-of-line functions.
1053 (reported by Mika Pihlajamäki)
1054 * bug fix: LISTEN sometimes returned T even in cases where no data was
1055 immediately available from the stream
1056 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1057 were reversed. (reported by Levente Mészáros)
1058 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1059 when given a vector displaced to an adjustable array. (reported
1061 * bug fix: derivation of float boundaries from numbers outside the
1062 appropriate float range (reported by John Wiseman)
1063 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1065 * bug fix: precomputing cache entries for generic functions with
1066 some subclasses of specializers as yet invalid does not attempt to
1067 fill a cache line with a negative offset. (reported by Levente
1069 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1070 allow condition class objects as arguments to DOCUMENTATION and
1071 (SETF DOCUMENTATION); only find and set documentation for
1072 structure names for the STRUCTURE doc-type. (suggested by Gary
1074 * improvements to the Win32/x86 port:
1075 ** (user-homedir-pathname) and default initialization file
1076 locations now know about the user's "Documents and Settings"
1077 directory (thanks to Yaroslav Kavenchuk)
1078 ** run-program is implemented (thanks to Mike Thomas)
1079 ** sockets support (thanks to Timothy Ritchey)
1080 ** better backtrace support (thanks to Alastair Bridgewater)
1081 ** sb-grovel supported
1082 ** asdf-install and sb-posix work somewhat
1083 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1084 * improvements to the Solaris/x86 port:
1085 ** works on Solaris 11/Solaris Express
1086 ** floating-point exception handling support
1087 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1088 * fixed some bugs revealed by Paul Dietz' test suite:
1089 ** REMOVE-METHOD returns its generic function argument even when
1090 the method is not one of the generic functions' methods.
1091 ** objects declared as MEMBER types can be admissible arguments to
1092 structure accessors.
1093 ** printing characters should simply be printed by the FORMAT ~:C
1095 ** compiler failure when compiling functions with hairy constant
1096 defaults for optional parameters.
1097 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1098 function, which is already optimized.
1100 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1101 * new platform: experimental support for SBCL x86/Darwin, including
1102 MacOS X 10.4.5 on Intel.
1103 * new feature: Unicode character names are now known to the system
1104 (through CHAR-NAME and NAME-CHAR).
1105 * new feature: the filesystem location of SBCL's core file is
1106 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1107 * minor incompatible change: the contrib modules SB-POSIX and
1108 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1109 this change is to make it easier to distribute
1110 location-independent binaries.
1111 * bug fix: as implied by AMOP, standardized classes no longer have
1112 slots named by external symbols of public packages. (reported by
1114 * optimization: calling functions via constant symbols -- as in
1115 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1116 via the function object as in (FUNCALL #'FOO).
1117 * optimization: on x86 and x86-64, the calling convention for the
1118 case of dealing with an unknown number of values has been altered
1119 to be friendlier to the prediction heuristics implemented,
1120 particularly with reference to CALL and RET pairing. (thanks to
1121 Alastair Bridgewater)
1122 * optimization: on x86, the code for access to thread-local symbol
1123 values and binding/unbinding of thread-local symbols is smaller.
1124 (thanks to Alastair Bridgewater)
1125 * enhancement: CONSTANTP is now able to determine constantness of
1126 more complex forms, including calls to constant-foldable standardized
1127 functions and some special forms beyond QUOTE.
1128 * fixed bug: occasional GC crashes on Solaris/x86
1129 * optimization: x86-64 supports stack allocation of results of simple
1130 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1131 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1132 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1133 (thanks to James Knight)
1134 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1135 by the standard (thanks to Stephen Compall)
1137 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1138 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1139 be used for bundling the runtime and the core file into one
1140 executable binary. This feature is not currently supported on all SBCL
1141 platforms. (thanks to James Bielman and NIIMI Satoshi)
1142 * new feature: a generational or ephemeral garbage collector is now
1143 the default on the PowerPC platform (both Linux and Darwin). The
1144 old Cheney (stop and copy) collector is a build-time option.
1145 (thanks to Cyrus Harmon, after Raymond Toy)
1146 * minor incompatible change: the method by which SBCL finds its
1147 contributed modules has changed; it no longer relies on symbolic
1148 links from an $SBCL_HOME/systems directory, but searches directly
1149 in the subdirectories of $SBCL_HOME.
1150 * enhancement: the dynamic heap size on the Linux/PPC platform is
1151 markedly larger, even using the older Cheney garbage collector.
1152 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1153 certain complicated string types. (reported by Gary King)
1154 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1155 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1156 * fixed bug: DOCUMENTATION on structures no longer signals an
1157 error if no structure type of the right name exists. (reported by
1159 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1160 applicable non-standard primary or :AROUND method on
1161 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1162 no longer cause unbound variable errors. (reported by Kevin Reid)
1163 * fixed bug: metacircle resolution in cases where methods have slots
1164 added before the slots from STANDARD-METHOD. (reported by Jean
1166 * fixed bug: the Power PC assembly code for calling into lisp
1167 sometimes computed its return address wrongly (depending on the
1168 15th bit of the address, fixed at link time). This used to
1169 manifest itself in a segmentation violation while building PCL.
1170 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1171 many others over the years)
1172 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1173 the stack pointer (treated as a signed integer) is negative.
1174 (thanks to Peter van Eynde)
1176 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1177 * new platform: experimental support for the Windows operating
1178 system has been added. (thanks to Alastair Bridgewater)
1179 * fixed several bugs in and robustified the PPC FFI (including
1180 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1181 * bug fix: allow non-simple string symbol names (reported by Paul
1183 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1185 * bug fix: saving a core corrupted callbacks on x86/x86-64
1186 * bug fix: closed a loophole in metacircularity detection and
1187 grounding in the PCL implementation of CLOS.
1188 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1189 (reported by tomppa on #lisp)
1190 * bug fix: clear the direction flag on Lisp -> C transitions, as
1191 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1192 (reported by Andrej Grozin and Hendrik Maryns)
1193 * optimization: major improvements to GC efficiency on GENCGC platforms
1194 * optimization: faster implementation of EQUAL
1195 * optimization: emit more efficient opcodes for some common
1196 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1197 * optimization: several other minor code-generation improvements on x86-64
1198 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1200 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1201 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1202 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1203 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1204 by ANSI as a generic function, and for consistency (SETF
1205 GENERIC-FUNCTION-NAME) is treated likewise.
1206 * fixed bug #233.b: make constraint propagation notice when a variable
1207 value is changed after it is referenced but before it is used
1208 * fixed bug #296: no more arbitrary behaviour from filenames with
1209 odd characters as --load arguments. (reported by Adam Warner)
1210 * fixed bug #390: :CHARACTER-SET pathname components now work as
1211 expected. (reported by Tim Daly Jr)
1212 * fixed bug #391: complicated :TYPE intersections in slot
1213 definitions no longer cause an error in PCL internals.
1214 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1215 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1216 returning the number of octets which would be written to the
1217 file-stream. (thanks to Robert J. Macomber)
1218 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1219 arguments to the internal stream functions. (thanks to David
1221 * bug fix: stack exhaustion is now handled on the x86/SunOS
1222 platform. (thanks to Daisuke Homma)
1223 * bug fix: more accurate ROOM results on GENCGC platforms
1224 * optimization: improved type inference for arithmetic-for
1225 index variables in LOOP
1226 * optimization: faster floating-point SQRT on x86-64
1227 * fixed some bugs revealed by Paul Dietz' test suite:
1228 ** DOCUMENTATION returns NIL instead of "" for method combinations
1229 that don't have a docstring
1231 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1232 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1233 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1234 therefore are not customizeable by user code (as seems to be at
1235 least permitted and maybe required by AMOP). As a consolation,
1236 however, the SBCL implementation of these functions now calls
1237 REINITIALIZE-INSTANCE as specified by AMOP.
1238 * bug fix: slot-definition documentation is propagated as per ANSI
1239 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1240 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1241 argument, as do eslotd initialization methods. (from Pascal
1242 Costanza's "Closer" project)
1243 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1244 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1246 * bug fix: it is now possible to have more than one subclass of
1247 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1248 (reported by Bruno Haible, Pascal Costanza and others)
1249 * bug fix: the dependent update protocol now works for generic
1250 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1251 and Pascal Costanza)
1252 * bug fix: condition-class instances corresponding to
1253 DEFINE-CONDITION forms are now created eagerly. (reported by
1254 Kalle Olavi Niemitalo on comp.lang.lisp)
1255 * bug fix: floating point printing is more accurate in some
1256 circumstances. (thanks to Simon Alexander)
1257 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1258 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1259 * bug fix: callbacks on OS X now preserve stack-alignment.
1260 (thanks to Cyrus Harmon)
1261 * enhancement: the x86-64 disassembler is much better at
1262 disassembling SSE instructions. (thanks to Lutz Euler)
1263 * enhancement: CHECK-TYPE on a variable now allows type-inference
1264 (thanks to Vincent Arkesteijn)
1265 * optimization: improved performance of EUC-JP external format.
1266 (thanks to NIIMI Satoshi)
1267 * optimization: performance improvements to IO on file streams of
1268 :ELEMENT-TYPE CHARACTER
1269 * optimization: much faster memory allocation on x86-64
1270 * optimization: faster garbage collections (latency and throughput)
1272 * optimization: faster bignum subtraction and division on x86 and x86-64
1273 (thanks to Lutz Euler)
1274 * optimization: more accurate type-inference for loops that use a
1275 floating point index variable or a negative step.
1277 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1278 * bug fix: add a workaround to SBCL looping infinitely at startup on
1279 Linux kernels with apparently buggy implementations of personality().
1280 (thanks to Svein Ove Aas)
1281 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1282 (thanks to David Lichteblau)
1283 * bug fix: local bindings shadow global symbol macros for
1284 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1285 * optimization: non-open coded uses of numeric comparison operators
1286 (e.g. >) no longer cons when called with more than one parameter
1287 on platforms supporting dynamic-extent allocation.
1288 * enhancement: saving cores with foreign code loaded is now supported
1289 on MIPS/Linux in addition to the previously supported platforms.
1290 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1291 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1292 * bug fix: division by zero in sb-sprof when no samples were collected
1293 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1294 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1295 (thanks to Alexander Kjeldaas)
1296 * new feature: ignoring the return values of destructive functions
1297 like NREVERSE signals a compile-time style-warning.
1298 (thanks to Kevin Reid)
1299 * new feature: the alignment of alien structure fields can be
1300 explicitly specified. (thanks to Cyrus Harmon)
1301 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1302 and dump core on SIGQUIT
1304 ** incompatible change: threads do not inherit values of specials
1305 from their parents (see manual)
1306 ** bug fix: threads stacks belonging to dead threads are freed by the
1307 next exiting thread, no need to gc to collect thread stacks anymore
1308 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1309 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1310 ** bug fix: dynamic variable and thread start related gc lossage
1311 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1313 * fixed some bugs revealed by Paul Dietz' test suite:
1314 ** SUBTYPEP is slightly more accurate on heinously complicated
1315 CONS types where some of the members have uncertain (in the
1316 NIL, NIL sense) type relationships to each other.
1318 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1319 no longer purified unless :PURIFY T is explicitly specified.
1320 ** Non-purified cores are significantly smaller than before
1322 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1323 * new feature: timers based on Zach Beane's excellent timer package
1324 * added support for the following external formats: koi8-u,
1325 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1326 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1327 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1328 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1329 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1330 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1331 * incompatible change: a threaded SBCL will no longer revert to
1332 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1333 * bug fix: interrupts are disabled until startup is complete; no
1334 more sigsegvs when receiving a signal to soon
1335 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1337 * bug fix: add a workaround for the memory randomization features in
1338 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1339 workaround will only be in effect on systems with the proc filesystem
1341 * bug fix: printing objects of type HASH-TABLE signals a
1342 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1344 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1345 have been running for over 50 days. (reported by Gilbert Baumann)
1346 * bug fix: the logic for getting names of functions gets less
1347 confused when confronded with alternate-metaclass
1348 funcallable-instances. (reported by Cyrus Harmon)
1349 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1350 consistent, even on internal alternate-metaclass objects.
1351 * bug fix: flush closure information collected by physical
1352 environment analysis prepass before the main pass. (bug reported
1354 * bug fix: compiler pack phase does not modify a hashtable, which is
1355 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1356 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1357 may be expanded. (reported by Alexander <malishev> on c.l.l)
1358 * bug fix: no more style-warnings when compiling code using LAST
1359 under high-SPEED low-SPACE settings. (reported by David Wallin
1361 * fixed some incompatibilities between SBCL's MOP and the MOP
1363 ** the METAOBJECT class is now implemented;
1364 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1365 STANDARD-OBJECT, as required;
1366 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1367 now compatible (as required by VALIDATE-SUPERCLASS); there
1368 remains a consistency requirement over the presence or absence
1369 of the FUNCTION class in the superclass list of finalized
1370 classes; see the manual for more details;
1371 ** the :AROUND method for COMPUTE-SLOTS on
1372 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1373 requested slot ordering.
1375 ** bug fix: parent thread now can be gc'ed even with a live
1377 ** bug fix: binding a special with PROGV to no value is not
1378 the same as that symbol not having been bound (thanks to
1380 ** bug fix: binding specials is thread safe (thanks to
1382 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1383 and SB-SPROF do not die with 'no handler for signal XX in
1384 interrupt_handle_now(..)' anymore
1385 ** bug fix: WITH-TIMEOUT works with multiple running threads
1386 * fixed some bugs revealed by Paul Dietz' test suite:
1387 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1388 the :method-class keyword argument.
1390 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1391 * new port: the Solaris operating system on x86 processors is now
1392 mostly supported, though some rough edges in the environment
1393 remain. (thanks to Daisuke Homma)
1394 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1395 space for its heap. (thanks to Thiemo Seufer)
1396 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1397 closures. (thanks to Thiemo Seufer)
1398 * minor incompatible change: eof selects abort in the debugger.
1399 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1400 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1402 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1403 deprecated, and will go away in a future revision of SBCL.
1404 * minor incompatible change: GC-ON and GC-OFF are no longer
1405 implemented with a counter, it does not matter how many times gc
1406 is switched on or off
1407 * bug fix: discriminating functions for generic function classes
1408 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1409 make invalid assumptions about method precedence order. (reported
1411 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1412 platforms. (thanks to Thiemo Seufer)
1413 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1414 * bug fix: correct pathname printing with printer escaping is on.
1415 (thanks to Kevin Reid)
1416 * bug fix: complex VOP definitions in "user-space" no longer trigger
1417 package locks. (reported by Zach Beane)
1418 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1419 for funcallable-instance functions, and is no different from
1421 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1422 characters in the current readtable. (reported by Nicholas Neuss)
1423 * bug fix: SUBTYPEP on various CONS types returns more a more
1424 accurate acknowledgment of its certainty.
1425 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1426 lists in some cases. This partially fixes bug 384.
1427 * flush all standard streams before prompting in the REPL and the
1429 * bug fix: signal handling and triggering gc do not conflict
1430 directly anymore, in particular a high frequency sb-sprof does
1431 not prevent gc from running
1432 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1433 approximation for timezone and DST information between the
1434 universal time epoch and the smallest negative 32-bit time_t.
1435 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1436 year 1899 so long as the encoded time is non-negative.
1437 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1438 the correct answer for non-constant rotate amounts when performing
1439 an inline 32-bit rotation.
1441 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1442 there is only one thread in the session
1443 ** bug fix: memory leak for streams created in one thread and
1444 written to in another
1445 ** bug fix: lockup when compiled with gcc4
1446 ** bug fix: race that allows the gc to be triggered when gc is
1448 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1449 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1451 ** bug fix: finished threads are gc'ed properly
1452 * fixed some bugs revealed by Paul Dietz' test suite:
1453 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1454 has arguments with a different set of applicable methods from
1455 the orignal arguments.
1456 ** The type error thrown by MAP now has a correct expected-type
1458 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1459 name a compiled function.
1460 ** Three MISC tests where a large bignum was improperly coerced to
1461 a float that couldn't represent that bignum during type
1462 derivation were fixed.
1463 ** SUBTYPEP can now handle types involving the negation of a
1464 list-form FUNCTION type.
1465 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1466 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1467 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1469 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1470 * New feature: Experimental support for bivalent streams: streams
1471 opened with :element-type :default now allow character and binary
1472 (unsigned-byte 8) I/O
1473 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1474 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1476 * bug fix: no more highly sporadic "couldn't check whether ~S is
1477 readable" when reading a stream and an interrupt hits in the middle
1478 of a select system call
1479 * compiler better recognizes complex arrays (reported by Eduardo
1481 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1482 for variables on SBCL built with linkage-tables. (reported by Luis
1484 * various error reporting improvements.
1485 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1486 (Thanks to Thiemo Seufer)
1487 * bug fix: nested reader invokations work correctly
1488 * bug fix: it is possible to have simultaneous references to foreign
1489 code and foreign data with the same name.
1491 ** added x86-64 support
1492 ** incompatible change: the threading api now works with thread
1493 objects instead of thread ids
1494 ** bug fix: threads are protected from signals and interruption when
1495 starting up or going down
1496 ** bug fix: a race where an exiting thread could lose its stack to gc
1497 ** bug fix: don't halt on infinite error in threads if possible
1498 ** fixed numerous gc deadlocks introduced in the pthread merge
1499 ** bug fix: fixed thread safety issues in read and print
1500 ** bug fix: debugger doesn't hang on session lock if interrupted at
1501 an inappropriate moment
1502 ** bug fix: run-program is now thread safe(r)
1503 ** bug fix: inner with-recursive-lock no longer releases the mutex
1504 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1505 * fixed some bugs revealed by Paul Dietz' test suite:
1506 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1507 EXPECTED-TYPE slots filled.
1508 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1510 ** MAKE-CONDITION accepts classes as type-designators.
1511 ** COMPILE may never return NIL.
1512 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1513 range before calling Unix time functions
1515 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1516 * numerous signal handling fixes to increase stability
1517 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1518 * minor incompatible change: we now correctly canonize default
1519 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1520 as per the MOP, rather than the historical (INITARG INITFUNCTION
1521 INITFORM). (reported by Bruno Haible)
1522 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1523 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1524 for more information.
1525 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1526 pathname is a directory pathname.
1527 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1528 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1530 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1531 :METACLASS argument in addition to a class name. (reported by
1532 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1533 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1534 (reported by Bruno Haible; patch from Gerd Moellmann)
1535 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1537 * bug fix: more cleanups to the floating point exception handling on
1538 x86-64 (thanks to James Knight)
1539 * bug fix: the compiler does not try to constant fold calls of
1540 COERCE and BIT-* functions when they return freshly constructed
1541 objects. (reported by Nikodemus Siivola and Paul Dietz)
1542 * optimization: improved the MIPS versions of generic (in the
1543 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1544 * optimization: direct conversion of (unsigned-byte 32) to floats on
1545 the PowerPC platform.
1546 * optimization: structure instances with raw slots now use less
1547 memory, and probably show better memory locality. (thanks to
1549 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1550 are treated as requests for stack allocation on the x86-64,
1551 Alpha-32, PPC and SPARC platforms.
1552 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1553 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1554 (thanks to Tony Martinez)
1556 ** gcing a dead thread can no longer lead to lockups
1557 ** threads block signals until they are set up properly
1558 ** errno is no longer shared by threads
1559 ** interrupt-thread restores the eflags register on x86
1560 ** fixed some lockups due to gc/thread interaction
1561 * fixed some bugs revealed by Paul Dietz' test suite:
1562 ** invalid dotted lists no longer raise a read error when
1563 *READ-SUPPRESS* is T
1564 ** signal an error if a symbol that names a declaration is used
1565 as the name of a type, or vice versa
1566 ** allow using the (declare (typespec var*)) abbreviation for
1567 (declare (type typespec var*)) with all type specifiers
1568 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1569 ** always return NIL from PPRINT-POP when OBJECT is NIL
1570 ** don't signal errors when pretty-printing malformed LABELS,
1571 FLET or MACROLET forms
1572 ** declarations in a DOLIST body are also in scope for the
1574 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1576 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1579 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1580 * fixed cross-compiler leakages that prevented building a 32-bit
1581 target with a 64-bit host compiler.
1582 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1583 opened with :IF-EXISTS :OVERWRITE.
1584 * fixed bug 281: error for an invalid qualifier in a short-form method
1585 combination method is not signalled until the faulty method is called.
1586 * bug fix: iteration variable type inferrer failed to deal with open
1587 intervals. (reported by Alan Shields)
1588 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1589 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1590 or more runtime options were provided to the sbcl binary.
1591 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1592 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1593 * optimization: there's now a fast-path for fixnum arguments in the
1594 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1595 * optimization: the code generated on x86-64 is more compact thanks
1596 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1597 * bug fix: floating-point exception handling now partly works on
1598 x86-64. (Thanks to James Knight)
1599 * improvement to the MIPS backend from Thiemo Seufer: C-style
1600 64-bit long long arguments and return values to alien functions
1602 * fixed some bugs revealed by Paul Dietz' test suite:
1603 ** the type-error signalled from WARN has a filled-in DATUM slot.
1604 ** the type-error required when a stream is not associated with
1605 a file has the stream as its datum.
1606 ** type-errors on single-floats on x86-64 no longer have
1607 :INVALID-OBJECT as the datum
1608 ** the type-errors signalled for invalid function names now have
1609 a correct expected type
1610 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1611 for structures defined with a :TYPE. Documentation strings for
1612 typed structures are no longer immediately discarded
1613 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1614 broadcast streams and synonym streams. FILE-LENGTH now also works
1615 on broadcast streams.
1617 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1618 * incompatible change: the --noprogrammer option, deprecated since
1619 version 0.7.5, has been removed. Please use the equivalent
1620 --disable-debugger option instead.
1621 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1623 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1624 inoperational for a while now) has been completely removed.
1625 * null lexical environments are now printed as #<NULL-LEXENV>,
1626 significantly reducing the amount of clutter in typical backtraces.
1627 * documentation on weak pointers, finalization, and after GC hooks
1628 has been added to the manual.
1629 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1630 as other specialized array types, is much faster. SUBSEQ and
1631 COPY-SEQ on such arrays have also been sped up.
1632 * optimization: EQL is now more efficient when at least other argument
1633 is known to be of type (OR FIXNUM (NOT NUMBER)).
1634 * fixed bug: the runtime is now less vulnerable to changes in the
1635 size of the SBCL object on OS X, and virtual memory is reserved for
1636 all spaces that need to be at a fixed address.
1637 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1638 * fixed bug: finalizers and after GC hooks that cause consing are now
1640 * fixed bug: compiler error messages and summaries are now printed to
1641 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1642 * fixed inference of the upper bound of an iteration variable.
1643 (reported by Rajat Datta).
1644 * fixed bug 211e: calling local functions with duplicated constant
1645 keyword argument no longer causes a bogus style warning about an
1647 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1648 ftype declaration to be disregarded. (reported by Dave Roberts)
1649 * fixed bug 373: caused by erronous compilation of references to alien
1650 variables in the runtime on ppc/darwin.
1651 * fixed bug 376: CONJUGATE type deriver.
1652 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1653 (reported by Baughn on #lisp)
1654 * a cross-compiler bug on non-x86 platforms has been identified and
1655 fixed. (thanks to Bruno Haible)
1656 * improvements to the MIPS runtime code for increased stability.
1657 (thanks to Thiemo Seufer)
1658 * increased the maximimum compact environment size to allow
1659 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1660 * partial workaround for bug 135: don't name the function we're
1661 calling for hairy cases of EVAL, so as not to accumulate one environment
1662 entry per call to EVAL. (reported by Kevin Reid)
1663 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1664 * optimization: type testing for non-vector arrays should be faster.
1665 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1667 * bug fix: setting 31st element of a bit vector to zero did not work
1669 * bug fix: redefining a class definition which failed due to a
1670 previous accessor / function clash now works (but see BUGS entry
1671 #380 for more problems in this area). (thanks to Zach Beane)
1672 * the long form of DEFINE-METHOD-COMBINATION disables method group
1673 checking when given a single method group with pattern *. (thanks
1675 * on x86 compiler supports stack allocation of results of simple
1676 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1677 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1678 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1679 * fixed some bugs related to Unicode integration:
1680 ** the restarts for recovering from input and output encoding
1681 errors only appear when there is in fact such an error to
1683 * fixed some bugs revealed by Paul Dietz' test suite:
1684 ** MISC.549 and similar: late transformation of unsafe type
1685 assertions into derived types caused unexpected code
1687 ** SCALE-FLOAT type deriver is less wrong.
1688 ** type derivers for EXP, LOG and similar functions compute result
1689 types for complex arguments better.
1690 ** (MISC.563) CONJUGATE type deriver works for very restricted
1692 ** out-of-line type testers for character strings are available.
1693 ** EQUAL compiler transform understands specialness of objects
1695 ** accessing double-floats stored on the stack now works on x86-64.
1696 ** debugger internals could sometimes create invalid lispobjs,
1697 resulting in GC crashes.
1698 ** MISC.548: type check weakening can convert required type into
1700 ** initialization forms for bindings are not in scope of free special
1703 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1704 * incompatible change: thread support for non-NPTL systems has
1705 been removed - locking is buggy and unreliable. A threaded
1706 SBCL build will now warn at startup and refuse to create
1707 new threads, unless futex support is detected in the kernel
1708 * incompatible change: the top level REPL now has only an ABORT
1709 restart associated with it, not TOPLEVEL and ABORT as it used to.
1710 TOP and TOPLEVEL are now available as debugger commands for
1711 returning to the top level.
1712 * incompatible change: forms evaluated in the REPL now use the
1713 global optimization policy.
1714 * incompatible change: user- and system-initialization files are
1715 no longer processed with LOAD, but by READ and EVAL; hence the
1716 global optimization policy, startup package, readtable, etc,
1718 * The .fasl file format number has been incremented because of
1719 various incompatible changes.
1720 * internal entry point details and argument counts no longer appear
1721 in backtraces unless explicitly requested by setting
1722 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1723 * built-in and standard functions no longer have names like "top
1724 level local call to FOO".
1725 * fixed bug 32: functions defined in non-null lexical environments
1726 now have more legible printed representation
1727 * fixed bug 33: functions defined in non-null lexical environemnts
1728 are now more amenable to inspection by INSPECT.
1729 * workaround for bug 354: XEPs no longer appear in backtraces unless
1730 explicitly requested.
1731 * fixed bug: receiving the signal which results from attempting to
1732 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1733 SIGBUS on other *BSDs) on architectures where the C stack is also
1734 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1735 to debugger confusion or wild instability. (reported by Cheuksan
1737 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1738 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1739 (reported by Lutz Euler)
1740 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1741 compiled code. (reported by Paul Dietz)
1742 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1743 specialized parameter to a method to something that is not TYPEP
1744 the specializer is now possible.
1745 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1746 face of package deletion.
1747 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1748 pathname. (reported by David Tolpin on c.l.l)
1749 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1750 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1751 STRING-STREAM and FILE-STREAM, respectively.
1752 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1753 than previously; in addition, its test suite is now run on build.
1754 (thanks to Luke Gorrie)
1755 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1756 * optimization: added a immediate representation for single-floats
1758 * fixed some bugs revealed by Paul Dietz' test suite:
1759 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1760 ** attempting to create a package with a colliding nickname causes
1761 correctable errors to be signalled.
1762 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1763 ** several x86-64 backend bugs related to sign-extension of immediate
1766 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1767 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1768 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1770 * fixed loading of multiply forward-referenced layouts.
1771 (thanks to Cheuksan Wang)
1772 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1773 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1774 (see COMPILE-FILE documentation for details of :PRINT option)
1775 * fixed bugs 19 and 317: fixed-format floating point printing is
1776 more accurate. This also fixes a bug reported by Adam Warner
1777 related to the ~@F format directive.
1778 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1780 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1781 dispatch table if the from-char is a dispatch macro character.
1782 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1783 by the wordsize no longer ignore the last word. (reported by Lutz
1785 * fixed bug in type checking of dynamic-extent variables. (reported
1787 * optimization: sequence traversal functions use their freedom to
1788 coerce function designators to functions.
1789 * optimization: code with many calls to CLOS methods specialized on
1790 CLOS classes has had redundant type checks removed.
1791 * optimization: added declarations to speed up operations that access
1792 the internal character database (for example STRING-UPCASE)
1793 * optimization: comparison operations between floats and sufficiently small
1794 fixnums no longer create extra rationals
1795 * fixed some bugs related to Unicode integration:
1796 ** portions of multibyte characters at the end of buffers for
1797 character-based file input are correctly transferred to the
1798 start of the buffer at the next read.
1799 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1800 passing it through to OPEN.
1801 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1802 argument given to internal calls to OPEN.
1803 ** the built-in comment readers (introduced by character sequences
1804 ";" and "#|") are more forgiving to encoding errors; they will
1805 STYLE-WARN and then attempt to resync the stream at a character
1806 boundary. (thanks to Teemu Kalvas)
1807 * fixed some bugs in the x86-64 port:
1808 ** Negative short int return values from c-calls are sign-extended
1810 ** The stack is aligned to 16-bytes for c-calls, as required by
1811 the ABI. (thanks to Cheuksan Wang)
1812 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1813 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1814 for enums. (thanks to Vincent Arkesteijn)
1815 ** Multiple small optimizations and bugfixes for floating point
1817 * fixed some bugs revealed by Paul Dietz' test suite:
1818 ** Space, Tab, Linefeed, Return and Page have the invalid
1819 secondary constituent character trait.
1820 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1822 ** WITH-INPUT-FROM-STRING should only update the index place on
1824 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1825 works more reliably.
1826 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1827 with one-dimensional arrays with fill pointers.
1828 ** TYPE-OF failed on a complex with an integer realpart and a
1830 ** compiler failure during type inference for the code of form
1831 (IF (EQL X (THE ...)) ...) (MISC.535).
1833 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1834 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1835 platform. The port supports 61-bit fixnums, large memory spaces
1836 and reloading shared object files.
1837 * enhancement: saving cores with foreign code loaded is now
1838 supported on ppc/Darwin in addition to the previously supported
1840 * enhancement: the statistical profiler now walks deeper into the
1841 call stack for more meaningful call-graphs and accrued time
1842 reports (x86/x86-64 only). It also now reports time spent in
1844 * enhancement: it is now possible to trace most individual methods
1845 of a generic function in addition to tracing the generic function
1847 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1849 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1850 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1851 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1852 produces an error. (thanks to Vincent Arkesteijn)
1853 * bug fix: NAMESTRING on pathnames with :WILD components in their
1854 directories works correctly. (thanks to Artem V. Andreev)
1855 * fixed bug 125: compiler preserves identity of closures. (reported
1857 * bug fixed: functions with &REST arguments sometimes failed with
1858 "Undefined function" when compiled with (DEBUG 3). (reported by
1860 * bug fix: overflow during compiling of setting element of a bit
1861 vector with constant index and value. (reported by Timmy Douglas)
1862 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1863 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1864 * on x86 compiler partially supports stack allocation of dynamic-extent
1866 * GO and RETURN-FROM do not check the extent of their exit points
1867 when compiled with SAFETY 0.
1868 * fixed some bugs related to Unicode integration:
1869 ** encoding and decoding errors are now much more robustly
1870 handled; it should now be possible to recover even from invalid
1871 input or output to the terminal. (thanks to Teemu Kalvas)
1872 ** provided a first cut at implementing STRING-TO-OCTETS and
1873 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1874 ** altered the SB-MD5 contributed module slightly, changing the
1875 interface just enough for it to be supportable for builds where
1876 lisp characters are not eight bits.
1877 * fixed some bugs revealed by Paul Dietz' test suite:
1878 ** the FORMATTER-generated functions for ~V[ conditionals require
1879 the correct number of arguments.
1880 ** READ-FROM-STRING returns the mandated second value when applied
1881 to displaced strings.
1882 ** the #\Rubout and #\Backspace characters are treated as invalid
1883 constituent characters by the tokenizer.
1885 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1886 * new feature: reloading changed shared object files with
1887 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1888 * new feature: references to foreign variables and functions
1889 can now be compiled and loaded before the corresponding shared
1890 object file is loaded, as long as the foreign definitions are
1891 available at runtime.
1892 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1893 supported. (thanks to Dan Debertin)
1894 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1895 just keywords. (thanks to Vincent Arkesteijn)
1896 * fixed bug #331: structure-class instances corresponding to
1897 DEFSTRUCT forms are now created eagerly.
1898 * fixed bug #345: backtraces from calls to undefined functions work
1899 on x86 as well. Related bug #61 is now also partially fixed on x86
1900 (backtraces from throws to unknown catch tags.)
1901 * bug fix: lambda-list parsing is now stricter vrt. order and number
1902 of lambda-list keywords.
1903 * bug fix: as specified by AMOP, an error is signalled if a
1904 class-option appears multiple times in a DEFCLASS form. (reported
1906 * bug fix: RANDOM can be compiled when the compiler derives the type
1907 of its numeric argument as a disjoint set of small integers.
1908 (reported by Paul Dietz)
1909 * bug fix: starting a core saved with shared objects loaded when
1910 those objects are not available no longer causes threaded SBCL to
1911 hang. (reported by Sean Ross)
1912 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1914 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1915 argument, which must always be NIL. (reported by Kalle Niemitalo)
1916 * bug fix: printing 1.0d+23 no longer results in an error.
1917 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1918 * bug fix: structure slot setters preserve evaluation order. (thanks
1920 * fixed some bugs related to Unicode integration:
1921 ** RUN-PROGRAM can allow its child to take input from a Lisp
1922 stream. (reported by Stefan Scholl)
1923 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1924 locales. (reported by Ken Causey)
1925 * fixed some bugs revealed by Paul Dietz' test suite:
1926 ** INCF, DECF and REMF evaluate their place form as specified in
1928 ** FORMATTER expands ~{ ~} iteration directives with V or #
1929 parameters correctly.
1930 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1931 consequent uses no arguments correctly.
1932 ** the system has a partial understanding of the (COMPLEX RATIO)
1934 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1937 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1938 * new feature: a build-time option (controlled by the :SB-UNICODE
1939 keyword feature, enabled by default) for building the system with
1940 support for the entire 21-bit character space defined by the
1942 * new feature: the system now has rudimentary external-format
1943 support; the primary user-visible change at this time is that
1944 characters with the high bit set (such as lower-case-e-acute) will
1945 print correctly to a terminal in a UTF-8 environment.
1946 * minor incompatible change: BASE-CHAR no longer names a class;
1947 however, CHARACTER continues to do so, as required by ANSI.
1948 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1949 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1950 has been moved to the SB-EXT package (temporarily re-exported from
1952 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1953 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1955 * fixed bug #351: better error-handlind and reporting for malformed
1957 * fixed bug #350: bignum-printing is now more memory-efficient,
1958 allowing printing of very large bignums, eg. (expt 2 10000000).
1959 (reported by Bruno Haible)
1960 * fixed bug #302: better primitive-type selection for intersection
1962 * fixed bug #308: non-graphic characters now all have names, as
1963 required. (reported by Bruno Haible)
1964 * bug fix: redefining a class with different superclasses now correctly
1965 removes it from the direct-subclasses of its previous superclasses.
1966 (reported by David Morse)
1967 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1968 new value now works. (reported by Bruno Haible)
1969 * bug fix: correct canonicalization of multiple non-standard slot
1970 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1971 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1972 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1973 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1974 now exists, an signals an error.
1975 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1976 during macroexpansion for non-iteration variables as well. (reported
1977 by Bruno Haible for CMUCL)
1978 * bug fix: Cyclic structures and unprintable objects in compiler
1979 messages no longer cause errors. (reported by Bruno Haible)
1980 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1981 and READ-FROM-STRING all now return a primary value of NIL if
1982 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1983 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1984 (reported by Bruno Haible for CMUCL)
1985 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1986 do not designate a condition. (reported by Bruno Haible for
1988 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1989 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1990 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1991 permitting resolution in favour of any of the conflicting symbols.
1992 (reported by Bruno Haible for CMUCL)
1993 * bug fix: EQUAL compiler optimizations is less aggressive on
1994 strings which can potentially compare true despite having distinct
1995 specialized array element types.
1996 * bug fix: unit enumerations can be defined without dividing by
1997 zero. (thanks to Vincent Arkesteijn)
1998 * FORMAT compile-time argument count checking has been enhanced.
1999 (report from Bruno Haible for CMUCL)
2000 * a partial workaround for the bug 262: the compiler does not try to
2001 inline-expand a local function doing RETURN-FROM from a deleted
2002 BLOCK. (thanks to Peter Denno for the bug report and to David
2003 Wragg for the simple test case)
2004 * fixed some bugs revealed by Paul Dietz' test suite:
2005 ** RENAME-PACKAGE allows all package designators as new package
2007 ** constraint propagation and lambda variable substitution are
2008 more cautious in dealing with partially deleted code.
2009 ** compiler sometimes lost reoptimization passes.
2010 ** CERROR, given a condition as condition designator, passes its
2011 remaining arguments to the continue format control without
2013 ** Case-altering FORMAT directives work correctly on non-ASCII
2015 ** The REAL type specifier handles bounds outside the floating
2016 point ranges without signalling FLOATING-POINT-OVERFLOW.
2017 ** Functions with IR1-transformations can create intercomponent
2018 references to global functions.
2019 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2021 ** FORMAT ~R treats a nil value for its first parameter correctly.
2023 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2024 * enhancement: saving cores with foreign code loaded is now
2025 supported on x86/NetBSD and sparc/Linux in addition to the previously
2026 supported platforms.
2027 * bug fix: on some platforms repeated installations caused multiple
2028 copies of HTML documentation to be installed -- should not happen
2029 any more. (reported by Stefan Scholl)
2030 * bug fix: parsing self-recursive alien record types multiple times
2031 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2032 original patch by Helmut Eller for CMUCL)
2033 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2034 (thanks to Richard Kreuter)
2035 * bug fix: defining classes whose accessors are methods on existing
2036 generic functions in other (locked) packages no longer signals
2037 bogus package lock violations. (reported by François-René Rideau)
2038 * bug fix: special variables as DEFMETHOD parameters no longer have
2039 associated bogus type declarations. (reported by David Wragg and
2041 * bug fix: read-write consistency on streams of element-type
2042 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2043 * bug fix: redefiniton of the only method of a generic function with
2044 no DEFGENERIC no longer emits a full WARNING. In addition,
2045 redefinition of generic functions with no DEFGENERIC to an
2046 incompatible lambda list now signals an error. (thanks to Zach
2048 * bug fix: DEFGENERIC now works even when there's a function of the
2049 same name in an enclosing lexical environment. (thanks to Zach
2051 * fixed compiler failure, caused by instrumenting code during
2052 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2053 * optimization: added loop analysis and improved register allocation
2054 * fixed some bugs revealed by Paul Dietz' test suite:
2055 ** POSITION on displaced vectors with non-zero displacement
2056 returns the right answer.
2057 ** (SIMPLE-STRING) is a valid type specifier for sequence
2059 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2061 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2062 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2064 ** The justification version of the FORMAT ~< directive treats
2065 non-zero minpad parameter correctly.
2067 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2068 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2069 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2070 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2071 the supported interface.
2072 * new feature: Single-stepping of code compiled with DEBUG 2 or
2073 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2074 * new feature: saving cores with foreign code loaded is now
2075 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2076 Timothy Moore's work for CMUCL)
2077 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2078 parameters to * instead of NIL if no initform is supplied.
2079 (reported by Johan Bockgaard)
2080 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2081 correctly when FUNCALL appears as the car of the form. Note:
2082 despite this FUNCALL forms are not currently subject to
2083 compiler-macro expansion. (port of Raymond Toy's fix for the
2084 same from CMUCL, reported by Johan Bockgaard)
2085 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2086 (thanks for Teemu Kalvas)
2087 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2088 case instead of terminating the iteration (thanks for Julian
2089 Squires, Sean Champ and Raymond Toy)
2090 * bug fix: incorrect expansion of defgeneric that caused a style
2091 warning. (thanks for Zach Beane)
2092 * bug fix: slot accessor effective method computation works properly
2093 for classes with multiple non-standard applicable methods on
2094 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2095 * on x86 compiler supports stack allocation of results of LIST and
2096 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2097 CMUCL implementation by Gerd Moellmann)
2098 * fixed some bugs revealed by Paul Dietz' test suite:
2099 ** FORMAT strings with both the ~<~:;~> form of the justification
2100 directive and pretty-printing directives cause an error.
2102 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2103 * incompatible change: the internal functions
2104 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2105 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2106 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2107 instead of the old functions.
2108 * new feature: on platforms where "dladdr" is available foreign
2109 function names now appear in backtraces. (based on Helmut Eller's
2111 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2112 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2114 * documentation: ASDF manual is now distributed with SBCL.
2115 * bug fix: SBCL can now load its contributed modules with REQUIRE
2116 even if the system-provided entries have been removed from
2117 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2119 * bug fix: backtraces involving undefined functions or assembly
2120 routines are more informative. (thanks to Brian Downing)
2121 * bug fix: mutually referent alien structures now work correctly.
2122 (reported by Rick Taube)
2123 * bug fix: structures defined by WITH-ALIEN can be referred to
2124 within other definitions in the same WITH-ALIEN.
2125 * bug fix: division operators (MOD, TRUNCATE and the like) with
2126 constant zero divisors and integer dividends no longer generate
2128 * bug fix: provide default methods for INPUT-STREAM-P and
2129 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2130 * bug fix: improve the detection and resolution of MOP metacycles
2131 (where effective method computation depends on the generic function
2132 having its effective method computed). (reported by Bruno Haible)
2133 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2134 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2135 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2136 * bug fix: pathnames with relative directory components can be
2137 represented relative to default pathnames.
2138 * optimization: in taking the GCD of bignums, reduce the two bignums
2139 to approximately the same size (using Euclid's algorithm) before
2140 applying the more sophisticated binary GCD. (thanks to Juho
2142 * optimization: COUNT on bitvectors now operates word-at-a-time.
2143 * optimization: ASH with a positive, but not necessarily constant,
2144 (leftwards) shift, when in a modular context, is compiled to a
2146 * fixed some bugs revealed by Paul Dietz' test suite:
2147 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2148 if the corresponding argument is NIL.
2149 ** FORMAT directives accepting a minpad parameter treat negative
2151 ** literal commas as character parameters to FORMAT directives are
2153 ** literal spaces directly after ~<Newline> directives within a
2154 format-logical-block (~:< ~@:>) do not induce :FILL-style
2155 conditional newlines.
2156 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2157 as specified: it no longer includes conditional newlines).
2158 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2160 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2161 :PER-LINE-PREFIX argument does not evaluate to a string.
2163 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2164 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2165 the manual for details; for now, package locks can be disabled by
2166 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2167 if no major problems are found then it is likely that they will be
2168 compiled in unconditionally.
2169 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2170 unsupported operators on all platforms. To load a shared library
2171 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2172 object file, link it into a shared library outside of SBCL and
2173 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2175 * bug workaround: ROOM T and the GC were not working together
2176 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2177 reporting facility aren't preserved. That reporting has been
2178 disabled, so now until and unless someone figures out how to make it
2179 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2180 * minor incompatible change: as threatened around sbcl-0.8.0, the
2181 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2182 The symbols are also exported from SB-PCL for backwards
2183 compatibility, but more so than before SB-PCL should be treated as
2184 an implementation-internal package.
2185 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2187 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2188 It is known as of this release not to work on the Alpha, however.
2189 * fixed bug #167: errors signalled due to illegal syntax in method
2190 bodies are now more legible.
2191 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2192 designators and can hence be used with TYPEP.
2193 * fixed bug #333: CHECK-TYPE now ensures that the type error
2194 signalled, if any, has the right object to be accessed by
2195 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2196 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2197 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2199 * fixed bug #334: programmatic addition of slots using specialized
2200 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2201 and :ALLOCATION :CLASS slots.
2202 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2203 float exponents. (rereported by Peter Seibel)
2204 * fixed a bug: #\Space (and other whitespace characters) are no
2205 longer considered to be macro characters in standard syntax by
2206 GET-MACRO-CHARACTER.
2207 * fixed bug: initialization of condition class metaobjects no longer
2208 causes an instance of the condition to be created. (reported by
2210 * fixed bug: it is now possible to have slots such that
2211 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2212 description is neither :INSTANCE nor :CLASS.
2213 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2214 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2215 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2216 non-standard SLOT-DEFINITION-ALLOCATION values.
2217 * fixed bugs: various race conditions handling exiting threads.
2218 CL-PPCRE's thread torture test now passes.
2219 * fixed arguably-a-bug: GC time proportional to number of threads in
2220 system even when most of them are idle
2221 * optimization: improved performance of BIT and SBIT on bit-vectors.
2222 * .fasl file incompatibility: The fasl file version number has
2223 been incremented because of changes associated with package locks.
2225 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2226 * minor incompatible change: the system no longer provides
2227 optimization hints (conditions which are TYPEP
2228 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2229 mode; these hints are still emitted when higher SPEED optimization
2231 * new contrib module: a sampling profiler (profiling by statistical
2232 sampling, rather than by instrumenting functions) is available as
2233 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2234 * the behaviour of the standard function ED is now customizeable by
2235 third parties through a hook variable: see ED's documentation
2236 string for information on the protocol.
2237 * the compiler no longer emits efficiency notes for (FUNCALL X)
2238 when the type of X is uncertain under default optimization
2240 * fixed bug 276: mutating a binding of a specialized parameter to a
2241 method to something that is not TYPEP the specializer is now
2243 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2244 exhibit the required behaviour on the x86 platform. (thanks to
2245 Peter van Eynde, Eric Marsden and Bruno Haible)
2246 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2247 even for difficult arguments. (reported by Peter Graves)
2248 * fixed bug 141a: the backquote printer now descends quoted
2250 * fixed another bug in backquote printing: no more destructive
2251 modification of the form's list structure. (reported by Brian
2253 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2254 move between its address being taken and the call to
2255 interrupt_thread, fixing a crashing race condition.
2256 * the SB-POSIX contrib implementation has been adjusted so that it
2257 no longer exhibits ridiculously poor performance when constructing
2258 instances corresponding to C structs.
2260 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2261 * minor incompatible change: the sb-grovel contrib now treats C
2262 structures as alien (in the sense of SB-ALIEN) objects rather than
2263 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2264 has implications for memory management of client code
2265 (sb-grovel-returned objects must now be manually managed) and for
2266 type safety (alien objects now have full types).
2267 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2268 used to control emission of compiler diagnostics, rather than the
2269 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2270 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2271 quality should be considered deprecated.
2272 * install.sh now installs the user manual as well
2273 * (not quite a new documentable feature, but worth considering in
2274 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2275 beginnings of a semantically meaningful condition hierarchy is
2276 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2278 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2279 designator as the defaults argument.
2280 * fixed bug: Displaced arrays whose displaced-to array has become
2281 too small now cause ARRAY-DIMENSION to signal an error, providing
2282 for safer bounds-checking. (reported by Bruno Haible)
2283 * fixed bug: DEFCLASS slot definitions with identical :READER and
2284 :WRITER names now signal a reasonable error. (reported by Thomas
2286 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2288 * fixed bug: Invalid slot specification errors now print correctly.
2289 (thanks to Zach Beane)
2290 * fixed bug 320: Shared to local slot value transfers in class
2291 redefinitions now happen corrently from superclasses as
2292 well. (reported by Bruno Haible)
2293 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2295 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2296 improper lists correctly. (reported by Bruno Haible)
2297 * fixed bug 313: source-transform for <fun-name> was erroneously
2298 applied to a call of a value of a variable with name <fun-name>.
2299 (reported by Antonio Menezes Leitao)
2300 * fixed bug 307: The obsolete instance protocol ensures that
2301 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2302 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2303 remove dead unknown-values globs from the middle of the stack.
2304 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2305 easier to resume long computations after using *BREAK-ON-SIGNALS*
2306 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2307 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2308 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2309 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2310 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2311 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2313 * fixed a (fairly theoretical) bug in string printing: if
2314 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2315 does not have array-element-type equal to the most general string
2317 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2318 function name. (thanks to Nikodemus Siivola)
2319 * fixed bug: compile-time format string checker failed on
2320 non-closed ~{. (reported by Thomas F Burdick)
2321 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2322 don't warn on legal loop code involving more than one aggregate
2323 boolean. (thanks to Nikodemus Siivola)
2324 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2325 in keyword position to MAKE-PACKAGE was not being reported as
2327 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2328 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2330 * fixed bugs 280 and 312: the checking for multiple definitions in a
2331 file is less likely to become confused by uses of inline
2333 * fixed bug: the #S reader macro performs the keyword coercion
2334 specified for slot names. (reported by Kalle Niemitalo)
2335 * fixed bug: lambda lists may contain symbols whose names start with
2336 & but are not lambda-list-keywords; their occurrence triggers a
2338 * fixed bug 321: define-method-combination argument lambda lists do
2339 not cause invalid code to be generated when &optional or &aux
2340 variables have default values. (reported by Bruno Haible)
2341 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2342 have CLOS classes; this appears also to have fixed reported
2343 instability in stack exhaustion detection.
2344 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2345 MUFFLE-WARNING when no associated restart is present is now
2347 * optimization: rearranged the expansion of various defining macros
2348 so that each expands into only one top-level form in a
2349 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2351 * optimization: used a previously little-used slot in symbols to
2352 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2354 * fixed some bugs revealed by Paul Dietz' test suite:
2355 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2357 ** Fixed an optimization bug related to inheritance of initforms
2358 from local to shared slots.
2359 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2360 ** CAST optimizer forgot to flush argument derived type.
2361 ** print/read consistency on floats is now orders of magnitude
2362 more likely. (thanks also to Bruno Haible for a similar report
2364 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2365 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2366 for objects of type REAL. Make it so.
2367 ** SXHASH is specified (infelicitously) to respect similarity,
2368 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2369 it so. (thanks to Markus Ziegler)
2370 ** on the Alpha, the compiler succeeds in compiling functions
2371 returning a known number of arguments greater than 63.
2372 ** fixed handling of invalid NIL arguments in keyword position
2374 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2376 ** provided a readably-printable representation for RANDOM-STATE
2378 ** ensured that pathnames, where they have a namestring, always
2379 print using #P"..." syntax.
2381 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2382 * Support for the forthcoming 2.0 version of the NetBSD kernel
2383 running on x86 hardware has been added. (thanks to Perry
2384 E. Metzger most immediately, and others for their past work)
2385 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2386 current release version; SBCL's OpenBSD support had been broken
2387 since about the time of OpenBSD's switch to ELF binary format.)
2388 * [placeholder for DX summary]
2389 ** user code with &REST lists declared dynamic-extent, under high
2390 speed or space and low safety and debug optimization policy.
2391 * The manual has been converted to Texinfo format and the debugger
2392 chapter from the cmucl manual has been added.
2393 * A facility has been added to extract documentation strings from
2394 sbcl and store them as Texinfo-formatted snippets for inclusion in
2395 the manual (via Texinfo's @include directive)
2396 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2397 the test case to Dave Roberts)
2398 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2399 pointers. (reported by Sean Ross)
2400 * bug fix: PROFILE output is printed nicely even for large numerical
2401 values. (thanks to Zach Beane)
2402 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2403 greater than 32 handle EOF correctly.
2404 * bug fix: on X86 an immediate argument of the IMUL instruction is
2405 correctly printed in disassembly. (thanks to Lutz Euler)
2406 * bug fix: class slots in redefined classes preserve their old
2407 values. (thanks to Bruno Haible and Nikodemus Siivola)
2408 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2409 succeeds. (reported by Marco Baringer)
2410 * fixed some bugs revealed by Paul Dietz' test suite:
2411 ** READ-SEQUENCE now works on ECHO-STREAMs.
2412 ** RATIONALIZE works more according to its specification. (thanks
2415 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2416 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2417 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2418 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2419 should matter to you only if you rebind the printer control
2420 variables and then find you want different bindings in the
2421 debugger than in the ordinary execution of your program.)
2422 * The runtime build system has been tweaked to support building
2423 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2424 assembler and linker. (thanks to Nikodemus Siivola)
2425 * Unbound, undefined, undeclared variables now trigger full
2426 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2427 more often programmer error than deliberate exploitation of undefined
2429 * optimization: the hash algorithm for strings has changed to one
2430 that is less vulnerable to spurious collisions. (thanks to Juho
2432 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2433 needless bounds checking. (thanks to Juho Snellman)
2434 * optimization: implemented multiplication as a modular
2435 (UNSIGNED-BYTE 32) operation on the PPC backend.
2436 * fixed some bugs revealed by Paul Dietz' test suite:
2437 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2439 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2440 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2441 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2443 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2445 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2447 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2448 their output stream on EOF from read.
2449 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2450 have been read to end-of-file.
2451 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2453 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2454 description of determination of which consecutive characters
2456 ** Printing the "Space" character with escaping on now yields "#\\ ",
2457 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2458 ** Reading floating-point numbers with *READ-BASE* set to a number
2459 less than 10 works correctly.
2460 ** Reading floating-point numbers with *READ-BASE* set to a number
2461 more than 10 works correctly.
2462 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2463 the readtable currently in effect.
2465 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2466 * minor incompatible change: parsing of namestrings on a physical
2467 (Unix) host has changed; numbers after the final #\. in a
2468 namestring are no longer interpreted as a version field. This is
2469 intented to be largely invisible to the user, except that the
2470 meaning of the namestring "*.*.*" has changed: it now refers to a
2471 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2472 should usually be replaced by
2473 (make-pathname :name :wild :type :wild :version :wild)
2474 with the added benefit that this is more likely to be portable.
2475 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2476 option to OPEN now signals an error if the file being opened
2477 exists; this may have an impact on existing code.
2478 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2479 signals received, on the PowerPC platforms (both Linux and
2480 Darwin). (thanks to Pierre Mai for pointing out the location of
2482 * several fixes on OS X: The system now builds and runs cleanly on
2483 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2484 (thanks to Brian Mastenbrook)
2485 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2486 second-resolution: integer multiples of 1/3600 between -24 and 24.
2487 (thanks to Vincent Arkesteijn)
2488 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2489 type when called with 1 argument; PEEK-CHAR checked type of
2490 PEEK-TYPE only after having read first character from a
2491 stream. (reported by Peter Graves)
2492 * bug fix: the garbage collector now has much better locality
2493 behaviour, and in particular no longer treats all memory as being
2494 exhausted when many small objects point to each other in a deeply
2496 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2497 recognized as being TYPEP their class.
2498 * bug fix: the PUSHNEW documentation string has been corrected.
2499 (thanks to Vincent Arkesteijn)
2500 * bug fix: defaulting of the value for the last of an atypically
2501 large number of multiple values being bound was not being
2502 performed correctly on the Alpha or PPC platforms
2503 * optimization: implemented multiplication as a modular
2504 (UNSIGNED-BYTE 32) operation on the x86 backend.
2505 * optimization: SEARCH on simple-base-strings can now be open-coded.
2506 (see also contrib/compiler-extras.lisp for inspiration for
2507 teaching the compiler about the Boyer-Moore algorithm).
2508 * value, returned by MAX (and MIN) called with several EQUALP, but
2509 not EQL, arguments now does not depend on compiler settings.
2510 * fixed some bugs revealed by Paul Dietz' test suite:
2511 ** in stack analysis liveness information is propagated from
2512 non-local entry points.
2513 ** pathwise CAST removing failed when the CAST node did not start
2515 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2516 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2518 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2519 host is already defined.
2520 ** RENAME-FILE works on streams instead of signalling an internal
2522 ** PEEK-CHAR uses the current readtable when determining whether
2523 or not a character is whitespace.
2524 ** MERGE-PATHNAMES handles the case when the pathname does not
2525 specify a name while the default-pathname specifies a version.
2526 ** Pathnames now stand a better chance of respecting print/read
2528 ** Attempting to use standardized file system operators with a
2529 pathname with invalid :DIRECTORY components signals a
2531 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2533 ** DIRECTORY on logical pathnames is more correct.
2534 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2535 signal a TYPE-ERROR if their argument is not a stream.
2536 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2537 designator argument does not designate a stream.
2538 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2539 examining the synonym.
2540 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2542 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2543 element-type larger than ([UN]SIGNED-BYTE 32).
2545 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2546 * When built with the :SB-FUTEX feature, threaded builds now take
2547 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2548 for faster/more reliable mutex and condition variable support.
2549 * Incompatible change (but one you probably shouldn't have been using
2550 anyway): the interface and code for arbitrating between multiple
2551 threads in the same user session has been redesigned.
2552 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2553 error when called without an explicit environment argument.
2554 (thanks to Nikodemus Siivola)
2555 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2556 confused about their position. (thanks to Adam Warner and Gerd
2558 * bug fix: Pretty printing backquoted forms with unquotations in the
2559 argument list position of various code constructs such as LAMBDA
2560 now works correctly. (reported by Paul Dietz)
2561 * bug fix: Pretty printing unquotations no longer loses all
2562 stream position information.
2563 * optimization: performance of string output streams is now less
2564 poor for multiple small sequence writes.
2565 * optimization: performance of CSUBTYPEP in the presence of complex
2566 expressions involving CONS and NOT many times has been improved.
2567 (reported by Paul Dietz)
2568 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2570 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2571 streams for the REPL, for the debugger, and so on. (thanks to
2573 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2574 * fixed some bugs revealed by Paul Dietz' test suite:
2575 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2576 is ignored if it is too small, rather than propagating through
2577 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2578 ** extremely complex negations of CONS types were not being
2579 sufficiently canonicalized, leading to inconsistencies in
2581 ** VALUES tranformer lost derived type.
2583 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2584 * fixed a bootstrapping bug: the build process no longer assumes
2585 that the various BOOLE-related constants have the same value in
2586 host and target lisps. (noted by Paul Dietz' test suite on an
2587 SBCL binary built from CLISP)
2588 * The system can now be dynamically linked on the MIPS platform,
2589 which enables dynamic loading of foreign code from Lisp. (thanks
2590 to Ralf Baechle for discussions on the MIPS ABI)
2591 * The system now records debugging information for its own source
2592 files in a filesystem-position-independent manner, relative to
2593 the "SYS" logical host.
2594 * fixed a compiler bug: MV-LET convertion did not check references
2595 to the "max args" entry point. (reported by Brian Downing)
2596 * tweaked disassembly notes to be less confident about proclaiming
2597 some instruction as an LRA. (thanks to Brian Downing)
2598 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2599 now each have their own history, command character, and other
2600 characteristics. (thanks to David Lichteblau)
2601 * fixed some bugs revealed by Paul Dietz' test suite:
2602 ** compiler failure in compiling LOGAND expressions including a
2604 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2605 shift greater than 32.
2606 ** FLUSH-DEST did not mark blocks for type check regeneration.
2607 ** HANDLER-CASE failed to accept declarations in handler clauses
2608 in some circumstances.
2610 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2611 * New code in contrib/sb-introspect (still probably not entirely
2612 stable yet) provides some support for smart Lisp development
2613 environments like SLIME.
2614 * The conditions signalled for errors occurring when loading .fasl
2615 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2616 in a way which should help ASDF recover gracefully.
2617 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2618 stuff has been cleaned up. If you code contrib/ stuff, this might
2619 affect you, and you can look at contrib/README, contrib/STANDARDS,
2620 and/or the 0.8.4.27 diff to check.
2621 * In full calls the compiler now does not generate checks for declared
2622 argument types for all arguments.
2623 * various threading fixes
2624 ** and some experimental patches which didn't make it into
2625 the main tree for this release, but which are shipped in
2626 contrib/experimental-thread.patch as a possible fix for some
2627 failures (deadlock, spinning...) in GC-intensive multithreaded
2629 * fixed PPC build problem (source code incompatibility of different
2630 library versions): added offsetof() hackery which attempts to divine
2631 where glibc maintainers put uc_mcontext today
2632 * fixed bug 282: compiler does not trust type assertions while passing
2633 arguments to a full call.
2634 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2635 &OPTIONAL VALUES type parameter.
2636 * fix bug 214: algorithm for noting rejected templates is now more
2637 similar to that of template seletion. (also reported by rydis on
2639 * fixed bug 141b: printing backquoted information readably and prettily
2640 inserts a space where necessary.
2641 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2642 (CHARACTER 10) are now reported as errors, rather than propagated
2643 as unknown types. (reported by piso on #lisp)
2644 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2645 with values NIL and :ERROR. (thanks to Milan Zamazal)
2646 * fixed bug 191c: CLOS now does proper keyword argument checking as
2647 described in CLHS 7.6.5 and 7.6.5.1.
2648 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2649 longer raise an error; further, using a list as a for-as-arithmetic
2650 counter now raises a meaningful error.
2651 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2652 now understood by sequence creation functions such as MAKE-SEQUENCE
2654 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2655 a STRING-INPUT-STREAM.
2656 * compiler enhancement: SIGNUM is now better able to derive the type
2658 * type declarations inside WITH-SLOTS are checked. (reported by
2660 * fixed some bugs revealed by Paul Dietz' test suite:
2661 ** incorrect optimization of TRUNCATE for a positive first
2662 argument and negative second.
2663 ** compiler failure in let-convertion during flushing dead code.
2664 ** compiler failure while deriving type of TRUNCATE on an
2665 interval, containing 0.
2666 ** ASH of a negative bignum by a negative bignum count now returns
2668 ** intersection of CONS types now canonicalizes properly, fixing
2669 inconsistencies in SUBTYPEP.
2671 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2672 * incompatible change: The --disable-debugger command line
2673 option now clobbers the debugger at a more fundamental
2674 level, by redefining #'INVOKE-DEBUGGER instead of by
2675 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2676 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2677 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2678 enter the debugger REPL and then suffer recursive errors
2679 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2680 while under the new system, BREAK in --disable-debugger mode
2681 terminates the system just as an unhandled error would.
2682 * fixed compiler performance when processing loops with a step >1;
2683 * bug fix: DOCUMENTATION now retrieves generic function
2684 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2685 support has been systematized, and now supports the methods
2686 specified by ANSI, along with a default method and a method for
2687 slot documentation. (reported by Nathan Froyd)
2688 * bug fix: effective methods associated with a generic function are
2689 no longer cached over a change of that generic function's method
2690 combination. (reported by Andreas Fuchs)
2691 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2692 names a non-existent directory. (thanks to Andreas Fuchs)
2693 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2694 the PPC platform, lead to stack corruption; this has been fixed.
2695 (reported by Rainer Joswig)
2696 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2697 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2698 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2699 arithmetic in general, is now much more reliable on the PPC
2701 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2702 platform now returns the right answer.
2703 * optimization: restored some effective method precomputation in
2704 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2705 precomputation is now tunable.
2706 * optimization: compiler-internal data structure use has been
2707 reviewed, and changes have been made that should improve the
2708 performance of the compiler by about 20%.
2709 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2710 simple-base-strings and simple-bit-vectors is improved.
2711 * optimization: the optimization of 32-bit logical and arithmetic
2712 functions introduced in version 0.8.3 on the x86 has been
2713 implemented on the mips, ppc and sparc platforms; an
2714 implementation of the same facility, but for 64-bit arithmetic,
2715 has been added for the alpha.
2716 * microoptimization: the compiler is better able to make use of the
2717 x86 LEA instruction for multiplication by constants.
2718 * bug fix: in some situations compiler did not report usage of
2719 generic arithmetic in (SPEED 3) policy.
2720 * bug 145b fix: compiler used wrong type specifier while converting
2721 MEMBER-types to numeric.
2722 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2724 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2725 index argument now works properly on the Alpha platform.
2726 * bug fix: floating point exception treatment on the Alpha platform
2728 * bug fix: FILE-POSITION works much better on string input and
2729 output streams. (thanks to Nikodemus Siivola)
2730 * bug fix: many threading/garbage collection symptoms sorted.
2731 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2732 might be pseudo-atomic.
2733 * internal change: Stopping for GC is now done with signals not
2734 ptrace. GC is now done in whichever thread wanted it, instead of
2736 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2738 * bug fix: result form in DO is not contained in the implicit
2740 * incompatible change: ICR structure is changed; the value part of
2741 CONTINUATION is now called LVAR; corresponding functions are
2742 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2743 * added type deriver for ISQRT (thanks to Robert E. Brown).
2744 * bug fix: better support for loading from the command line when an
2745 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2747 * fixed some bugs revealed by Paul Dietz' test suite:
2748 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2749 ** ROUND and FROUND now give the right answer when given very
2750 small float arguments.
2751 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2753 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2754 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2755 ** LCM should return a non-negative integer.
2756 ** PARSE-INTEGER returned the index of a terminator instead of the
2757 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2758 ** PARSE-INTEGER returned an incorrect index being applied to a
2760 ** LCM with two arguments of 0 returns 0 rather than signalling
2762 ** unsigned addition of a 32-bit constant with the high bit set no
2763 longer causes an internal compiler error on the x86.
2764 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2765 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2766 with negative last argument.
2767 ** byte specifiers with zero size and position no longer cause
2768 an error during type derivation.
2769 ** bignum multiplication on the Alpha platform now returns the
2771 * porting: The system now builds on SuSE AMD64, although it still
2772 generates a 32-bit binary.
2773 * .fasl file incompatibility: The fasl file version number has
2774 been incremented (because of the changes to internal compiler
2775 data structures referred to above).
2777 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2778 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2779 more accurately, on the Darwin kernel running on PowerPC hardware.
2780 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2781 * Compiler code deletion notes now signal a condition of type
2782 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2783 an associated MUFFLE-WARNING restart.
2784 * The compiler now performs limited argument count validation of
2785 constant format strings in FORMAT, and where appropriate in ERROR,
2786 CERROR and WARN. (thanks to Gerd Moellmann)
2787 * New ASDF-INSTALL contrib can be used for automatic download and
2788 installation of third-party Lisp code from CCLAN or other sites
2790 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2791 which forces another thread to execute a function supplied by the
2793 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2794 now accept and act upon their :ELEMENT-TYPE keyword argument.
2795 (reported by Martin Atzmueller, Edi Weitz)
2796 * bug fix: FILE-POSITION now accepts position designators up to
2797 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2798 is the greater. (thanks to Patrik Nordebo)
2799 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2800 Kalle Olavi Niemitalo)
2801 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2802 printable. (reported by Eric Marsden)
2803 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2804 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2805 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2806 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2807 (thanks to Stig Sandoe)
2808 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2809 UNREAD-CHAR now correctly includes the unread character in the
2810 target sequence. (thanks to Gerd Moellmann)
2811 * bug fix in threaded builds: the system can now be suspended and
2812 resumed by shell job control with minimal disruption.
2813 * bug fixes in times and timezones >2038AD
2814 * better handling of "where is GNU make?" problem in build scripts
2815 (thanks to Nikodemus Siivola)
2816 * new optimization: inside a named function any reference to a
2817 function with the same name is considered to be a self-reference;
2818 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2819 optimization quality.
2820 * new optimization on x86: logical functions and + now have
2821 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2822 used when the result is truncated to 32 bits.
2823 * VALUES declaration is partially enabled.
2824 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2825 * fixed some bugs revealed by Paul Dietz' test suite:
2826 ** The system now obeys the constraint imposed by
2827 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2828 types form a lattice under type intersection.
2829 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2830 ** ASSOC now ignores NIL elements in an alist.
2831 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2832 and (1+ MOST-POSITIVE-FIXNUM) answers.
2833 ** The addition of a method with invalid qualifiers to a generic
2834 function does not cause an error to be signalled immediately;
2835 a warning is signalled, and the error is generated only on
2836 calling the generic function.
2837 * changed .fasl file version number, in part to add type codes for
2838 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2839 obscure ANSI requirements
2841 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2842 * fixed bug 148: failure to inline-expand a local function left
2843 garbage, confusing the compiler.
2844 * fixed bugs 3cd: structure slot readers perform type check if the
2845 slot can have an invalid value (i.e. it is either not initialized
2846 or can be written with a less specific slot writer).
2847 * bug fix: the compiler now traps array references to elements off
2848 the end of an array; previously, the bounds checking in some
2849 circumstances could go off-by-one.
2850 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2852 * type declarations for array element types now obey the description
2853 on the CLHS page "Declaration TYPE", as per discussions on
2854 sbcl-help around 2003-05-08. This means that a declaration
2855 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2856 declaration, all references to BAR will be asserted or assumed
2857 (with THE, so dependent on compiler policy) to involve objects of
2858 type FOO. Note that no such declaration is implied in
2859 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2860 * declared types of functions from the "Conditions"
2861 chapter. (reported by Paul Dietz)
2862 * bug fix: CERROR accepts a function as its first argument.
2863 * bug fix: NTH an NTHCDR accept a bignum as index
2864 arguments. (reported by Adam Warner)
2865 * optimization: character compare routines now optimize comparing
2866 against a constant character. (reported by Gilbert Baumann)
2867 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2868 argument now works properly on the MIPS platform.
2869 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2870 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2871 declared to have a functional type, does not cause an error
2873 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2874 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2875 * fixed bug: if last continuation of a deleted block has a
2876 destination, this destination should be deleted too. (reported by
2878 * fixed a bug in the bootstrap process: the host compiler's values
2879 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2880 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2881 test case from Patrik Nordebo)
2882 * improved the ability of the disassembler on the PPC platform to
2883 provide helpful disassembly notes.
2884 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2885 the class in more cases than previously.
2886 * bug fix: FILE-POSITION now understands :START and :END for
2887 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2888 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2889 8) data. (thanks to David Lichteblau)
2890 * bug fix: it is possible to add a method to a generic function
2891 without lambda list.
2892 * bug fix: reader failed to signal END-OF-FILE inside an
2893 object representation. (reported by Nikodemus Siivola)
2894 * fixed some bugs revealed by Paul Dietz' test suite:
2895 ** LAST and [N]BUTLAST should accept a bignum.
2896 ** condition slot accessors are methods.
2897 ** (VECTOR NIL) is a subtype of STRING.
2899 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2900 * minor incompatible change: some nonsensical specialized lambda
2901 lists (used in DEFMETHOD) which were previously ignored now signal
2903 * minor incompatible change: the system is now aware of the types of
2904 variables in the COMMON-LISP package, and will signal errors for
2905 most violations of these type constraints (where previously they
2906 were silently accepted).
2907 * minor incompatible change: COMPILE-FILE now uses the freedom
2908 afforded (ANSI 3.2.2.3) to use derived function types for
2909 functions defined in the same file. This also permits the system
2910 to warn on static type mismatches and function
2911 redefinition. (Currently it does not work with high DEBUG level.)
2912 * minor incompatible change: VALUES declaration is disabled.
2913 * When issuing notes, the compiler now signals a condition of type
2914 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2915 restart for use in user handlers. It is expected that the
2916 COMPILER-NOTE condition will eventually become a condition
2917 supertype to a hierarchy of note types, which will then be
2918 handleable in a similar fashion. However, at the moment, no such
2919 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2920 but it's an implementation detail, not a classification for the
2922 * Changes in type checking closed the following bugs:
2923 ** type checking of unused values (192b, 194d, 203);
2924 ** template selection based on unsafe type assertions (192c, 236);
2925 ** type checking in branches (194bc).
2926 * A short form of VALUES type specifier has ANSI meaning (it has
2927 increased the number of situations when SBCL cannot perform type
2929 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2930 keywords or constants is permissible.
2931 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2932 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2933 classes. (thanks to Antonio Martinez)
2934 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2935 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2936 argument to a pretty-print stream. Instead, it leaves any such
2937 operations to DESCRIBE-OBJECT methods.
2938 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2940 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2941 lambda lists. (thanks to David Lichteblau)
2942 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2943 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2944 (thanks to Gerd Moellmann)
2945 * fixed bug 246: increased compilation speed of long
2946 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2948 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2950 * DEFCONSTANT now throws a condition of type
2951 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2952 constant to a non-EQL value; CONTINUE and ABORT restarts
2953 respectively change and preserve the value.
2954 * fixed bug 63: The code walker, part of the implementation of CLOS,
2955 is now better at handling symbol macros.
2956 * bug fix: There is no longer an internal implementation type named
2957 CL:LENGTH. (reported by Raymond Toy)
2958 * bug fix: In macro-like defining macros/special operators the
2959 implicit block does not enclose the lambda list.
2960 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2961 enclosed in parentheses) are not suitable as type specifiers, and
2962 their use properly signals an error now.
2963 * bug fix: An argument count mismatch for a type specifier in code
2964 being compiled no longer causes an unhandled error at compile
2965 time, but signals a compile-time warning.
2966 * fixed simple vector readable printing
2967 * bug fix: DESCRIBE takes more care over whether the class
2968 precedence list slot of a class is bound before accessing it.
2969 (reported by Markus Krummenacker)
2970 * bug fix: FORMATTER can successfully compile pretty-printer format
2971 strings which use variants of the ~* directive inside.
2972 * bug fix: SEARCH now applies its TEST predicate to the elements of
2973 the arguments in the correct order. (thanks to Wolfhard Buss)
2974 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2975 in type inference. (thanks to Robert E. Brown)
2976 * bug fix: Reading in symbols with an explicit package name of ""
2977 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2978 * fixed some bugs revealed by Paul Dietz' test suite:
2979 ** NIL is now allowed as a structure slot name.
2980 ** Arbitrary numbers, not just REALs, are allowed in certain
2981 circumstances in LOOP for-as-arithmetic clauses.
2982 ** Multiple class redefinitions before slot access no longer
2983 causes a type error.
2984 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2985 association between the name and a class.
2986 ** Generic functions with non-standard method-combination and over
2987 five methods all of which return constants no longer return NIL
2988 after the first few invocations. (thanks to Gerd Moellmann)
2989 ** CALL-NEXT-METHOD with no arguments now passes the original
2990 values of the arguments, even in the presence of assignment.
2991 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2993 ** DISASSEMBLE works with closures and funcallable instances.
2994 ** ADD-METHOD now returns the generic function, not the new method.
2995 ** FIND-METHOD signals an error if the lengths of the specializers
2996 is incompatible with the generic function, even if the ERRORP
2998 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2999 which its argument is a member.
3000 ** DEFCLASS only redefines the class named by its class-name
3001 argument if that name is the proper name of the class;
3002 otherwise, it creates a new class.
3003 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3004 of the UNBOUND-SLOT condition to the name of the slot.
3005 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3006 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3007 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3008 SLOT-MAKUNBOUND in the specified fashion.
3010 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3011 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3012 cross-compilation host. As a consequence, we can now bootstrap our
3013 way up to SBCL starting with a bare gcc toolchain and human-readable
3014 source code (first the source to CLISP, then the source to SBCL).
3015 * A contributed module containing a partial implementation of the
3016 simple-streams interface has been included. (thanks to Rudi
3018 * A contributed module implementing the RFC1321 Message Digest
3019 Algorithm, known as MD5, has been included.
3020 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3021 no longer has any effect, as the code controlled by this feature
3022 has been deleted. (As far as we know, no-one has ever built using
3023 this feature, and its semantics were confused in any case).
3024 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3025 work on conditions (as required by the ANSI specification),
3026 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3027 expected behaviour on conditions. Users should note, however,
3028 that such behaviour is not required by the ANSI specification,
3029 and so use of this behaviour may render their code unportable.
3030 * fixed some bugs revealed by Paul Dietz' test suite:
3031 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3033 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3034 ** SXHASH on condition objects no longer returns NIL.
3035 ** :ALLOCATION :CLASS slots are better treated; their values are
3036 updated on class redefinition, and initforms inherited from
3037 superclasses are applied.
3038 ** REMOVE-METHOD returns its generic function argument even when
3039 no method was removed.
3040 ** SHARED-INITIALIZE now initializes the values of the requested
3041 slots, including those with :ALLOCATION :CLASS.
3042 ** ALLOCATE-INSTANCE now works on structure classes defined via
3043 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3045 ** SLOT-EXISTS-P now works on conditions, as well as structures
3047 ** MAKE-LOAD-FORM now has the required methods on
3048 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3049 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3050 keyword, and now implements the SLOT-NAMES argument.
3051 ** methods with &OPTIONAL arguments no longer allow too many
3052 arguments to be passed in the call without error.
3053 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3054 option is consistent with the required arguments of the generic
3055 function lambda list.
3056 * bug fix: REQUIRE accepts a string designator. (Thanks to
3058 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3059 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3060 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3061 * bug fix: The compiler now checks for duplicated variables in macro
3063 * bug fix: SETQ on globals returns the correct value.
3064 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3065 not a COMPILER-ERROR (followed by some other strange error on
3066 choosing the CONTINUE restart).
3067 * bug fix: make.sh and friends are now more consistent in the way that
3068 they look for GNU "make".
3070 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3071 * experimental native threads support (on x86 Linux >=2.4 only).
3072 This is not compiled in by default: you need to add :SB-THREAD to
3073 the target features. See the "Beyond ANSI" chapter of the manual
3075 * fix for longstanding nonANSIism: The old distinction between
3076 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3077 The return value from CL:FIND-CLASS is now a CLOS class, and
3078 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3079 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3080 * An interface to the MetaObject Protocol, as described in Kiczales,
3081 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3082 MIT Press, 1991, is available from the SB-MOP package.
3083 * incompatible change: the SB-PCL package should now be considered
3084 a private implementation detail, and no longer a semi-private MOP
3086 * minor incompatible change: due to rearrangement for threads, the
3087 control stack and binding stack are now allocated at arbitrary
3088 addresses instead of being hardcoded per-port. Users affected by
3089 this probably have to be doing advanced things with shared
3090 libraries, and will know who they are.
3091 * minor incompatible change: Previously, all --eval forms used were
3092 processed with READ before any of them were processed with EVAL.
3093 Now each --eval form is processed with both READ and EVAL before
3094 the next --eval form is processed. (Thus package operations like
3095 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3096 work as the user might reasonably expect.)
3097 * minor incompatible change: *STANDARD-INPUT* is now only an
3098 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3100 * minor incompatible change: Y-OR-N-P is now character-oriented,
3101 not line oriented. Also, YES-OR-NO-P now works without errors.
3102 (thanks to Antonio Martinez)
3103 * sb-aclrepl module improvements: an integrated inspector, added
3104 repl features, and a bug fix to :trace command.
3105 * Known functions, which cannot be open coded by the backend, are
3106 considered to be able to check types of their arguments. (fixing
3107 a bug report by Nathan J. Froyd)
3108 * fixed a bug in computing method discriminating functions: It is
3109 now possible to define methods specialized on classes which have
3110 forward-referenced superclasses. (thanks to Gerd Moellmann)
3111 * fixed evaluation order in optional entries (reported by Gilbert
3113 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3114 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3115 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3116 required-by-AMOP NAME argument, as well as CLASS and
3117 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3118 * fixed bug 20: DEFMETHOD can define methods using names that are
3119 not the proper names of classes to designate class specializers.
3120 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3121 Unix file descriptors, instead of blowing up. (thanks to Antonio
3123 * Garbage collection refactoring: user-visible change is that a
3124 call to the GC function during WITHOUT-GCING will not do garbage
3125 collection until the end of the WITHOUT-GCING. If you were doing
3126 this you were probably losing anyway.
3127 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3128 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3129 TYPEP the latter but not the former.
3130 * The compiler issues a full WARNING for calls to undefined functions
3131 with names from the CL package.
3132 * MAP-INTO for a vector destination is open coded. (reported by
3133 Brian Downing on c.l.l)
3134 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3135 documentation string.
3136 * fixed some bugs revealed by Paul Dietz' test suite:
3137 ** COPY-ALIST now signals an error if its argument is a dotted
3139 ** Condition slots are now accessed more correctly in the presence
3140 of multiple initargs for a given slot.
3141 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3142 exclude restarts of the same name associated with a different
3144 ** DEFCLASS of forward-referenced classes with another
3145 forward-referenced class in the superclasses list no longer
3147 ** Condition slots are now initialized once each, not multiple
3148 times. (thanks to Gerd Moellmann)
3149 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3151 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3152 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3153 arguments contain duplicated elements.
3154 ** RESTART-CASE understands local macros.
3155 ** RESTART-CASE associates exactly its own restarts with a condition.
3156 ** ENDP in safe mode checks its argument to be of type LIST.
3157 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3158 in question is unbound.
3159 ** Optimized MAKE-INSTANCE functions no longer cause internal
3160 assertion failures in the presence of duplicate initargs.
3161 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3162 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3163 propagated even in the ctor optimized implementation of
3165 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3167 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3168 * a better implementation of SXHASH on (simple) bit vectors,
3169 measured both in execution speed and in distribution of results
3170 over the positive fixnums, has been installed. Likewise, a better
3171 implementation of EQUAL for simple bit vectors is now available.
3172 * fixed CEILING optimization for a divisor of form 2^k.
3173 * fixed bug 240 (emitting extra style warnings "using the lexical
3174 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3175 by Antonio Martinez)
3176 * fixed SXHASH, giving different results for NIL depending on type
3177 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3178 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3179 variable. (found by Rolf Wester)
3180 * SBCL does not ignore type declarations for special
3181 variables. (reported by rif on c.l.l 2003-03-05)
3182 * some bug fixes in contrib/sb-aclrepl/
3183 * fixed some bugs revealed by Paul Dietz' test suite:
3184 ** a bug in the CONS type specifier, whereby the CAR and CDR
3185 types got intertwined, has been fixed;
3186 ** the type system is now able to reason about the interaction
3187 between INTEGER and RATIO types more completely;
3188 ** APPEND, [N]REVERSE and NRECONC check that those their
3189 arguments, which must be proper lists, are really so;
3190 ** An array specialized to be unable to hold elements has been
3191 implemented, as required -- yes, really -- by ANSI;
3192 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3193 on malformed property lists;
3195 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3196 * incompatible packaging change: in line with Unix convention,
3197 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3198 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3199 * REQUIRE and PROVIDE are now optionally capable of doing something
3200 useful. See the documentation string for REQUIRE.
3201 * infrastructure for a managed SBCL contrib system: contributed
3202 modules in this release include:
3203 ** the ASDF system definition facility;
3204 ** an interface to the BSD Sockets API;
3205 ** an ACL-like convenience interface to the repl;
3206 (thanks to Kevin Rosenberg)
3207 ** an implementation of ROTATE-BYTE, with efficient implementation
3209 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3210 gives a closer approximation to the right answer than previously.
3211 (thanks to Raymond Toy)
3212 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3213 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3214 optional environment arguments, as required by ANSI.
3215 * fixed bugs in other functions taking environment objects, allowing
3216 calls with an explicit NIL environment argument to be compiled
3218 * fixed bug 228: primary return values from
3219 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3220 COMPILE or FUNCTION.
3221 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3222 :TYPE will no longer signal errors on innocuous objects.
3223 * fixed bug 231b: SETQ is better at respecting type declarations in
3224 the lexical environment.
3225 * fixed a bug in DEFCLASS: classes named by symbols with no or
3226 unprintable packages can now be defined.
3227 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3228 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3229 * fixed compiler failure related to checking types of functions.
3230 (reported by Robert E. Brown)
3231 * the compiler is now much more consistent in its error-checking
3232 treatment of bounding index arguments to sequence functions: in
3233 (SAFETY 3) code, errors will be signalled in almost all cases if
3234 invalid sequence bounding indices are passed to functions defined
3235 by ANSI to operate on sequences.
3236 * fixed a bug in the build procedure: documentation of SBCL-specific
3237 packages is now preserved and available in the final Lisp image.
3238 * lifted FDEFINITION lookup out of loops in the implementation of
3239 many list operations. (thanks to Robert E. Brown)
3240 * fixed a bug in the reader: the #n# reader macro now works for
3241 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3242 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3243 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3244 E. Brown and Tony Martinez respectively)
3245 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3246 SLOT-BOUNDP: in particular, it is now able to optimize them much
3247 better, and is now not vulnerable to having packages renamed.
3248 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3249 always called when a slot is not present in an instance. (thanks
3251 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3252 longer signal an error on unreading a character following EOF on
3253 the previous constituent stream. (thanks to Tony Martinez)
3254 * fixed some bugs revealed by Paul Dietz' test suite:
3255 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3256 not just nonnegative fixnums;
3257 ** the logical bit-array operators such as BIT-AND now accept an
3258 explicit NIL for their "opt-arg" argument (to indicate a
3259 freshly-consed result bit-array);
3260 ** ELT now signals an error on an invalid sequence index in safe
3262 ** the type system is now cleverer about negations of numeric
3263 types, and consequently understands the BIGNUM and RATIO types
3265 ** the type system is now cleverer about the interaction between
3266 INTEGER and RATIO types: while bugs still remain, many more
3267 cases are accurately computed;
3268 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3269 if it is in the last clause;
3270 ** CONSTANTLY now correctly returns a side-effect-free function in
3272 ** DECLARE is no longer treated as a special-operator; in
3273 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3274 * incremented fasl file version number due to the change in the
3275 DEFSTRUCT-SLOT-DESCRIPTION structure.
3277 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3278 * minor incompatible change: code processed by the "interpreter" or
3279 EVAL now has a compilation optimization policy of (DEBUG 2)
3280 (changed from (DEBUG 1)) to improve debuggability of interactive
3281 development, and to allow the use of the debug RETURN command in
3283 * an experimental implementation of the RETURN command for the
3284 debugger has been included. (thanks to Frederik Kuivinen)
3285 * fixed bug 62: constraints were not propagated into a loop.
3286 * fixed bug in embedded calls of SORT (reported and investigated by
3288 * fixed some bugs revealed by Paul Dietz' test suite:
3289 ** printing and reading of arrays with some dimensions having
3290 length 0 (thanks to Gerd Moellmann);
3291 ** BOA constructor with &AUX argument without a default value does
3292 not cause a type error;
3293 ** CONSTANTP now returns true for all self-evaluating objects.
3295 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3296 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3297 accessors that are related by inheritance, as specified in the
3298 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3299 Valtteri Vuorikoski)
3300 * The compiler is now able to inline functions that were defined in
3301 a complex lexical environment (e.g. inside a MACROLET).
3302 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3304 * Support for the upcoming FreeBSD-5.0 release has been included.
3305 (thanks to Dag-Erling Smorgrav)
3306 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3307 effect when it is not in a toplevel context.
3308 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3309 stand a better chance of being correct. (thanks to Gerd
3311 * fixed bug in COERCE, which now signals an error on coercing a
3312 rational to a bounded real type which excludes the expected
3314 * The compiler is now able to derive types more accurately from the
3315 COERCE and COMPILE functions.
3316 * fixed bug 223: functional binding is considered to be constant
3317 only for symbols in the CL package.
3318 * fixed bug 231: SETQ did not check the type of a variable being set
3319 (reported by Robert E. Brown)
3320 * A new optimization for MAKE-INSTANCE has been included, fixing
3321 various bugs (including relating to :ALLOCATION :CLASS slots and
3322 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3324 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3325 ** As required by ANSI, LOOP now disallows anonymous collection
3326 clauses such as COLLECT I in conjunction with aggregate boolean
3327 clauses such as THEREIS (= I 1);
3328 ** LOOP now signals an error when any variable is reused in the
3329 same loop (including the potentially useful construct analogous
3330 to WITH A = 1 WITH A = (1+ A);
3331 ** IT is only a special loop symbol within the first clause of a
3332 conditional loop clause;
3333 ** LOOP with a typed iteration variable over a hashtable now
3334 signals a type error iff it should.
3335 * fixed some other bugs revealed by Paul Dietz' test suite:
3336 ** FILE-STREAM now names the class previously known as FD-STREAM;
3337 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3338 argument) no longer signals an error;
3339 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3340 of the slot symbol, rather than using the current package
3341 ((:CONC-NAME "") continues to intern the slot's name in the
3343 * incremented fasl file version number, because of the incompatible
3344 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3345 of the new implementation of DEFINE-COMPILER-MACRO.
3347 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3348 * Support for building SBCL for MIPS platforms running in
3349 little-endian mode has now been checked in, and basic
3350 functionality on said platforms verified.
3351 * minor incompatible change: PCL now records the pathname of a file
3352 in which methods and the like are defined, rather than its
3354 * minor incompatible change: TRUENAME now considers the truename of
3355 a file naming a directory to be the pathname with :DIRECTORY
3356 component indicating that directory.
3357 * minor incompatible change: a NAMED clause in the extended form of
3358 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3359 reason for the previous behaviour is unclear.
3360 * more systematization and improvement of CLOS and MOP conformance
3361 in PCL (thanks to Gerd Moellman and Pierre Mai):
3362 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3364 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3365 primary methods with no specializers;
3366 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3368 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3369 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3370 FORWARD-REFERENCED-CLASSes; error reporting on
3371 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3373 ** SXHASH on CLOS instances now uses a slot internal to the
3374 instance to return different numbers on distinct instances,
3375 while preserving the same return value through invocations of
3377 ** DEFMETHOD signals errors when methods with longer incongruent
3378 lambda lists are added to generic functions;
3379 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3380 CLASS, as specified in AMOP;
3381 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3382 on the order returned by the primary method for classes of
3383 class STANDARD-CLASS;
3384 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3385 * fixed some bugs shown by Paul Dietz' test suite:
3386 ** DOLIST puts its body in TAGBODY;
3387 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3389 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3390 value producing form;
3391 ** if more variables are given to PROGV than values, extra
3392 variables are bound and made to have no value;
3393 ** NSUBSTITUTE on list arguments gets the right answer with
3395 ** ELT signals an error of type TYPE-ERROR when the index argument
3396 is not a valid sequence index;
3397 ** LOOP signals (at macroexpansion time) an error of type
3398 PROGRAM-ERROR when duplicate variable names are found;
3399 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3400 ** FUNCALL of special-operators now cause an error of type
3402 ** PSETQ now works as required in the presence of side-effecting
3403 symbol-macro places;
3404 ** NCONC accepts any object as its last argument;
3405 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3407 ** loop-for-as-package does not require a package to be explicitely
3409 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3411 * fixed bug 166: compiler preserves "there is a way to go"
3412 invariant when deleting code.
3413 * fixed bug 172: macro lambda lists with required arguments after
3414 &REST arguments now cause an error to be signalled. (thanks to
3416 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3417 bug. (thanks to Matthew Danish)
3418 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3420 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3421 arguments are lexically rebound. (thanks to Gerd Moellmann and
3423 * fixed bug 194: error messages are now more informative when there
3424 is no primary method applicable in a call to a generic
3425 function. (thanks to Gerd Moellmann)
3426 * fixed bug in command line argument checking (thanks to Julian
3428 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3429 (thanks to Matthew Danish)
3430 * incremented fasl file version number, because of the
3431 SXHASH-related changes in the layout of CLOS data structures
3433 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3434 * minor incompatible change: The runtime (the Unix executable named
3435 "sbcl") is now much pickier about the .core files it will load.
3436 Essentially it now requires .core files to descend from the same
3437 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3438 as the runtime does. (The intent is to prevent the crashes which
3439 can occur, and which can even be reported as mysterious failures,
3440 when people patch the sources or change the build parameters
3441 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3442 sbcl and .core files.)
3443 * fixed bug: VALUES-LIST is no longer optimized away.
3444 * fixed bug 142: The FFI conversion of C string values to Lisp
3445 string values no longer conses excessively. (thanks to Nathan
3446 Froyd porting Raymond Toy's fix to CMU CL)
3447 * began to systematize and improve MOP conformance in PCL (thanks to
3448 Nathan Froyd, Gerd Moellman and Pierre Mai):
3449 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3451 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3452 ** FINALIZE-INHERITANCE is now called on class finalization;
3453 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3454 argument precedence order.
3455 * fixed bug 202: The compiler no longer fails on functions whose
3456 derived types contradict their declared type.
3457 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3458 so it can be non-toplevel.
3459 * The fasl file version number has changed (because of the new
3460 implementation of DEFMACRO).
3461 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3462 safe code, that any length requirement by their type-specifier
3463 argument is valid. The exceptions to this are described in bug
3465 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3466 functions now check the types of their inputs as required by ANSI.
3467 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3468 introduced symbol is DECLAREd to be SPECIAL.
3469 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3470 * fixed bug: :COUNT argument to sequence functions may be negative
3471 * fixed bug: body of DO-SYMBOLS may contain declarations
3472 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3473 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3474 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3475 symbol macro only once
3476 * fixed printing of call frame when argument list is unavailable
3477 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3478 * compiler no longer signals WARNING on unknown keyword
3481 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3482 * A beta-quality port to the mips architecture running Linux,
3483 based on the old CMUCL backend, has been made. It has been tested
3484 on a big-endian kernel, and works sufficiently well to be able to
3485 rebuild itself; it has not been tested in little-endian mode.
3486 * fixed an inconsistency between gencgc.c and purify.c which made
3487 dumping/loading .core files unreliable
3488 * fixed bug 120a: The compiler now deals correctly with IFs where
3489 the consequent is the same as the alternative, instead of
3490 misderiving the return type. (thanks to Alexey Dejneka)
3491 * fixed bug 113: Logical pathnames are now dumpable (the logical
3492 host is resolved at load-time, throwing an error if it is not
3494 * fixed bug 174: FORMAT's error message is slightly clearer when a
3495 non-printing character is used in a format directive.
3496 * fixed several bugs in compiler checking of type declarations, i.e.
3497 violations of the Python "declarations are assertions" principle
3498 (thanks to Alexey Dejneka)
3499 * fixed several bugs in PCL's error checking (thanks to Gerd
3501 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3503 * fixed bug in compilation of functions as first class values
3504 (thanks to Antonio Martinez-Shotton)
3505 * The compiler's handling TYPE-ERRORs which it can prove will
3506 inevitably happen at runtime has been cleaned up and corrected
3507 in several ways. (thanks to Alexey Dejneka)
3508 * improved argument type checking for various basic arithmetic
3509 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3510 lavished on them in the past that they can be compiled in many
3511 ways in different special cases
3512 * fixed bug 181: compiler checks validity of user supplied type
3514 * cleaned up code flushing in optimization: Function calls which
3515 should signal errors for safety purposes (e.g. which ANSI says
3516 should signal errors when their arguments are of incorrect type)
3517 are no longer optimized away.
3518 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3519 * incremented fasl file version number, because changes in the
3520 implementation of sequence functions like COERCE caused
3521 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3522 in old inline expansions) to become undefined. (Actually these
3523 changes were later undone, so we might very well be binary
3524 compatible with 0.7.7 after all, but leaving the version number
3525 incremented seemed like the simplest and most conservative
3528 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3529 * An alpha-quality port to the parisc architecture running Linux,
3530 based on the old CMUCL backend, has been made. This, even more so
3531 than the other backends, should be considered still a work in
3532 progress; known problems include that the Linux kernel in 64-bit
3533 mode does not propagate the correct sigcontext structure to
3534 userspace, and consequently SBCL on a parisc64 kernel will not
3536 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3537 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3538 suppressed.) Also now that the compiler is looking at declarations
3539 in the environment, it checks optimization declarations as well,
3540 and suppresses inlining when (> DEBUG SPEED).
3541 * More fixes have been made to treatment of floating point exception
3542 treatment and other Unix signals. In particular, floating point
3543 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3544 * The detection and handling of control stack exhaustion (infinite
3545 or very deeply nested recursion) has changed. Stack exhaustion
3546 detection is now done by write-protecting pages at the OS level
3547 and applies at all optimization settings; when found, a
3548 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3549 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3550 be caught using IGNORE-ERRORS.
3551 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3552 correctly when transforming recursive LABELS functions to LETs.
3553 (thanks to Alexey Dejneka)
3554 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3555 that are names of constants or global variables.
3556 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3557 alien routines with docstrings.
3558 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3559 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3561 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3562 (thanks to Eric Marsden)
3563 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3564 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3565 * Bug fix: The SPARC backend can now compile functions involving
3566 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3567 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3568 to a functions expecting a PATHNAME-DESIGNATOR.
3569 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3570 lambda lists. (thanks to Alexey Dejneka)
3571 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3572 functions so that EXPORTing the name of the class doesn't cause
3573 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3574 to Antonio Martinez for reporting this)
3575 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3576 constant-folded without failing an assertion. (thanks to Einar
3577 Floystad Dorum for reporting this)
3578 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3579 types are now dealt with more correctly by the compiler.
3580 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3581 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3582 behaviour analogous to RENAME-FILE. This puts its behaviour more
3583 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3585 * The fasl file version number has changed again. (because of the
3586 bug fix involving the names of PCL MAKE-INSTANCE functions)
3588 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3589 * bug fix: Floating point exceptions are treated much more
3590 consistently on the x86/Linux and PPC/Linux platforms.
3591 * Array initialization with :INITIAL-ELEMENT is now much faster for
3592 cases when the compiler cannot open code the array creation, but
3593 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3594 array accesses have also seen a speed increase.
3595 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3596 specified. (This was at the root of some bad interactions between
3597 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3599 * bug fix: Internal error arguments for undefined functions are now
3600 computed correctly on the PPC/Linux platform.
3601 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3602 Raymond Toy's patch for CMU CL)
3603 * Support for the Solaris 9 operating environment has been included
3604 (thanks to Daniel Merritt)
3605 * A very ugly but hopefully complete draft of the missing FFI chapter
3606 of the manual has been created by reformatting the corresponding
3607 CMU CL manual chapter into (currently very ugly and incoherent)
3608 DocBook and bringing it up to date for SBCL behavior. Thus, the
3609 manual is now essentially complete, at least by my extreme
3610 once-and-only-once standards, whereby it's acceptable to refer to
3611 the doc strings of SB-EXT functions as the primary documentation.
3612 * The fasl file version number has changed again, due to cleanup of
3613 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3614 is no longer a static symbol.)
3616 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3617 * SBCL now builds with OpenMCL (version 0.12) as the
3618 cross-compilation host; also, more progress has been made toward
3619 bootstrapping under CLISP.
3620 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3622 * bug 158 fixed: The compiler can now deal with integer loop
3623 increments different from 1; fixing this turned out also to fix
3625 * bug 169 fixed: no more bogus warnings about using lexical bindings
3626 despite the presence of perfectly good SPECIAL declarations (thanks
3627 to David Lichteblau)
3628 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3629 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3630 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3632 * bug 180 fixed: Method combination specifications no longer ignore
3633 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3634 * bug fix: Structure type predicate functions now check their argument
3635 count as they should.
3636 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3637 correctly. (thanks to Pierre Mai)
3638 * minor incompatible change: The --noprogrammer option is deprecated
3639 in favor of the new --disable-debugger option, which is very similar.
3640 (The major difference is that it takes effect at a slightly different
3641 time at startup, causing handling of errors in --sysinit and
3642 --userinit files will be affected differently.) The
3643 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3644 been added to allow this functionality to be controlled from ordinary
3645 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3646 maintainers, who might want to run non-interactive scripts to
3647 build SBCL cores which will later be used interactively.)
3648 * minor incompatible change: The LOAD function no longer, when given
3649 a wild pathname to load, loads all files matching that pathname.
3650 Instead, an error of type FILE-ERROR is signalled.
3652 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3653 * bug 147 fixed: The compiler preserves its block link/count
3654 invariants more correctly now so that it doesn't crash. (thanks
3656 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3658 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3659 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3660 no-value-for-this-character as NIL (as specified by ANSI).
3661 * HOST-NAMESTRING on physical pathnames now returns a string that is
3662 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3663 (thanks to Christophe Rhodes)
3664 * The Alpha port handles icache flushing more correctly. (thanks to
3666 * More progress has been made toward bootstrapping under CLISP. (thanks
3667 to Christophe Rhodes)
3668 * The fasl file format has changed again, because dynamic loading
3669 on OpenBSD (which has non-ELF object files) motivated some cleanups
3670 in the way that foreign symbols are transformed and passed around.
3671 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3672 is no longer treated as whitespace by the reader, but instead as
3673 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3674 |A
\7fB|, instead of A as it used to.
3676 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3677 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3678 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3679 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3680 this as of 0.7.1.45, but was left out of the previous news section
3681 (thanks to Dan Barlow)
3682 * SBCL now runs on the Solaris operating system on SPARC architectures
3683 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3684 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3685 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3686 * SPARC backend cleanups, allowing builds of cores optimized for V8
3687 and V9 SPARCS, and also emission of code targeted to a particular
3688 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3690 * SBCL is closer to bootstrapping under CLISP, thanks to various
3691 fixes by Christophe Rhodes.
3692 * The fasl file format has changed again, to allow the compiler's
3693 INFO database to support symbol macros.
3694 * The user manual (in doc/) is formatted into HTML more nicely.
3695 (thanks to coreythomas)
3696 * The system is smarter about SUBTYPEP relationships, especially
3697 those involving NOT types (including types such as ATOM which are
3698 represented internally using NOT types). Thus SUBTYPEP is less
3699 likely to return (VALUES NIL NIL) in general, and in particular
3700 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3702 * The fasl file format has changed again, because the internal
3703 representation of types now includes a new slot to support the new
3704 SUBTYPEP-of-NOT-types logic.
3705 * (not a change in the main branch of SBCL, but a related prototype
3706 which can hopefully be merged into the main branch of SBCL in the
3707 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3708 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3709 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3710 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3711 dispatch macro character combinations. (thanks to Alexey Dejneka)
3712 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3713 that superficially look like logical namestrings correctly.
3714 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3716 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3717 default supplied-p) for &optional and &key arguments. (thanks to
3720 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3721 * incompatible change: The compiler is now less aggressive about
3722 tail call optimization, doing it only when (> SPACE DEBUG) or
3723 (> SPEED DEBUG). (This is an incompatible change because there are
3724 programs which relied on the old CMU-CL-style behavior to optimize
3725 away their unbounded recursion which will now die of stack overflow.)
3726 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3727 for non-GENCGC systems has been increased to 20M (since that
3728 seems much closer to the likely performance optimum for modern
3729 systems than the old 4M value was)
3730 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3731 and *DEBUG-PRINT-LEVEL*
3732 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3733 of CMU CL's support for SPARC, and various endianness and other
3734 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3735 * new syntactic sugar for the Unix command line: --load foo.bar is now
3736 an alternate notation for --eval '(load "foo.bar")'.
3738 ** The system now detects stack overflow and handles it gracefully,
3739 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3740 optimization settings. (This is a good thing in general, and
3741 its introduction in this version should be particularly timely
3742 for anyone whose code fails because of suppression of tail
3744 ** The system now hunts for the C variable "environ" in a more
3745 devious way, to avoid segfaults when the C library version differs
3746 between compile time and run time. (thanks to Christophe Rhodes)
3747 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3748 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3749 ** The compiler no longer issues bogus style warnings for undefined
3750 classes in the same source file as the DEFCLASSes which defined
3751 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3753 ** fixes in CONDITION class precedence list for undefined function
3754 errors (thanks to Alexei Dejneka)
3755 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3756 correctly. (thanks to Dan Barlow)
3757 ** portability fixes aiming at bootstrapping under CLISP (thanks
3758 to Dave McDonald and Christophe Rhodes)
3759 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3760 ** fixes in type translation and and type inference (thanks to
3762 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3763 ** various fixes in the new ports (thanks to Dan Barlow)
3764 * several changes related to debugging:
3765 ** suppression of tail recursion, as noted above
3766 ** stack overflow detection, as noted above
3767 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3768 is now the default. (For some time encapsulation has been more
3769 reliable than the breakpoint-based :ENCAPSULATE NIL
3770 implementation, at least on X86 systems; and I just noticed that
3771 encapsulation also seems closer to the spirit of the ANSI
3774 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3776 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3777 up properly again. (There was a packaging bug in 0.7.0 which
3778 left their definitions in SB-SYS::LOAD-FOREIGN and
3779 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3780 vital for most things which interface to C-level interfaces,
3781 like extensions working with sockets or databases or
3782 Perl-compatible regexes or whatever, and the need to fix
3783 this bug is the main reason that 0.7.1 was released so
3785 ** DEFGENERIC is now choosier about the methods it redefines, so that
3786 reLOADing a previously-LOADed file containing DEFGENERICs does
3787 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3788 cycle now works as it should. (thanks to Alexey Dejneka)
3789 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3790 fixed by Christophe Rhodes. (He actually submitted this patch
3791 months ago, and I delayed until after 0.7.0.)
3792 ** Bug 111 (internal compiler confusion about runtime checks on
3793 FUNCTION types) was fixed by Alexey Dejneka.
3794 * Some internal cleanups (getting rid of variables which aren't
3795 needed now that the byte interpreter is gone) caused the fasl
3796 file format number to change again.
3798 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3799 * major incompatible change: The default fasl file extension, i.e. the
3800 default extension for files produced by COMPILE-FILE, has changed
3801 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3803 ** There are many changes in the implementation of the compiler.
3804 SBCL is now essentially a compiler-only implementation of ANSI
3805 Common Lisp. EVAL still "interprets" a few special cases, but
3806 almost all the interesting cases are handled by creating
3807 a LAMBDA expression, calling COMPILE on it, then calling
3808 FUNCALL on the result.
3809 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3810 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3811 Since the code is newer, there might still be some new bugs
3812 (though not as many as before Martin Atzmueller's fixes:-). But
3813 the new code is substantially simpler and clearer, and hopefully
3814 any remaining bugs will be simpler, less fundamental, and more
3815 fixable then the bugs in the old code.
3816 ** The revised compiler is still a little unsteady on its feet.
3818 *** The debugging information it produces (particularly the names
3819 of FUNCTION objects) is sometimes much less useful than what
3820 the old compiler produced.
3821 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3822 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3823 within a MACROLET) has been temporarily weakened.
3824 ** There are new compiler optimizations for various functions:
3825 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3826 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3827 *** the math functions TRUNCATE, FLOOR, and CEILING
3828 *** the function-of-all-trades COERCE
3829 Mostly these should be transparent, but there's one
3830 potentially-annoying problem (bug 117): when the compiler
3831 inline-expands a function and does type analysis on the result,
3832 it can create control paths which have type mismatches, and
3833 when it can't prove that those control paths aren't taken,
3834 it will issue WARNINGs about the type mismatches. This is
3835 a particular problem in practice for the new sequence functions.
3836 It's not clear how this should be fixed, and for now, a
3837 workaround is given in the entry for 117 in the BUGS file.
3838 ** (Because of the interaction between the two previous items --
3839 occasional inlining problems and new inline expansions -- some
3840 of the new sequence function optimizations won't really kick in
3841 completely until debugging information, and then inlining, are
3842 straightened out in some future version.)
3843 * minor incompatible changes:
3844 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3845 DIRECTORY no longer implicitly promotes NIL slots of its
3846 pathname argument to :WILD. In particular, when you ask for the
3847 contents of a directory (which you used to be able to do without
3848 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3849 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3850 ** changes in behavior that ANSI explicitly defines to be
3851 implementation dependent:
3852 *** The new compiler-only implementation still conforms with ANSI,
3853 but acts a little different than before. Besides the obvious
3854 changes in performance tradeoffs (that the cost per form passed
3855 to EVAL has gone up, and the cost per form executed by EVAL
3856 has gone down), the behavior of the system changes a little
3857 because there are no longer any interpreted function objects.
3858 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3859 e.g. doing COMPILE on the output of interactive DEFUN is
3861 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3863 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3864 has changed from (:CL) to NIL.
3865 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3866 e.g. #\Newline and #\Tab, don't have names specified in the
3867 ANSI Common Lisp standard, is now based on their ASCII symbolic
3868 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3869 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3870 are no longer used for output.
3871 ** changes in internal implementation constants:
3872 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3873 4 million. (If your application spends a lot of time GCing and
3874 you have a lot of RAM, you might want to experiment with
3875 increasing it even more.)
3876 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3877 However, almost all old code should still continue to work without
3878 immediate update, as SB-C-CALL is now a (deprecated) nickname
3880 ** Old operator names in the style DEF-FOO are now deprecated in
3881 favor of new corresponding names DEFINE-FOO, for consistency with
3882 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3883 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3884 internal symbols, but a few supported extensions like
3885 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3886 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3887 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3888 etc. as you get deeper into recursive calls to the debugger
3889 command loop, instead of the old "5]", "5]]", "5]]]"
3890 sequence. (I was motivated to do this when squabbles between
3891 ILISP and SBCL left me very deeply nested in the debugger. In the
3892 short term, this change will probably provoke more ILISP/SBCL
3893 squabbles, but hopefully it will be an improvement in the long run.)
3894 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3895 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3896 function, since declaiming return types involving aliens is
3897 (1) annoyingly messy to do by hand and (2) vital to efficient
3898 compilation of code which calls such functions.
3899 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3900 longer reexported by the SB-EXT package. They're solely useful
3901 for alien code, so it seems more logical that you should get
3902 them from the SB-ALIEN package, not in SB-EXT.
3903 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3904 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3905 features. Instead, the code that they used to control is always
3906 built into the system.
3907 * many other bug fixes
3908 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3909 advantage of the new EVAL-WHEN stuff and to clean them up in
3910 general, and they are now more ANSI-compliant in a number of
3911 ways. Martin Atzmueller is responsible for a lot of this.
3912 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3914 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3915 *** correct ERROR type for various file operations
3916 *** some fixes for Lisp streams
3917 *** DEFMETHOD syntax checking
3918 *** changing old weird representation of debug information as
3919 strings (which, among their other deficiencies, don't transform
3920 correctly when you rename packages, and don't change their
3921 print representation when you change things like *PACKAGE*
3922 and *PRINT-LENGTH*) to symbols and lists of symbols
3923 He also made several improvements and fixed several bugs in DESCRIBE.
3924 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3926 *** misbehavior of WRITE-STRING/WRITE-LINE
3927 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3928 and several other LOOP problems as well
3929 *** DIRECTORY when similar filenames are present
3930 *** DEFGENERIC with :METHOD options
3931 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3932 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3933 *** argument ordering in FIND with :TEST option
3934 *** mishandled package designator argument in APROPOS-LIST
3935 *** various problems in the backquote readmacro
3936 *** a bug in APROPOS
3937 *** probably some others that I'm not describing very well here,
3938 since the CVS log documents them by reference to sbcl-devel
3939 messages, and the SourceForge archives aren't working well.:-(
3940 ** Dan Barlow improved the Alpha port (and is making progress on the
3941 PPC port, for those of you who think different).
3942 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3943 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3944 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3945 continues to work on the SPARC port (for those of you in a position
3946 to look down upon our little PC-compatible boxes from a great height).
3947 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3948 and uses that copy, rather than the current dynamic value, when
3949 it's trying to decide whether to truncate output. Thus e.g.
3950 (let ((*print-lines* 50))
3951 (pprint-logical-block (stream nil)
3953 (let ((*print-lines* 8))
3954 (print (aref possiblybigthings i) stream)))))
3955 should now truncate the logical block only at 50 lines, instead of
3956 often truncating it at 8 lines, as it did before.
3957 * The doc/cmucl/ directory, containing old CMU CL documentation
3958 from the time of the fork, is no longer part of the base system.
3959 SourceForge has shut down its anonymous FTP service, and with it
3960 my original plan for distributing the old CMU CL documentation
3961 there. For now, if you need these files you can download an old
3962 SBCL source release and extract them from it.
3963 * The fasl file version number changed again, for dozens of reasons,
3964 some of which are apparent above.
3966 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3967 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3968 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3969 patch, so that DISASSEMBLE output is much nicer.
3970 * The code in the SB-PROFILE package now seems reasonably stable.
3971 I still haven't decided what the final interface should look like
3972 (I'd like PROFILE to interact cleanly with TRACE, since both
3973 facilities use function encapsulation) but if you have a need
3974 for profiling now, you can probably use it successfully with
3975 the current CMU-CL-style interface.
3976 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3977 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3978 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3979 link now returns the dangling link itself, and for similar
3980 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3981 link itself. (In these cases the old code signalled an error and
3982 looped endlessly, respectively.) Thus, DIRECTORY now works even
3983 in the presence of dangling and cyclic symbolic links.
3984 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3985 is now a supported extension again, since the consensus on
3986 sbcl-devel was that it can be useful for ordinary development
3987 work, not just for debugging SBCL itself.
3988 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3989 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3990 that currently-defined functions might be redefined later with
3991 different return types.
3992 * Hash tables can be printed readably, as inspired by CMU CL code
3993 of Eric Marsden and SBCL code of Martin Atzmueller.
3994 * better error handling in CLOS method combination, thanks to
3995 Martin Atzmueller porting Pierre Mai's CMU CL patches
3996 * more overflow fixes for >16Mbyte I/O buffers
3997 * A bug in READ has been fixed, so that now a single Ctrl-D
3998 character suffices to cause end-of-file on character streams.
3999 In particular, now you only need one Ctrl-D at the command
4000 line (not two) to exit SBCL.
4001 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4003 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4004 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4005 does the right thing.
4006 * The implementation of some type tests, especially for CONDITION
4007 types, is now tidier and maybe faster, due to CMU CL code
4008 originally by Douglas Crosher, ported by Martin Atzmueller.
4009 * Some math functions have been fixed, and there are new
4010 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4011 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4012 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4013 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4014 likely to be built into sbcl-0.7.0.)
4015 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4016 CL, ported to SBCL by Martin Atzmueller)
4017 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4018 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4019 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4020 * There's a new slam.sh hack to shorten the edit/compile/debug
4021 cycle for low-level changes to SBCL itself, and a new
4022 :SB-AFTER-XC-CORE target feature to control the generation of
4023 the after-xc.core file needed by slam.sh.
4024 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4025 COMPILE-FILE is no longer supported, so that now every function
4026 gets an entry point, so that block compilation looks a little
4027 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4028 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4029 returns the number of bytes consed since the system started,
4030 rather than the number consed since the first time the function
4031 was called. (The new definition parallels ANSI functions like
4032 CL:GET-INTERNAL-RUN-TIME.)
4033 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4034 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4035 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4036 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4037 * Fasl file version numbers are now independent of the target CPU,
4038 since historically most system changes which required version
4039 number changes have affected all CPUs equally. Similarly,
4040 the byte fasl file version is now equal to the ordinary
4043 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4044 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4045 is no longer recognized. I apologize for this, because it was
4046 listed in SB-EXT as a supported extension, but I found that
4047 its existing behavior was poorly specified, as well as incorrectly
4048 specified, and it looked like too much of a mess to straighten it
4049 out. I have enough on my hands trying to get ANSI stuff to work..
4050 * many patches ported from CMU CL by Martin Atzmueller, with
4051 half a dozen bug fixes in pretty-printing and the debugger, and
4052 half a dozen others elsewhere
4053 * fixed bug 13: Floating point infinities are now supported again.
4054 They might still be a little bit flaky, but thanks to bug reports
4055 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4056 as flaky as they were.
4057 * The --noprogrammer command line option is now supported. (Its
4058 behavior is slightly different in detail from what the old man
4059 page claimed it would do, but it's still appropriate under the
4060 same circumstances that the man page talks about.)
4061 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4062 are now supported, and enabled by default. Thus, the compiler can
4063 handle many floating point and complex operations much less
4064 inefficiently. (Thus e.g. you can implement a complex FFT
4066 * The compiler now detects type mismatches between DECLAIM FTYPE
4067 and DEFUN better, and implements CHECK-TYPE more correctly, and
4068 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4069 thanks to patches from Martin Atzmueller.
4070 * various fixes to make the cross-compiler more portable to
4071 ANSI-conforming-but-different cross-compilation hosts (notably
4072 Lispworks for Windows, following bug reports from Arthur Lemmens)
4073 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4074 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4075 fixed thanks to Pierre Mai's CMU CL patches.
4076 * fixes to keep the system from overflowing internal counters when
4077 it tries to use i/o buffers larger than 16M bytes
4078 * fixed bug 45a: Various internal functions required to support
4079 complex special functions have been merged from CMU CL sources.
4080 (When I was first setting up SBCL, I misunderstood a compile-time
4081 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4082 * improved support for type intersection and union, fixing bug 12
4083 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4084 more obscure bugs as well
4085 * some steps toward byte-compiling non-performance-critical
4086 parts of the system, courtesy of patches from Martin Atzmueller
4087 * Christophe Rhodes has made some debian packages of sbcl at
4088 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4089 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4090 stable, but are nonetheless usable. When he's ready, I'd be happy
4091 to add them to the SourceForge "File Releases" section. (And if
4092 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4093 * new fasl file format version number (because of changes in
4094 internal representation of (OR ..) types to accommodate the new
4095 support for (AND ..) types, among other things)
4097 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4098 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4100 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4102 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4103 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4104 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4105 * DESCRIBE now works on structure objects again.
4106 * Most function call argument type mismatches are now handled as
4107 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4108 know whether the function will be redefined before the call is
4109 executed. (The compiler could flag local calls with full WARNINGs,
4110 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4111 it doesn't keep track of enough information to know whether calls
4112 are local in this sense.)
4113 * Compiler output is now more verbose, with messages truncated
4114 later than before. (There should be some supported way for users
4115 to override the default verbosity, but I haven't decided how to
4116 provide it yet, so this behavior is still controlled by the internal
4117 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4118 src/compiler/ir1util.lisp.)
4119 * Fasl file format version numbers have increased again, because
4120 support for the Gray streams extension changes the layout of the
4121 system's STREAM objects.
4122 * The Gray subclassable streams extension now works, thanks to a
4123 patch from Martin Atzmueller.
4124 * The full LOAD-FOREIGN extension (not just the primitive
4125 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4126 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4127 but like most other programs, it defaults to copying the Unix
4128 environment from the original process instead of starting the
4129 new process in an empty environment.
4130 * Extensions which manipulate the Unix environment now support
4131 an :ENVIRONMENT keyword option which doesn't smash case or
4132 do other bad things. The CMU-CL-style :ENV option is retained
4133 for porting convenience.
4134 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4135 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4137 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4139 * A patch from Martin Atzmueller seems to have solved the SIGINT
4140 problem, and as far as we know, signal-handling now works cleanly.
4141 (If you find any new bugs, please report them!)
4142 * The system no longer defaults Lisp source file names to types
4143 ".l", ".cl", or ".lsp", but only to ".lisp".
4144 * The compiler no longer uses special default file extensions for
4145 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4146 seems to expect a single default extension for all compiled code,
4147 and there's no compelling reason to try to stretch the standard
4148 to allow two different extensions.) Instead, byte-compiled files
4149 default to the same extension as native-compiled files.
4150 * Fasl file format version numbers have increased again, because
4151 a rearrangement of internal implementation packages made some
4152 dumped symbols in old fasl files unreadable in new cores.
4153 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4154 many fewer weird special cases.
4155 * Bug #17 (differing COMPILE-FILE behavior between logical and
4156 physical pathnames) has been fixed, and some related misbehavior too,
4157 thanks to a patch from Martin Atzmueller.
4158 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4159 by Tim Moore, ported to SBCL by Martin Atzmueller.
4160 * Martin Atzmueller fixed several filesystem-related problems,
4161 including bug #36, in part by porting CMU CL patches, which were
4162 written in part by Paul Werkowski.
4163 * More compiler warnings in src/runtime/ are gone, thanks to
4164 more patches from Martin Atzmueller.
4165 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4168 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4170 * DESCRIBE now works on CONDITION objects.
4171 * The debugger now handles errors which arise when trying to print
4172 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4174 * The build system now uses an additional file, customize-target-features.lisp,
4175 to allow local modifications to the target *FEATURES* list. (The point of
4176 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4177 debugging features enabled, without having to worry about propagating it
4178 into everyone's system when I do a "cvs update".) When no
4179 customize-target-features.lisp file exists, the target *FEATURES* list
4180 should be constructed the same way as before.
4181 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4182 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4183 warn about behavior which is undefined under the ANSI spec. Note: This
4184 is specified by ANSI, but it's not very popular with programmers.
4185 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4186 macro in the SBCL sources for an example of a workaround which you
4187 might use to make portable ANSI-standard code which does what you want.
4188 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4189 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4190 when it's used as a non-toplevel form. (This is required in order
4191 to implement the DEFCONSTANT-EQX macro.)
4192 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4193 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4194 funny things, due to bugs in the implementation of EVAL-WHEN
4195 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4196 non-toplevel DEFCONSTANTs are uncommon.)
4197 * The core file version number and fasl file version number have been
4198 incremented, because the old noncompliant DEFCONSTANT behavior involved
4199 calling functions which no longer exist, and because I also took the
4200 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4201 * fixed bug 1 (error handling before read-eval-print loop starts), and
4202 redid debugger restarts and related debugger commands somewhat while
4204 ** The QUIT debugger command is gone, since it did something
4205 rather different than the SB-EXT:QUIT command, and since it never
4206 worked properly outside the main toplevel read/eval/print loop.
4207 Invoking the new TOPLEVEL restart provides the same functionality.
4208 ** The GO debugger command is also gone, since you can just invoke
4209 the CONTINUE restart directly instead.
4210 ** The TOP debugger command is also gone, since it's redundant with the
4211 FRAME 0 command, and since it interfered with abbreviations for the
4213 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4215 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4216 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4217 some porting work by Martin Atzmueller)
4218 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4219 with a boatload of patches which he ported to SBCL. Now that those
4221 ** The system tries to make sure that its low-priority messages
4222 are prefixed by semicolons, to help people who like to use
4223 syntax highlighting in their ILISP buffer. (This patch
4224 was originally due to Raymond Toy.)
4225 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4226 patches originally written by Raymond Toy.
4227 ** The compiler understands coercion between single-value and
4228 multiple-VALUES type expressions better, getting rid of some very
4229 weird behavior, thanks to patches originally by Robert MacLachlan
4230 and Douglas Crosher.
4231 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4232 lambda lists, thanks to a patch originally by Pierre Mai.
4233 ** The system no longer bogusly warns about "abbreviated type
4235 ** The compiler gets less confused by inlining and RETURN-FROM,
4236 thanks to some patches originally by Tim Moore.
4237 ** The system no longer hangs when dumping circular lists to fasl
4238 files, thanks to a patch originally from Douglas Crosher.
4239 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4240 undefined function error.
4241 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4242 in structure slot initforms) for now, documented workaround instead:-|
4243 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4244 * fixed bug 5: added stubs for various Gray stream functions called
4245 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4246 installed, at least appropriate type errors are generated
4247 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4248 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4249 consistently in DEFMETHOD forms.
4250 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4251 it doesn't seem to affect SBCL after all
4252 * The C runtime system now builds with better optimization and many
4253 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4255 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4257 * The system is now under CVS at SourceForge (instead of the
4258 CVS repository on my home machine).
4259 * The new signal handling code has been tweaked to treat register
4260 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4261 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4262 downstream, e.g. in debug-int.lisp, has implicit dependencies
4263 on the unsignedness of integer representation of machine words,
4264 and that caused the system to bomb out with infinite regress
4265 when trying to recover from type errors involving signed values,
4266 e.g. (BUTLAST '(1 2 3) -1).)
4267 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4268 (This was one of the bugs Peter Van Eynde reported back in July.)
4269 * The system now uses code inspired by Colin Walters' O(N)
4270 implementation of MAP (from the cmucl-imp@cons.org mailing
4271 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4272 inline the MAP operation, and there is more than one
4273 sequence argument to the MAP call (so that it can't just
4274 do ETYPECASE once and for all based on the type of the
4275 single sequence argument). (The old non-inline implementation
4276 of the general M-argument sequence-of-length-N case required
4277 O(M*N*N) time when any of the sequence arguments were LISTs.)
4278 * The QUIT :UNIX-CODE keyword argument has been renamed to
4279 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4280 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4282 * Raymond Wiker's patches to port dynamic loading from Linux to
4283 FreeBSD have been added.
4284 * The BUGS file is now more nearly up to date, thanks in large part
4285 to Martin Atzmueller's review of it.
4286 * The debugger now flushes standard output streams before it begins
4287 its output ("debugger invoked" and so forth).
4288 * The core version number and fasl file version number have both
4289 been incremented, because of incompatible changes in the layout
4291 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4292 used to not be called for a saved Lisp image.)
4293 * Martin Atzmueller's version of a patch to fix a compiler crash,
4294 as posted on sbcl-devel 13 September 2000, has been installed.
4295 * Instead of installing Martin Atzmueller's patch for the
4296 compiler transform for SUBSEQ, I deleted the compiler transform,
4297 and transforms for some similar consing operations.
4298 * A bug in signal handling which kept TRACE from working on OpenBSD
4300 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4301 away by constant folding
4302 * The system now defines its address space constants in one place
4303 (in the Lisp sources), and propagates them automatically elsewhere
4304 (through GENESIS and the sbcl.h file). Therefore, patching the
4305 address map is less unnecessarily tedious and error-prone. The
4306 Lisp names of address space constants have also been systematized.
4307 * CVS tags like dollar-Header-dollar have been removed from
4308 the sources, because they have never saved me trouble and
4309 they've been source of trouble working with patches and other
4310 diff-related operations.
4311 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4312 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4314 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4316 * The system has been ported to OpenBSD.
4317 * The system now compiles with a simple "sh make.sh" on the systems
4318 that it's supported on. I.e., now you no longer need to tweak
4319 text in the target-features.lisp-expr and symlinks in src/runtime/
4320 by hand, the make.sh takes care of it for you.
4321 * The system is no longer so grossly inefficient when compiling code
4322 involving vectors implemented as general (not simple) vectors (VECTOR T),
4323 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4324 which dares to use the various sequence functions on non-simple
4325 vectors, takes less of a performance hit.
4326 * There is now a primitive type predicate VECTOR-T-P
4327 to test for the (VECTOR T) type, so that e.g.
4328 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4329 can now be compiled with some semblance of efficiency. (The old code
4330 turned the type declaration into a full call to %TYPEP at runtime!)
4331 * AREF on (VECTOR T) is still not fast, since it's still compiled
4332 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4333 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4334 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4335 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4336 without making full calls to SUBTYPEP at runtime.
4337 (Some analogous efficiency issues for non-simple vectors specialized to
4338 element types other than T, or for non-simple multidimensional arrays,
4339 have not been addressed. They could almost certainly be handled the
4340 same way if anyone is motivated to do so.)
4341 * The changes in array handling break binary compatibility, so
4342 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4343 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4344 instead of (VALUES T T).
4345 * By following the instructions that Dan Barlow posted to sbcl-devel
4346 on 2 July 2000, I was able to enable primitive dynamic object
4347 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4348 functionality is not implemented (since it calls ld to resolve
4349 library references automatically, requiring RUN-PROGRAM for its
4350 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4351 try to resolve library references) is now supported.
4352 * The system now flushes the standard output streams when it terminates,
4353 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4354 them at several other probably-convenient times, e.g. in each pass of
4355 the toplevel read-eval-print loop, and after evaluating a form given
4356 as an "--eval" command-line option. (These changes were motivated by a
4357 discussion of stream flushing issues on cmucl-imp in August 2000.)
4358 * The source transform for TYPEP of array types no longer assumes
4359 that an array whose element type is a not-yet-defined type
4360 is implemented as an array of T, but instead punts, so that the
4361 type will be interpreted at runtime.
4362 * There is now some support for cross-compiling in make.sh: each of
4363 the phases of make.sh has its own script. (This should be transparent
4364 to people doing ordinary, non-cross-compile builds.)
4365 * Since my laptop doesn't have hundreds of megabytes of memory like
4366 my desktop machine, I became more motivated to do some items on
4367 my to-do list in order to reduce the size of the system a little:
4368 ** Arrange for various needed-only-at-cold-init things to be
4369 uninterned after cold init. To support this, those things have
4370 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4371 symbols with such names are now uninterned after cold init).
4372 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4373 fasl files for cold load.
4374 ** Remove the old compiler structure pooling code (which used to
4375 be conditional on the target feature :SB-ALLOC) completely.
4376 ** Redo the representation of some data in cold init to be more compact.
4377 (I also looked into supporting byte compiled code at bootstrap time,
4378 which would probably reduce the size of the system a lot, but that
4379 looked too complicated, so I punted for now.)
4380 * The maximum signal nesting depth in the src/runtime/ support code has
4381 been reduced from 4096 to 256. (I don't know any reason for the very
4382 large old value. If the new smaller value turns out to break something,
4383 I'll probably just bump it back up.)
4384 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4386 * Many, many bugs reported by Peter Van Eynde have been added to
4387 the BUGS list; some have even been fixed.
4388 * While enabling dynamic object file loading, I tried to make the
4389 code easier to understand, renaming various functions and variables
4390 with less ambiguous names, and changing some function calling
4391 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4392 * While trying to figure out how to do the OpenBSD port, I tried to
4393 clean up some of the code in src/runtime/. In particular, I dropped
4394 support for non-POSIX signal handling, added various comments,
4395 tweaked the code to reduce the number of compilation warnings, and
4396 renamed some files to increase consistency.
4397 * To support the new automatic configuration functionality in make.sh,
4398 the source file target-features.lisp-expr has been replaced with the
4399 source file base-target-features.lisp-expr and the machine-generated
4400 file local-target-features.lisp-expr.
4401 * fixed a stupid quoting error in make.sh so that using CMU CL
4402 "lisp -batch" as cross-compilation host works again
4404 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4406 * DESCRIBE no longer tries to call itself recursively to describe
4407 bound/fbound values, so that it no longer fails on symbols which are
4408 bound to themselves (like keywords, T, and NIL).
4409 * DESCRIBE now works on generic functions.
4410 * The printer now prints less-screwed-up representations of closures
4411 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4412 * A private symbol is used instead of the :EMPTY keyword previously
4413 used to mark empty slots in hash tables. Thus
4414 (DEFVAR *HT* (MAKE-HASH-TABLE))
4415 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4416 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4417 now does what ANSI says that it should. (You can still get
4418 similar noncompliant behavior if bang on the hash table
4419 implementation with all the symbols you get back from
4420 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4421 This breaks binary compatibility, since tests for equality to
4422 :EMPTY are wired into things like the macroexpansion of
4423 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4425 * There's now a minimal placeholder implementation for CL:STEP,
4426 as required by ANSI.
4427 * An obscure bug in the interaction of the normal compiler, the byte
4428 compiler, inlining, and structure predicates has been patched
4429 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4430 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4432 * Missing ordinary arguments in a macro call are now detected even
4433 when the macro lambda list contains &KEY or &REST.
4434 * The debugger no longer complains about encountering the top of the
4435 stack when you type "FRAME 0" to explicitly instruct it to go to
4436 the top of the stack. And it now prints the frame you request even
4437 if it's the current frame (instead of saying "You are here.").
4438 * As specified by ANSI, the system now always prints keywords
4439 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4441 * The default initial SIZE of HASH-TABLEs is now smaller.
4442 * Type information from CLOS class dispatch is now propagated
4443 into DEFMETHOD bodies, so that e.g.
4444 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4446 is now basically equivalent to
4447 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4448 (DECLARE (TYPE SINGLE-FLOAT X))
4450 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4451 operation, without having to do run-time type dispatch.
4452 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4453 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4454 * Since I don't seem to be making big file reorganizations very often
4455 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4456 is overflowing my ability to conveniently back them up), I've finally
4457 checked the system into CVS. (The CVS repository is on my home system,
4458 not at SourceForge -- putting it on SourceForge might come later.)
4459 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4460 high-level GC-NOTIFY-FOO functions send their output. (There's
4461 still very little control of where low-level verbose GC functions
4462 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4463 less than it used to -- the GC-NOTIFY-FOO functions are now under
4464 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4465 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4466 in only one place in the source code, and propagates it automatically
4467 everywhere that it's needed. Thus e.g. when I bump the version from
4468 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4469 * The C source files now include boilerplate legalese and documentation
4470 at the head of each file (just as the Lisp source files already did).
4471 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4472 to his page will be replaced with a link to his new CLiki service.
4474 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4476 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4477 * The build process now looks for GNU make under the default name "gmake",
4478 instead of "make" as it used to. If GNU make is not available as "gmake"
4479 on your system, you can change this default behavior by setting the
4480 GNUMAKE environment variable.
4481 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4482 can build without error under CMU CL.
4484 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4486 * There is now a partial SBCL user manual (with some new text and some
4487 text cribbed from the CMU CL manual).
4488 * The beginnings of a profiler have been added (starting with the
4489 CMU CL profiler and simplifying and cleaning up). Eventually the
4490 main interface should be through the TRACE macro, but for now,
4491 it's still accessed through vaguely CMU-CL-style functions and macros
4492 exported from the package SB-PROFILE.
4493 * Some problems left over from porting CMU CL to the new
4494 cross-compilation bootstrap process have been cleaned up:
4495 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4496 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4497 part when getting rid of the compiler *BACKEND* stuff.)
4498 ** The value of *NULL-TYPE* was screwed up, because it was
4499 being initialized before the type system knew the final
4500 definition of the 'NULL type. This screwed up several key
4501 optimizations in the compiler, causing inefficiency in all sorts
4502 of places. (I found it because I wanted to understand why
4503 GET-INTERNAL-RUN-TIME was consing.)
4504 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4505 PROCLAIM FTYPE information. Unfortunately this broke binary
4506 compatibility again, since now the forms output by DEFGENERIC
4507 to refer to functions which didn't exist in 0.6.3.
4508 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4509 can use the new (as of 0.6.3) transform for SOME into MAP into
4511 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4512 values to (INTEGER 0 1000000), so that the time code will no longer
4513 occasionally get blown up by Linux returning 1000000 microseconds
4514 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4515 its output conform to the ANSI spec. (Alas, this makes its output
4516 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4517 * A full call to MAP NIL with a single sequence argument no longer conses.
4518 * fixes to problems pointed out by Martin Atzmueller:
4519 * The manual page no longer talks about multiprocessing as though
4520 it were currently supported.
4521 * The ILISP support patches have been removed from the distribution,
4522 because as of version 5.10.1, ILISP now supports SBCL without us
4523 having to maintain patches.
4524 * added a modified version of Raymond Toy's recent CMU CL patch for
4525 EQUALP comparison of HASH-TABLE
4527 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4529 * The system still can't cross-compile itself with
4530 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4531 entails), but at least it can compile more of itself that way
4532 than it used to be able to, and various buglets which were uncovered
4533 by trying to cross-compile itself that way have now been fixed.
4534 * This release breaks binary compatibility again. This time
4535 at least I've incremented the FASL file format version to 2, so that the
4536 problem can be detected reliably instead of just causing weird errors.
4537 * various new style warnings:
4538 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4539 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4540 ** using the lexical binding of a variable named in the *FOO* style
4541 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4542 as specified by ANSI.
4543 * *RANDOM-STATE* is no longer automatically initialized from
4544 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4545 default behavior of the system is to repeat its behavior every time
4546 it's run. If you'd like to change this behavior, you can always
4547 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4548 old convention there was no comparably easy way to get the system to
4549 repeat its behavior every time it was run.
4550 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4551 FTYPE declarations has been removed, in favor of their ANSI
4552 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4553 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4554 the types of their sequence arguments can be determined at compile time.
4555 This is done through a new open code expansion for MAP which eliminates
4556 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4557 argument types can be determined at compile time.
4558 * The optimizer now transforms COERCE into an identity operation when it
4559 can prove that the coerced object is already of the correct type. (This
4560 can be a win for machine generated code, including the output of other
4561 optimization transforms, such as the MAP transform above.)
4562 * Credit information has been moved from source file headers into CREDITS.
4563 * Source file headers have been made more standard.
4564 * The CASE macro now compiles without complaining even when it has
4567 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4569 * (Note that the way that the PCL macroexpansions were rewritten
4570 to accommodate the change in DEFGENERIC below breaks binary
4571 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4572 not run under sbcl-0.6.2. Once we get out of alpha releases,
4573 i.e. hit release 1.0.0, we'll probably try to maintain binary
4574 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4575 and sbcl-1.4.4. Until then, however, it might be fairly common
4576 for maintenance releases to break binary compatibility.)
4577 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4579 * The reporting of unhandled signals has been changed to print some
4580 explanatory text as well as the report form. (Previously only
4581 the report form was printed.)
4582 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4583 it defines, so that the compiler no longer issues undefined function
4584 warnings for compiled-but-not-yet-loaded generic functions.
4585 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4586 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4587 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4588 section of the ANSI spec are supported.
4589 * The "" nickname for the "KEYWORD" package has been removed.
4590 The reader still handles symbol tokens which begin with a package marker
4591 as keywords, but it doesn't expose its mechanism for doing so in the
4592 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4593 * The system now issues STYLE-WARNINGs for contradictory TYPE
4594 proclamations. (Warnings for contradictory FTYPE proclamations would
4595 be nice too, but those can't be done usefully unless the type system
4596 is made smarter about FUNCTION types.)
4597 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4598 systematized, so that "*target-*.lisp is supposed to exist only on the
4599 target and imply that there's a related file which exists on the
4600 host, and *host-*.lisp is supposed to exist only on the host and imply
4601 that there's a related file which exists on the target. This involves a
4602 lot of renaming. Hopefully the acute confusion caused by the renaming
4603 will be justified by the reduction in chronic confusion..
4604 ** runtime-type.lisp -> early-target-type.lisp
4605 ** target-type.lisp -> late-target-type.lisp
4606 ** early-host-format.lisp -> early-format.lisp
4607 ** late-host-format.lisp -> late-format.lisp
4608 ** host-error.lisp -> misc-error.lisp
4609 ** early-error.lisp -> early-target-error.lisp
4610 ** late-error.lisp -> late-target-error.lisp
4611 ** host-defboot.lisp -> early-defboot.lisp
4612 ** code/misc.lisp -> code/target-misc.lisp
4613 ** code/host-misc.lisp -> code/misc.lisp
4614 ** code/numbers.lisp -> code/target-numbers.lisp
4615 ** code/early-numbers.lisp -> numbers.lisp
4616 ** early-host-type.lisp -> early-type.lisp
4617 ** late-host-type.lisp -> late-type.lisp
4618 ** host-typep.lisp -> typep.lisp
4619 ** load.lisp -> target-load.lisp
4620 ** host-load.lisp -> load.lisp
4621 ** host-disassem.lisp -> disassem.lisp
4622 ** host-insts.lisp -> insts.lisp
4623 ** byte-comp.lisp -> target-byte-comp.lisp
4624 ** host-byte-comp.lisp -> byte-comp.lisp
4625 ** host-signal.lisp -> signal.lisp
4626 ** host-defstruct.lisp -> defstruct.lisp
4627 ** late-target-type.lisp -> deftypes-for-target.lisp
4628 Furthermore, several other previously target-only files foo.lisp (e.g.
4629 hash-table.lisp and random.lisp) have been split into a target-and-host
4630 foo.lisp file and a target-only target-foo.lisp file, with their key type
4631 definitions in the target-and-host part, so that the cross-compiler will
4632 know more about target types.
4633 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4634 gone away. In their place are various *BACKEND-FOO* variables
4635 corresponding to the slots of the old structure.
4636 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4637 into the target SBCL has been fixed.
4638 * The chill.lisp system for loading cold code into a running SBCL
4640 * Support for the CMU CL "scavenger hook" extension has been removed.
4641 (It was undocumented and unused in the CMU CL sources that SBCL was
4642 derived from, and stale in sbcl-0.6.1.)
4643 * Various errors in the cross-compiler type system were detected
4644 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4645 (enabling various consistency checks). Many of them were fixed,
4646 but some hard problems remain, so the compiler is back to
4647 running without *TYPE-SYSTEM-INITIALIZED* for now.
4648 * As part of the cross-compiler type system cleanup, I implemented
4649 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4650 * I have started adding UNCROSS calls throughout the type system
4651 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4652 on cross-compiler input files will be able to go away, and various
4654 * CONSTANTP now returns true for quoted forms (as explicitly required
4657 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4659 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4660 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4661 (MAKE-LIST -1). In the longer term, it ought to become true
4662 that declarations are assertions even at SAFETY 1. For now, it's not
4663 quite true even at SAFETY 3, but it's at least more nearly true..
4664 (Note that this change seems to increases the size of the system by
4665 O(5%) and to decrease the speed of the compiler by 20% or more.)
4666 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4667 for the problem of printing dozens of lines of distracting information
4668 about low-level system machinery as part of the top stack frame
4669 on entry to the debugger when an undefined function was called.
4670 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4671 is not reset to COMMON-LISP-USER.
4672 * Compilation of stuff related to dyncount.lisp has been made conditional
4673 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4674 smaller. The various dyncount-related symbols have been moved into
4675 a new "SB-DYNCOUNT" package.
4676 * tty-inspect.lisp has been renamed to inspect.lisp.
4677 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4678 feature has gone away. (When we eventually port to other flavors of
4679 libc and/or Unix, we'll try to make the differences between flavors
4680 invisible at the user level.)
4681 * Various other *FEATURES* tags, and/or their associated conditionals,
4682 have been removed if obsolescent, or given better documentation, or
4683 sometimes given more-mnemonic names.
4685 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4687 * tidied up "make.sh" script
4688 * tidied up system directory structure
4689 * better "clean.sh" behavior
4690 * added doc/FOR-CMUCL-DEVELOPERS
4691 * many many small tweaks to output format, e.g. removing possibly-confusing
4692 trailing #\. character in DESCRIBE-INSTANCE
4693 * (EQUALP #\A 'A) no longer signals an error.
4694 * new hashing code, including EQUALP hashing
4695 * tidied up Lisp initialization and toplevel
4696 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4697 * command line argument processing
4698 * added POSIX-GETENV function to deal with Unix-ish environment variables
4699 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4700 terminating SBCL on EOF
4701 * non-verbose GC by default
4702 * There is no more "sbcl" shell script; the sbcl file is now the C
4703 runtime executable (just like CMU CL).
4704 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4706 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4707 symbol and package information as Lisp native symbol and package objects
4708 instead of strings naming symbols and strings naming packages. This way,
4709 whenever packages are renamed (as in warm init), debug information is
4710 transformed along with everything else.
4711 * tweaked the optimization policy declarations which control the building
4712 of SBCL itself. Now, among other things, the system no longer saves
4713 source location debugging information. (This helps two problems at once
4714 by reducing SBCL size and by keeping SBCL from trying to look for its
4715 sources -- which may not exist -- when reporting errors.)
4716 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4717 debugging and testing purposes
4718 * cleaned up printing, making the printer call PRINT-OBJECT for
4719 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4720 methods, giving nearly-ANSI behavior
4721 * converted almost all special variables to use *FOO* naming convention
4722 * deleted PARSE-TIME functionality, since it can be done portably
4723 * moved some files out of cold init into warm init
4724 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4726 * regularized formatting of source files
4727 * added an install.sh script
4728 * fixed ridiculous memory usage of cross-compiler by making
4729 compiler/alloc.lisp not try to do pooling unless it can hook
4730 itself into the GC of the cross-compilation host. Now the system
4731 builds nicely on my old laptop.
4732 * added :SB-ALLOC in target-features.lisp-expr
4733 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4734 was not implemented)
4735 * re-did condition handling and note reporting in the compiler. Notes
4736 are no longer handled by signalling conditions. Style warnings
4737 and warnings are handled more correctly and reported in such a way
4738 that it's easy to find one or the other in your output (so that you
4739 can e.g. figure out which of many problems caused COMPILE-FILE to
4741 * changed the severity of several compiler warnings from full WARNING
4742 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4743 compiler note reporting so that it doesn't use the condition system
4744 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4745 COMPILE-FILE command)
4746 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4747 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4748 no longer supported, since their implementation was deeply intertwingled
4749 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4750 * removed bogus "support" for compiler macros named (SETF FOO), and
4751 removed the compiler macro for SETF INFO (but only after making a fool
4752 of myself on the cmucl-imp mailing list by posting a bogus patch for
4753 DEFINE-COMPILER-MACRO..)
4754 * Compiled files containing forms which have side effects on the Lisp
4755 reader (such as DEFPACKAGE forms) are now handled more correctly.
4756 (Compiler queuing of top level lambdas has been suppressed by setting
4757 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4758 * deleted various currently-unused source files, e.g. gengc.lisp. They
4759 may be added back at some point e.g. when porting to other architectures,
4760 but until they are it's distracting to distribute them and to try to
4762 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4763 there were so many of them they're just distractions, and UNCROSS is
4764 known to be able to handle the current sources
4765 * moved PROFILE functionality into TRACE, so that it will be clear
4766 how the wrapping and unwrapping of functions when you profile them
4767 interacts with the wrapping and unwrapping of functions when you
4768 trace them. (Actually, the functionality isn't there yet, but at least
4769 the interface specification is there. Hopefully, the functionality will
4770 arrive with some future maintenance release.)
4771 * removed host-oops.lisp
4772 * changed signature of QUIT function to allow UNIX-CODE argument
4773 * fixed READ-SEQUENCE bug
4774 * tweaked verbose GC output so that it looks more like the progress
4775 output that ANSI specifies for functions like LOAD
4776 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4777 * added <http://sbcl.sourceforge.com> to the banner information printed by