1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.20 relative to 1.0.19:
3 * minor incompatible change: OPTIMIZE qualities
4 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
5 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
6 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
7 * bug fix: fixed #427: unused local aliens no longer cause compiler
8 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
10 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
11 alien-stack leakage. (reported by Andy Hefner)
12 * bug fix: PROGV signals an error when an attempt to violate declared
13 type of a variable or bind a constant is made.
14 * bug fix: SET signals an error when an attempt to violate declared
15 type of a variable is made.
16 * bug fix: restart computation during the execution of a restart
17 test function no longer causes infinite recursion. (reported by
19 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
20 signals an error. (thanks to Pierre Mai)
22 changes in sbcl-1.0.19 relative to 1.0.18:
23 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
24 warnings that go otherwise unhandled will be muffled if they are
25 of the type that's the value of this variable.
26 * optimization: stack allocation is slightly more efficient on x86
28 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
29 where a raw slot always is initialized using the initform whose
30 type is not know sufficiently well a compile-time are now compiled
31 correctly. (reported by John Morrison)
32 * bug fix: compiler no longer makes erronous assumptions in the
33 presense of non-foldable SATISFIES types.
34 * bug fix: stack analysis missed cleanups of dynamic-extent
35 arguments in non-let-converted calls to local functions.
36 * improvements to the Windows port:
37 ** adjusted address spaces for building on both Win32 and
38 Win64. (thanks for John Connors)
39 * fixed some bugs revealed by Paul Dietz' test suite:
40 ** interval arithmetic during type derivation used inexact integer
41 to single-float coercions.
42 ** arithmetic operations involving large integers and single
43 floats give the same results in compiled and interpreted code.
44 ** deriving the result type of COERCE no longer signals an error
45 if the derived type of the second argument is a MEMBER type
46 containing invalid type specifiers.
47 ** ADJOIN with constant NIL as second argument works correctly.
48 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
50 changes in sbcl-1.0.18 relative to 1.0.17:
51 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
52 profiles only the current thread.
53 * minor incompatible change: changes to SYMBOL-VALUE of constants
54 defined with DEFCONSTANT now signal an error.
55 * enhancement: SB-SPROF now has support for wallclock profiling,
56 and is also able to profile specific threads. REPORT output
57 has also additional sorting options.
58 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
60 * optimization: structure allocation has been improved
61 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
62 ** out of line constructors are ~10% faster.
63 ** inline constructors are ~15% faster.
64 ** inline constructors are capable of dynamic extent allocation
65 (generally on x86 and x86-64, in some cases on other platforms
67 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
69 * optimization: file compiler is now able to coalesce non-circular
70 lists, non-base strings, and bit-vectors. Additionally, constants
71 are never referenced through SYMBOL-VALUE at runtime.
72 * optimization: code defining methods on PRINT-OBJECT (and other
73 generic functions in the COMMON-LISP package) now loads faster.
74 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
76 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
77 fasl is now deleted. (reported by Attila Lendvai)
78 * bug fix: READ-LINE always returned NIL for the last line in files.
79 (reported by Yoshinori Tahara)
80 * bug fix: more accurate disassembly annotations of foreign function
81 calls. (thanks to Andy Hefner)
82 * bug fix: trimming non-simple strings and non-string string
83 designators when the there is nothing to trim works properly.
84 (thanks to James Knight)
85 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
86 (thanks to Travis Cross)
87 * fixed some bugs revealed by Paul Dietz' test suite:
88 ** NIL is a valid function name (regression at 1.0.13.38)
89 ** FILL on lists was missing its return value (regression at 1.0.12.27)
90 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
91 fill pointers properly (regression at 1.0.12.23)
92 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
93 displacement indices properly (regression at 1.0.12.23)
95 changes in sbcl-1.0.17 relative to 1.0.16:
96 * temporary regression: user code can no longer allocate closure
97 variable storage on stack, due to bug 419 without explicitly
98 requesting it. Please consult sbcl-devel for advice if you need to
99 use this feature in the meanwhile.
100 * new feature: runtime argument --control-stack-size can be used to
101 adjust thread default control stack size.
102 * enhancement: improved TIME output
103 ** all times are reported using the measured accuracy (milliseconds
104 for real and GC times, microseconds for everything else.)
105 ** processor cycle counts on x86 and x86-64.
106 ** interpreted forms are counted for both evaluator modes.
107 ** number of lambdas converted by the compiler is reported.
108 ** CPU percentage report (computed from real and total run time.)
109 ** more comprehensive run time reporting, using a condenced format
110 ** interperted form, lambda, and page fault counts are omitted
112 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
114 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
115 * optimization: two argument forms of LAST are upto ~10% faster
116 in normal SPEED policies.
117 * optimization: NCONC no longer needs to heap cons its &REST list
118 in normal SPEED policies.
119 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
121 * bug fix: UNION and NUNION work with :TEST-NOT once more,
122 regression since 1.0.9.1. (thanks to Eric Marsden)
123 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
124 since 1.0.15.36. (thanks to Paul Khuong)
125 * bug fix: LAST when always returned the whole list when given a bignum
126 as the second argument.
127 * bug fix: dynamic extent allocation of nested lists and vectors
128 could leak to otherwise accessible parts.
129 * bug fix: invalid optimization of heap-allocated alien variable
131 * bug fix: fasl header checking is less vulnerable to different
132 platform word lengths.
133 * bug fix: more correct assembler syntax for GNU binutils
134 2.18.50.0.4 support. (thanks to Marijn Schouten)
135 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
137 * bug fix: the fopcompiler can handle LOCALLY forms (with no
138 declarations) successfully. (reported by Attila Lendvai)
140 changes in sbcl-1.0.16 relative to 1.0.15:
141 * minor incompatible change: revert the changes to sb-posix's error
142 signaling added in 1.0.14.
143 * minor incompatible change: change PROBE-FILE back to returning
144 NIL whenever we can't get a truename, as was the case before 1.0.14.
145 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
146 inherits from ERROR instead of just CONDITION.
147 * new feature: SB-INTROSPECT can provide source locations for instances
148 as well. (thanks to Tobian Ritterweiler)
149 * optimization: binding special variables now generates smaller code
150 on threaded platforms.
151 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
152 and cases where no :TEST is given but the compiler can infer that
153 the element to search is of type (OR FIXNUM (NOT NUMBER)).
154 * optimization: better LOGNOT on fixnums.
155 * optimization: modular arithmetic for a particular requested width
156 is implemented using a tagged representation unless a better
157 representation is available.
158 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
159 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
160 for wellformedness and returns a specialized vector. (reported by
161 Francois-Rene Rideau)
162 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
163 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
164 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
165 unscheduling timer at the same time as another timer fires.
166 * bug fix: don't reschedule timers for dead threads.
167 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
168 * bug fix: copying output from RUN-PROGRAM to a stream signalled
169 bogus errors if select() was interrupted.
170 * enhancement: add support for fcntl's struct flock to SB-POSIX.
172 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
173 * enhancement: cleaner backtraces for interactive interrupts, as
174 well as other cases where the interesting frames used to be
175 obscured by interrupt handling frames.
176 * enhancement: untracing a whole package using (UNTRACE "FOO") is
177 now supported, and tracing a whole package using (TRACE "FOO") now
178 traces SETF-functions as well.
179 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
180 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
181 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
182 * unreadably printed representation of hash-tables now includes
184 * bug fix: partially fixed #188: type propagation from assignments
185 is now more efficient.
186 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
187 'DOUBLE-FLOAT) are not flushable.
188 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
189 full address of the object, and none of the tag bits.
190 * bug fix: readably printing hash-tables now respects other printer
191 control variables. (reported by Cedric St-Jean)
192 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
193 keyword with MAKE-HASH-TABLE.
194 * bug fix: export SB-POSIX:MKSTEMP.
195 * bug fix: SORT was not interrupt safe.
196 * bug fix: XREF accounts for the last node of each basic-block as
198 * bug fix: MAKE-INSTANCE optimizations interacted badly with
199 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
200 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
201 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
202 create obsolete instances in the case of redefinition or
203 obsoletion of a superclass. (thanks to Andy Hefner)
204 * bug fix: Support for the Alpha architecture has been revived; it had
205 suffered somewhat from lack of maintenance since sbcl-1.0.
206 * improvements to the Windows port:
207 ** The system detects the codepage to use at startup, instead of
208 using the value from the saved core. (thanks to Kei Suzuki)
210 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
211 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
212 (see documentation for details.)
213 * revived support for OpenBSD (contributed by Josh Elsasser)
214 * partially fixed bug #108: ROOM no longer suffers from occasional
215 (AVER (SAP= CURRENT END)) failures.
216 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
217 bodies no longer cause a WARNING to be signalled. (reported by
219 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
220 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
221 the bug in TRUNCATE.) (reported by Michael Weber)
222 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
223 no samples. (reported by Andy Hefner)
224 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
225 no longer appear as (NIL ...) frames in backtraces.
226 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
227 is now actually exported from SB-EXT as documented. (reported by
229 * bug fix: sb-aclrepl now correctly understands how to inspect
230 single-floats on 64-bit platforms where single-floats are not boxed.
231 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
232 yet been finalized. (reported by Levente Meszaros)
233 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
234 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
235 SBCL-specific optimize qualities.
237 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
238 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
239 an executable in the search path, and does so in the child
240 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
241 been removed; it can be found in the CVS history, for somebody who
242 needs that search behavior (see the manual).
243 * minor incompatible change: compiler policy re. weakening type
244 checks has changed: now type checks are weakened only if SAFETY < 2
246 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
247 unparsing of directory pathnames as files. Analogously,
248 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
249 filename to parse into a directory pathname.
250 * enhancement: implicit generic function creation now signals a
251 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
252 users can bind handlers for and muffle around calls to LOAD.
253 * enhancement: RUN-PROGRAM allows unicode arguments and environments
254 to be used (using the default stream external format), and allows
255 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
256 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
257 strings and vectors whose element-type or simplicity is not fully
258 known at compile-time.
259 * optimization: STRING-TRIM and related functions no longer allocate
260 a new string when no trimming needs to be performed. These functions
261 are also faster than before when the input string has been declared
263 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
264 * optimization: READ-LINE is significantly faster for files containing
266 * optimization: non-open coded uses of character comparison operators
267 (e.g. char=) no longer cons when called with more than one parameter
268 on platforms supporting dynamic-extent allocation.
269 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
270 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
271 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
273 * bug fix: some sequence functions elided bounds checking when
275 * bug fix: too liberal weakening of union-type checks when SPEED >
277 * bug fix: more bogus fixnum declarations in ROOM implementation
280 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
281 * new feature: MAKE-HASH-TABLE now experimentally accepts a
282 :SYNCHRONIZED argument, which makes the hash-table safe for
283 concurrent accesses (but not iteration.) See also:
284 SB-EXT:WITH-LOCKED-HASH-TABLE, and
285 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
286 * optimization: CONCATENATE on strings is an order of magnitude faster
287 in code compiled with (> SPEED SPACE).
288 * optimization: SUBSEQ is ~50% faster on lists.
289 * bug fix: bug 417 fixed -- source location reporting is now more robust.
290 * bug fix: SUBSEQ on a list will now correctly signal an error if if
291 END is smaller then START.
292 * bug fix: SB-PROFILE will no longer report extra consing for nested
293 calls to profiled functions.
294 * bug fix: ROOM implementation had bogus fixnum declarations which
295 could cause type-errors when calling ROOM on large images.
296 * bug fix: if file compilation is aborted, the partial fasl is now
297 deleted, and COMPILE-FILE returns NIL as the primary value.
298 * bug fix: number of thread safety issues relating to SBCL's internal
299 hash-table usage have been fixed.
300 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
301 be returned from its body when the values were being returned
302 using unknown-values return convection and the W-P-O was wrapped
303 inside an UNWIND-PROTECT.
304 * bug fix: sb-posix should now compile again under Windows, enabling
307 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
308 * incompatible change: hash-table accessor functions are no longer
309 automatically protected by locks. Concurrent accesses on the same hash-table
310 from multiple threads can give inconsistent results or even corrupt the
311 hash-table completely. Multi-threaded applications should do their own
312 locking at the correct granularity. In the current implementation it is
313 still safe to have multiple readers access the same table, but it's not
314 guaranteed that this property will be maintained in future releases.
315 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
316 and will signal an error at runtime.
317 * enhancement: SB-THREAD package now exports a semaphore interface.
318 * enhancement: CONS can now stack-allocate on x86 and
319 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
321 * enhancement: nested lists can now be stack allocated on
322 platforms providing stack allocation support.
323 * enhancement: dynamic-extent support has been extended to support
324 cases where there are multiple possible sources for the stack
326 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
327 if the mutex is uncontested on Linux.
328 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
329 as specified by the CLHS. (thanks to Tobias Rittweiler)
330 * bug fix: NaN comparison now works on x86-64.
331 * bug fix: CLOSE :ABORT T on a stream with pending output now
333 * bug fix: instances of non-standard metaclasses using standard
334 instance structure protocol sometimes missed the slot type checks
336 * bug fix: known functions can be inlined in high-debug code.
338 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
339 * minor incompatible change: the MSI installer on Windows no longer
340 associates .lisp and .fasl files with the installed SBCL.
341 * minor incompatible change: :UNIX is no longer present in *FEATURES*
342 on Windows. (thanks to Luis Oliviera)
343 * new feature: SB-CLTL2 contrib module now implements
344 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
345 * optimization: scavenging weak pointers is now more efficient,
346 requiring O(1) instead of O(N) per weak pointer to identify
347 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
348 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
349 method bodies using constant slot-names when the first argument is
350 a specializer parameter for the method.
351 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
352 names, when the class of the instance is a direct instance
353 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
354 as slow as the constant slot-name case.)
355 * optimization: member type construction is now O(N) instead
357 * optimization: UNION and NUNION are now O(N+M) for large
358 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
360 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
361 now more readable in environments like Slime which display it.
362 (thanks to Tobias C. Rittweiler)
363 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
364 was able to infer the correct argument type for the object on which
365 the CAS operation was being performed.
366 * bug fix: copy propagation interfered with parallel assignment
367 semantics in local calls. (reported by Paul Khuong)
368 * bug fix: the signed modular fixnum shift compiled to wrong code on
369 x86 and x86-64. (spotted by a slight modification to some of
372 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
373 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
374 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
375 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
377 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
378 and generic function now print the object identity for anonymous
379 instances. (thanks to Attila Lendvai)
380 * enhancement: as an extension to MOP, SBCL now supports SETF of
381 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
382 (thanks to Attila Lendvai)
383 * enhancement: sb-cover has an alternate annotation mode which puts
384 the form annotations at the car of the form, instead of annotating
385 it completely. The benefit of this mode is that it shows better
386 which forms were instrumented by the compiler.
387 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
388 by printer control variables. (reported by Dan Corkill)
389 * bug fix: system leaked memory when delayed output was performed by
390 the OS in smaller chunks then expected. (thanks to David Smith)
391 * bug fix: system leaked memory when file streams were not closed
393 * bug fix: large objects written to slow streams that were modified
394 after the write could end up with the modified state written to
395 the underlying file descriptor.
396 * bug fix: multiple threads operating in parallel on the same stream
397 could cause buffer-overflows.
398 * bug fix: source location information is stored correctly
399 (broken since 1.0.6). This bug would generally show up as the
400 Slime debugger highlighting the wrong form when the sldb-show-source
402 * bug fix: Branch forms are again annotated as branches in the sb-cover
404 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
405 on x86-64. (reported by Gregory Vanuxem)
406 * bug fix: EQUALP could return wrong results for structures with raw
407 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
408 word). (reported by Vjacheslav Fyodorov)
409 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
412 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
413 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
414 atomic compare-and-swap operations on threaded platforms.
415 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
416 allows assigning a global minimum value to optimization qualities
417 (overriding proclamations and declarations).
418 * enhancement: closed over variables can be stack-allocated on x86
420 * performance bug fix: GETHASH and (SETF GETHASH) are once again
422 * optimization: slot definition lookup is now O(1). This speeds up
423 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
424 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
425 * optimization: ASSOC and MEMBER can now be open-coded for all
426 combinations of keyword arguments when second argument is constant
427 and SPEED >= SPACE. In other cases a specialized version is
429 * bug fix: using obsoleted structure instances with TYPEP and
430 generic functions now signals a sensible error.
431 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
432 (reported by Kristoffer Kvello)
433 * bug fix: backtrace construction is now more careful when making
434 lisp-objects from pointers on the stack, to avoid creating bogus
435 objects that can be seen by the GC.
436 * bug fix: defaulting of values in contexts expecting more than 7
437 variables now works on x86-64. (reported by Christopher Laux)
438 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
440 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
441 as the property-list of a symbol.
442 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
443 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
444 situations requested, are once again file-compileable. (reported
447 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
448 * MOP improvement: support for user-defined subclasses of
449 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
450 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
451 the expansion of DEFMETHOD, is responsible for generating a form
452 which creates a list of specializers when evaluated. Additional
453 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
454 debugging and introspective support.
455 * minor incompatible change: the (unsupported) spinlock interface
456 has changed: free spinlock now has the value NIL, and a held spinlock
457 has the owning thread as its value.
458 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
459 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
461 * enhancement: name of a socket-stream is now "a socket" instead of
463 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
464 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
465 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
466 for generic functions. (Thanks to Tobias C. Rittweiler)
467 * optimization: bignum printing speed has been improved by 20-40%
468 (depending on the bignum size.)
469 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
471 * bug fix: the cache used by the CLOS to store precomputed effective
472 methods, slot offsets, and constant return values is now thread and
474 * bug fix: generic function dispatch function updating is now thread
475 and interrupt safe (in the sense that the known issues have been
477 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
478 * bug fix: interning EQL-specializers is now thread and interrupt safe.
479 * bug fix: asdf systems with dependencies to the SB-POSIX or
480 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
481 * bug fix: interrupt safety of applicable method computation has been
484 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
485 * new contrib: sb-cover, an experimental code coverage tool, is included
487 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
488 significantly faster.
489 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
490 produce more efficient code on x86-64 if the compiler can determine
491 that (+ INDEX <constant>) does not require a bounds check and FOO
492 has an element type at least 8 bits wide.
493 * enhancement: a new, experimental synchronous timeout facility is
494 provided. Refer to SB-SYS:WITH-DEADLINE for details.
495 * enhancement: when a symbol name conflict error arises, the
496 conflicting symbols are always printed with a package prefix.
497 (thanks to Kevin Reid)
498 * enhancement: stepping is now once again supported on the SPARC and
499 MIPS platforms. (It is also now more likely to work on CheneyGC
501 * enhancement: sb-sprof can now also track and report accurate call
503 * bug fixes: the treatment of non-standard subclasses of
504 SB-MOP:SPECIALIZER is more correct.
505 * incompatible change: PURIFY no longer copies the data from the
506 dynamic space into the static and read-only spaces on platforms
507 that use the generational garbage collector
508 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
510 * bug fix: GC race condition occasionally resulting in crashes with
511 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
513 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
515 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
516 a STYLE-WARNING to be signalled (regression from 1.0.4.)
517 * bug fix: an asynchronous interrupt could previously leave the
518 system running with GC inhibited.
519 * bug fix: a DECLARE form evaluated at top-level now causes an error
520 rather than silently (or verbosely) returning NIL.
521 * bug fix: trying to dissassemble functions compiled at high DEBUG could
522 result in a error being signalled due to source form lookup errors.
523 (reported by Peter Graves)
525 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
526 * incompatible change: removed writer methods for host-ent-name,
527 host-ent-addresses -- changing the values did not update the DNS
529 * minor incompatible change: changed experimental JOIN-THREAD interface
530 * documentation: the manual now lists reader and writer methods
531 in class slot documentation sections. (thanks to Richard M Kreuter)
532 * documentation: unwinding from asyncronous events has been
533 documented as unsafe.
534 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
535 in multithreaded application code.
536 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
537 platforms. (thanks to James Anderson for the optimization hint)
538 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
540 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
541 variants no longer cons.
542 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
543 their NOT- variants no longer cons.
544 * optimization: EQUAL hash tables no longer use SXHASH for objects
545 of all data types, but instead use an EQL hash for types for which
546 EQUAL is the same as EQL.
547 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
548 are significantly faster.
549 * optimization: new STRING-OUTPUT-STREAM implementation is much
550 faster for large outputs and conses 30% less on average.
551 * enhancement: XREF information is now collected for references made
552 to global variables using SYMBOL-VALUE and a constant argument.
553 * enhancement: SIGINT now causes a specific condition
554 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
555 * bug fix: FILE-POSITION used to signal an error in some cases where
556 ANSI requires it to return NIL.
557 * bug fix: ADJUST-ARRAY is now interrupt-safe.
558 * bug fix: adding and removing fd-handlers is now interrupt-safe.
559 * bug fix: inlined calls to C now ensure 16byte stack alignment on
561 * bug fix: bad type declaration in the CLOS implementation has
562 been fixed. (reported by James Anderson)
563 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
564 (thanks to Lutz Euler)
565 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
566 required by the CLHS. (reported by Samium Gromoff)
567 * bug fix: dead unbound variable references now signal an error.
568 (reported by Marco Monteiro)
569 * bug fix: / with an unused value was being deleted in safe code.
570 (thanks to Marco Monteiro and Kevin Reid)
571 * bug fix: number of characters that can be written onto a single
572 line in a file is unlimited.
573 * bug fix: some GC deadlocks caused by asynchronous interrupts have
574 been fixed by inhibiting interrupts for when GC is disabled.
575 * bug fix: some interrupt safety issues with GETHASH, (SETF
576 GETHASH), CLRHASH and REMHASH have been fixed.
577 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
578 specifier no longer causes infinite recursion.
579 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
580 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
581 * bug fix: an &environment argument with the correct variable information
582 is passed to macros that are expanded during byte compilation.
583 (reported by Samium Gromoff)
584 * bug fix: a logic error was causing the hash function for bignums to
585 have excessive amounts of collisions. (reported by Faré Rideau)
586 * bug fix: modifying the contents of an array could change the return
587 value of SXHASH on that array, which is only allowed for strings
588 and bit vectors. (bug introduced in 0.9.16)
589 * improvement: the x86-64/darwin port now passes all tests (except
590 for the debugger tests) but should still be considered
591 experimental until this is fixed.
592 * improvement: a style-warning is signaled for CASE (etc) clauses with
593 duplicate keys. (thanks to Kevin Reid)
594 * improvement: macos/x86 and macos/x86-64 now use mach exception
595 handlers for dealing with illegal instructions (for trapping and
596 error handling) and memory protection violations (for GC).
598 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
599 * new platform: experimental support for x86-64/darwin (MacOS).
600 * incompatible change: the thread-safe (on most platforms) getaddrinfo
601 and getnameinfo sockets functions are used instead of gethostbyaddr
602 and gethostbyname, on platforms where the newer functions are available.
603 As a result, the ALIASES field of HOST-ENT will always be NIL on these
605 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
606 don't print the startup banner, but behave as if --noinform was passed
607 as a command line argument. (thanks to Kevin Reid)
608 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
609 * optimization: code using alien values with undeclared types is much faster.
610 * optimization: the compiler is now able to open code SEARCH in more cases.
611 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
612 * bug fix: using standardized COMMON-LISP special variables as loop
613 variables no longer signals bogus package lock violations. (reported
615 * bug fix: declaring local loop variables to be of a range-limited type
616 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
617 (reported by Andras Simon)
618 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
619 bugs remain on x86-64.)
620 * bug fix: the #= and ## reader macros now interact reasonably with
621 funcallable instances.
622 * bug fix: type-checks for function arguments were compiled using the
623 compiler policy settings of the wrong lexical environment (in 1.0.2
625 * bug fix: SHADOW accepts characters as string designators, as required
626 by the spec (thanks to Eric Marsden)
627 * bug fix: fixed GC safety issues when foreign functions are called with
628 non-base strings as arguments
629 * bug fix: more consistent error output for fatal-compiler-warnings, like
631 * bug fix: functions with proper names were showing up as NIL in
632 backtraces (reported by Edi Weitz, regression in 1.0.2)
634 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
635 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
636 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
637 produce more efficient code on the x86 if the compiler can determine
638 that (+ INDEX <constant>) does not require a bounds check and FOO
639 has an element type at least 8 bits wide.
640 * bug fix: references to undefined variables are handled the same way
641 in toplevel forms as in the normal compiler
642 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
644 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
645 (thanks to Jon Buller)
646 * improvement: faster compilation times for complex functions
647 * improvement: added readlink support to SB-POSIX (thanks to Richard
650 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
651 * improvement: experimental support for mach exception handling on
652 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
654 * improvement: support for GBK external format.
655 (thanks to Chun Tian (binghe))
656 * improvement: the debugger now displays variables that have been closed
657 over, in code compiled with (DEBUG 2) or higher
658 * improvement: support for executable cores on NetBSD (thanks to
660 * new feature: added a RESTART-FRAME debugger command
661 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
662 be used to provide an implementation for FILE-POSITION on Gray streams
663 (thanks to Eric Marsden)
664 * new feature: add syslog, openlog and closelog support to SB-POSIX
665 (thanks to Richard Kreuter)
666 * optimization: the function call overhead in code compiled with
667 a high DEBUG optimization setting is significantly
668 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
669 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
670 (thanks to Tony Martinez)
671 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
672 evaluated once (reported by Kevin Reid)
673 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
674 works again (thanks to Kilian Sprotte)
675 * bug fix: an error is signaled for tagbodies with duplicate tags
676 (thanks to Stephen Wilson)
677 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
678 * bug fix: Win32 port can now handle foreign code unwinding Lisp
679 stack frames from alien callbacks.
680 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
681 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
682 65k pages (thanks to David Woodhouse)
683 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
685 changes in sbcl-1.0.1 relative to sbcl-1.0:
686 * new platform: FreeBSD/x86-64, including support for threading.
687 * new feature: the compiler stores cross-referencing information
688 about function calls (who-calls), macroexpansion (who-macroexpands)
689 and special variables (who-binds, who-sets, who-references) for code
690 compiled with (< SPACE 3). This information is available through the
691 sb-introspect contrib.
692 * new feature: users may subclass SEQUENCE, and have instances of
693 these classes interoperate with standard Common Lisp functions if
694 a number of methods are defined. (This feature is experimental
695 and the interface subject to change based on feedback from SBCL
696 users and the general community)
697 * improvement: sb-sprof traces call stacks to an arbitrary depth on
698 x86/x86-64, rather than the previous fixed depth of 8
699 * improvement: another pthread back-end of mutex "pthread-futex".
700 * bug fix: non-ascii command-line arguments are processed correctly
701 (thanks to Yaroslav Kavenchuk)
702 * bug fix: non-required arguments were not passed correctly when a method
703 defined using DEFMETHOD was called from a mop-generated method using
704 CALL-NEXT-METHOD (reported by Pascal Costanza)
705 * bug fix: recursion is now permitted in accessors through
706 SLOT-UNBOUND. (reported by Pascal Costanza)
707 * bug fix: an error was signaled at startup if the HOME environment
708 variable was defined, but had an empty value (reported by Peter Van Eynde)
709 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
710 signal a PROGRAM-ERROR, not a BUG.
711 * bug fix: SB-POSIX:READDIR works when built with large file support.
712 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
714 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
715 for better automatic indentation support. (thanks to Matt Pillsbury)
716 * optimization: loading generic functions no longer takes O(n^2) time,
717 proportional to the amount of methods in the generic function
718 (reported by Todd Sabin and Jeremy Brown)
719 * optimization: the FIND and POSITION family of sequence functions
720 are significantly faster on arrays whose element types have been
722 * improvements to the Windows port:
723 ** Intermittent heap corruption problems have been fixed. (thanks
724 to Alastair Bridgewater)
725 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
727 ** Lisp is able to unwind foreign exception frames from alien
728 callbacks. (thanks to Alastair Bridgewater)
730 changes in sbcl-1.0 relative to sbcl-0.9.18:
731 * improvement: experimental support for threading on FreeBSD/x86.
732 (thanks to NIIMI Satoshi)
733 * improvement: runtime option --dynamic-space-size can be used
734 to set the size of the dynamic space reserved on startup.
735 * improvement: floating point modes in effect are now saved in
736 core, and restored on startup.
737 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
738 startup, not time since first call to GET-INTERNAL-REAL-TIME.
739 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
740 threads are not running after *SAVE-HOOKS* have run.
741 * improvement: writes to CLOS instance slots are type-checked in code
742 compiled with (SAFETY 3)
743 * improvement: floating-point exception handling on FreeBSD (thanks to
745 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
746 (thanks to Zach Beane)
747 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
749 * improvement: added support for the Shift-JIS external format.
750 (contributed by NIIMI Satoshi)
751 * improvement: callbacks are supported on Linux/PPC. (thanks to
753 * bug fix: compiler bug triggered by a (non-standard) VALUES
754 declaration in a LET* was fixed. (reported by Kaersten Poeck)
755 * bug fix: file compiler no longer confuses validated and already
756 dumped structurres. (reported by Kaersten Poeck)
757 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
758 fill-pointer signals a type-error as required. (thanks to
760 * bug fix: disassemly of funcallable instances works.
761 * bug fix: single stepping on PPC.
762 * bug fix: fix thread-safety problems in the type system (generally
763 manifesting as nonsensical errors like "STRING is a bad type specifier
764 for sequences" or "The value 1 is not of type FIXNUM")
765 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
766 (reported by Josip Gracin)
767 * bug fix: an error is signaled for attempts to displace arrays with
768 incompatible element types (thanks to Mario Mommer)
769 * bug fix: more correct handling of wide characters in debug info
770 (bug reported by Attila Lendvai and fixed by Juho Snellman)
771 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
772 and don't cause extra consing
773 * optimization: MAP and MAP-INTO are significantly faster on vectors
774 whose elements types have been declared.
775 * Improvements to SB-SPROF:
776 ** Support for allocation profiling
777 ** Reduced profiling overhead, especially for long profiling runs
778 * Improvements to the Windows port:
779 ** floating point exceptions are now reported correctly.
780 ** stack exhaustion detection works partially.
781 ** more accurate GET-INTERNAL-REAL-TIME.
782 ** better RUN-PROGRAM behaviour in case of errors in the child
784 ** PROBE-FILE now simplifies pathnames correctly.
785 ** DIRECTORY now works correctly with :WILD-INFERIORS.
786 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
787 1st 1970 (although time timezone data will be unreliable).
788 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
790 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
791 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
792 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
794 * bug fix: two potential GC deadlocks affecting threaded builds.
795 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
796 returns T (reported by Anton Kazennikov)
797 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
798 with non-variable places
799 * bug fix: remove a race condition in the setting of
800 funcallable-instance functions, this should make threaded CLOS
801 code more stable against memory faults.
802 * bug fix: corruption of specials when unbinding is interrupted by an
803 asynchronous unwind (reported by Hannu Koivisto)
804 * improvement: the debugger will now also display local variables that
805 are only used once, for code compiled with a DEBUG optimization quality
808 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
809 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
810 * incompatible change: External-format support for FFI calls. The
811 SB-ALIEN:C-STRING no longer implies an ASCII
812 external-format. Instead the string is subject to external-format
813 conversion. Additionally return values of type SB-ALIEN:C-STRING
814 are by default of type (SIMPLE-ARRAY CHARACTER), not
815 SIMPLE-BASE-STRING. For an alien type that behaves like the old
816 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
817 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
818 * incompatible change: SB-EXT package no longer contains the
819 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
820 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
822 * incompatible change: the single-stepper is no longer available
823 on Alpha, Mips and Sparc platforms.
824 * minor incompatible change: the direct superclasses of
825 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
826 not (STANDARD-OBJECT FUNCTION). This makes the
827 class-precedence-lists of GENERIC-FUNCTION and
828 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
830 * minor incompatible change: the default stream external format on
831 non-unicode SBCL is no longer detected from the locale, but is always
833 * new feature: Add a version of evaluator that uses an interpreter instead
834 of the compiler. EVAL still uses the compiler by default, to switch it
835 to use the interpreter, set the value of the variable
836 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
837 * minor incompatible change: the single-stepper REPL has been merged
838 with the normal debugger (see the "Stepping" heading of the debugger help
839 for more details). The debugger command STEP will no longer switch
840 to the single-stepper REPL.
841 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
842 for a type now works.
843 * bug fix: loading of default sysinit file works. (thanks to Leonid
845 * bug fix: better detection of circularities in the file-compiler.
846 (reported by Marco Monteiro)
847 * bug fix: the CL pathname functions now work with files that have
848 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
849 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
850 systems with Unix98 pty semantics.
851 * bug fix: ASDF-INSTALL will now work with bsd tar.
852 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
854 * bug fix: timers expiring in dead threads no longer cause a
855 type-error (reported by Paul "Nonny Mouse").
856 * bug fix: thanks to more lightweight single-stepper instrumentation,
857 code compiled with (DEBUG 3) will compile and execute significantly faster,
858 and will have more accurate type-inferencing than before
859 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
860 whose bindings are modified
861 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
862 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
863 ** replace dummy implementations of CL:MACHINE-INSTANCE and
864 CL:SOFTWARE-VERSION with proper version
866 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
867 * feature: implemented the READER-METHOD-CLASS and
868 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
869 as specified by AMOP.
870 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
872 * optimization: faster LOGCOUNT implementation on x86 and x86-64
873 (thanks to Lutz Euler)
874 * optimization: hashing of general arrays and vectors has been
875 improved. (reported by Any Fingerhut)
876 * enhancement: SB-INTROSPECT is now able to find definitions of
877 profiled functions. (thanks to Troels Henriksen)
878 * enhancement: compiler-macro expansion applies now to FUNCALL forms
879 as well, allowing compiler-macros for SETF-functions to expand.
880 * enhancement: step-instrumentation no longer wraps known
881 single-value functions with multiple-value context, allowing
882 better type inference.
883 * fixed bug #337: use of MAKE-METHOD in method combination now works
884 even in the presence of user-defined method classes. (reported by
885 Bruno Haible and Pascal Costanza)
886 * fixed bug #339(c): if there are applicable methods not part of any
887 long-form method-combination group, call INVALID-METHOD-ERROR.
888 (reported by Bruno Haible)
889 * fixed bug #361: the :FUNCTION initarg in the protocol for
890 initialization of methods can now be used to override
891 internally-produced optimized functions. (reported by Bruno
893 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
894 system-provided lambda expression no longer cause warnings about
895 unbound #:|pv-table| symbols.
896 * bug fix: improved the handling of type declarations and the
897 detection of violations for keyword arguments with non-constant
899 * bug fix: potentially erronous calls to PATHNAME and
900 MERGE-PATHNAMES were being flushed in some cases.
901 (reported by Richard Kreuter)
902 * bug fix: compiled calls to TYPEP were mishandling obsolete
903 instances. (reported by James Bielman and Attila Lendvai)
904 * bug fix: format strings with ~> without matching ~< no longer
905 trigger an AVER, but signal an understandable error instead.
906 (reported by Antonio Martinez)
907 * bug fix: specifying an output-file without a pathname-type for
908 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
909 of input-file instead of "fasl". (reported by Robert Dodier)
910 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
911 (reported by James Y Knight).
912 * bug fix: compiler-macros-function did not consider the environment
913 argument for shadowing by local functions.
914 * bug fix: compiler-macros expansion was inhibited by local INLINE
916 * bug fix: inline expansions of known functions were subject to
917 step-instrumentation in high debug policies, leading to problems
919 * bug fix: compiler failed to differentiate between different CONS
921 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
922 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
923 * bug fix: #368: incorrect use of expressed vs. upgraded array
925 * bug fix: #306a: more precise unions of array types.
926 * thread-safety improvements:
927 ** CONDITION-WAIT could return early on Linux, if the thread was
928 interrupted and subsequently continued with SIGCONT.
929 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
931 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
932 * added support for the ucs-2 external format. (contributed by Ivan
934 * minor incompatible change: pretty printing of objects of type
935 (cons symbol) is, in the default pprint-dispatch-table, now
936 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
938 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
939 called later in a class's lifetime, possibly as late as when the
940 first instance of the class is created. Previously,
941 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
942 class became finalizeable.
943 * fixed bug: FILE-POSITION sometimes returned inconsistent results
944 for multibyte external-format streams. (thanks to "vbzoli")
945 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
946 with :ALLOCATION :CLASS inherited from superclasses of the
948 * fixed bug: anonymous classes can now be created using the :NAME
949 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
950 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
951 * fixed bug: core-files saved with :EXECUTABLE T can again be
952 executed when SBCL_HOME isn't set. (reported by James Knight)
953 * fixed bug: toplevel LOCALLY forms with declarations could
954 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
955 * fixed bug: printing from several different threads using different
956 values of *print-case* could cause invalid output, due to
957 some internal special variables of the printer not being bound
958 thread-locally (reported by Max Mikhanosha)
959 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
960 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
961 (reported by Pascal Costanza)
962 * fixed bug: FINALIZE-INHERITANCE is called from
963 REINITIALIZE-INSTANCE on classes when the class has previously
964 been finalized, as required by AMOP.
965 * minor code generation optimizations:
966 ** better register allocation in CLOS dispatching functions
967 ** overflow detection when coercing signed bytes to fixnums on x86-64
968 ** is now implemented with one IMUL instruction instead of three shifts
969 ** more efficient bit-vector access on x86 and x86-64
970 ** more efficient access to raw structure slots on x86 and x86-64
971 * fixed some bugs revealed by Paul Dietz' test suite:
972 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
975 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
976 * feature: thread support on Solaris/x86, and experimental thread support
978 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
980 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
981 default method for SLOT-UNBOUND.
982 * minor incompatible change: prevent the user from specializing the
983 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
984 somewhat counter to the intent of the protocol, I (CSR) think, and
985 additionally it just doesn't work in SBCL as currently
986 implemented, thanks to optimizations (that are always valid for
987 the other three SLOT-VALUEish functions, but not for the setter).
988 * minor incompatible change: the :SB-LDB feature is now enabled by
989 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
990 the low-level debugger.
991 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
992 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
993 down to an SB-SYS:FD-STREAM, allowing direct communication in
994 more cases. Temporary files and pipes are still used as a fallback
996 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
997 on the non-reentrant C functions localtime(3) and gmtime(3).
998 * bug fix: global optimization policy was not visible in LOCALLY and
1000 * bug fix: class objects can be used as specializers in methods.
1001 (reported by Pascal Costanza)
1002 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1003 with a type error. (reported by blitz_ on #lisp)
1004 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1005 failed with a type error. (reported by Pascal Bourguignon)
1006 * bug fix: merging pathnames against defaults with :DIRECTORY
1007 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1008 (reported by James Y Knight)
1009 * bug fix: saving large (>2GB) cores on x86-64 now works
1010 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1011 constant index and a (simple-array (signed-byte 32)) array
1012 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1013 error (patch by Robert J. Macomber)
1014 * feature: TIME now displays the amount of run-time spent in GC
1015 * bug fix: The debugger now does a better job of respecting
1016 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1017 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1018 workaround for bug 403.)
1019 * bug fix: floating point exception handling now works on FreeBSD
1020 (thanks to NIIMI Satoshi)
1021 * fixed some bugs revealed by Paul Dietz' test suite:
1022 ** MISC.641: LET-conversion were not supposed to work in late
1024 * improvements to the Win32/x86 port:
1025 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1026 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1028 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1029 * new feature: source path information is generated for macro-expansion
1030 errors for use in IDE's like Slime (thanks to Helmut Eller)
1031 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1032 * bug fix: compiler does not loop forever on an invalid type in
1034 * improvement: compilation of most CLOS applications is significantly
1036 * optimization: added a limited bytecode compiler for simple toplevel
1037 forms, speeding up compilation and FASL loading
1038 * bug fix: the statistical profiler now properly distinguishes anonymous
1041 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1042 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1043 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1044 into subclasses of the finalized class. Now user calls to
1045 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1046 system return before any subclasses are finalized.
1047 * minor incompatible change: The reader no longer ignores errors
1048 regarding non-existent packages in #+ and #- feature tests.
1049 * new feature: command line options --no-sysinit, --no-userinit to
1050 inhibit loading the corresponding init files
1051 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1052 for sending data through UDP sockets (thanks to François-René Rideau)
1053 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1054 error is signalled instead (thanks to François-René Rideau)
1055 * minor incompatible change: Error signalling behaviour of lexical
1056 operations violating package locks has changed slightly. Refer to
1057 documentation on package locks for details.
1058 * enhancement: EVAL can process IF-expressions without resorting to the
1060 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1061 constant-folded properly due to missing out-of-line functions.
1062 (reported by Mika Pihlajamäki)
1063 * bug fix: LISTEN sometimes returned T even in cases where no data was
1064 immediately available from the stream
1065 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1066 were reversed. (reported by Levente Mészáros)
1067 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1068 when given a vector displaced to an adjustable array. (reported
1070 * bug fix: derivation of float boundaries from numbers outside the
1071 appropriate float range (reported by John Wiseman)
1072 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1074 * bug fix: precomputing cache entries for generic functions with
1075 some subclasses of specializers as yet invalid does not attempt to
1076 fill a cache line with a negative offset. (reported by Levente
1078 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1079 allow condition class objects as arguments to DOCUMENTATION and
1080 (SETF DOCUMENTATION); only find and set documentation for
1081 structure names for the STRUCTURE doc-type. (suggested by Gary
1083 * improvements to the Win32/x86 port:
1084 ** (user-homedir-pathname) and default initialization file
1085 locations now know about the user's "Documents and Settings"
1086 directory (thanks to Yaroslav Kavenchuk)
1087 ** run-program is implemented (thanks to Mike Thomas)
1088 ** sockets support (thanks to Timothy Ritchey)
1089 ** better backtrace support (thanks to Alastair Bridgewater)
1090 ** sb-grovel supported
1091 ** asdf-install and sb-posix work somewhat
1092 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1093 * improvements to the Solaris/x86 port:
1094 ** works on Solaris 11/Solaris Express
1095 ** floating-point exception handling support
1096 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1097 * fixed some bugs revealed by Paul Dietz' test suite:
1098 ** REMOVE-METHOD returns its generic function argument even when
1099 the method is not one of the generic functions' methods.
1100 ** objects declared as MEMBER types can be admissible arguments to
1101 structure accessors.
1102 ** printing characters should simply be printed by the FORMAT ~:C
1104 ** compiler failure when compiling functions with hairy constant
1105 defaults for optional parameters.
1106 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1107 function, which is already optimized.
1109 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1110 * new platform: experimental support for SBCL x86/Darwin, including
1111 MacOS X 10.4.5 on Intel.
1112 * new feature: Unicode character names are now known to the system
1113 (through CHAR-NAME and NAME-CHAR).
1114 * new feature: the filesystem location of SBCL's core file is
1115 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1116 * minor incompatible change: the contrib modules SB-POSIX and
1117 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1118 this change is to make it easier to distribute
1119 location-independent binaries.
1120 * bug fix: as implied by AMOP, standardized classes no longer have
1121 slots named by external symbols of public packages. (reported by
1123 * optimization: calling functions via constant symbols -- as in
1124 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1125 via the function object as in (FUNCALL #'FOO).
1126 * optimization: on x86 and x86-64, the calling convention for the
1127 case of dealing with an unknown number of values has been altered
1128 to be friendlier to the prediction heuristics implemented,
1129 particularly with reference to CALL and RET pairing. (thanks to
1130 Alastair Bridgewater)
1131 * optimization: on x86, the code for access to thread-local symbol
1132 values and binding/unbinding of thread-local symbols is smaller.
1133 (thanks to Alastair Bridgewater)
1134 * enhancement: CONSTANTP is now able to determine constantness of
1135 more complex forms, including calls to constant-foldable standardized
1136 functions and some special forms beyond QUOTE.
1137 * fixed bug: occasional GC crashes on Solaris/x86
1138 * optimization: x86-64 supports stack allocation of results of simple
1139 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1140 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1141 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1142 (thanks to James Knight)
1143 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1144 by the standard (thanks to Stephen Compall)
1146 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1147 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1148 be used for bundling the runtime and the core file into one
1149 executable binary. This feature is not currently supported on all SBCL
1150 platforms. (thanks to James Bielman and NIIMI Satoshi)
1151 * new feature: a generational or ephemeral garbage collector is now
1152 the default on the PowerPC platform (both Linux and Darwin). The
1153 old Cheney (stop and copy) collector is a build-time option.
1154 (thanks to Cyrus Harmon, after Raymond Toy)
1155 * minor incompatible change: the method by which SBCL finds its
1156 contributed modules has changed; it no longer relies on symbolic
1157 links from an $SBCL_HOME/systems directory, but searches directly
1158 in the subdirectories of $SBCL_HOME.
1159 * enhancement: the dynamic heap size on the Linux/PPC platform is
1160 markedly larger, even using the older Cheney garbage collector.
1161 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1162 certain complicated string types. (reported by Gary King)
1163 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1164 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1165 * fixed bug: DOCUMENTATION on structures no longer signals an
1166 error if no structure type of the right name exists. (reported by
1168 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1169 applicable non-standard primary or :AROUND method on
1170 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1171 no longer cause unbound variable errors. (reported by Kevin Reid)
1172 * fixed bug: metacircle resolution in cases where methods have slots
1173 added before the slots from STANDARD-METHOD. (reported by Jean
1175 * fixed bug: the Power PC assembly code for calling into lisp
1176 sometimes computed its return address wrongly (depending on the
1177 15th bit of the address, fixed at link time). This used to
1178 manifest itself in a segmentation violation while building PCL.
1179 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1180 many others over the years)
1181 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1182 the stack pointer (treated as a signed integer) is negative.
1183 (thanks to Peter van Eynde)
1185 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1186 * new platform: experimental support for the Windows operating
1187 system has been added. (thanks to Alastair Bridgewater)
1188 * fixed several bugs in and robustified the PPC FFI (including
1189 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1190 * bug fix: allow non-simple string symbol names (reported by Paul
1192 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1194 * bug fix: saving a core corrupted callbacks on x86/x86-64
1195 * bug fix: closed a loophole in metacircularity detection and
1196 grounding in the PCL implementation of CLOS.
1197 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1198 (reported by tomppa on #lisp)
1199 * bug fix: clear the direction flag on Lisp -> C transitions, as
1200 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1201 (reported by Andrej Grozin and Hendrik Maryns)
1202 * optimization: major improvements to GC efficiency on GENCGC platforms
1203 * optimization: faster implementation of EQUAL
1204 * optimization: emit more efficient opcodes for some common
1205 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1206 * optimization: several other minor code-generation improvements on x86-64
1207 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1209 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1210 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1211 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1212 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1213 by ANSI as a generic function, and for consistency (SETF
1214 GENERIC-FUNCTION-NAME) is treated likewise.
1215 * fixed bug #233.b: make constraint propagation notice when a variable
1216 value is changed after it is referenced but before it is used
1217 * fixed bug #296: no more arbitrary behaviour from filenames with
1218 odd characters as --load arguments. (reported by Adam Warner)
1219 * fixed bug #390: :CHARACTER-SET pathname components now work as
1220 expected. (reported by Tim Daly Jr)
1221 * fixed bug #391: complicated :TYPE intersections in slot
1222 definitions no longer cause an error in PCL internals.
1223 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1224 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1225 returning the number of octets which would be written to the
1226 file-stream. (thanks to Robert J. Macomber)
1227 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1228 arguments to the internal stream functions. (thanks to David
1230 * bug fix: stack exhaustion is now handled on the x86/SunOS
1231 platform. (thanks to Daisuke Homma)
1232 * bug fix: more accurate ROOM results on GENCGC platforms
1233 * optimization: improved type inference for arithmetic-for
1234 index variables in LOOP
1235 * optimization: faster floating-point SQRT on x86-64
1236 * fixed some bugs revealed by Paul Dietz' test suite:
1237 ** DOCUMENTATION returns NIL instead of "" for method combinations
1238 that don't have a docstring
1240 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1241 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1242 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1243 therefore are not customizeable by user code (as seems to be at
1244 least permitted and maybe required by AMOP). As a consolation,
1245 however, the SBCL implementation of these functions now calls
1246 REINITIALIZE-INSTANCE as specified by AMOP.
1247 * bug fix: slot-definition documentation is propagated as per ANSI
1248 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1249 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1250 argument, as do eslotd initialization methods. (from Pascal
1251 Costanza's "Closer" project)
1252 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1253 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1255 * bug fix: it is now possible to have more than one subclass of
1256 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1257 (reported by Bruno Haible, Pascal Costanza and others)
1258 * bug fix: the dependent update protocol now works for generic
1259 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1260 and Pascal Costanza)
1261 * bug fix: condition-class instances corresponding to
1262 DEFINE-CONDITION forms are now created eagerly. (reported by
1263 Kalle Olavi Niemitalo on comp.lang.lisp)
1264 * bug fix: floating point printing is more accurate in some
1265 circumstances. (thanks to Simon Alexander)
1266 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1267 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1268 * bug fix: callbacks on OS X now preserve stack-alignment.
1269 (thanks to Cyrus Harmon)
1270 * enhancement: the x86-64 disassembler is much better at
1271 disassembling SSE instructions. (thanks to Lutz Euler)
1272 * enhancement: CHECK-TYPE on a variable now allows type-inference
1273 (thanks to Vincent Arkesteijn)
1274 * optimization: improved performance of EUC-JP external format.
1275 (thanks to NIIMI Satoshi)
1276 * optimization: performance improvements to IO on file streams of
1277 :ELEMENT-TYPE CHARACTER
1278 * optimization: much faster memory allocation on x86-64
1279 * optimization: faster garbage collections (latency and throughput)
1281 * optimization: faster bignum subtraction and division on x86 and x86-64
1282 (thanks to Lutz Euler)
1283 * optimization: more accurate type-inference for loops that use a
1284 floating point index variable or a negative step.
1286 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1287 * bug fix: add a workaround to SBCL looping infinitely at startup on
1288 Linux kernels with apparently buggy implementations of personality().
1289 (thanks to Svein Ove Aas)
1290 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1291 (thanks to David Lichteblau)
1292 * bug fix: local bindings shadow global symbol macros for
1293 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1294 * optimization: non-open coded uses of numeric comparison operators
1295 (e.g. >) no longer cons when called with more than one parameter
1296 on platforms supporting dynamic-extent allocation.
1297 * enhancement: saving cores with foreign code loaded is now supported
1298 on MIPS/Linux in addition to the previously supported platforms.
1299 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1300 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1301 * bug fix: division by zero in sb-sprof when no samples were collected
1302 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1303 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1304 (thanks to Alexander Kjeldaas)
1305 * new feature: ignoring the return values of destructive functions
1306 like NREVERSE signals a compile-time style-warning.
1307 (thanks to Kevin Reid)
1308 * new feature: the alignment of alien structure fields can be
1309 explicitly specified. (thanks to Cyrus Harmon)
1310 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1311 and dump core on SIGQUIT
1313 ** incompatible change: threads do not inherit values of specials
1314 from their parents (see manual)
1315 ** bug fix: threads stacks belonging to dead threads are freed by the
1316 next exiting thread, no need to gc to collect thread stacks anymore
1317 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1318 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1319 ** bug fix: dynamic variable and thread start related gc lossage
1320 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1322 * fixed some bugs revealed by Paul Dietz' test suite:
1323 ** SUBTYPEP is slightly more accurate on heinously complicated
1324 CONS types where some of the members have uncertain (in the
1325 NIL, NIL sense) type relationships to each other.
1327 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1328 no longer purified unless :PURIFY T is explicitly specified.
1329 ** Non-purified cores are significantly smaller than before
1331 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1332 * new feature: timers based on Zach Beane's excellent timer package
1333 * added support for the following external formats: koi8-u,
1334 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1335 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1336 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1337 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1338 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1339 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1340 * incompatible change: a threaded SBCL will no longer revert to
1341 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1342 * bug fix: interrupts are disabled until startup is complete; no
1343 more sigsegvs when receiving a signal to soon
1344 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1346 * bug fix: add a workaround for the memory randomization features in
1347 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1348 workaround will only be in effect on systems with the proc filesystem
1350 * bug fix: printing objects of type HASH-TABLE signals a
1351 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1353 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1354 have been running for over 50 days. (reported by Gilbert Baumann)
1355 * bug fix: the logic for getting names of functions gets less
1356 confused when confronded with alternate-metaclass
1357 funcallable-instances. (reported by Cyrus Harmon)
1358 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1359 consistent, even on internal alternate-metaclass objects.
1360 * bug fix: flush closure information collected by physical
1361 environment analysis prepass before the main pass. (bug reported
1363 * bug fix: compiler pack phase does not modify a hashtable, which is
1364 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1365 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1366 may be expanded. (reported by Alexander <malishev> on c.l.l)
1367 * bug fix: no more style-warnings when compiling code using LAST
1368 under high-SPEED low-SPACE settings. (reported by David Wallin
1370 * fixed some incompatibilities between SBCL's MOP and the MOP
1372 ** the METAOBJECT class is now implemented;
1373 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1374 STANDARD-OBJECT, as required;
1375 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1376 now compatible (as required by VALIDATE-SUPERCLASS); there
1377 remains a consistency requirement over the presence or absence
1378 of the FUNCTION class in the superclass list of finalized
1379 classes; see the manual for more details;
1380 ** the :AROUND method for COMPUTE-SLOTS on
1381 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1382 requested slot ordering.
1384 ** bug fix: parent thread now can be gc'ed even with a live
1386 ** bug fix: binding a special with PROGV to no value is not
1387 the same as that symbol not having been bound (thanks to
1389 ** bug fix: binding specials is thread safe (thanks to
1391 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1392 and SB-SPROF do not die with 'no handler for signal XX in
1393 interrupt_handle_now(..)' anymore
1394 ** bug fix: WITH-TIMEOUT works with multiple running threads
1395 * fixed some bugs revealed by Paul Dietz' test suite:
1396 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1397 the :method-class keyword argument.
1399 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1400 * new port: the Solaris operating system on x86 processors is now
1401 mostly supported, though some rough edges in the environment
1402 remain. (thanks to Daisuke Homma)
1403 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1404 space for its heap. (thanks to Thiemo Seufer)
1405 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1406 closures. (thanks to Thiemo Seufer)
1407 * minor incompatible change: eof selects abort in the debugger.
1408 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1409 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1411 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1412 deprecated, and will go away in a future revision of SBCL.
1413 * minor incompatible change: GC-ON and GC-OFF are no longer
1414 implemented with a counter, it does not matter how many times gc
1415 is switched on or off
1416 * bug fix: discriminating functions for generic function classes
1417 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1418 make invalid assumptions about method precedence order. (reported
1420 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1421 platforms. (thanks to Thiemo Seufer)
1422 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1423 * bug fix: correct pathname printing with printer escaping is on.
1424 (thanks to Kevin Reid)
1425 * bug fix: complex VOP definitions in "user-space" no longer trigger
1426 package locks. (reported by Zach Beane)
1427 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1428 for funcallable-instance functions, and is no different from
1430 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1431 characters in the current readtable. (reported by Nicholas Neuss)
1432 * bug fix: SUBTYPEP on various CONS types returns more a more
1433 accurate acknowledgment of its certainty.
1434 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1435 lists in some cases. This partially fixes bug 384.
1436 * flush all standard streams before prompting in the REPL and the
1438 * bug fix: signal handling and triggering gc do not conflict
1439 directly anymore, in particular a high frequency sb-sprof does
1440 not prevent gc from running
1441 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1442 approximation for timezone and DST information between the
1443 universal time epoch and the smallest negative 32-bit time_t.
1444 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1445 year 1899 so long as the encoded time is non-negative.
1446 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1447 the correct answer for non-constant rotate amounts when performing
1448 an inline 32-bit rotation.
1450 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1451 there is only one thread in the session
1452 ** bug fix: memory leak for streams created in one thread and
1453 written to in another
1454 ** bug fix: lockup when compiled with gcc4
1455 ** bug fix: race that allows the gc to be triggered when gc is
1457 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1458 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1460 ** bug fix: finished threads are gc'ed properly
1461 * fixed some bugs revealed by Paul Dietz' test suite:
1462 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1463 has arguments with a different set of applicable methods from
1464 the orignal arguments.
1465 ** The type error thrown by MAP now has a correct expected-type
1467 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1468 name a compiled function.
1469 ** Three MISC tests where a large bignum was improperly coerced to
1470 a float that couldn't represent that bignum during type
1471 derivation were fixed.
1472 ** SUBTYPEP can now handle types involving the negation of a
1473 list-form FUNCTION type.
1474 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1475 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1476 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1478 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1479 * New feature: Experimental support for bivalent streams: streams
1480 opened with :element-type :default now allow character and binary
1481 (unsigned-byte 8) I/O
1482 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1483 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1485 * bug fix: no more highly sporadic "couldn't check whether ~S is
1486 readable" when reading a stream and an interrupt hits in the middle
1487 of a select system call
1488 * compiler better recognizes complex arrays (reported by Eduardo
1490 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1491 for variables on SBCL built with linkage-tables. (reported by Luis
1493 * various error reporting improvements.
1494 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1495 (Thanks to Thiemo Seufer)
1496 * bug fix: nested reader invokations work correctly
1497 * bug fix: it is possible to have simultaneous references to foreign
1498 code and foreign data with the same name.
1500 ** added x86-64 support
1501 ** incompatible change: the threading api now works with thread
1502 objects instead of thread ids
1503 ** bug fix: threads are protected from signals and interruption when
1504 starting up or going down
1505 ** bug fix: a race where an exiting thread could lose its stack to gc
1506 ** bug fix: don't halt on infinite error in threads if possible
1507 ** fixed numerous gc deadlocks introduced in the pthread merge
1508 ** bug fix: fixed thread safety issues in read and print
1509 ** bug fix: debugger doesn't hang on session lock if interrupted at
1510 an inappropriate moment
1511 ** bug fix: run-program is now thread safe(r)
1512 ** bug fix: inner with-recursive-lock no longer releases the mutex
1513 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1514 * fixed some bugs revealed by Paul Dietz' test suite:
1515 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1516 EXPECTED-TYPE slots filled.
1517 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1519 ** MAKE-CONDITION accepts classes as type-designators.
1520 ** COMPILE may never return NIL.
1521 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1522 range before calling Unix time functions
1524 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1525 * numerous signal handling fixes to increase stability
1526 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1527 * minor incompatible change: we now correctly canonize default
1528 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1529 as per the MOP, rather than the historical (INITARG INITFUNCTION
1530 INITFORM). (reported by Bruno Haible)
1531 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1532 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1533 for more information.
1534 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1535 pathname is a directory pathname.
1536 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1537 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1539 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1540 :METACLASS argument in addition to a class name. (reported by
1541 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1542 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1543 (reported by Bruno Haible; patch from Gerd Moellmann)
1544 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1546 * bug fix: more cleanups to the floating point exception handling on
1547 x86-64 (thanks to James Knight)
1548 * bug fix: the compiler does not try to constant fold calls of
1549 COERCE and BIT-* functions when they return freshly constructed
1550 objects. (reported by Nikodemus Siivola and Paul Dietz)
1551 * optimization: improved the MIPS versions of generic (in the
1552 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1553 * optimization: direct conversion of (unsigned-byte 32) to floats on
1554 the PowerPC platform.
1555 * optimization: structure instances with raw slots now use less
1556 memory, and probably show better memory locality. (thanks to
1558 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1559 are treated as requests for stack allocation on the x86-64,
1560 Alpha-32, PPC and SPARC platforms.
1561 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1562 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1563 (thanks to Tony Martinez)
1565 ** gcing a dead thread can no longer lead to lockups
1566 ** threads block signals until they are set up properly
1567 ** errno is no longer shared by threads
1568 ** interrupt-thread restores the eflags register on x86
1569 ** fixed some lockups due to gc/thread interaction
1570 * fixed some bugs revealed by Paul Dietz' test suite:
1571 ** invalid dotted lists no longer raise a read error when
1572 *READ-SUPPRESS* is T
1573 ** signal an error if a symbol that names a declaration is used
1574 as the name of a type, or vice versa
1575 ** allow using the (declare (typespec var*)) abbreviation for
1576 (declare (type typespec var*)) with all type specifiers
1577 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1578 ** always return NIL from PPRINT-POP when OBJECT is NIL
1579 ** don't signal errors when pretty-printing malformed LABELS,
1580 FLET or MACROLET forms
1581 ** declarations in a DOLIST body are also in scope for the
1583 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1585 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1588 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1589 * fixed cross-compiler leakages that prevented building a 32-bit
1590 target with a 64-bit host compiler.
1591 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1592 opened with :IF-EXISTS :OVERWRITE.
1593 * fixed bug 281: error for an invalid qualifier in a short-form method
1594 combination method is not signalled until the faulty method is called.
1595 * bug fix: iteration variable type inferrer failed to deal with open
1596 intervals. (reported by Alan Shields)
1597 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1598 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1599 or more runtime options were provided to the sbcl binary.
1600 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1601 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1602 * optimization: there's now a fast-path for fixnum arguments in the
1603 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1604 * optimization: the code generated on x86-64 is more compact thanks
1605 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1606 * bug fix: floating-point exception handling now partly works on
1607 x86-64. (Thanks to James Knight)
1608 * improvement to the MIPS backend from Thiemo Seufer: C-style
1609 64-bit long long arguments and return values to alien functions
1611 * fixed some bugs revealed by Paul Dietz' test suite:
1612 ** the type-error signalled from WARN has a filled-in DATUM slot.
1613 ** the type-error required when a stream is not associated with
1614 a file has the stream as its datum.
1615 ** type-errors on single-floats on x86-64 no longer have
1616 :INVALID-OBJECT as the datum
1617 ** the type-errors signalled for invalid function names now have
1618 a correct expected type
1619 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1620 for structures defined with a :TYPE. Documentation strings for
1621 typed structures are no longer immediately discarded
1622 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1623 broadcast streams and synonym streams. FILE-LENGTH now also works
1624 on broadcast streams.
1626 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1627 * incompatible change: the --noprogrammer option, deprecated since
1628 version 0.7.5, has been removed. Please use the equivalent
1629 --disable-debugger option instead.
1630 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1632 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1633 inoperational for a while now) has been completely removed.
1634 * null lexical environments are now printed as #<NULL-LEXENV>,
1635 significantly reducing the amount of clutter in typical backtraces.
1636 * documentation on weak pointers, finalization, and after GC hooks
1637 has been added to the manual.
1638 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1639 as other specialized array types, is much faster. SUBSEQ and
1640 COPY-SEQ on such arrays have also been sped up.
1641 * optimization: EQL is now more efficient when at least other argument
1642 is known to be of type (OR FIXNUM (NOT NUMBER)).
1643 * fixed bug: the runtime is now less vulnerable to changes in the
1644 size of the SBCL object on OS X, and virtual memory is reserved for
1645 all spaces that need to be at a fixed address.
1646 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1647 * fixed bug: finalizers and after GC hooks that cause consing are now
1649 * fixed bug: compiler error messages and summaries are now printed to
1650 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1651 * fixed inference of the upper bound of an iteration variable.
1652 (reported by Rajat Datta).
1653 * fixed bug 211e: calling local functions with duplicated constant
1654 keyword argument no longer causes a bogus style warning about an
1656 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1657 ftype declaration to be disregarded. (reported by Dave Roberts)
1658 * fixed bug 373: caused by erronous compilation of references to alien
1659 variables in the runtime on ppc/darwin.
1660 * fixed bug 376: CONJUGATE type deriver.
1661 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1662 (reported by Baughn on #lisp)
1663 * a cross-compiler bug on non-x86 platforms has been identified and
1664 fixed. (thanks to Bruno Haible)
1665 * improvements to the MIPS runtime code for increased stability.
1666 (thanks to Thiemo Seufer)
1667 * increased the maximimum compact environment size to allow
1668 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1669 * partial workaround for bug 135: don't name the function we're
1670 calling for hairy cases of EVAL, so as not to accumulate one environment
1671 entry per call to EVAL. (reported by Kevin Reid)
1672 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1673 * optimization: type testing for non-vector arrays should be faster.
1674 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1676 * bug fix: setting 31st element of a bit vector to zero did not work
1678 * bug fix: redefining a class definition which failed due to a
1679 previous accessor / function clash now works (but see BUGS entry
1680 #380 for more problems in this area). (thanks to Zach Beane)
1681 * the long form of DEFINE-METHOD-COMBINATION disables method group
1682 checking when given a single method group with pattern *. (thanks
1684 * on x86 compiler supports stack allocation of results of simple
1685 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1686 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1687 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1688 * fixed some bugs related to Unicode integration:
1689 ** the restarts for recovering from input and output encoding
1690 errors only appear when there is in fact such an error to
1692 * fixed some bugs revealed by Paul Dietz' test suite:
1693 ** MISC.549 and similar: late transformation of unsafe type
1694 assertions into derived types caused unexpected code
1696 ** SCALE-FLOAT type deriver is less wrong.
1697 ** type derivers for EXP, LOG and similar functions compute result
1698 types for complex arguments better.
1699 ** (MISC.563) CONJUGATE type deriver works for very restricted
1701 ** out-of-line type testers for character strings are available.
1702 ** EQUAL compiler transform understands specialness of objects
1704 ** accessing double-floats stored on the stack now works on x86-64.
1705 ** debugger internals could sometimes create invalid lispobjs,
1706 resulting in GC crashes.
1707 ** MISC.548: type check weakening can convert required type into
1709 ** initialization forms for bindings are not in scope of free special
1712 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1713 * incompatible change: thread support for non-NPTL systems has
1714 been removed - locking is buggy and unreliable. A threaded
1715 SBCL build will now warn at startup and refuse to create
1716 new threads, unless futex support is detected in the kernel
1717 * incompatible change: the top level REPL now has only an ABORT
1718 restart associated with it, not TOPLEVEL and ABORT as it used to.
1719 TOP and TOPLEVEL are now available as debugger commands for
1720 returning to the top level.
1721 * incompatible change: forms evaluated in the REPL now use the
1722 global optimization policy.
1723 * incompatible change: user- and system-initialization files are
1724 no longer processed with LOAD, but by READ and EVAL; hence the
1725 global optimization policy, startup package, readtable, etc,
1727 * The .fasl file format number has been incremented because of
1728 various incompatible changes.
1729 * internal entry point details and argument counts no longer appear
1730 in backtraces unless explicitly requested by setting
1731 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1732 * built-in and standard functions no longer have names like "top
1733 level local call to FOO".
1734 * fixed bug 32: functions defined in non-null lexical environments
1735 now have more legible printed representation
1736 * fixed bug 33: functions defined in non-null lexical environemnts
1737 are now more amenable to inspection by INSPECT.
1738 * workaround for bug 354: XEPs no longer appear in backtraces unless
1739 explicitly requested.
1740 * fixed bug: receiving the signal which results from attempting to
1741 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1742 SIGBUS on other *BSDs) on architectures where the C stack is also
1743 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1744 to debugger confusion or wild instability. (reported by Cheuksan
1746 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1747 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1748 (reported by Lutz Euler)
1749 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1750 compiled code. (reported by Paul Dietz)
1751 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1752 specialized parameter to a method to something that is not TYPEP
1753 the specializer is now possible.
1754 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1755 face of package deletion.
1756 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1757 pathname. (reported by David Tolpin on c.l.l)
1758 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1759 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1760 STRING-STREAM and FILE-STREAM, respectively.
1761 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1762 than previously; in addition, its test suite is now run on build.
1763 (thanks to Luke Gorrie)
1764 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1765 * optimization: added a immediate representation for single-floats
1767 * fixed some bugs revealed by Paul Dietz' test suite:
1768 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1769 ** attempting to create a package with a colliding nickname causes
1770 correctable errors to be signalled.
1771 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1772 ** several x86-64 backend bugs related to sign-extension of immediate
1775 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1776 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1777 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1779 * fixed loading of multiply forward-referenced layouts.
1780 (thanks to Cheuksan Wang)
1781 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1782 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1783 (see COMPILE-FILE documentation for details of :PRINT option)
1784 * fixed bugs 19 and 317: fixed-format floating point printing is
1785 more accurate. This also fixes a bug reported by Adam Warner
1786 related to the ~@F format directive.
1787 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1789 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1790 dispatch table if the from-char is a dispatch macro character.
1791 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1792 by the wordsize no longer ignore the last word. (reported by Lutz
1794 * fixed bug in type checking of dynamic-extent variables. (reported
1796 * optimization: sequence traversal functions use their freedom to
1797 coerce function designators to functions.
1798 * optimization: code with many calls to CLOS methods specialized on
1799 CLOS classes has had redundant type checks removed.
1800 * optimization: added declarations to speed up operations that access
1801 the internal character database (for example STRING-UPCASE)
1802 * optimization: comparison operations between floats and sufficiently small
1803 fixnums no longer create extra rationals
1804 * fixed some bugs related to Unicode integration:
1805 ** portions of multibyte characters at the end of buffers for
1806 character-based file input are correctly transferred to the
1807 start of the buffer at the next read.
1808 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1809 passing it through to OPEN.
1810 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1811 argument given to internal calls to OPEN.
1812 ** the built-in comment readers (introduced by character sequences
1813 ";" and "#|") are more forgiving to encoding errors; they will
1814 STYLE-WARN and then attempt to resync the stream at a character
1815 boundary. (thanks to Teemu Kalvas)
1816 * fixed some bugs in the x86-64 port:
1817 ** Negative short int return values from c-calls are sign-extended
1819 ** The stack is aligned to 16-bytes for c-calls, as required by
1820 the ABI. (thanks to Cheuksan Wang)
1821 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1822 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1823 for enums. (thanks to Vincent Arkesteijn)
1824 ** Multiple small optimizations and bugfixes for floating point
1826 * fixed some bugs revealed by Paul Dietz' test suite:
1827 ** Space, Tab, Linefeed, Return and Page have the invalid
1828 secondary constituent character trait.
1829 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1831 ** WITH-INPUT-FROM-STRING should only update the index place on
1833 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1834 works more reliably.
1835 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1836 with one-dimensional arrays with fill pointers.
1837 ** TYPE-OF failed on a complex with an integer realpart and a
1839 ** compiler failure during type inference for the code of form
1840 (IF (EQL X (THE ...)) ...) (MISC.535).
1842 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1843 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1844 platform. The port supports 61-bit fixnums, large memory spaces
1845 and reloading shared object files.
1846 * enhancement: saving cores with foreign code loaded is now
1847 supported on ppc/Darwin in addition to the previously supported
1849 * enhancement: the statistical profiler now walks deeper into the
1850 call stack for more meaningful call-graphs and accrued time
1851 reports (x86/x86-64 only). It also now reports time spent in
1853 * enhancement: it is now possible to trace most individual methods
1854 of a generic function in addition to tracing the generic function
1856 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1858 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1859 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1860 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1861 produces an error. (thanks to Vincent Arkesteijn)
1862 * bug fix: NAMESTRING on pathnames with :WILD components in their
1863 directories works correctly. (thanks to Artem V. Andreev)
1864 * fixed bug 125: compiler preserves identity of closures. (reported
1866 * bug fixed: functions with &REST arguments sometimes failed with
1867 "Undefined function" when compiled with (DEBUG 3). (reported by
1869 * bug fix: overflow during compiling of setting element of a bit
1870 vector with constant index and value. (reported by Timmy Douglas)
1871 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1872 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1873 * on x86 compiler partially supports stack allocation of dynamic-extent
1875 * GO and RETURN-FROM do not check the extent of their exit points
1876 when compiled with SAFETY 0.
1877 * fixed some bugs related to Unicode integration:
1878 ** encoding and decoding errors are now much more robustly
1879 handled; it should now be possible to recover even from invalid
1880 input or output to the terminal. (thanks to Teemu Kalvas)
1881 ** provided a first cut at implementing STRING-TO-OCTETS and
1882 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1883 ** altered the SB-MD5 contributed module slightly, changing the
1884 interface just enough for it to be supportable for builds where
1885 lisp characters are not eight bits.
1886 * fixed some bugs revealed by Paul Dietz' test suite:
1887 ** the FORMATTER-generated functions for ~V[ conditionals require
1888 the correct number of arguments.
1889 ** READ-FROM-STRING returns the mandated second value when applied
1890 to displaced strings.
1891 ** the #\Rubout and #\Backspace characters are treated as invalid
1892 constituent characters by the tokenizer.
1894 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1895 * new feature: reloading changed shared object files with
1896 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1897 * new feature: references to foreign variables and functions
1898 can now be compiled and loaded before the corresponding shared
1899 object file is loaded, as long as the foreign definitions are
1900 available at runtime.
1901 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1902 supported. (thanks to Dan Debertin)
1903 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1904 just keywords. (thanks to Vincent Arkesteijn)
1905 * fixed bug #331: structure-class instances corresponding to
1906 DEFSTRUCT forms are now created eagerly.
1907 * fixed bug #345: backtraces from calls to undefined functions work
1908 on x86 as well. Related bug #61 is now also partially fixed on x86
1909 (backtraces from throws to unknown catch tags.)
1910 * bug fix: lambda-list parsing is now stricter vrt. order and number
1911 of lambda-list keywords.
1912 * bug fix: as specified by AMOP, an error is signalled if a
1913 class-option appears multiple times in a DEFCLASS form. (reported
1915 * bug fix: RANDOM can be compiled when the compiler derives the type
1916 of its numeric argument as a disjoint set of small integers.
1917 (reported by Paul Dietz)
1918 * bug fix: starting a core saved with shared objects loaded when
1919 those objects are not available no longer causes threaded SBCL to
1920 hang. (reported by Sean Ross)
1921 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1923 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1924 argument, which must always be NIL. (reported by Kalle Niemitalo)
1925 * bug fix: printing 1.0d+23 no longer results in an error.
1926 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1927 * bug fix: structure slot setters preserve evaluation order. (thanks
1929 * fixed some bugs related to Unicode integration:
1930 ** RUN-PROGRAM can allow its child to take input from a Lisp
1931 stream. (reported by Stefan Scholl)
1932 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1933 locales. (reported by Ken Causey)
1934 * fixed some bugs revealed by Paul Dietz' test suite:
1935 ** INCF, DECF and REMF evaluate their place form as specified in
1937 ** FORMATTER expands ~{ ~} iteration directives with V or #
1938 parameters correctly.
1939 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1940 consequent uses no arguments correctly.
1941 ** the system has a partial understanding of the (COMPLEX RATIO)
1943 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1946 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1947 * new feature: a build-time option (controlled by the :SB-UNICODE
1948 keyword feature, enabled by default) for building the system with
1949 support for the entire 21-bit character space defined by the
1951 * new feature: the system now has rudimentary external-format
1952 support; the primary user-visible change at this time is that
1953 characters with the high bit set (such as lower-case-e-acute) will
1954 print correctly to a terminal in a UTF-8 environment.
1955 * minor incompatible change: BASE-CHAR no longer names a class;
1956 however, CHARACTER continues to do so, as required by ANSI.
1957 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1958 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1959 has been moved to the SB-EXT package (temporarily re-exported from
1961 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1962 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1964 * fixed bug #351: better error-handlind and reporting for malformed
1966 * fixed bug #350: bignum-printing is now more memory-efficient,
1967 allowing printing of very large bignums, eg. (expt 2 10000000).
1968 (reported by Bruno Haible)
1969 * fixed bug #302: better primitive-type selection for intersection
1971 * fixed bug #308: non-graphic characters now all have names, as
1972 required. (reported by Bruno Haible)
1973 * bug fix: redefining a class with different superclasses now correctly
1974 removes it from the direct-subclasses of its previous superclasses.
1975 (reported by David Morse)
1976 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1977 new value now works. (reported by Bruno Haible)
1978 * bug fix: correct canonicalization of multiple non-standard slot
1979 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1980 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1981 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1982 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1983 now exists, an signals an error.
1984 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1985 during macroexpansion for non-iteration variables as well. (reported
1986 by Bruno Haible for CMUCL)
1987 * bug fix: Cyclic structures and unprintable objects in compiler
1988 messages no longer cause errors. (reported by Bruno Haible)
1989 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1990 and READ-FROM-STRING all now return a primary value of NIL if
1991 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1992 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1993 (reported by Bruno Haible for CMUCL)
1994 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1995 do not designate a condition. (reported by Bruno Haible for
1997 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1998 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1999 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2000 permitting resolution in favour of any of the conflicting symbols.
2001 (reported by Bruno Haible for CMUCL)
2002 * bug fix: EQUAL compiler optimizations is less aggressive on
2003 strings which can potentially compare true despite having distinct
2004 specialized array element types.
2005 * bug fix: unit enumerations can be defined without dividing by
2006 zero. (thanks to Vincent Arkesteijn)
2007 * FORMAT compile-time argument count checking has been enhanced.
2008 (report from Bruno Haible for CMUCL)
2009 * a partial workaround for the bug 262: the compiler does not try to
2010 inline-expand a local function doing RETURN-FROM from a deleted
2011 BLOCK. (thanks to Peter Denno for the bug report and to David
2012 Wragg for the simple test case)
2013 * fixed some bugs revealed by Paul Dietz' test suite:
2014 ** RENAME-PACKAGE allows all package designators as new package
2016 ** constraint propagation and lambda variable substitution are
2017 more cautious in dealing with partially deleted code.
2018 ** compiler sometimes lost reoptimization passes.
2019 ** CERROR, given a condition as condition designator, passes its
2020 remaining arguments to the continue format control without
2022 ** Case-altering FORMAT directives work correctly on non-ASCII
2024 ** The REAL type specifier handles bounds outside the floating
2025 point ranges without signalling FLOATING-POINT-OVERFLOW.
2026 ** Functions with IR1-transformations can create intercomponent
2027 references to global functions.
2028 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2030 ** FORMAT ~R treats a nil value for its first parameter correctly.
2032 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2033 * enhancement: saving cores with foreign code loaded is now
2034 supported on x86/NetBSD and sparc/Linux in addition to the previously
2035 supported platforms.
2036 * bug fix: on some platforms repeated installations caused multiple
2037 copies of HTML documentation to be installed -- should not happen
2038 any more. (reported by Stefan Scholl)
2039 * bug fix: parsing self-recursive alien record types multiple times
2040 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2041 original patch by Helmut Eller for CMUCL)
2042 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2043 (thanks to Richard Kreuter)
2044 * bug fix: defining classes whose accessors are methods on existing
2045 generic functions in other (locked) packages no longer signals
2046 bogus package lock violations. (reported by François-René Rideau)
2047 * bug fix: special variables as DEFMETHOD parameters no longer have
2048 associated bogus type declarations. (reported by David Wragg and
2050 * bug fix: read-write consistency on streams of element-type
2051 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2052 * bug fix: redefiniton of the only method of a generic function with
2053 no DEFGENERIC no longer emits a full WARNING. In addition,
2054 redefinition of generic functions with no DEFGENERIC to an
2055 incompatible lambda list now signals an error. (thanks to Zach
2057 * bug fix: DEFGENERIC now works even when there's a function of the
2058 same name in an enclosing lexical environment. (thanks to Zach
2060 * fixed compiler failure, caused by instrumenting code during
2061 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2062 * optimization: added loop analysis and improved register allocation
2063 * fixed some bugs revealed by Paul Dietz' test suite:
2064 ** POSITION on displaced vectors with non-zero displacement
2065 returns the right answer.
2066 ** (SIMPLE-STRING) is a valid type specifier for sequence
2068 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2070 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2071 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2073 ** The justification version of the FORMAT ~< directive treats
2074 non-zero minpad parameter correctly.
2076 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2077 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2078 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2079 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2080 the supported interface.
2081 * new feature: Single-stepping of code compiled with DEBUG 2 or
2082 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2083 * new feature: saving cores with foreign code loaded is now
2084 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2085 Timothy Moore's work for CMUCL)
2086 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2087 parameters to * instead of NIL if no initform is supplied.
2088 (reported by Johan Bockgaard)
2089 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2090 correctly when FUNCALL appears as the car of the form. Note:
2091 despite this FUNCALL forms are not currently subject to
2092 compiler-macro expansion. (port of Raymond Toy's fix for the
2093 same from CMUCL, reported by Johan Bockgaard)
2094 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2095 (thanks for Teemu Kalvas)
2096 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2097 case instead of terminating the iteration (thanks for Julian
2098 Squires, Sean Champ and Raymond Toy)
2099 * bug fix: incorrect expansion of defgeneric that caused a style
2100 warning. (thanks for Zach Beane)
2101 * bug fix: slot accessor effective method computation works properly
2102 for classes with multiple non-standard applicable methods on
2103 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2104 * on x86 compiler supports stack allocation of results of LIST and
2105 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2106 CMUCL implementation by Gerd Moellmann)
2107 * fixed some bugs revealed by Paul Dietz' test suite:
2108 ** FORMAT strings with both the ~<~:;~> form of the justification
2109 directive and pretty-printing directives cause an error.
2111 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2112 * incompatible change: the internal functions
2113 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2114 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2115 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2116 instead of the old functions.
2117 * new feature: on platforms where "dladdr" is available foreign
2118 function names now appear in backtraces. (based on Helmut Eller's
2120 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2121 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2123 * documentation: ASDF manual is now distributed with SBCL.
2124 * bug fix: SBCL can now load its contributed modules with REQUIRE
2125 even if the system-provided entries have been removed from
2126 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2128 * bug fix: backtraces involving undefined functions or assembly
2129 routines are more informative. (thanks to Brian Downing)
2130 * bug fix: mutually referent alien structures now work correctly.
2131 (reported by Rick Taube)
2132 * bug fix: structures defined by WITH-ALIEN can be referred to
2133 within other definitions in the same WITH-ALIEN.
2134 * bug fix: division operators (MOD, TRUNCATE and the like) with
2135 constant zero divisors and integer dividends no longer generate
2137 * bug fix: provide default methods for INPUT-STREAM-P and
2138 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2139 * bug fix: improve the detection and resolution of MOP metacycles
2140 (where effective method computation depends on the generic function
2141 having its effective method computed). (reported by Bruno Haible)
2142 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2143 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2144 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2145 * bug fix: pathnames with relative directory components can be
2146 represented relative to default pathnames.
2147 * optimization: in taking the GCD of bignums, reduce the two bignums
2148 to approximately the same size (using Euclid's algorithm) before
2149 applying the more sophisticated binary GCD. (thanks to Juho
2151 * optimization: COUNT on bitvectors now operates word-at-a-time.
2152 * optimization: ASH with a positive, but not necessarily constant,
2153 (leftwards) shift, when in a modular context, is compiled to a
2155 * fixed some bugs revealed by Paul Dietz' test suite:
2156 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2157 if the corresponding argument is NIL.
2158 ** FORMAT directives accepting a minpad parameter treat negative
2160 ** literal commas as character parameters to FORMAT directives are
2162 ** literal spaces directly after ~<Newline> directives within a
2163 format-logical-block (~:< ~@:>) do not induce :FILL-style
2164 conditional newlines.
2165 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2166 as specified: it no longer includes conditional newlines).
2167 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2169 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2170 :PER-LINE-PREFIX argument does not evaluate to a string.
2172 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2173 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2174 the manual for details; for now, package locks can be disabled by
2175 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2176 if no major problems are found then it is likely that they will be
2177 compiled in unconditionally.
2178 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2179 unsupported operators on all platforms. To load a shared library
2180 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2181 object file, link it into a shared library outside of SBCL and
2182 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2184 * bug workaround: ROOM T and the GC were not working together
2185 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2186 reporting facility aren't preserved. That reporting has been
2187 disabled, so now until and unless someone figures out how to make it
2188 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2189 * minor incompatible change: as threatened around sbcl-0.8.0, the
2190 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2191 The symbols are also exported from SB-PCL for backwards
2192 compatibility, but more so than before SB-PCL should be treated as
2193 an implementation-internal package.
2194 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2196 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2197 It is known as of this release not to work on the Alpha, however.
2198 * fixed bug #167: errors signalled due to illegal syntax in method
2199 bodies are now more legible.
2200 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2201 designators and can hence be used with TYPEP.
2202 * fixed bug #333: CHECK-TYPE now ensures that the type error
2203 signalled, if any, has the right object to be accessed by
2204 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2205 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2206 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2208 * fixed bug #334: programmatic addition of slots using specialized
2209 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2210 and :ALLOCATION :CLASS slots.
2211 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2212 float exponents. (rereported by Peter Seibel)
2213 * fixed a bug: #\Space (and other whitespace characters) are no
2214 longer considered to be macro characters in standard syntax by
2215 GET-MACRO-CHARACTER.
2216 * fixed bug: initialization of condition class metaobjects no longer
2217 causes an instance of the condition to be created. (reported by
2219 * fixed bug: it is now possible to have slots such that
2220 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2221 description is neither :INSTANCE nor :CLASS.
2222 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2223 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2224 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2225 non-standard SLOT-DEFINITION-ALLOCATION values.
2226 * fixed bugs: various race conditions handling exiting threads.
2227 CL-PPCRE's thread torture test now passes.
2228 * fixed arguably-a-bug: GC time proportional to number of threads in
2229 system even when most of them are idle
2230 * optimization: improved performance of BIT and SBIT on bit-vectors.
2231 * .fasl file incompatibility: The fasl file version number has
2232 been incremented because of changes associated with package locks.
2234 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2235 * minor incompatible change: the system no longer provides
2236 optimization hints (conditions which are TYPEP
2237 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2238 mode; these hints are still emitted when higher SPEED optimization
2240 * new contrib module: a sampling profiler (profiling by statistical
2241 sampling, rather than by instrumenting functions) is available as
2242 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2243 * the behaviour of the standard function ED is now customizeable by
2244 third parties through a hook variable: see ED's documentation
2245 string for information on the protocol.
2246 * the compiler no longer emits efficiency notes for (FUNCALL X)
2247 when the type of X is uncertain under default optimization
2249 * fixed bug 276: mutating a binding of a specialized parameter to a
2250 method to something that is not TYPEP the specializer is now
2252 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2253 exhibit the required behaviour on the x86 platform. (thanks to
2254 Peter van Eynde, Eric Marsden and Bruno Haible)
2255 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2256 even for difficult arguments. (reported by Peter Graves)
2257 * fixed bug 141a: the backquote printer now descends quoted
2259 * fixed another bug in backquote printing: no more destructive
2260 modification of the form's list structure. (reported by Brian
2262 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2263 move between its address being taken and the call to
2264 interrupt_thread, fixing a crashing race condition.
2265 * the SB-POSIX contrib implementation has been adjusted so that it
2266 no longer exhibits ridiculously poor performance when constructing
2267 instances corresponding to C structs.
2269 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2270 * minor incompatible change: the sb-grovel contrib now treats C
2271 structures as alien (in the sense of SB-ALIEN) objects rather than
2272 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2273 has implications for memory management of client code
2274 (sb-grovel-returned objects must now be manually managed) and for
2275 type safety (alien objects now have full types).
2276 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2277 used to control emission of compiler diagnostics, rather than the
2278 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2279 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2280 quality should be considered deprecated.
2281 * install.sh now installs the user manual as well
2282 * (not quite a new documentable feature, but worth considering in
2283 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2284 beginnings of a semantically meaningful condition hierarchy is
2285 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2287 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2288 designator as the defaults argument.
2289 * fixed bug: Displaced arrays whose displaced-to array has become
2290 too small now cause ARRAY-DIMENSION to signal an error, providing
2291 for safer bounds-checking. (reported by Bruno Haible)
2292 * fixed bug: DEFCLASS slot definitions with identical :READER and
2293 :WRITER names now signal a reasonable error. (reported by Thomas
2295 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2297 * fixed bug: Invalid slot specification errors now print correctly.
2298 (thanks to Zach Beane)
2299 * fixed bug 320: Shared to local slot value transfers in class
2300 redefinitions now happen corrently from superclasses as
2301 well. (reported by Bruno Haible)
2302 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2304 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2305 improper lists correctly. (reported by Bruno Haible)
2306 * fixed bug 313: source-transform for <fun-name> was erroneously
2307 applied to a call of a value of a variable with name <fun-name>.
2308 (reported by Antonio Menezes Leitao)
2309 * fixed bug 307: The obsolete instance protocol ensures that
2310 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2311 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2312 remove dead unknown-values globs from the middle of the stack.
2313 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2314 easier to resume long computations after using *BREAK-ON-SIGNALS*
2315 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2316 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2317 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2318 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2319 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2320 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2322 * fixed a (fairly theoretical) bug in string printing: if
2323 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2324 does not have array-element-type equal to the most general string
2326 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2327 function name. (thanks to Nikodemus Siivola)
2328 * fixed bug: compile-time format string checker failed on
2329 non-closed ~{. (reported by Thomas F Burdick)
2330 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2331 don't warn on legal loop code involving more than one aggregate
2332 boolean. (thanks to Nikodemus Siivola)
2333 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2334 in keyword position to MAKE-PACKAGE was not being reported as
2336 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2337 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2339 * fixed bugs 280 and 312: the checking for multiple definitions in a
2340 file is less likely to become confused by uses of inline
2342 * fixed bug: the #S reader macro performs the keyword coercion
2343 specified for slot names. (reported by Kalle Niemitalo)
2344 * fixed bug: lambda lists may contain symbols whose names start with
2345 & but are not lambda-list-keywords; their occurrence triggers a
2347 * fixed bug 321: define-method-combination argument lambda lists do
2348 not cause invalid code to be generated when &optional or &aux
2349 variables have default values. (reported by Bruno Haible)
2350 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2351 have CLOS classes; this appears also to have fixed reported
2352 instability in stack exhaustion detection.
2353 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2354 MUFFLE-WARNING when no associated restart is present is now
2356 * optimization: rearranged the expansion of various defining macros
2357 so that each expands into only one top-level form in a
2358 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2360 * optimization: used a previously little-used slot in symbols to
2361 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2363 * fixed some bugs revealed by Paul Dietz' test suite:
2364 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2366 ** Fixed an optimization bug related to inheritance of initforms
2367 from local to shared slots.
2368 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2369 ** CAST optimizer forgot to flush argument derived type.
2370 ** print/read consistency on floats is now orders of magnitude
2371 more likely. (thanks also to Bruno Haible for a similar report
2373 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2374 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2375 for objects of type REAL. Make it so.
2376 ** SXHASH is specified (infelicitously) to respect similarity,
2377 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2378 it so. (thanks to Markus Ziegler)
2379 ** on the Alpha, the compiler succeeds in compiling functions
2380 returning a known number of arguments greater than 63.
2381 ** fixed handling of invalid NIL arguments in keyword position
2383 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2385 ** provided a readably-printable representation for RANDOM-STATE
2387 ** ensured that pathnames, where they have a namestring, always
2388 print using #P"..." syntax.
2390 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2391 * Support for the forthcoming 2.0 version of the NetBSD kernel
2392 running on x86 hardware has been added. (thanks to Perry
2393 E. Metzger most immediately, and others for their past work)
2394 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2395 current release version; SBCL's OpenBSD support had been broken
2396 since about the time of OpenBSD's switch to ELF binary format.)
2397 * [placeholder for DX summary]
2398 ** user code with &REST lists declared dynamic-extent, under high
2399 speed or space and low safety and debug optimization policy.
2400 * The manual has been converted to Texinfo format and the debugger
2401 chapter from the cmucl manual has been added.
2402 * A facility has been added to extract documentation strings from
2403 sbcl and store them as Texinfo-formatted snippets for inclusion in
2404 the manual (via Texinfo's @include directive)
2405 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2406 the test case to Dave Roberts)
2407 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2408 pointers. (reported by Sean Ross)
2409 * bug fix: PROFILE output is printed nicely even for large numerical
2410 values. (thanks to Zach Beane)
2411 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2412 greater than 32 handle EOF correctly.
2413 * bug fix: on X86 an immediate argument of the IMUL instruction is
2414 correctly printed in disassembly. (thanks to Lutz Euler)
2415 * bug fix: class slots in redefined classes preserve their old
2416 values. (thanks to Bruno Haible and Nikodemus Siivola)
2417 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2418 succeeds. (reported by Marco Baringer)
2419 * fixed some bugs revealed by Paul Dietz' test suite:
2420 ** READ-SEQUENCE now works on ECHO-STREAMs.
2421 ** RATIONALIZE works more according to its specification. (thanks
2424 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2425 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2426 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2427 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2428 should matter to you only if you rebind the printer control
2429 variables and then find you want different bindings in the
2430 debugger than in the ordinary execution of your program.)
2431 * The runtime build system has been tweaked to support building
2432 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2433 assembler and linker. (thanks to Nikodemus Siivola)
2434 * Unbound, undefined, undeclared variables now trigger full
2435 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2436 more often programmer error than deliberate exploitation of undefined
2438 * optimization: the hash algorithm for strings has changed to one
2439 that is less vulnerable to spurious collisions. (thanks to Juho
2441 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2442 needless bounds checking. (thanks to Juho Snellman)
2443 * optimization: implemented multiplication as a modular
2444 (UNSIGNED-BYTE 32) operation on the PPC backend.
2445 * fixed some bugs revealed by Paul Dietz' test suite:
2446 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2448 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2449 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2450 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2452 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2454 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2456 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2457 their output stream on EOF from read.
2458 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2459 have been read to end-of-file.
2460 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2462 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2463 description of determination of which consecutive characters
2465 ** Printing the "Space" character with escaping on now yields "#\\ ",
2466 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2467 ** Reading floating-point numbers with *READ-BASE* set to a number
2468 less than 10 works correctly.
2469 ** Reading floating-point numbers with *READ-BASE* set to a number
2470 more than 10 works correctly.
2471 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2472 the readtable currently in effect.
2474 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2475 * minor incompatible change: parsing of namestrings on a physical
2476 (Unix) host has changed; numbers after the final #\. in a
2477 namestring are no longer interpreted as a version field. This is
2478 intented to be largely invisible to the user, except that the
2479 meaning of the namestring "*.*.*" has changed: it now refers to a
2480 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2481 should usually be replaced by
2482 (make-pathname :name :wild :type :wild :version :wild)
2483 with the added benefit that this is more likely to be portable.
2484 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2485 option to OPEN now signals an error if the file being opened
2486 exists; this may have an impact on existing code.
2487 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2488 signals received, on the PowerPC platforms (both Linux and
2489 Darwin). (thanks to Pierre Mai for pointing out the location of
2491 * several fixes on OS X: The system now builds and runs cleanly on
2492 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2493 (thanks to Brian Mastenbrook)
2494 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2495 second-resolution: integer multiples of 1/3600 between -24 and 24.
2496 (thanks to Vincent Arkesteijn)
2497 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2498 type when called with 1 argument; PEEK-CHAR checked type of
2499 PEEK-TYPE only after having read first character from a
2500 stream. (reported by Peter Graves)
2501 * bug fix: the garbage collector now has much better locality
2502 behaviour, and in particular no longer treats all memory as being
2503 exhausted when many small objects point to each other in a deeply
2505 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2506 recognized as being TYPEP their class.
2507 * bug fix: the PUSHNEW documentation string has been corrected.
2508 (thanks to Vincent Arkesteijn)
2509 * bug fix: defaulting of the value for the last of an atypically
2510 large number of multiple values being bound was not being
2511 performed correctly on the Alpha or PPC platforms
2512 * optimization: implemented multiplication as a modular
2513 (UNSIGNED-BYTE 32) operation on the x86 backend.
2514 * optimization: SEARCH on simple-base-strings can now be open-coded.
2515 (see also contrib/compiler-extras.lisp for inspiration for
2516 teaching the compiler about the Boyer-Moore algorithm).
2517 * value, returned by MAX (and MIN) called with several EQUALP, but
2518 not EQL, arguments now does not depend on compiler settings.
2519 * fixed some bugs revealed by Paul Dietz' test suite:
2520 ** in stack analysis liveness information is propagated from
2521 non-local entry points.
2522 ** pathwise CAST removing failed when the CAST node did not start
2524 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2525 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2527 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2528 host is already defined.
2529 ** RENAME-FILE works on streams instead of signalling an internal
2531 ** PEEK-CHAR uses the current readtable when determining whether
2532 or not a character is whitespace.
2533 ** MERGE-PATHNAMES handles the case when the pathname does not
2534 specify a name while the default-pathname specifies a version.
2535 ** Pathnames now stand a better chance of respecting print/read
2537 ** Attempting to use standardized file system operators with a
2538 pathname with invalid :DIRECTORY components signals a
2540 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2542 ** DIRECTORY on logical pathnames is more correct.
2543 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2544 signal a TYPE-ERROR if their argument is not a stream.
2545 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2546 designator argument does not designate a stream.
2547 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2548 examining the synonym.
2549 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2551 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2552 element-type larger than ([UN]SIGNED-BYTE 32).
2554 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2555 * When built with the :SB-FUTEX feature, threaded builds now take
2556 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2557 for faster/more reliable mutex and condition variable support.
2558 * Incompatible change (but one you probably shouldn't have been using
2559 anyway): the interface and code for arbitrating between multiple
2560 threads in the same user session has been redesigned.
2561 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2562 error when called without an explicit environment argument.
2563 (thanks to Nikodemus Siivola)
2564 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2565 confused about their position. (thanks to Adam Warner and Gerd
2567 * bug fix: Pretty printing backquoted forms with unquotations in the
2568 argument list position of various code constructs such as LAMBDA
2569 now works correctly. (reported by Paul Dietz)
2570 * bug fix: Pretty printing unquotations no longer loses all
2571 stream position information.
2572 * optimization: performance of string output streams is now less
2573 poor for multiple small sequence writes.
2574 * optimization: performance of CSUBTYPEP in the presence of complex
2575 expressions involving CONS and NOT many times has been improved.
2576 (reported by Paul Dietz)
2577 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2579 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2580 streams for the REPL, for the debugger, and so on. (thanks to
2582 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2583 * fixed some bugs revealed by Paul Dietz' test suite:
2584 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2585 is ignored if it is too small, rather than propagating through
2586 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2587 ** extremely complex negations of CONS types were not being
2588 sufficiently canonicalized, leading to inconsistencies in
2590 ** VALUES tranformer lost derived type.
2592 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2593 * fixed a bootstrapping bug: the build process no longer assumes
2594 that the various BOOLE-related constants have the same value in
2595 host and target lisps. (noted by Paul Dietz' test suite on an
2596 SBCL binary built from CLISP)
2597 * The system can now be dynamically linked on the MIPS platform,
2598 which enables dynamic loading of foreign code from Lisp. (thanks
2599 to Ralf Baechle for discussions on the MIPS ABI)
2600 * The system now records debugging information for its own source
2601 files in a filesystem-position-independent manner, relative to
2602 the "SYS" logical host.
2603 * fixed a compiler bug: MV-LET convertion did not check references
2604 to the "max args" entry point. (reported by Brian Downing)
2605 * tweaked disassembly notes to be less confident about proclaiming
2606 some instruction as an LRA. (thanks to Brian Downing)
2607 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2608 now each have their own history, command character, and other
2609 characteristics. (thanks to David Lichteblau)
2610 * fixed some bugs revealed by Paul Dietz' test suite:
2611 ** compiler failure in compiling LOGAND expressions including a
2613 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2614 shift greater than 32.
2615 ** FLUSH-DEST did not mark blocks for type check regeneration.
2616 ** HANDLER-CASE failed to accept declarations in handler clauses
2617 in some circumstances.
2619 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2620 * New code in contrib/sb-introspect (still probably not entirely
2621 stable yet) provides some support for smart Lisp development
2622 environments like SLIME.
2623 * The conditions signalled for errors occurring when loading .fasl
2624 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2625 in a way which should help ASDF recover gracefully.
2626 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2627 stuff has been cleaned up. If you code contrib/ stuff, this might
2628 affect you, and you can look at contrib/README, contrib/STANDARDS,
2629 and/or the 0.8.4.27 diff to check.
2630 * In full calls the compiler now does not generate checks for declared
2631 argument types for all arguments.
2632 * various threading fixes
2633 ** and some experimental patches which didn't make it into
2634 the main tree for this release, but which are shipped in
2635 contrib/experimental-thread.patch as a possible fix for some
2636 failures (deadlock, spinning...) in GC-intensive multithreaded
2638 * fixed PPC build problem (source code incompatibility of different
2639 library versions): added offsetof() hackery which attempts to divine
2640 where glibc maintainers put uc_mcontext today
2641 * fixed bug 282: compiler does not trust type assertions while passing
2642 arguments to a full call.
2643 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2644 &OPTIONAL VALUES type parameter.
2645 * fix bug 214: algorithm for noting rejected templates is now more
2646 similar to that of template seletion. (also reported by rydis on
2648 * fixed bug 141b: printing backquoted information readably and prettily
2649 inserts a space where necessary.
2650 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2651 (CHARACTER 10) are now reported as errors, rather than propagated
2652 as unknown types. (reported by piso on #lisp)
2653 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2654 with values NIL and :ERROR. (thanks to Milan Zamazal)
2655 * fixed bug 191c: CLOS now does proper keyword argument checking as
2656 described in CLHS 7.6.5 and 7.6.5.1.
2657 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2658 longer raise an error; further, using a list as a for-as-arithmetic
2659 counter now raises a meaningful error.
2660 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2661 now understood by sequence creation functions such as MAKE-SEQUENCE
2663 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2664 a STRING-INPUT-STREAM.
2665 * compiler enhancement: SIGNUM is now better able to derive the type
2667 * type declarations inside WITH-SLOTS are checked. (reported by
2669 * fixed some bugs revealed by Paul Dietz' test suite:
2670 ** incorrect optimization of TRUNCATE for a positive first
2671 argument and negative second.
2672 ** compiler failure in let-convertion during flushing dead code.
2673 ** compiler failure while deriving type of TRUNCATE on an
2674 interval, containing 0.
2675 ** ASH of a negative bignum by a negative bignum count now returns
2677 ** intersection of CONS types now canonicalizes properly, fixing
2678 inconsistencies in SUBTYPEP.
2680 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2681 * incompatible change: The --disable-debugger command line
2682 option now clobbers the debugger at a more fundamental
2683 level, by redefining #'INVOKE-DEBUGGER instead of by
2684 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2685 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2686 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2687 enter the debugger REPL and then suffer recursive errors
2688 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2689 while under the new system, BREAK in --disable-debugger mode
2690 terminates the system just as an unhandled error would.
2691 * fixed compiler performance when processing loops with a step >1;
2692 * bug fix: DOCUMENTATION now retrieves generic function
2693 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2694 support has been systematized, and now supports the methods
2695 specified by ANSI, along with a default method and a method for
2696 slot documentation. (reported by Nathan Froyd)
2697 * bug fix: effective methods associated with a generic function are
2698 no longer cached over a change of that generic function's method
2699 combination. (reported by Andreas Fuchs)
2700 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2701 names a non-existent directory. (thanks to Andreas Fuchs)
2702 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2703 the PPC platform, lead to stack corruption; this has been fixed.
2704 (reported by Rainer Joswig)
2705 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2706 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2707 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2708 arithmetic in general, is now much more reliable on the PPC
2710 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2711 platform now returns the right answer.
2712 * optimization: restored some effective method precomputation in
2713 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2714 precomputation is now tunable.
2715 * optimization: compiler-internal data structure use has been
2716 reviewed, and changes have been made that should improve the
2717 performance of the compiler by about 20%.
2718 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2719 simple-base-strings and simple-bit-vectors is improved.
2720 * optimization: the optimization of 32-bit logical and arithmetic
2721 functions introduced in version 0.8.3 on the x86 has been
2722 implemented on the mips, ppc and sparc platforms; an
2723 implementation of the same facility, but for 64-bit arithmetic,
2724 has been added for the alpha.
2725 * microoptimization: the compiler is better able to make use of the
2726 x86 LEA instruction for multiplication by constants.
2727 * bug fix: in some situations compiler did not report usage of
2728 generic arithmetic in (SPEED 3) policy.
2729 * bug 145b fix: compiler used wrong type specifier while converting
2730 MEMBER-types to numeric.
2731 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2733 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2734 index argument now works properly on the Alpha platform.
2735 * bug fix: floating point exception treatment on the Alpha platform
2737 * bug fix: FILE-POSITION works much better on string input and
2738 output streams. (thanks to Nikodemus Siivola)
2739 * bug fix: many threading/garbage collection symptoms sorted.
2740 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2741 might be pseudo-atomic.
2742 * internal change: Stopping for GC is now done with signals not
2743 ptrace. GC is now done in whichever thread wanted it, instead of
2745 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2747 * bug fix: result form in DO is not contained in the implicit
2749 * incompatible change: ICR structure is changed; the value part of
2750 CONTINUATION is now called LVAR; corresponding functions are
2751 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2752 * added type deriver for ISQRT (thanks to Robert E. Brown).
2753 * bug fix: better support for loading from the command line when an
2754 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2756 * fixed some bugs revealed by Paul Dietz' test suite:
2757 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2758 ** ROUND and FROUND now give the right answer when given very
2759 small float arguments.
2760 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2762 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2763 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2764 ** LCM should return a non-negative integer.
2765 ** PARSE-INTEGER returned the index of a terminator instead of the
2766 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2767 ** PARSE-INTEGER returned an incorrect index being applied to a
2769 ** LCM with two arguments of 0 returns 0 rather than signalling
2771 ** unsigned addition of a 32-bit constant with the high bit set no
2772 longer causes an internal compiler error on the x86.
2773 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2774 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2775 with negative last argument.
2776 ** byte specifiers with zero size and position no longer cause
2777 an error during type derivation.
2778 ** bignum multiplication on the Alpha platform now returns the
2780 * porting: The system now builds on SuSE AMD64, although it still
2781 generates a 32-bit binary.
2782 * .fasl file incompatibility: The fasl file version number has
2783 been incremented (because of the changes to internal compiler
2784 data structures referred to above).
2786 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2787 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2788 more accurately, on the Darwin kernel running on PowerPC hardware.
2789 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2790 * Compiler code deletion notes now signal a condition of type
2791 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2792 an associated MUFFLE-WARNING restart.
2793 * The compiler now performs limited argument count validation of
2794 constant format strings in FORMAT, and where appropriate in ERROR,
2795 CERROR and WARN. (thanks to Gerd Moellmann)
2796 * New ASDF-INSTALL contrib can be used for automatic download and
2797 installation of third-party Lisp code from CCLAN or other sites
2799 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2800 which forces another thread to execute a function supplied by the
2802 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2803 now accept and act upon their :ELEMENT-TYPE keyword argument.
2804 (reported by Martin Atzmueller, Edi Weitz)
2805 * bug fix: FILE-POSITION now accepts position designators up to
2806 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2807 is the greater. (thanks to Patrik Nordebo)
2808 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2809 Kalle Olavi Niemitalo)
2810 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2811 printable. (reported by Eric Marsden)
2812 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2813 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2814 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2815 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2816 (thanks to Stig Sandoe)
2817 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2818 UNREAD-CHAR now correctly includes the unread character in the
2819 target sequence. (thanks to Gerd Moellmann)
2820 * bug fix in threaded builds: the system can now be suspended and
2821 resumed by shell job control with minimal disruption.
2822 * bug fixes in times and timezones >2038AD
2823 * better handling of "where is GNU make?" problem in build scripts
2824 (thanks to Nikodemus Siivola)
2825 * new optimization: inside a named function any reference to a
2826 function with the same name is considered to be a self-reference;
2827 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2828 optimization quality.
2829 * new optimization on x86: logical functions and + now have
2830 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2831 used when the result is truncated to 32 bits.
2832 * VALUES declaration is partially enabled.
2833 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2834 * fixed some bugs revealed by Paul Dietz' test suite:
2835 ** The system now obeys the constraint imposed by
2836 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2837 types form a lattice under type intersection.
2838 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2839 ** ASSOC now ignores NIL elements in an alist.
2840 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2841 and (1+ MOST-POSITIVE-FIXNUM) answers.
2842 ** The addition of a method with invalid qualifiers to a generic
2843 function does not cause an error to be signalled immediately;
2844 a warning is signalled, and the error is generated only on
2845 calling the generic function.
2846 * changed .fasl file version number, in part to add type codes for
2847 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2848 obscure ANSI requirements
2850 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2851 * fixed bug 148: failure to inline-expand a local function left
2852 garbage, confusing the compiler.
2853 * fixed bugs 3cd: structure slot readers perform type check if the
2854 slot can have an invalid value (i.e. it is either not initialized
2855 or can be written with a less specific slot writer).
2856 * bug fix: the compiler now traps array references to elements off
2857 the end of an array; previously, the bounds checking in some
2858 circumstances could go off-by-one.
2859 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2861 * type declarations for array element types now obey the description
2862 on the CLHS page "Declaration TYPE", as per discussions on
2863 sbcl-help around 2003-05-08. This means that a declaration
2864 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2865 declaration, all references to BAR will be asserted or assumed
2866 (with THE, so dependent on compiler policy) to involve objects of
2867 type FOO. Note that no such declaration is implied in
2868 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2869 * declared types of functions from the "Conditions"
2870 chapter. (reported by Paul Dietz)
2871 * bug fix: CERROR accepts a function as its first argument.
2872 * bug fix: NTH an NTHCDR accept a bignum as index
2873 arguments. (reported by Adam Warner)
2874 * optimization: character compare routines now optimize comparing
2875 against a constant character. (reported by Gilbert Baumann)
2876 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2877 argument now works properly on the MIPS platform.
2878 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2879 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2880 declared to have a functional type, does not cause an error
2882 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2883 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2884 * fixed bug: if last continuation of a deleted block has a
2885 destination, this destination should be deleted too. (reported by
2887 * fixed a bug in the bootstrap process: the host compiler's values
2888 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2889 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2890 test case from Patrik Nordebo)
2891 * improved the ability of the disassembler on the PPC platform to
2892 provide helpful disassembly notes.
2893 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2894 the class in more cases than previously.
2895 * bug fix: FILE-POSITION now understands :START and :END for
2896 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2897 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2898 8) data. (thanks to David Lichteblau)
2899 * bug fix: it is possible to add a method to a generic function
2900 without lambda list.
2901 * bug fix: reader failed to signal END-OF-FILE inside an
2902 object representation. (reported by Nikodemus Siivola)
2903 * fixed some bugs revealed by Paul Dietz' test suite:
2904 ** LAST and [N]BUTLAST should accept a bignum.
2905 ** condition slot accessors are methods.
2906 ** (VECTOR NIL) is a subtype of STRING.
2908 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2909 * minor incompatible change: some nonsensical specialized lambda
2910 lists (used in DEFMETHOD) which were previously ignored now signal
2912 * minor incompatible change: the system is now aware of the types of
2913 variables in the COMMON-LISP package, and will signal errors for
2914 most violations of these type constraints (where previously they
2915 were silently accepted).
2916 * minor incompatible change: COMPILE-FILE now uses the freedom
2917 afforded (ANSI 3.2.2.3) to use derived function types for
2918 functions defined in the same file. This also permits the system
2919 to warn on static type mismatches and function
2920 redefinition. (Currently it does not work with high DEBUG level.)
2921 * minor incompatible change: VALUES declaration is disabled.
2922 * When issuing notes, the compiler now signals a condition of type
2923 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2924 restart for use in user handlers. It is expected that the
2925 COMPILER-NOTE condition will eventually become a condition
2926 supertype to a hierarchy of note types, which will then be
2927 handleable in a similar fashion. However, at the moment, no such
2928 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2929 but it's an implementation detail, not a classification for the
2931 * Changes in type checking closed the following bugs:
2932 ** type checking of unused values (192b, 194d, 203);
2933 ** template selection based on unsafe type assertions (192c, 236);
2934 ** type checking in branches (194bc).
2935 * A short form of VALUES type specifier has ANSI meaning (it has
2936 increased the number of situations when SBCL cannot perform type
2938 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2939 keywords or constants is permissible.
2940 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2941 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2942 classes. (thanks to Antonio Martinez)
2943 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2944 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2945 argument to a pretty-print stream. Instead, it leaves any such
2946 operations to DESCRIBE-OBJECT methods.
2947 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2949 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2950 lambda lists. (thanks to David Lichteblau)
2951 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2952 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2953 (thanks to Gerd Moellmann)
2954 * fixed bug 246: increased compilation speed of long
2955 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2957 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2959 * DEFCONSTANT now throws a condition of type
2960 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2961 constant to a non-EQL value; CONTINUE and ABORT restarts
2962 respectively change and preserve the value.
2963 * fixed bug 63: The code walker, part of the implementation of CLOS,
2964 is now better at handling symbol macros.
2965 * bug fix: There is no longer an internal implementation type named
2966 CL:LENGTH. (reported by Raymond Toy)
2967 * bug fix: In macro-like defining macros/special operators the
2968 implicit block does not enclose the lambda list.
2969 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2970 enclosed in parentheses) are not suitable as type specifiers, and
2971 their use properly signals an error now.
2972 * bug fix: An argument count mismatch for a type specifier in code
2973 being compiled no longer causes an unhandled error at compile
2974 time, but signals a compile-time warning.
2975 * fixed simple vector readable printing
2976 * bug fix: DESCRIBE takes more care over whether the class
2977 precedence list slot of a class is bound before accessing it.
2978 (reported by Markus Krummenacker)
2979 * bug fix: FORMATTER can successfully compile pretty-printer format
2980 strings which use variants of the ~* directive inside.
2981 * bug fix: SEARCH now applies its TEST predicate to the elements of
2982 the arguments in the correct order. (thanks to Wolfhard Buss)
2983 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2984 in type inference. (thanks to Robert E. Brown)
2985 * bug fix: Reading in symbols with an explicit package name of ""
2986 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2987 * fixed some bugs revealed by Paul Dietz' test suite:
2988 ** NIL is now allowed as a structure slot name.
2989 ** Arbitrary numbers, not just REALs, are allowed in certain
2990 circumstances in LOOP for-as-arithmetic clauses.
2991 ** Multiple class redefinitions before slot access no longer
2992 causes a type error.
2993 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2994 association between the name and a class.
2995 ** Generic functions with non-standard method-combination and over
2996 five methods all of which return constants no longer return NIL
2997 after the first few invocations. (thanks to Gerd Moellmann)
2998 ** CALL-NEXT-METHOD with no arguments now passes the original
2999 values of the arguments, even in the presence of assignment.
3000 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3002 ** DISASSEMBLE works with closures and funcallable instances.
3003 ** ADD-METHOD now returns the generic function, not the new method.
3004 ** FIND-METHOD signals an error if the lengths of the specializers
3005 is incompatible with the generic function, even if the ERRORP
3007 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3008 which its argument is a member.
3009 ** DEFCLASS only redefines the class named by its class-name
3010 argument if that name is the proper name of the class;
3011 otherwise, it creates a new class.
3012 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3013 of the UNBOUND-SLOT condition to the name of the slot.
3014 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3015 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3016 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3017 SLOT-MAKUNBOUND in the specified fashion.
3019 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3020 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3021 cross-compilation host. As a consequence, we can now bootstrap our
3022 way up to SBCL starting with a bare gcc toolchain and human-readable
3023 source code (first the source to CLISP, then the source to SBCL).
3024 * A contributed module containing a partial implementation of the
3025 simple-streams interface has been included. (thanks to Rudi
3027 * A contributed module implementing the RFC1321 Message Digest
3028 Algorithm, known as MD5, has been included.
3029 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3030 no longer has any effect, as the code controlled by this feature
3031 has been deleted. (As far as we know, no-one has ever built using
3032 this feature, and its semantics were confused in any case).
3033 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3034 work on conditions (as required by the ANSI specification),
3035 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3036 expected behaviour on conditions. Users should note, however,
3037 that such behaviour is not required by the ANSI specification,
3038 and so use of this behaviour may render their code unportable.
3039 * fixed some bugs revealed by Paul Dietz' test suite:
3040 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3042 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3043 ** SXHASH on condition objects no longer returns NIL.
3044 ** :ALLOCATION :CLASS slots are better treated; their values are
3045 updated on class redefinition, and initforms inherited from
3046 superclasses are applied.
3047 ** REMOVE-METHOD returns its generic function argument even when
3048 no method was removed.
3049 ** SHARED-INITIALIZE now initializes the values of the requested
3050 slots, including those with :ALLOCATION :CLASS.
3051 ** ALLOCATE-INSTANCE now works on structure classes defined via
3052 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3054 ** SLOT-EXISTS-P now works on conditions, as well as structures
3056 ** MAKE-LOAD-FORM now has the required methods on
3057 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3058 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3059 keyword, and now implements the SLOT-NAMES argument.
3060 ** methods with &OPTIONAL arguments no longer allow too many
3061 arguments to be passed in the call without error.
3062 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3063 option is consistent with the required arguments of the generic
3064 function lambda list.
3065 * bug fix: REQUIRE accepts a string designator. (Thanks to
3067 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3068 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3069 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3070 * bug fix: The compiler now checks for duplicated variables in macro
3072 * bug fix: SETQ on globals returns the correct value.
3073 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3074 not a COMPILER-ERROR (followed by some other strange error on
3075 choosing the CONTINUE restart).
3076 * bug fix: make.sh and friends are now more consistent in the way that
3077 they look for GNU "make".
3079 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3080 * experimental native threads support (on x86 Linux >=2.4 only).
3081 This is not compiled in by default: you need to add :SB-THREAD to
3082 the target features. See the "Beyond ANSI" chapter of the manual
3084 * fix for longstanding nonANSIism: The old distinction between
3085 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3086 The return value from CL:FIND-CLASS is now a CLOS class, and
3087 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3088 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3089 * An interface to the MetaObject Protocol, as described in Kiczales,
3090 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3091 MIT Press, 1991, is available from the SB-MOP package.
3092 * incompatible change: the SB-PCL package should now be considered
3093 a private implementation detail, and no longer a semi-private MOP
3095 * minor incompatible change: due to rearrangement for threads, the
3096 control stack and binding stack are now allocated at arbitrary
3097 addresses instead of being hardcoded per-port. Users affected by
3098 this probably have to be doing advanced things with shared
3099 libraries, and will know who they are.
3100 * minor incompatible change: Previously, all --eval forms used were
3101 processed with READ before any of them were processed with EVAL.
3102 Now each --eval form is processed with both READ and EVAL before
3103 the next --eval form is processed. (Thus package operations like
3104 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3105 work as the user might reasonably expect.)
3106 * minor incompatible change: *STANDARD-INPUT* is now only an
3107 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3109 * minor incompatible change: Y-OR-N-P is now character-oriented,
3110 not line oriented. Also, YES-OR-NO-P now works without errors.
3111 (thanks to Antonio Martinez)
3112 * sb-aclrepl module improvements: an integrated inspector, added
3113 repl features, and a bug fix to :trace command.
3114 * Known functions, which cannot be open coded by the backend, are
3115 considered to be able to check types of their arguments. (fixing
3116 a bug report by Nathan J. Froyd)
3117 * fixed a bug in computing method discriminating functions: It is
3118 now possible to define methods specialized on classes which have
3119 forward-referenced superclasses. (thanks to Gerd Moellmann)
3120 * fixed evaluation order in optional entries (reported by Gilbert
3122 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3123 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3124 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3125 required-by-AMOP NAME argument, as well as CLASS and
3126 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3127 * fixed bug 20: DEFMETHOD can define methods using names that are
3128 not the proper names of classes to designate class specializers.
3129 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3130 Unix file descriptors, instead of blowing up. (thanks to Antonio
3132 * Garbage collection refactoring: user-visible change is that a
3133 call to the GC function during WITHOUT-GCING will not do garbage
3134 collection until the end of the WITHOUT-GCING. If you were doing
3135 this you were probably losing anyway.
3136 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3137 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3138 TYPEP the latter but not the former.
3139 * The compiler issues a full WARNING for calls to undefined functions
3140 with names from the CL package.
3141 * MAP-INTO for a vector destination is open coded. (reported by
3142 Brian Downing on c.l.l)
3143 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3144 documentation string.
3145 * fixed some bugs revealed by Paul Dietz' test suite:
3146 ** COPY-ALIST now signals an error if its argument is a dotted
3148 ** Condition slots are now accessed more correctly in the presence
3149 of multiple initargs for a given slot.
3150 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3151 exclude restarts of the same name associated with a different
3153 ** DEFCLASS of forward-referenced classes with another
3154 forward-referenced class in the superclasses list no longer
3156 ** Condition slots are now initialized once each, not multiple
3157 times. (thanks to Gerd Moellmann)
3158 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3160 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3161 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3162 arguments contain duplicated elements.
3163 ** RESTART-CASE understands local macros.
3164 ** RESTART-CASE associates exactly its own restarts with a condition.
3165 ** ENDP in safe mode checks its argument to be of type LIST.
3166 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3167 in question is unbound.
3168 ** Optimized MAKE-INSTANCE functions no longer cause internal
3169 assertion failures in the presence of duplicate initargs.
3170 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3171 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3172 propagated even in the ctor optimized implementation of
3174 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3176 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3177 * a better implementation of SXHASH on (simple) bit vectors,
3178 measured both in execution speed and in distribution of results
3179 over the positive fixnums, has been installed. Likewise, a better
3180 implementation of EQUAL for simple bit vectors is now available.
3181 * fixed CEILING optimization for a divisor of form 2^k.
3182 * fixed bug 240 (emitting extra style warnings "using the lexical
3183 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3184 by Antonio Martinez)
3185 * fixed SXHASH, giving different results for NIL depending on type
3186 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3187 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3188 variable. (found by Rolf Wester)
3189 * SBCL does not ignore type declarations for special
3190 variables. (reported by rif on c.l.l 2003-03-05)
3191 * some bug fixes in contrib/sb-aclrepl/
3192 * fixed some bugs revealed by Paul Dietz' test suite:
3193 ** a bug in the CONS type specifier, whereby the CAR and CDR
3194 types got intertwined, has been fixed;
3195 ** the type system is now able to reason about the interaction
3196 between INTEGER and RATIO types more completely;
3197 ** APPEND, [N]REVERSE and NRECONC check that those their
3198 arguments, which must be proper lists, are really so;
3199 ** An array specialized to be unable to hold elements has been
3200 implemented, as required -- yes, really -- by ANSI;
3201 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3202 on malformed property lists;
3204 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3205 * incompatible packaging change: in line with Unix convention,
3206 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3207 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3208 * REQUIRE and PROVIDE are now optionally capable of doing something
3209 useful. See the documentation string for REQUIRE.
3210 * infrastructure for a managed SBCL contrib system: contributed
3211 modules in this release include:
3212 ** the ASDF system definition facility;
3213 ** an interface to the BSD Sockets API;
3214 ** an ACL-like convenience interface to the repl;
3215 (thanks to Kevin Rosenberg)
3216 ** an implementation of ROTATE-BYTE, with efficient implementation
3218 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3219 gives a closer approximation to the right answer than previously.
3220 (thanks to Raymond Toy)
3221 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3222 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3223 optional environment arguments, as required by ANSI.
3224 * fixed bugs in other functions taking environment objects, allowing
3225 calls with an explicit NIL environment argument to be compiled
3227 * fixed bug 228: primary return values from
3228 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3229 COMPILE or FUNCTION.
3230 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3231 :TYPE will no longer signal errors on innocuous objects.
3232 * fixed bug 231b: SETQ is better at respecting type declarations in
3233 the lexical environment.
3234 * fixed a bug in DEFCLASS: classes named by symbols with no or
3235 unprintable packages can now be defined.
3236 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3237 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3238 * fixed compiler failure related to checking types of functions.
3239 (reported by Robert E. Brown)
3240 * the compiler is now much more consistent in its error-checking
3241 treatment of bounding index arguments to sequence functions: in
3242 (SAFETY 3) code, errors will be signalled in almost all cases if
3243 invalid sequence bounding indices are passed to functions defined
3244 by ANSI to operate on sequences.
3245 * fixed a bug in the build procedure: documentation of SBCL-specific
3246 packages is now preserved and available in the final Lisp image.
3247 * lifted FDEFINITION lookup out of loops in the implementation of
3248 many list operations. (thanks to Robert E. Brown)
3249 * fixed a bug in the reader: the #n# reader macro now works for
3250 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3251 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3252 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3253 E. Brown and Tony Martinez respectively)
3254 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3255 SLOT-BOUNDP: in particular, it is now able to optimize them much
3256 better, and is now not vulnerable to having packages renamed.
3257 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3258 always called when a slot is not present in an instance. (thanks
3260 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3261 longer signal an error on unreading a character following EOF on
3262 the previous constituent stream. (thanks to Tony Martinez)
3263 * fixed some bugs revealed by Paul Dietz' test suite:
3264 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3265 not just nonnegative fixnums;
3266 ** the logical bit-array operators such as BIT-AND now accept an
3267 explicit NIL for their "opt-arg" argument (to indicate a
3268 freshly-consed result bit-array);
3269 ** ELT now signals an error on an invalid sequence index in safe
3271 ** the type system is now cleverer about negations of numeric
3272 types, and consequently understands the BIGNUM and RATIO types
3274 ** the type system is now cleverer about the interaction between
3275 INTEGER and RATIO types: while bugs still remain, many more
3276 cases are accurately computed;
3277 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3278 if it is in the last clause;
3279 ** CONSTANTLY now correctly returns a side-effect-free function in
3281 ** DECLARE is no longer treated as a special-operator; in
3282 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3283 * incremented fasl file version number due to the change in the
3284 DEFSTRUCT-SLOT-DESCRIPTION structure.
3286 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3287 * minor incompatible change: code processed by the "interpreter" or
3288 EVAL now has a compilation optimization policy of (DEBUG 2)
3289 (changed from (DEBUG 1)) to improve debuggability of interactive
3290 development, and to allow the use of the debug RETURN command in
3292 * an experimental implementation of the RETURN command for the
3293 debugger has been included. (thanks to Frederik Kuivinen)
3294 * fixed bug 62: constraints were not propagated into a loop.
3295 * fixed bug in embedded calls of SORT (reported and investigated by
3297 * fixed some bugs revealed by Paul Dietz' test suite:
3298 ** printing and reading of arrays with some dimensions having
3299 length 0 (thanks to Gerd Moellmann);
3300 ** BOA constructor with &AUX argument without a default value does
3301 not cause a type error;
3302 ** CONSTANTP now returns true for all self-evaluating objects.
3304 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3305 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3306 accessors that are related by inheritance, as specified in the
3307 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3308 Valtteri Vuorikoski)
3309 * The compiler is now able to inline functions that were defined in
3310 a complex lexical environment (e.g. inside a MACROLET).
3311 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3313 * Support for the upcoming FreeBSD-5.0 release has been included.
3314 (thanks to Dag-Erling Smorgrav)
3315 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3316 effect when it is not in a toplevel context.
3317 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3318 stand a better chance of being correct. (thanks to Gerd
3320 * fixed bug in COERCE, which now signals an error on coercing a
3321 rational to a bounded real type which excludes the expected
3323 * The compiler is now able to derive types more accurately from the
3324 COERCE and COMPILE functions.
3325 * fixed bug 223: functional binding is considered to be constant
3326 only for symbols in the CL package.
3327 * fixed bug 231: SETQ did not check the type of a variable being set
3328 (reported by Robert E. Brown)
3329 * A new optimization for MAKE-INSTANCE has been included, fixing
3330 various bugs (including relating to :ALLOCATION :CLASS slots and
3331 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3333 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3334 ** As required by ANSI, LOOP now disallows anonymous collection
3335 clauses such as COLLECT I in conjunction with aggregate boolean
3336 clauses such as THEREIS (= I 1);
3337 ** LOOP now signals an error when any variable is reused in the
3338 same loop (including the potentially useful construct analogous
3339 to WITH A = 1 WITH A = (1+ A);
3340 ** IT is only a special loop symbol within the first clause of a
3341 conditional loop clause;
3342 ** LOOP with a typed iteration variable over a hashtable now
3343 signals a type error iff it should.
3344 * fixed some other bugs revealed by Paul Dietz' test suite:
3345 ** FILE-STREAM now names the class previously known as FD-STREAM;
3346 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3347 argument) no longer signals an error;
3348 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3349 of the slot symbol, rather than using the current package
3350 ((:CONC-NAME "") continues to intern the slot's name in the
3352 * incremented fasl file version number, because of the incompatible
3353 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3354 of the new implementation of DEFINE-COMPILER-MACRO.
3356 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3357 * Support for building SBCL for MIPS platforms running in
3358 little-endian mode has now been checked in, and basic
3359 functionality on said platforms verified.
3360 * minor incompatible change: PCL now records the pathname of a file
3361 in which methods and the like are defined, rather than its
3363 * minor incompatible change: TRUENAME now considers the truename of
3364 a file naming a directory to be the pathname with :DIRECTORY
3365 component indicating that directory.
3366 * minor incompatible change: a NAMED clause in the extended form of
3367 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3368 reason for the previous behaviour is unclear.
3369 * more systematization and improvement of CLOS and MOP conformance
3370 in PCL (thanks to Gerd Moellman and Pierre Mai):
3371 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3373 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3374 primary methods with no specializers;
3375 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3377 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3378 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3379 FORWARD-REFERENCED-CLASSes; error reporting on
3380 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3382 ** SXHASH on CLOS instances now uses a slot internal to the
3383 instance to return different numbers on distinct instances,
3384 while preserving the same return value through invocations of
3386 ** DEFMETHOD signals errors when methods with longer incongruent
3387 lambda lists are added to generic functions;
3388 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3389 CLASS, as specified in AMOP;
3390 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3391 on the order returned by the primary method for classes of
3392 class STANDARD-CLASS;
3393 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3394 * fixed some bugs shown by Paul Dietz' test suite:
3395 ** DOLIST puts its body in TAGBODY;
3396 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3398 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3399 value producing form;
3400 ** if more variables are given to PROGV than values, extra
3401 variables are bound and made to have no value;
3402 ** NSUBSTITUTE on list arguments gets the right answer with
3404 ** ELT signals an error of type TYPE-ERROR when the index argument
3405 is not a valid sequence index;
3406 ** LOOP signals (at macroexpansion time) an error of type
3407 PROGRAM-ERROR when duplicate variable names are found;
3408 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3409 ** FUNCALL of special-operators now cause an error of type
3411 ** PSETQ now works as required in the presence of side-effecting
3412 symbol-macro places;
3413 ** NCONC accepts any object as its last argument;
3414 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3416 ** loop-for-as-package does not require a package to be explicitely
3418 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3420 * fixed bug 166: compiler preserves "there is a way to go"
3421 invariant when deleting code.
3422 * fixed bug 172: macro lambda lists with required arguments after
3423 &REST arguments now cause an error to be signalled. (thanks to
3425 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3426 bug. (thanks to Matthew Danish)
3427 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3429 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3430 arguments are lexically rebound. (thanks to Gerd Moellmann and
3432 * fixed bug 194: error messages are now more informative when there
3433 is no primary method applicable in a call to a generic
3434 function. (thanks to Gerd Moellmann)
3435 * fixed bug in command line argument checking (thanks to Julian
3437 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3438 (thanks to Matthew Danish)
3439 * incremented fasl file version number, because of the
3440 SXHASH-related changes in the layout of CLOS data structures
3442 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3443 * minor incompatible change: The runtime (the Unix executable named
3444 "sbcl") is now much pickier about the .core files it will load.
3445 Essentially it now requires .core files to descend from the same
3446 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3447 as the runtime does. (The intent is to prevent the crashes which
3448 can occur, and which can even be reported as mysterious failures,
3449 when people patch the sources or change the build parameters
3450 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3451 sbcl and .core files.)
3452 * fixed bug: VALUES-LIST is no longer optimized away.
3453 * fixed bug 142: The FFI conversion of C string values to Lisp
3454 string values no longer conses excessively. (thanks to Nathan
3455 Froyd porting Raymond Toy's fix to CMU CL)
3456 * began to systematize and improve MOP conformance in PCL (thanks to
3457 Nathan Froyd, Gerd Moellman and Pierre Mai):
3458 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3460 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3461 ** FINALIZE-INHERITANCE is now called on class finalization;
3462 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3463 argument precedence order.
3464 * fixed bug 202: The compiler no longer fails on functions whose
3465 derived types contradict their declared type.
3466 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3467 so it can be non-toplevel.
3468 * The fasl file version number has changed (because of the new
3469 implementation of DEFMACRO).
3470 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3471 safe code, that any length requirement by their type-specifier
3472 argument is valid. The exceptions to this are described in bug
3474 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3475 functions now check the types of their inputs as required by ANSI.
3476 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3477 introduced symbol is DECLAREd to be SPECIAL.
3478 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3479 * fixed bug: :COUNT argument to sequence functions may be negative
3480 * fixed bug: body of DO-SYMBOLS may contain declarations
3481 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3482 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3483 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3484 symbol macro only once
3485 * fixed printing of call frame when argument list is unavailable
3486 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3487 * compiler no longer signals WARNING on unknown keyword
3490 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3491 * A beta-quality port to the mips architecture running Linux,
3492 based on the old CMUCL backend, has been made. It has been tested
3493 on a big-endian kernel, and works sufficiently well to be able to
3494 rebuild itself; it has not been tested in little-endian mode.
3495 * fixed an inconsistency between gencgc.c and purify.c which made
3496 dumping/loading .core files unreliable
3497 * fixed bug 120a: The compiler now deals correctly with IFs where
3498 the consequent is the same as the alternative, instead of
3499 misderiving the return type. (thanks to Alexey Dejneka)
3500 * fixed bug 113: Logical pathnames are now dumpable (the logical
3501 host is resolved at load-time, throwing an error if it is not
3503 * fixed bug 174: FORMAT's error message is slightly clearer when a
3504 non-printing character is used in a format directive.
3505 * fixed several bugs in compiler checking of type declarations, i.e.
3506 violations of the Python "declarations are assertions" principle
3507 (thanks to Alexey Dejneka)
3508 * fixed several bugs in PCL's error checking (thanks to Gerd
3510 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3512 * fixed bug in compilation of functions as first class values
3513 (thanks to Antonio Martinez-Shotton)
3514 * The compiler's handling TYPE-ERRORs which it can prove will
3515 inevitably happen at runtime has been cleaned up and corrected
3516 in several ways. (thanks to Alexey Dejneka)
3517 * improved argument type checking for various basic arithmetic
3518 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3519 lavished on them in the past that they can be compiled in many
3520 ways in different special cases
3521 * fixed bug 181: compiler checks validity of user supplied type
3523 * cleaned up code flushing in optimization: Function calls which
3524 should signal errors for safety purposes (e.g. which ANSI says
3525 should signal errors when their arguments are of incorrect type)
3526 are no longer optimized away.
3527 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3528 * incremented fasl file version number, because changes in the
3529 implementation of sequence functions like COERCE caused
3530 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3531 in old inline expansions) to become undefined. (Actually these
3532 changes were later undone, so we might very well be binary
3533 compatible with 0.7.7 after all, but leaving the version number
3534 incremented seemed like the simplest and most conservative
3537 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3538 * An alpha-quality port to the parisc architecture running Linux,
3539 based on the old CMUCL backend, has been made. This, even more so
3540 than the other backends, should be considered still a work in
3541 progress; known problems include that the Linux kernel in 64-bit
3542 mode does not propagate the correct sigcontext structure to
3543 userspace, and consequently SBCL on a parisc64 kernel will not
3545 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3546 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3547 suppressed.) Also now that the compiler is looking at declarations
3548 in the environment, it checks optimization declarations as well,
3549 and suppresses inlining when (> DEBUG SPEED).
3550 * More fixes have been made to treatment of floating point exception
3551 treatment and other Unix signals. In particular, floating point
3552 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3553 * The detection and handling of control stack exhaustion (infinite
3554 or very deeply nested recursion) has changed. Stack exhaustion
3555 detection is now done by write-protecting pages at the OS level
3556 and applies at all optimization settings; when found, a
3557 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3558 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3559 be caught using IGNORE-ERRORS.
3560 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3561 correctly when transforming recursive LABELS functions to LETs.
3562 (thanks to Alexey Dejneka)
3563 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3564 that are names of constants or global variables.
3565 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3566 alien routines with docstrings.
3567 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3568 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3570 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3571 (thanks to Eric Marsden)
3572 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3573 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3574 * Bug fix: The SPARC backend can now compile functions involving
3575 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3576 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3577 to a functions expecting a PATHNAME-DESIGNATOR.
3578 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3579 lambda lists. (thanks to Alexey Dejneka)
3580 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3581 functions so that EXPORTing the name of the class doesn't cause
3582 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3583 to Antonio Martinez for reporting this)
3584 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3585 constant-folded without failing an assertion. (thanks to Einar
3586 Floystad Dorum for reporting this)
3587 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3588 types are now dealt with more correctly by the compiler.
3589 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3590 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3591 behaviour analogous to RENAME-FILE. This puts its behaviour more
3592 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3594 * The fasl file version number has changed again. (because of the
3595 bug fix involving the names of PCL MAKE-INSTANCE functions)
3597 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3598 * bug fix: Floating point exceptions are treated much more
3599 consistently on the x86/Linux and PPC/Linux platforms.
3600 * Array initialization with :INITIAL-ELEMENT is now much faster for
3601 cases when the compiler cannot open code the array creation, but
3602 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3603 array accesses have also seen a speed increase.
3604 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3605 specified. (This was at the root of some bad interactions between
3606 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3608 * bug fix: Internal error arguments for undefined functions are now
3609 computed correctly on the PPC/Linux platform.
3610 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3611 Raymond Toy's patch for CMU CL)
3612 * Support for the Solaris 9 operating environment has been included
3613 (thanks to Daniel Merritt)
3614 * A very ugly but hopefully complete draft of the missing FFI chapter
3615 of the manual has been created by reformatting the corresponding
3616 CMU CL manual chapter into (currently very ugly and incoherent)
3617 DocBook and bringing it up to date for SBCL behavior. Thus, the
3618 manual is now essentially complete, at least by my extreme
3619 once-and-only-once standards, whereby it's acceptable to refer to
3620 the doc strings of SB-EXT functions as the primary documentation.
3621 * The fasl file version number has changed again, due to cleanup of
3622 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3623 is no longer a static symbol.)
3625 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3626 * SBCL now builds with OpenMCL (version 0.12) as the
3627 cross-compilation host; also, more progress has been made toward
3628 bootstrapping under CLISP.
3629 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3631 * bug 158 fixed: The compiler can now deal with integer loop
3632 increments different from 1; fixing this turned out also to fix
3634 * bug 169 fixed: no more bogus warnings about using lexical bindings
3635 despite the presence of perfectly good SPECIAL declarations (thanks
3636 to David Lichteblau)
3637 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3638 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3639 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3641 * bug 180 fixed: Method combination specifications no longer ignore
3642 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3643 * bug fix: Structure type predicate functions now check their argument
3644 count as they should.
3645 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3646 correctly. (thanks to Pierre Mai)
3647 * minor incompatible change: The --noprogrammer option is deprecated
3648 in favor of the new --disable-debugger option, which is very similar.
3649 (The major difference is that it takes effect at a slightly different
3650 time at startup, causing handling of errors in --sysinit and
3651 --userinit files will be affected differently.) The
3652 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3653 been added to allow this functionality to be controlled from ordinary
3654 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3655 maintainers, who might want to run non-interactive scripts to
3656 build SBCL cores which will later be used interactively.)
3657 * minor incompatible change: The LOAD function no longer, when given
3658 a wild pathname to load, loads all files matching that pathname.
3659 Instead, an error of type FILE-ERROR is signalled.
3661 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3662 * bug 147 fixed: The compiler preserves its block link/count
3663 invariants more correctly now so that it doesn't crash. (thanks
3665 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3667 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3668 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3669 no-value-for-this-character as NIL (as specified by ANSI).
3670 * HOST-NAMESTRING on physical pathnames now returns a string that is
3671 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3672 (thanks to Christophe Rhodes)
3673 * The Alpha port handles icache flushing more correctly. (thanks to
3675 * More progress has been made toward bootstrapping under CLISP. (thanks
3676 to Christophe Rhodes)
3677 * The fasl file format has changed again, because dynamic loading
3678 on OpenBSD (which has non-ELF object files) motivated some cleanups
3679 in the way that foreign symbols are transformed and passed around.
3680 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3681 is no longer treated as whitespace by the reader, but instead as
3682 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3683 |A
\7fB|, instead of A as it used to.
3685 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3686 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3687 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3688 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3689 this as of 0.7.1.45, but was left out of the previous news section
3690 (thanks to Dan Barlow)
3691 * SBCL now runs on the Solaris operating system on SPARC architectures
3692 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3693 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3694 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3695 * SPARC backend cleanups, allowing builds of cores optimized for V8
3696 and V9 SPARCS, and also emission of code targeted to a particular
3697 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3699 * SBCL is closer to bootstrapping under CLISP, thanks to various
3700 fixes by Christophe Rhodes.
3701 * The fasl file format has changed again, to allow the compiler's
3702 INFO database to support symbol macros.
3703 * The user manual (in doc/) is formatted into HTML more nicely.
3704 (thanks to coreythomas)
3705 * The system is smarter about SUBTYPEP relationships, especially
3706 those involving NOT types (including types such as ATOM which are
3707 represented internally using NOT types). Thus SUBTYPEP is less
3708 likely to return (VALUES NIL NIL) in general, and in particular
3709 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3711 * The fasl file format has changed again, because the internal
3712 representation of types now includes a new slot to support the new
3713 SUBTYPEP-of-NOT-types logic.
3714 * (not a change in the main branch of SBCL, but a related prototype
3715 which can hopefully be merged into the main branch of SBCL in the
3716 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3717 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3718 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3719 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3720 dispatch macro character combinations. (thanks to Alexey Dejneka)
3721 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3722 that superficially look like logical namestrings correctly.
3723 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3725 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3726 default supplied-p) for &optional and &key arguments. (thanks to
3729 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3730 * incompatible change: The compiler is now less aggressive about
3731 tail call optimization, doing it only when (> SPACE DEBUG) or
3732 (> SPEED DEBUG). (This is an incompatible change because there are
3733 programs which relied on the old CMU-CL-style behavior to optimize
3734 away their unbounded recursion which will now die of stack overflow.)
3735 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3736 for non-GENCGC systems has been increased to 20M (since that
3737 seems much closer to the likely performance optimum for modern
3738 systems than the old 4M value was)
3739 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3740 and *DEBUG-PRINT-LEVEL*
3741 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3742 of CMU CL's support for SPARC, and various endianness and other
3743 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3744 * new syntactic sugar for the Unix command line: --load foo.bar is now
3745 an alternate notation for --eval '(load "foo.bar")'.
3747 ** The system now detects stack overflow and handles it gracefully,
3748 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3749 optimization settings. (This is a good thing in general, and
3750 its introduction in this version should be particularly timely
3751 for anyone whose code fails because of suppression of tail
3753 ** The system now hunts for the C variable "environ" in a more
3754 devious way, to avoid segfaults when the C library version differs
3755 between compile time and run time. (thanks to Christophe Rhodes)
3756 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3757 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3758 ** The compiler no longer issues bogus style warnings for undefined
3759 classes in the same source file as the DEFCLASSes which defined
3760 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3762 ** fixes in CONDITION class precedence list for undefined function
3763 errors (thanks to Alexei Dejneka)
3764 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3765 correctly. (thanks to Dan Barlow)
3766 ** portability fixes aiming at bootstrapping under CLISP (thanks
3767 to Dave McDonald and Christophe Rhodes)
3768 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3769 ** fixes in type translation and and type inference (thanks to
3771 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3772 ** various fixes in the new ports (thanks to Dan Barlow)
3773 * several changes related to debugging:
3774 ** suppression of tail recursion, as noted above
3775 ** stack overflow detection, as noted above
3776 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3777 is now the default. (For some time encapsulation has been more
3778 reliable than the breakpoint-based :ENCAPSULATE NIL
3779 implementation, at least on X86 systems; and I just noticed that
3780 encapsulation also seems closer to the spirit of the ANSI
3783 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3785 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3786 up properly again. (There was a packaging bug in 0.7.0 which
3787 left their definitions in SB-SYS::LOAD-FOREIGN and
3788 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3789 vital for most things which interface to C-level interfaces,
3790 like extensions working with sockets or databases or
3791 Perl-compatible regexes or whatever, and the need to fix
3792 this bug is the main reason that 0.7.1 was released so
3794 ** DEFGENERIC is now choosier about the methods it redefines, so that
3795 reLOADing a previously-LOADed file containing DEFGENERICs does
3796 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3797 cycle now works as it should. (thanks to Alexey Dejneka)
3798 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3799 fixed by Christophe Rhodes. (He actually submitted this patch
3800 months ago, and I delayed until after 0.7.0.)
3801 ** Bug 111 (internal compiler confusion about runtime checks on
3802 FUNCTION types) was fixed by Alexey Dejneka.
3803 * Some internal cleanups (getting rid of variables which aren't
3804 needed now that the byte interpreter is gone) caused the fasl
3805 file format number to change again.
3807 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3808 * major incompatible change: The default fasl file extension, i.e. the
3809 default extension for files produced by COMPILE-FILE, has changed
3810 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3812 ** There are many changes in the implementation of the compiler.
3813 SBCL is now essentially a compiler-only implementation of ANSI
3814 Common Lisp. EVAL still "interprets" a few special cases, but
3815 almost all the interesting cases are handled by creating
3816 a LAMBDA expression, calling COMPILE on it, then calling
3817 FUNCALL on the result.
3818 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3819 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3820 Since the code is newer, there might still be some new bugs
3821 (though not as many as before Martin Atzmueller's fixes:-). But
3822 the new code is substantially simpler and clearer, and hopefully
3823 any remaining bugs will be simpler, less fundamental, and more
3824 fixable then the bugs in the old code.
3825 ** The revised compiler is still a little unsteady on its feet.
3827 *** The debugging information it produces (particularly the names
3828 of FUNCTION objects) is sometimes much less useful than what
3829 the old compiler produced.
3830 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3831 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3832 within a MACROLET) has been temporarily weakened.
3833 ** There are new compiler optimizations for various functions:
3834 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3835 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3836 *** the math functions TRUNCATE, FLOOR, and CEILING
3837 *** the function-of-all-trades COERCE
3838 Mostly these should be transparent, but there's one
3839 potentially-annoying problem (bug 117): when the compiler
3840 inline-expands a function and does type analysis on the result,
3841 it can create control paths which have type mismatches, and
3842 when it can't prove that those control paths aren't taken,
3843 it will issue WARNINGs about the type mismatches. This is
3844 a particular problem in practice for the new sequence functions.
3845 It's not clear how this should be fixed, and for now, a
3846 workaround is given in the entry for 117 in the BUGS file.
3847 ** (Because of the interaction between the two previous items --
3848 occasional inlining problems and new inline expansions -- some
3849 of the new sequence function optimizations won't really kick in
3850 completely until debugging information, and then inlining, are
3851 straightened out in some future version.)
3852 * minor incompatible changes:
3853 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3854 DIRECTORY no longer implicitly promotes NIL slots of its
3855 pathname argument to :WILD. In particular, when you ask for the
3856 contents of a directory (which you used to be able to do without
3857 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3858 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3859 ** changes in behavior that ANSI explicitly defines to be
3860 implementation dependent:
3861 *** The new compiler-only implementation still conforms with ANSI,
3862 but acts a little different than before. Besides the obvious
3863 changes in performance tradeoffs (that the cost per form passed
3864 to EVAL has gone up, and the cost per form executed by EVAL
3865 has gone down), the behavior of the system changes a little
3866 because there are no longer any interpreted function objects.
3867 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3868 e.g. doing COMPILE on the output of interactive DEFUN is
3870 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3872 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3873 has changed from (:CL) to NIL.
3874 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3875 e.g. #\Newline and #\Tab, don't have names specified in the
3876 ANSI Common Lisp standard, is now based on their ASCII symbolic
3877 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3878 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3879 are no longer used for output.
3880 ** changes in internal implementation constants:
3881 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3882 4 million. (If your application spends a lot of time GCing and
3883 you have a lot of RAM, you might want to experiment with
3884 increasing it even more.)
3885 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3886 However, almost all old code should still continue to work without
3887 immediate update, as SB-C-CALL is now a (deprecated) nickname
3889 ** Old operator names in the style DEF-FOO are now deprecated in
3890 favor of new corresponding names DEFINE-FOO, for consistency with
3891 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3892 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3893 internal symbols, but a few supported extensions like
3894 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3895 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3896 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3897 etc. as you get deeper into recursive calls to the debugger
3898 command loop, instead of the old "5]", "5]]", "5]]]"
3899 sequence. (I was motivated to do this when squabbles between
3900 ILISP and SBCL left me very deeply nested in the debugger. In the
3901 short term, this change will probably provoke more ILISP/SBCL
3902 squabbles, but hopefully it will be an improvement in the long run.)
3903 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3904 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3905 function, since declaiming return types involving aliens is
3906 (1) annoyingly messy to do by hand and (2) vital to efficient
3907 compilation of code which calls such functions.
3908 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3909 longer reexported by the SB-EXT package. They're solely useful
3910 for alien code, so it seems more logical that you should get
3911 them from the SB-ALIEN package, not in SB-EXT.
3912 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3913 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3914 features. Instead, the code that they used to control is always
3915 built into the system.
3916 * many other bug fixes
3917 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3918 advantage of the new EVAL-WHEN stuff and to clean them up in
3919 general, and they are now more ANSI-compliant in a number of
3920 ways. Martin Atzmueller is responsible for a lot of this.
3921 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3923 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3924 *** correct ERROR type for various file operations
3925 *** some fixes for Lisp streams
3926 *** DEFMETHOD syntax checking
3927 *** changing old weird representation of debug information as
3928 strings (which, among their other deficiencies, don't transform
3929 correctly when you rename packages, and don't change their
3930 print representation when you change things like *PACKAGE*
3931 and *PRINT-LENGTH*) to symbols and lists of symbols
3932 He also made several improvements and fixed several bugs in DESCRIBE.
3933 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3935 *** misbehavior of WRITE-STRING/WRITE-LINE
3936 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3937 and several other LOOP problems as well
3938 *** DIRECTORY when similar filenames are present
3939 *** DEFGENERIC with :METHOD options
3940 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3941 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3942 *** argument ordering in FIND with :TEST option
3943 *** mishandled package designator argument in APROPOS-LIST
3944 *** various problems in the backquote readmacro
3945 *** a bug in APROPOS
3946 *** probably some others that I'm not describing very well here,
3947 since the CVS log documents them by reference to sbcl-devel
3948 messages, and the SourceForge archives aren't working well.:-(
3949 ** Dan Barlow improved the Alpha port (and is making progress on the
3950 PPC port, for those of you who think different).
3951 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3952 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3953 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3954 continues to work on the SPARC port (for those of you in a position
3955 to look down upon our little PC-compatible boxes from a great height).
3956 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3957 and uses that copy, rather than the current dynamic value, when
3958 it's trying to decide whether to truncate output. Thus e.g.
3959 (let ((*print-lines* 50))
3960 (pprint-logical-block (stream nil)
3962 (let ((*print-lines* 8))
3963 (print (aref possiblybigthings i) stream)))))
3964 should now truncate the logical block only at 50 lines, instead of
3965 often truncating it at 8 lines, as it did before.
3966 * The doc/cmucl/ directory, containing old CMU CL documentation
3967 from the time of the fork, is no longer part of the base system.
3968 SourceForge has shut down its anonymous FTP service, and with it
3969 my original plan for distributing the old CMU CL documentation
3970 there. For now, if you need these files you can download an old
3971 SBCL source release and extract them from it.
3972 * The fasl file version number changed again, for dozens of reasons,
3973 some of which are apparent above.
3975 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3976 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3977 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3978 patch, so that DISASSEMBLE output is much nicer.
3979 * The code in the SB-PROFILE package now seems reasonably stable.
3980 I still haven't decided what the final interface should look like
3981 (I'd like PROFILE to interact cleanly with TRACE, since both
3982 facilities use function encapsulation) but if you have a need
3983 for profiling now, you can probably use it successfully with
3984 the current CMU-CL-style interface.
3985 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3986 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3987 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3988 link now returns the dangling link itself, and for similar
3989 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3990 link itself. (In these cases the old code signalled an error and
3991 looped endlessly, respectively.) Thus, DIRECTORY now works even
3992 in the presence of dangling and cyclic symbolic links.
3993 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3994 is now a supported extension again, since the consensus on
3995 sbcl-devel was that it can be useful for ordinary development
3996 work, not just for debugging SBCL itself.
3997 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3998 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3999 that currently-defined functions might be redefined later with
4000 different return types.
4001 * Hash tables can be printed readably, as inspired by CMU CL code
4002 of Eric Marsden and SBCL code of Martin Atzmueller.
4003 * better error handling in CLOS method combination, thanks to
4004 Martin Atzmueller porting Pierre Mai's CMU CL patches
4005 * more overflow fixes for >16Mbyte I/O buffers
4006 * A bug in READ has been fixed, so that now a single Ctrl-D
4007 character suffices to cause end-of-file on character streams.
4008 In particular, now you only need one Ctrl-D at the command
4009 line (not two) to exit SBCL.
4010 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4012 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4013 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4014 does the right thing.
4015 * The implementation of some type tests, especially for CONDITION
4016 types, is now tidier and maybe faster, due to CMU CL code
4017 originally by Douglas Crosher, ported by Martin Atzmueller.
4018 * Some math functions have been fixed, and there are new
4019 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4020 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4021 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4022 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4023 likely to be built into sbcl-0.7.0.)
4024 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4025 CL, ported to SBCL by Martin Atzmueller)
4026 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4027 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4028 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4029 * There's a new slam.sh hack to shorten the edit/compile/debug
4030 cycle for low-level changes to SBCL itself, and a new
4031 :SB-AFTER-XC-CORE target feature to control the generation of
4032 the after-xc.core file needed by slam.sh.
4033 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4034 COMPILE-FILE is no longer supported, so that now every function
4035 gets an entry point, so that block compilation looks a little
4036 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4037 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4038 returns the number of bytes consed since the system started,
4039 rather than the number consed since the first time the function
4040 was called. (The new definition parallels ANSI functions like
4041 CL:GET-INTERNAL-RUN-TIME.)
4042 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4043 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4044 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4045 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4046 * Fasl file version numbers are now independent of the target CPU,
4047 since historically most system changes which required version
4048 number changes have affected all CPUs equally. Similarly,
4049 the byte fasl file version is now equal to the ordinary
4052 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4053 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4054 is no longer recognized. I apologize for this, because it was
4055 listed in SB-EXT as a supported extension, but I found that
4056 its existing behavior was poorly specified, as well as incorrectly
4057 specified, and it looked like too much of a mess to straighten it
4058 out. I have enough on my hands trying to get ANSI stuff to work..
4059 * many patches ported from CMU CL by Martin Atzmueller, with
4060 half a dozen bug fixes in pretty-printing and the debugger, and
4061 half a dozen others elsewhere
4062 * fixed bug 13: Floating point infinities are now supported again.
4063 They might still be a little bit flaky, but thanks to bug reports
4064 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4065 as flaky as they were.
4066 * The --noprogrammer command line option is now supported. (Its
4067 behavior is slightly different in detail from what the old man
4068 page claimed it would do, but it's still appropriate under the
4069 same circumstances that the man page talks about.)
4070 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4071 are now supported, and enabled by default. Thus, the compiler can
4072 handle many floating point and complex operations much less
4073 inefficiently. (Thus e.g. you can implement a complex FFT
4075 * The compiler now detects type mismatches between DECLAIM FTYPE
4076 and DEFUN better, and implements CHECK-TYPE more correctly, and
4077 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4078 thanks to patches from Martin Atzmueller.
4079 * various fixes to make the cross-compiler more portable to
4080 ANSI-conforming-but-different cross-compilation hosts (notably
4081 Lispworks for Windows, following bug reports from Arthur Lemmens)
4082 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4083 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4084 fixed thanks to Pierre Mai's CMU CL patches.
4085 * fixes to keep the system from overflowing internal counters when
4086 it tries to use i/o buffers larger than 16M bytes
4087 * fixed bug 45a: Various internal functions required to support
4088 complex special functions have been merged from CMU CL sources.
4089 (When I was first setting up SBCL, I misunderstood a compile-time
4090 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4091 * improved support for type intersection and union, fixing bug 12
4092 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4093 more obscure bugs as well
4094 * some steps toward byte-compiling non-performance-critical
4095 parts of the system, courtesy of patches from Martin Atzmueller
4096 * Christophe Rhodes has made some debian packages of sbcl at
4097 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4098 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4099 stable, but are nonetheless usable. When he's ready, I'd be happy
4100 to add them to the SourceForge "File Releases" section. (And if
4101 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4102 * new fasl file format version number (because of changes in
4103 internal representation of (OR ..) types to accommodate the new
4104 support for (AND ..) types, among other things)
4106 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4107 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4109 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4111 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4112 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4113 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4114 * DESCRIBE now works on structure objects again.
4115 * Most function call argument type mismatches are now handled as
4116 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4117 know whether the function will be redefined before the call is
4118 executed. (The compiler could flag local calls with full WARNINGs,
4119 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4120 it doesn't keep track of enough information to know whether calls
4121 are local in this sense.)
4122 * Compiler output is now more verbose, with messages truncated
4123 later than before. (There should be some supported way for users
4124 to override the default verbosity, but I haven't decided how to
4125 provide it yet, so this behavior is still controlled by the internal
4126 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4127 src/compiler/ir1util.lisp.)
4128 * Fasl file format version numbers have increased again, because
4129 support for the Gray streams extension changes the layout of the
4130 system's STREAM objects.
4131 * The Gray subclassable streams extension now works, thanks to a
4132 patch from Martin Atzmueller.
4133 * The full LOAD-FOREIGN extension (not just the primitive
4134 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4135 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4136 but like most other programs, it defaults to copying the Unix
4137 environment from the original process instead of starting the
4138 new process in an empty environment.
4139 * Extensions which manipulate the Unix environment now support
4140 an :ENVIRONMENT keyword option which doesn't smash case or
4141 do other bad things. The CMU-CL-style :ENV option is retained
4142 for porting convenience.
4143 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4144 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4146 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4148 * A patch from Martin Atzmueller seems to have solved the SIGINT
4149 problem, and as far as we know, signal-handling now works cleanly.
4150 (If you find any new bugs, please report them!)
4151 * The system no longer defaults Lisp source file names to types
4152 ".l", ".cl", or ".lsp", but only to ".lisp".
4153 * The compiler no longer uses special default file extensions for
4154 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4155 seems to expect a single default extension for all compiled code,
4156 and there's no compelling reason to try to stretch the standard
4157 to allow two different extensions.) Instead, byte-compiled files
4158 default to the same extension as native-compiled files.
4159 * Fasl file format version numbers have increased again, because
4160 a rearrangement of internal implementation packages made some
4161 dumped symbols in old fasl files unreadable in new cores.
4162 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4163 many fewer weird special cases.
4164 * Bug #17 (differing COMPILE-FILE behavior between logical and
4165 physical pathnames) has been fixed, and some related misbehavior too,
4166 thanks to a patch from Martin Atzmueller.
4167 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4168 by Tim Moore, ported to SBCL by Martin Atzmueller.
4169 * Martin Atzmueller fixed several filesystem-related problems,
4170 including bug #36, in part by porting CMU CL patches, which were
4171 written in part by Paul Werkowski.
4172 * More compiler warnings in src/runtime/ are gone, thanks to
4173 more patches from Martin Atzmueller.
4174 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4177 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4179 * DESCRIBE now works on CONDITION objects.
4180 * The debugger now handles errors which arise when trying to print
4181 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4183 * The build system now uses an additional file, customize-target-features.lisp,
4184 to allow local modifications to the target *FEATURES* list. (The point of
4185 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4186 debugging features enabled, without having to worry about propagating it
4187 into everyone's system when I do a "cvs update".) When no
4188 customize-target-features.lisp file exists, the target *FEATURES* list
4189 should be constructed the same way as before.
4190 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4191 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4192 warn about behavior which is undefined under the ANSI spec. Note: This
4193 is specified by ANSI, but it's not very popular with programmers.
4194 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4195 macro in the SBCL sources for an example of a workaround which you
4196 might use to make portable ANSI-standard code which does what you want.
4197 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4198 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4199 when it's used as a non-toplevel form. (This is required in order
4200 to implement the DEFCONSTANT-EQX macro.)
4201 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4202 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4203 funny things, due to bugs in the implementation of EVAL-WHEN
4204 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4205 non-toplevel DEFCONSTANTs are uncommon.)
4206 * The core file version number and fasl file version number have been
4207 incremented, because the old noncompliant DEFCONSTANT behavior involved
4208 calling functions which no longer exist, and because I also took the
4209 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4210 * fixed bug 1 (error handling before read-eval-print loop starts), and
4211 redid debugger restarts and related debugger commands somewhat while
4213 ** The QUIT debugger command is gone, since it did something
4214 rather different than the SB-EXT:QUIT command, and since it never
4215 worked properly outside the main toplevel read/eval/print loop.
4216 Invoking the new TOPLEVEL restart provides the same functionality.
4217 ** The GO debugger command is also gone, since you can just invoke
4218 the CONTINUE restart directly instead.
4219 ** The TOP debugger command is also gone, since it's redundant with the
4220 FRAME 0 command, and since it interfered with abbreviations for the
4222 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4224 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4225 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4226 some porting work by Martin Atzmueller)
4227 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4228 with a boatload of patches which he ported to SBCL. Now that those
4230 ** The system tries to make sure that its low-priority messages
4231 are prefixed by semicolons, to help people who like to use
4232 syntax highlighting in their ILISP buffer. (This patch
4233 was originally due to Raymond Toy.)
4234 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4235 patches originally written by Raymond Toy.
4236 ** The compiler understands coercion between single-value and
4237 multiple-VALUES type expressions better, getting rid of some very
4238 weird behavior, thanks to patches originally by Robert MacLachlan
4239 and Douglas Crosher.
4240 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4241 lambda lists, thanks to a patch originally by Pierre Mai.
4242 ** The system no longer bogusly warns about "abbreviated type
4244 ** The compiler gets less confused by inlining and RETURN-FROM,
4245 thanks to some patches originally by Tim Moore.
4246 ** The system no longer hangs when dumping circular lists to fasl
4247 files, thanks to a patch originally from Douglas Crosher.
4248 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4249 undefined function error.
4250 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4251 in structure slot initforms) for now, documented workaround instead:-|
4252 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4253 * fixed bug 5: added stubs for various Gray stream functions called
4254 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4255 installed, at least appropriate type errors are generated
4256 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4257 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4258 consistently in DEFMETHOD forms.
4259 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4260 it doesn't seem to affect SBCL after all
4261 * The C runtime system now builds with better optimization and many
4262 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4264 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4266 * The system is now under CVS at SourceForge (instead of the
4267 CVS repository on my home machine).
4268 * The new signal handling code has been tweaked to treat register
4269 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4270 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4271 downstream, e.g. in debug-int.lisp, has implicit dependencies
4272 on the unsignedness of integer representation of machine words,
4273 and that caused the system to bomb out with infinite regress
4274 when trying to recover from type errors involving signed values,
4275 e.g. (BUTLAST '(1 2 3) -1).)
4276 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4277 (This was one of the bugs Peter Van Eynde reported back in July.)
4278 * The system now uses code inspired by Colin Walters' O(N)
4279 implementation of MAP (from the cmucl-imp@cons.org mailing
4280 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4281 inline the MAP operation, and there is more than one
4282 sequence argument to the MAP call (so that it can't just
4283 do ETYPECASE once and for all based on the type of the
4284 single sequence argument). (The old non-inline implementation
4285 of the general M-argument sequence-of-length-N case required
4286 O(M*N*N) time when any of the sequence arguments were LISTs.)
4287 * The QUIT :UNIX-CODE keyword argument has been renamed to
4288 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4289 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4291 * Raymond Wiker's patches to port dynamic loading from Linux to
4292 FreeBSD have been added.
4293 * The BUGS file is now more nearly up to date, thanks in large part
4294 to Martin Atzmueller's review of it.
4295 * The debugger now flushes standard output streams before it begins
4296 its output ("debugger invoked" and so forth).
4297 * The core version number and fasl file version number have both
4298 been incremented, because of incompatible changes in the layout
4300 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4301 used to not be called for a saved Lisp image.)
4302 * Martin Atzmueller's version of a patch to fix a compiler crash,
4303 as posted on sbcl-devel 13 September 2000, has been installed.
4304 * Instead of installing Martin Atzmueller's patch for the
4305 compiler transform for SUBSEQ, I deleted the compiler transform,
4306 and transforms for some similar consing operations.
4307 * A bug in signal handling which kept TRACE from working on OpenBSD
4309 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4310 away by constant folding
4311 * The system now defines its address space constants in one place
4312 (in the Lisp sources), and propagates them automatically elsewhere
4313 (through GENESIS and the sbcl.h file). Therefore, patching the
4314 address map is less unnecessarily tedious and error-prone. The
4315 Lisp names of address space constants have also been systematized.
4316 * CVS tags like dollar-Header-dollar have been removed from
4317 the sources, because they have never saved me trouble and
4318 they've been source of trouble working with patches and other
4319 diff-related operations.
4320 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4321 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4323 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4325 * The system has been ported to OpenBSD.
4326 * The system now compiles with a simple "sh make.sh" on the systems
4327 that it's supported on. I.e., now you no longer need to tweak
4328 text in the target-features.lisp-expr and symlinks in src/runtime/
4329 by hand, the make.sh takes care of it for you.
4330 * The system is no longer so grossly inefficient when compiling code
4331 involving vectors implemented as general (not simple) vectors (VECTOR T),
4332 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4333 which dares to use the various sequence functions on non-simple
4334 vectors, takes less of a performance hit.
4335 * There is now a primitive type predicate VECTOR-T-P
4336 to test for the (VECTOR T) type, so that e.g.
4337 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4338 can now be compiled with some semblance of efficiency. (The old code
4339 turned the type declaration into a full call to %TYPEP at runtime!)
4340 * AREF on (VECTOR T) is still not fast, since it's still compiled
4341 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4342 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4343 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4344 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4345 without making full calls to SUBTYPEP at runtime.
4346 (Some analogous efficiency issues for non-simple vectors specialized to
4347 element types other than T, or for non-simple multidimensional arrays,
4348 have not been addressed. They could almost certainly be handled the
4349 same way if anyone is motivated to do so.)
4350 * The changes in array handling break binary compatibility, so
4351 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4352 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4353 instead of (VALUES T T).
4354 * By following the instructions that Dan Barlow posted to sbcl-devel
4355 on 2 July 2000, I was able to enable primitive dynamic object
4356 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4357 functionality is not implemented (since it calls ld to resolve
4358 library references automatically, requiring RUN-PROGRAM for its
4359 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4360 try to resolve library references) is now supported.
4361 * The system now flushes the standard output streams when it terminates,
4362 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4363 them at several other probably-convenient times, e.g. in each pass of
4364 the toplevel read-eval-print loop, and after evaluating a form given
4365 as an "--eval" command-line option. (These changes were motivated by a
4366 discussion of stream flushing issues on cmucl-imp in August 2000.)
4367 * The source transform for TYPEP of array types no longer assumes
4368 that an array whose element type is a not-yet-defined type
4369 is implemented as an array of T, but instead punts, so that the
4370 type will be interpreted at runtime.
4371 * There is now some support for cross-compiling in make.sh: each of
4372 the phases of make.sh has its own script. (This should be transparent
4373 to people doing ordinary, non-cross-compile builds.)
4374 * Since my laptop doesn't have hundreds of megabytes of memory like
4375 my desktop machine, I became more motivated to do some items on
4376 my to-do list in order to reduce the size of the system a little:
4377 ** Arrange for various needed-only-at-cold-init things to be
4378 uninterned after cold init. To support this, those things have
4379 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4380 symbols with such names are now uninterned after cold init).
4381 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4382 fasl files for cold load.
4383 ** Remove the old compiler structure pooling code (which used to
4384 be conditional on the target feature :SB-ALLOC) completely.
4385 ** Redo the representation of some data in cold init to be more compact.
4386 (I also looked into supporting byte compiled code at bootstrap time,
4387 which would probably reduce the size of the system a lot, but that
4388 looked too complicated, so I punted for now.)
4389 * The maximum signal nesting depth in the src/runtime/ support code has
4390 been reduced from 4096 to 256. (I don't know any reason for the very
4391 large old value. If the new smaller value turns out to break something,
4392 I'll probably just bump it back up.)
4393 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4395 * Many, many bugs reported by Peter Van Eynde have been added to
4396 the BUGS list; some have even been fixed.
4397 * While enabling dynamic object file loading, I tried to make the
4398 code easier to understand, renaming various functions and variables
4399 with less ambiguous names, and changing some function calling
4400 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4401 * While trying to figure out how to do the OpenBSD port, I tried to
4402 clean up some of the code in src/runtime/. In particular, I dropped
4403 support for non-POSIX signal handling, added various comments,
4404 tweaked the code to reduce the number of compilation warnings, and
4405 renamed some files to increase consistency.
4406 * To support the new automatic configuration functionality in make.sh,
4407 the source file target-features.lisp-expr has been replaced with the
4408 source file base-target-features.lisp-expr and the machine-generated
4409 file local-target-features.lisp-expr.
4410 * fixed a stupid quoting error in make.sh so that using CMU CL
4411 "lisp -batch" as cross-compilation host works again
4413 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4415 * DESCRIBE no longer tries to call itself recursively to describe
4416 bound/fbound values, so that it no longer fails on symbols which are
4417 bound to themselves (like keywords, T, and NIL).
4418 * DESCRIBE now works on generic functions.
4419 * The printer now prints less-screwed-up representations of closures
4420 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4421 * A private symbol is used instead of the :EMPTY keyword previously
4422 used to mark empty slots in hash tables. Thus
4423 (DEFVAR *HT* (MAKE-HASH-TABLE))
4424 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4425 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4426 now does what ANSI says that it should. (You can still get
4427 similar noncompliant behavior if bang on the hash table
4428 implementation with all the symbols you get back from
4429 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4430 This breaks binary compatibility, since tests for equality to
4431 :EMPTY are wired into things like the macroexpansion of
4432 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4434 * There's now a minimal placeholder implementation for CL:STEP,
4435 as required by ANSI.
4436 * An obscure bug in the interaction of the normal compiler, the byte
4437 compiler, inlining, and structure predicates has been patched
4438 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4439 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4441 * Missing ordinary arguments in a macro call are now detected even
4442 when the macro lambda list contains &KEY or &REST.
4443 * The debugger no longer complains about encountering the top of the
4444 stack when you type "FRAME 0" to explicitly instruct it to go to
4445 the top of the stack. And it now prints the frame you request even
4446 if it's the current frame (instead of saying "You are here.").
4447 * As specified by ANSI, the system now always prints keywords
4448 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4450 * The default initial SIZE of HASH-TABLEs is now smaller.
4451 * Type information from CLOS class dispatch is now propagated
4452 into DEFMETHOD bodies, so that e.g.
4453 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4455 is now basically equivalent to
4456 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4457 (DECLARE (TYPE SINGLE-FLOAT X))
4459 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4460 operation, without having to do run-time type dispatch.
4461 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4462 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4463 * Since I don't seem to be making big file reorganizations very often
4464 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4465 is overflowing my ability to conveniently back them up), I've finally
4466 checked the system into CVS. (The CVS repository is on my home system,
4467 not at SourceForge -- putting it on SourceForge might come later.)
4468 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4469 high-level GC-NOTIFY-FOO functions send their output. (There's
4470 still very little control of where low-level verbose GC functions
4471 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4472 less than it used to -- the GC-NOTIFY-FOO functions are now under
4473 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4474 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4475 in only one place in the source code, and propagates it automatically
4476 everywhere that it's needed. Thus e.g. when I bump the version from
4477 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4478 * The C source files now include boilerplate legalese and documentation
4479 at the head of each file (just as the Lisp source files already did).
4480 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4481 to his page will be replaced with a link to his new CLiki service.
4483 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4485 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4486 * The build process now looks for GNU make under the default name "gmake",
4487 instead of "make" as it used to. If GNU make is not available as "gmake"
4488 on your system, you can change this default behavior by setting the
4489 GNUMAKE environment variable.
4490 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4491 can build without error under CMU CL.
4493 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4495 * There is now a partial SBCL user manual (with some new text and some
4496 text cribbed from the CMU CL manual).
4497 * The beginnings of a profiler have been added (starting with the
4498 CMU CL profiler and simplifying and cleaning up). Eventually the
4499 main interface should be through the TRACE macro, but for now,
4500 it's still accessed through vaguely CMU-CL-style functions and macros
4501 exported from the package SB-PROFILE.
4502 * Some problems left over from porting CMU CL to the new
4503 cross-compilation bootstrap process have been cleaned up:
4504 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4505 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4506 part when getting rid of the compiler *BACKEND* stuff.)
4507 ** The value of *NULL-TYPE* was screwed up, because it was
4508 being initialized before the type system knew the final
4509 definition of the 'NULL type. This screwed up several key
4510 optimizations in the compiler, causing inefficiency in all sorts
4511 of places. (I found it because I wanted to understand why
4512 GET-INTERNAL-RUN-TIME was consing.)
4513 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4514 PROCLAIM FTYPE information. Unfortunately this broke binary
4515 compatibility again, since now the forms output by DEFGENERIC
4516 to refer to functions which didn't exist in 0.6.3.
4517 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4518 can use the new (as of 0.6.3) transform for SOME into MAP into
4520 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4521 values to (INTEGER 0 1000000), so that the time code will no longer
4522 occasionally get blown up by Linux returning 1000000 microseconds
4523 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4524 its output conform to the ANSI spec. (Alas, this makes its output
4525 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4526 * A full call to MAP NIL with a single sequence argument no longer conses.
4527 * fixes to problems pointed out by Martin Atzmueller:
4528 * The manual page no longer talks about multiprocessing as though
4529 it were currently supported.
4530 * The ILISP support patches have been removed from the distribution,
4531 because as of version 5.10.1, ILISP now supports SBCL without us
4532 having to maintain patches.
4533 * added a modified version of Raymond Toy's recent CMU CL patch for
4534 EQUALP comparison of HASH-TABLE
4536 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4538 * The system still can't cross-compile itself with
4539 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4540 entails), but at least it can compile more of itself that way
4541 than it used to be able to, and various buglets which were uncovered
4542 by trying to cross-compile itself that way have now been fixed.
4543 * This release breaks binary compatibility again. This time
4544 at least I've incremented the FASL file format version to 2, so that the
4545 problem can be detected reliably instead of just causing weird errors.
4546 * various new style warnings:
4547 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4548 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4549 ** using the lexical binding of a variable named in the *FOO* style
4550 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4551 as specified by ANSI.
4552 * *RANDOM-STATE* is no longer automatically initialized from
4553 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4554 default behavior of the system is to repeat its behavior every time
4555 it's run. If you'd like to change this behavior, you can always
4556 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4557 old convention there was no comparably easy way to get the system to
4558 repeat its behavior every time it was run.
4559 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4560 FTYPE declarations has been removed, in favor of their ANSI
4561 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4562 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4563 the types of their sequence arguments can be determined at compile time.
4564 This is done through a new open code expansion for MAP which eliminates
4565 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4566 argument types can be determined at compile time.
4567 * The optimizer now transforms COERCE into an identity operation when it
4568 can prove that the coerced object is already of the correct type. (This
4569 can be a win for machine generated code, including the output of other
4570 optimization transforms, such as the MAP transform above.)
4571 * Credit information has been moved from source file headers into CREDITS.
4572 * Source file headers have been made more standard.
4573 * The CASE macro now compiles without complaining even when it has
4576 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4578 * (Note that the way that the PCL macroexpansions were rewritten
4579 to accommodate the change in DEFGENERIC below breaks binary
4580 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4581 not run under sbcl-0.6.2. Once we get out of alpha releases,
4582 i.e. hit release 1.0.0, we'll probably try to maintain binary
4583 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4584 and sbcl-1.4.4. Until then, however, it might be fairly common
4585 for maintenance releases to break binary compatibility.)
4586 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4588 * The reporting of unhandled signals has been changed to print some
4589 explanatory text as well as the report form. (Previously only
4590 the report form was printed.)
4591 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4592 it defines, so that the compiler no longer issues undefined function
4593 warnings for compiled-but-not-yet-loaded generic functions.
4594 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4595 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4596 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4597 section of the ANSI spec are supported.
4598 * The "" nickname for the "KEYWORD" package has been removed.
4599 The reader still handles symbol tokens which begin with a package marker
4600 as keywords, but it doesn't expose its mechanism for doing so in the
4601 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4602 * The system now issues STYLE-WARNINGs for contradictory TYPE
4603 proclamations. (Warnings for contradictory FTYPE proclamations would
4604 be nice too, but those can't be done usefully unless the type system
4605 is made smarter about FUNCTION types.)
4606 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4607 systematized, so that "*target-*.lisp is supposed to exist only on the
4608 target and imply that there's a related file which exists on the
4609 host, and *host-*.lisp is supposed to exist only on the host and imply
4610 that there's a related file which exists on the target. This involves a
4611 lot of renaming. Hopefully the acute confusion caused by the renaming
4612 will be justified by the reduction in chronic confusion..
4613 ** runtime-type.lisp -> early-target-type.lisp
4614 ** target-type.lisp -> late-target-type.lisp
4615 ** early-host-format.lisp -> early-format.lisp
4616 ** late-host-format.lisp -> late-format.lisp
4617 ** host-error.lisp -> misc-error.lisp
4618 ** early-error.lisp -> early-target-error.lisp
4619 ** late-error.lisp -> late-target-error.lisp
4620 ** host-defboot.lisp -> early-defboot.lisp
4621 ** code/misc.lisp -> code/target-misc.lisp
4622 ** code/host-misc.lisp -> code/misc.lisp
4623 ** code/numbers.lisp -> code/target-numbers.lisp
4624 ** code/early-numbers.lisp -> numbers.lisp
4625 ** early-host-type.lisp -> early-type.lisp
4626 ** late-host-type.lisp -> late-type.lisp
4627 ** host-typep.lisp -> typep.lisp
4628 ** load.lisp -> target-load.lisp
4629 ** host-load.lisp -> load.lisp
4630 ** host-disassem.lisp -> disassem.lisp
4631 ** host-insts.lisp -> insts.lisp
4632 ** byte-comp.lisp -> target-byte-comp.lisp
4633 ** host-byte-comp.lisp -> byte-comp.lisp
4634 ** host-signal.lisp -> signal.lisp
4635 ** host-defstruct.lisp -> defstruct.lisp
4636 ** late-target-type.lisp -> deftypes-for-target.lisp
4637 Furthermore, several other previously target-only files foo.lisp (e.g.
4638 hash-table.lisp and random.lisp) have been split into a target-and-host
4639 foo.lisp file and a target-only target-foo.lisp file, with their key type
4640 definitions in the target-and-host part, so that the cross-compiler will
4641 know more about target types.
4642 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4643 gone away. In their place are various *BACKEND-FOO* variables
4644 corresponding to the slots of the old structure.
4645 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4646 into the target SBCL has been fixed.
4647 * The chill.lisp system for loading cold code into a running SBCL
4649 * Support for the CMU CL "scavenger hook" extension has been removed.
4650 (It was undocumented and unused in the CMU CL sources that SBCL was
4651 derived from, and stale in sbcl-0.6.1.)
4652 * Various errors in the cross-compiler type system were detected
4653 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4654 (enabling various consistency checks). Many of them were fixed,
4655 but some hard problems remain, so the compiler is back to
4656 running without *TYPE-SYSTEM-INITIALIZED* for now.
4657 * As part of the cross-compiler type system cleanup, I implemented
4658 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4659 * I have started adding UNCROSS calls throughout the type system
4660 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4661 on cross-compiler input files will be able to go away, and various
4663 * CONSTANTP now returns true for quoted forms (as explicitly required
4666 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4668 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4669 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4670 (MAKE-LIST -1). In the longer term, it ought to become true
4671 that declarations are assertions even at SAFETY 1. For now, it's not
4672 quite true even at SAFETY 3, but it's at least more nearly true..
4673 (Note that this change seems to increases the size of the system by
4674 O(5%) and to decrease the speed of the compiler by 20% or more.)
4675 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4676 for the problem of printing dozens of lines of distracting information
4677 about low-level system machinery as part of the top stack frame
4678 on entry to the debugger when an undefined function was called.
4679 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4680 is not reset to COMMON-LISP-USER.
4681 * Compilation of stuff related to dyncount.lisp has been made conditional
4682 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4683 smaller. The various dyncount-related symbols have been moved into
4684 a new "SB-DYNCOUNT" package.
4685 * tty-inspect.lisp has been renamed to inspect.lisp.
4686 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4687 feature has gone away. (When we eventually port to other flavors of
4688 libc and/or Unix, we'll try to make the differences between flavors
4689 invisible at the user level.)
4690 * Various other *FEATURES* tags, and/or their associated conditionals,
4691 have been removed if obsolescent, or given better documentation, or
4692 sometimes given more-mnemonic names.
4694 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4696 * tidied up "make.sh" script
4697 * tidied up system directory structure
4698 * better "clean.sh" behavior
4699 * added doc/FOR-CMUCL-DEVELOPERS
4700 * many many small tweaks to output format, e.g. removing possibly-confusing
4701 trailing #\. character in DESCRIBE-INSTANCE
4702 * (EQUALP #\A 'A) no longer signals an error.
4703 * new hashing code, including EQUALP hashing
4704 * tidied up Lisp initialization and toplevel
4705 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4706 * command line argument processing
4707 * added POSIX-GETENV function to deal with Unix-ish environment variables
4708 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4709 terminating SBCL on EOF
4710 * non-verbose GC by default
4711 * There is no more "sbcl" shell script; the sbcl file is now the C
4712 runtime executable (just like CMU CL).
4713 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4715 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4716 symbol and package information as Lisp native symbol and package objects
4717 instead of strings naming symbols and strings naming packages. This way,
4718 whenever packages are renamed (as in warm init), debug information is
4719 transformed along with everything else.
4720 * tweaked the optimization policy declarations which control the building
4721 of SBCL itself. Now, among other things, the system no longer saves
4722 source location debugging information. (This helps two problems at once
4723 by reducing SBCL size and by keeping SBCL from trying to look for its
4724 sources -- which may not exist -- when reporting errors.)
4725 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4726 debugging and testing purposes
4727 * cleaned up printing, making the printer call PRINT-OBJECT for
4728 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4729 methods, giving nearly-ANSI behavior
4730 * converted almost all special variables to use *FOO* naming convention
4731 * deleted PARSE-TIME functionality, since it can be done portably
4732 * moved some files out of cold init into warm init
4733 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4735 * regularized formatting of source files
4736 * added an install.sh script
4737 * fixed ridiculous memory usage of cross-compiler by making
4738 compiler/alloc.lisp not try to do pooling unless it can hook
4739 itself into the GC of the cross-compilation host. Now the system
4740 builds nicely on my old laptop.
4741 * added :SB-ALLOC in target-features.lisp-expr
4742 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4743 was not implemented)
4744 * re-did condition handling and note reporting in the compiler. Notes
4745 are no longer handled by signalling conditions. Style warnings
4746 and warnings are handled more correctly and reported in such a way
4747 that it's easy to find one or the other in your output (so that you
4748 can e.g. figure out which of many problems caused COMPILE-FILE to
4750 * changed the severity of several compiler warnings from full WARNING
4751 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4752 compiler note reporting so that it doesn't use the condition system
4753 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4754 COMPILE-FILE command)
4755 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4756 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4757 no longer supported, since their implementation was deeply intertwingled
4758 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4759 * removed bogus "support" for compiler macros named (SETF FOO), and
4760 removed the compiler macro for SETF INFO (but only after making a fool
4761 of myself on the cmucl-imp mailing list by posting a bogus patch for
4762 DEFINE-COMPILER-MACRO..)
4763 * Compiled files containing forms which have side effects on the Lisp
4764 reader (such as DEFPACKAGE forms) are now handled more correctly.
4765 (Compiler queuing of top level lambdas has been suppressed by setting
4766 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4767 * deleted various currently-unused source files, e.g. gengc.lisp. They
4768 may be added back at some point e.g. when porting to other architectures,
4769 but until they are it's distracting to distribute them and to try to
4771 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4772 there were so many of them they're just distractions, and UNCROSS is
4773 known to be able to handle the current sources
4774 * moved PROFILE functionality into TRACE, so that it will be clear
4775 how the wrapping and unwrapping of functions when you profile them
4776 interacts with the wrapping and unwrapping of functions when you
4777 trace them. (Actually, the functionality isn't there yet, but at least
4778 the interface specification is there. Hopefully, the functionality will
4779 arrive with some future maintenance release.)
4780 * removed host-oops.lisp
4781 * changed signature of QUIT function to allow UNIX-CODE argument
4782 * fixed READ-SEQUENCE bug
4783 * tweaked verbose GC output so that it looks more like the progress
4784 output that ANSI specifies for functions like LOAD
4785 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4786 * added <http://sbcl.sourceforge.com> to the banner information printed by