1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.22 relative to 1.0.21:
3 * enhancement: inoccous calls to EVAL or generic functions dispatching
4 on subclasses of eg. STREAM no longer cause compiler notes to appear.
5 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
6 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
7 provided. (reported by Cedric St-Jean)
8 * bug fix: circularity handling in the reader did not treat raw
9 structure slots correctly. (reported by Cedric St-Jean)
10 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
11 file descriptors when there were none.
13 changes in sbcl-1.0.21 relative to 1.0.20:
14 * new feature: the compiler is able to track the effective type of a
15 generic function across method addition and removal.
16 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
17 appropriately typed structure slots without locking.
18 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
19 information like those gathered by TIME using a programming-friendly
21 * new feature: TIME reports time taken even if the form performs a
22 non-local transfer of control.
23 * enhancement: reduced conservativism on GENCGC platforms: on
24 average 45% less pages pinned (measured from SBCL self build).
25 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
26 mutate constant symbols or violate declaimed type of the symbol.
27 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
28 sleeping on the semaphore in heavily contested semaphores.
29 * bug fix: semaphores and condition variables were not interrupt
31 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
32 owned by other threads anymore.
33 * bug fix: FIND on lists called KEY outside the specified
34 subsequence. (reported by budden)
35 * bug fix: LOG doesn't use single-float intermediate results when
36 given mixed integer and double-float arguments, leading to better
37 precision. (reported by Bob Felts)
38 * bug fix: LOG with base zero returned values of inconsistent type.
40 changes in sbcl-1.0.20 relative to 1.0.19:
41 * minor incompatible change: OPTIMIZE qualities
42 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
43 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
44 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
45 * documentation: some slot access efficiency guidelines have been
46 added to the user manual.
47 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
48 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
50 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
51 to more efficient EQ-comparison versions more often.
52 * optimization: enhanced derivation of DOLIST iteration variable type
54 * optimization: constant folding of simple (LIST ...) forms as DOLIST
56 * optimization: runtime lookup of function definitions can be
57 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
58 * optimization: compiler is able to derive the return type of
59 (AREF (THE STRING X) Y) as being CHARACTER.
60 * optimization: CLRHASH on empty hash-tables no longer does pointless
61 work. (thanks to Alec Berryman)
62 * optimization: REPLACE deftransforms don't punt when :START1 or
64 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
65 no longer emits a code deletion note. (reported by Andrew
67 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
68 functions with an odd number of &OPTIONAL arguments, a &REST
69 argument, and one or more &KEY arguments at the call site.
70 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
71 arguments appeared at call sites as well.
72 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
73 READ-LINE &co can no longer read from them afterwards. (reported
75 * bug fix: fixed #427: unused local aliens no longer cause compiler
76 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
78 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
79 alien-stack leakage. (reported by Andy Hefner)
80 * bug fix: PROGV signals an error when an attempt to violate declared
81 type of a variable or bind a constant is made.
82 * bug fix: SET signals an error when an attempt to violate declared
83 type of a variable is made.
84 * bug fix: restart computation during the execution of a restart
85 test function no longer causes infinite recursion. (reported by
87 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
88 signals an error. (thanks to Pierre Mai)
89 * bug fix: EXPORT left symbol unexported in conflict situations.
90 (thanks to Michael Weber)
91 * bug fix: correctly handle name conflicts involving CL:NIL.
92 (thanks to Michael Weber)
93 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
94 conflicts arising from USEing package with conflicting symbols
95 correctly. (thanks to Michael Weber)
97 changes in sbcl-1.0.19 relative to 1.0.18:
98 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
99 warnings that go otherwise unhandled will be muffled if they are
100 of the type that's the value of this variable.
101 * optimization: stack allocation is slightly more efficient on x86
103 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
104 where a raw slot always is initialized using the initform whose
105 type is not know sufficiently well a compile-time are now compiled
106 correctly. (reported by John Morrison)
107 * bug fix: compiler no longer makes erronous assumptions in the
108 presense of non-foldable SATISFIES types.
109 * bug fix: stack analysis missed cleanups of dynamic-extent
110 arguments in non-let-converted calls to local functions.
111 * improvements to the Windows port:
112 ** adjusted address spaces for building on both Win32 and
113 Win64. (thanks for John Connors)
114 * fixed some bugs revealed by Paul Dietz' test suite:
115 ** interval arithmetic during type derivation used inexact integer
116 to single-float coercions.
117 ** arithmetic operations involving large integers and single
118 floats give the same results in compiled and interpreted code.
119 ** deriving the result type of COERCE no longer signals an error
120 if the derived type of the second argument is a MEMBER type
121 containing invalid type specifiers.
122 ** ADJOIN with constant NIL as second argument works correctly.
123 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
125 changes in sbcl-1.0.18 relative to 1.0.17:
126 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
127 profiles only the current thread.
128 * minor incompatible change: changes to SYMBOL-VALUE of constants
129 defined with DEFCONSTANT now signal an error.
130 * enhancement: SB-SPROF now has support for wallclock profiling,
131 and is also able to profile specific threads. REPORT output
132 has also additional sorting options.
133 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
135 * optimization: structure allocation has been improved
136 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
137 ** out of line constructors are ~10% faster.
138 ** inline constructors are ~15% faster.
139 ** inline constructors are capable of dynamic extent allocation
140 (generally on x86 and x86-64, in some cases on other platforms
142 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
144 * optimization: file compiler is now able to coalesce non-circular
145 lists, non-base strings, and bit-vectors. Additionally, constants
146 are never referenced through SYMBOL-VALUE at runtime.
147 * optimization: code defining methods on PRINT-OBJECT (and other
148 generic functions in the COMMON-LISP package) now loads faster.
149 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
151 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
152 fasl is now deleted. (reported by Attila Lendvai)
153 * bug fix: READ-LINE always returned NIL for the last line in files.
154 (reported by Yoshinori Tahara)
155 * bug fix: more accurate disassembly annotations of foreign function
156 calls. (thanks to Andy Hefner)
157 * bug fix: trimming non-simple strings and non-string string
158 designators when the there is nothing to trim works properly.
159 (thanks to James Knight)
160 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
161 (thanks to Travis Cross)
162 * fixed some bugs revealed by Paul Dietz' test suite:
163 ** NIL is a valid function name (regression at 1.0.13.38)
164 ** FILL on lists was missing its return value (regression at 1.0.12.27)
165 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
166 fill pointers properly (regression at 1.0.12.23)
167 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
168 displacement indices properly (regression at 1.0.12.23)
170 changes in sbcl-1.0.17 relative to 1.0.16:
171 * temporary regression: user code can no longer allocate closure
172 variable storage on stack, due to bug 419 without explicitly
173 requesting it. Please consult sbcl-devel for advice if you need to
174 use this feature in the meanwhile.
175 * new feature: runtime argument --control-stack-size can be used to
176 adjust thread default control stack size.
177 * enhancement: improved TIME output
178 ** all times are reported using the measured accuracy (milliseconds
179 for real and GC times, microseconds for everything else.)
180 ** processor cycle counts on x86 and x86-64.
181 ** interpreted forms are counted for both evaluator modes.
182 ** number of lambdas converted by the compiler is reported.
183 ** CPU percentage report (computed from real and total run time.)
184 ** more comprehensive run time reporting, using a condenced format
185 ** interperted form, lambda, and page fault counts are omitted
187 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
189 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
190 * optimization: two argument forms of LAST are upto ~10% faster
191 in normal SPEED policies.
192 * optimization: NCONC no longer needs to heap cons its &REST list
193 in normal SPEED policies.
194 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
196 * bug fix: UNION and NUNION work with :TEST-NOT once more,
197 regression since 1.0.9.1. (thanks to Eric Marsden)
198 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
199 since 1.0.15.36. (thanks to Paul Khuong)
200 * bug fix: LAST when always returned the whole list when given a bignum
201 as the second argument.
202 * bug fix: dynamic extent allocation of nested lists and vectors
203 could leak to otherwise accessible parts.
204 * bug fix: invalid optimization of heap-allocated alien variable
206 * bug fix: fasl header checking is less vulnerable to different
207 platform word lengths.
208 * bug fix: more correct assembler syntax for GNU binutils
209 2.18.50.0.4 support. (thanks to Marijn Schouten)
210 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
212 * bug fix: the fopcompiler can handle LOCALLY forms (with no
213 declarations) successfully. (reported by Attila Lendvai)
215 changes in sbcl-1.0.16 relative to 1.0.15:
216 * minor incompatible change: revert the changes to sb-posix's error
217 signaling added in 1.0.14.
218 * minor incompatible change: change PROBE-FILE back to returning
219 NIL whenever we can't get a truename, as was the case before 1.0.14.
220 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
221 inherits from ERROR instead of just CONDITION.
222 * new feature: SB-INTROSPECT can provide source locations for instances
223 as well. (thanks to Tobian Ritterweiler)
224 * optimization: binding special variables now generates smaller code
225 on threaded platforms.
226 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
227 and cases where no :TEST is given but the compiler can infer that
228 the element to search is of type (OR FIXNUM (NOT NUMBER)).
229 * optimization: better LOGNOT on fixnums.
230 * optimization: modular arithmetic for a particular requested width
231 is implemented using a tagged representation unless a better
232 representation is available.
233 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
234 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
235 for wellformedness and returns a specialized vector. (reported by
236 Francois-Rene Rideau)
237 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
238 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
239 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
240 unscheduling timer at the same time as another timer fires.
241 * bug fix: don't reschedule timers for dead threads.
242 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
243 * bug fix: copying output from RUN-PROGRAM to a stream signalled
244 bogus errors if select() was interrupted.
245 * enhancement: add support for fcntl's struct flock to SB-POSIX.
247 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
248 * enhancement: cleaner backtraces for interactive interrupts, as
249 well as other cases where the interesting frames used to be
250 obscured by interrupt handling frames.
251 * enhancement: untracing a whole package using (UNTRACE "FOO") is
252 now supported, and tracing a whole package using (TRACE "FOO") now
253 traces SETF-functions as well.
254 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
255 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
256 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
257 * unreadably printed representation of hash-tables now includes
259 * bug fix: partially fixed #188: type propagation from assignments
260 is now more efficient.
261 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
262 'DOUBLE-FLOAT) are not flushable.
263 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
264 full address of the object, and none of the tag bits.
265 * bug fix: readably printing hash-tables now respects other printer
266 control variables. (reported by Cedric St-Jean)
267 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
268 keyword with MAKE-HASH-TABLE.
269 * bug fix: export SB-POSIX:MKSTEMP.
270 * bug fix: SORT was not interrupt safe.
271 * bug fix: XREF accounts for the last node of each basic-block as
273 * bug fix: MAKE-INSTANCE optimizations interacted badly with
274 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
275 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
276 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
277 create obsolete instances in the case of redefinition or
278 obsoletion of a superclass. (thanks to Andy Hefner)
279 * bug fix: Support for the Alpha architecture has been revived; it had
280 suffered somewhat from lack of maintenance since sbcl-1.0.
281 * improvements to the Windows port:
282 ** The system detects the codepage to use at startup, instead of
283 using the value from the saved core. (thanks to Kei Suzuki)
285 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
286 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
287 (see documentation for details.)
288 * revived support for OpenBSD (contributed by Josh Elsasser)
289 * partially fixed bug #108: ROOM no longer suffers from occasional
290 (AVER (SAP= CURRENT END)) failures.
291 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
292 bodies no longer cause a WARNING to be signalled. (reported by
294 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
295 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
296 the bug in TRUNCATE.) (reported by Michael Weber)
297 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
298 no samples. (reported by Andy Hefner)
299 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
300 no longer appear as (NIL ...) frames in backtraces.
301 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
302 is now actually exported from SB-EXT as documented. (reported by
304 * bug fix: sb-aclrepl now correctly understands how to inspect
305 single-floats on 64-bit platforms where single-floats are not boxed.
306 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
307 yet been finalized. (reported by Levente Meszaros)
308 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
309 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
310 SBCL-specific optimize qualities.
312 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
313 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
314 an executable in the search path, and does so in the child
315 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
316 been removed; it can be found in the CVS history, for somebody who
317 needs that search behavior (see the manual).
318 * minor incompatible change: compiler policy re. weakening type
319 checks has changed: now type checks are weakened only if SAFETY < 2
321 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
322 unparsing of directory pathnames as files. Analogously,
323 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
324 filename to parse into a directory pathname.
325 * enhancement: implicit generic function creation now signals a
326 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
327 users can bind handlers for and muffle around calls to LOAD.
328 * enhancement: RUN-PROGRAM allows unicode arguments and environments
329 to be used (using the default stream external format), and allows
330 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
331 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
332 strings and vectors whose element-type or simplicity is not fully
333 known at compile-time.
334 * optimization: STRING-TRIM and related functions no longer allocate
335 a new string when no trimming needs to be performed. These functions
336 are also faster than before when the input string has been declared
338 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
339 * optimization: READ-LINE is significantly faster for files containing
341 * optimization: non-open coded uses of character comparison operators
342 (e.g. char=) no longer cons when called with more than one parameter
343 on platforms supporting dynamic-extent allocation.
344 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
345 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
346 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
348 * bug fix: some sequence functions elided bounds checking when
350 * bug fix: too liberal weakening of union-type checks when SPEED >
352 * bug fix: more bogus fixnum declarations in ROOM implementation
355 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
356 * new feature: MAKE-HASH-TABLE now experimentally accepts a
357 :SYNCHRONIZED argument, which makes the hash-table safe for
358 concurrent accesses (but not iteration.) See also:
359 SB-EXT:WITH-LOCKED-HASH-TABLE, and
360 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
361 * optimization: CONCATENATE on strings is an order of magnitude faster
362 in code compiled with (> SPEED SPACE).
363 * optimization: SUBSEQ is ~50% faster on lists.
364 * bug fix: bug 417 fixed -- source location reporting is now more robust.
365 * bug fix: SUBSEQ on a list will now correctly signal an error if if
366 END is smaller then START.
367 * bug fix: SB-PROFILE will no longer report extra consing for nested
368 calls to profiled functions.
369 * bug fix: ROOM implementation had bogus fixnum declarations which
370 could cause type-errors when calling ROOM on large images.
371 * bug fix: if file compilation is aborted, the partial fasl is now
372 deleted, and COMPILE-FILE returns NIL as the primary value.
373 * bug fix: number of thread safety issues relating to SBCL's internal
374 hash-table usage have been fixed.
375 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
376 be returned from its body when the values were being returned
377 using unknown-values return convection and the W-P-O was wrapped
378 inside an UNWIND-PROTECT.
379 * bug fix: sb-posix should now compile again under Windows, enabling
382 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
383 * incompatible change: hash-table accessor functions are no longer
384 automatically protected by locks. Concurrent accesses on the same hash-table
385 from multiple threads can give inconsistent results or even corrupt the
386 hash-table completely. Multi-threaded applications should do their own
387 locking at the correct granularity. In the current implementation it is
388 still safe to have multiple readers access the same table, but it's not
389 guaranteed that this property will be maintained in future releases.
390 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
391 and will signal an error at runtime.
392 * enhancement: SB-THREAD package now exports a semaphore interface.
393 * enhancement: CONS can now stack-allocate on x86 and
394 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
396 * enhancement: nested lists can now be stack allocated on
397 platforms providing stack allocation support.
398 * enhancement: dynamic-extent support has been extended to support
399 cases where there are multiple possible sources for the stack
401 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
402 if the mutex is uncontested on Linux.
403 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
404 as specified by the CLHS. (thanks to Tobias Rittweiler)
405 * bug fix: NaN comparison now works on x86-64.
406 * bug fix: CLOSE :ABORT T on a stream with pending output now
408 * bug fix: instances of non-standard metaclasses using standard
409 instance structure protocol sometimes missed the slot type checks
411 * bug fix: known functions can be inlined in high-debug code.
413 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
414 * minor incompatible change: the MSI installer on Windows no longer
415 associates .lisp and .fasl files with the installed SBCL.
416 * minor incompatible change: :UNIX is no longer present in *FEATURES*
417 on Windows. (thanks to Luis Oliviera)
418 * new feature: SB-CLTL2 contrib module now implements
419 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
420 * optimization: scavenging weak pointers is now more efficient,
421 requiring O(1) instead of O(N) per weak pointer to identify
422 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
423 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
424 method bodies using constant slot-names when the first argument is
425 a specializer parameter for the method.
426 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
427 names, when the class of the instance is a direct instance
428 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
429 as slow as the constant slot-name case.)
430 * optimization: member type construction is now O(N) instead
432 * optimization: UNION and NUNION are now O(N+M) for large
433 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
435 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
436 now more readable in environments like Slime which display it.
437 (thanks to Tobias C. Rittweiler)
438 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
439 was able to infer the correct argument type for the object on which
440 the CAS operation was being performed.
441 * bug fix: copy propagation interfered with parallel assignment
442 semantics in local calls. (reported by Paul Khuong)
443 * bug fix: the signed modular fixnum shift compiled to wrong code on
444 x86 and x86-64. (spotted by a slight modification to some of
447 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
448 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
449 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
450 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
452 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
453 and generic function now print the object identity for anonymous
454 instances. (thanks to Attila Lendvai)
455 * enhancement: as an extension to MOP, SBCL now supports SETF of
456 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
457 (thanks to Attila Lendvai)
458 * enhancement: sb-cover has an alternate annotation mode which puts
459 the form annotations at the car of the form, instead of annotating
460 it completely. The benefit of this mode is that it shows better
461 which forms were instrumented by the compiler.
462 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
463 by printer control variables. (reported by Dan Corkill)
464 * bug fix: system leaked memory when delayed output was performed by
465 the OS in smaller chunks then expected. (thanks to David Smith)
466 * bug fix: system leaked memory when file streams were not closed
468 * bug fix: large objects written to slow streams that were modified
469 after the write could end up with the modified state written to
470 the underlying file descriptor.
471 * bug fix: multiple threads operating in parallel on the same stream
472 could cause buffer-overflows.
473 * bug fix: source location information is stored correctly
474 (broken since 1.0.6). This bug would generally show up as the
475 Slime debugger highlighting the wrong form when the sldb-show-source
477 * bug fix: Branch forms are again annotated as branches in the sb-cover
479 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
480 on x86-64. (reported by Gregory Vanuxem)
481 * bug fix: EQUALP could return wrong results for structures with raw
482 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
483 word). (reported by Vjacheslav Fyodorov)
484 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
487 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
488 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
489 atomic compare-and-swap operations on threaded platforms.
490 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
491 allows assigning a global minimum value to optimization qualities
492 (overriding proclamations and declarations).
493 * enhancement: closed over variables can be stack-allocated on x86
495 * performance bug fix: GETHASH and (SETF GETHASH) are once again
497 * optimization: slot definition lookup is now O(1). This speeds up
498 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
499 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
500 * optimization: ASSOC and MEMBER can now be open-coded for all
501 combinations of keyword arguments when second argument is constant
502 and SPEED >= SPACE. In other cases a specialized version is
504 * bug fix: using obsoleted structure instances with TYPEP and
505 generic functions now signals a sensible error.
506 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
507 (reported by Kristoffer Kvello)
508 * bug fix: backtrace construction is now more careful when making
509 lisp-objects from pointers on the stack, to avoid creating bogus
510 objects that can be seen by the GC.
511 * bug fix: defaulting of values in contexts expecting more than 7
512 variables now works on x86-64. (reported by Christopher Laux)
513 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
515 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
516 as the property-list of a symbol.
517 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
518 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
519 situations requested, are once again file-compileable. (reported
522 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
523 * MOP improvement: support for user-defined subclasses of
524 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
525 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
526 the expansion of DEFMETHOD, is responsible for generating a form
527 which creates a list of specializers when evaluated. Additional
528 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
529 debugging and introspective support.
530 * minor incompatible change: the (unsupported) spinlock interface
531 has changed: free spinlock now has the value NIL, and a held spinlock
532 has the owning thread as its value.
533 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
534 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
536 * enhancement: name of a socket-stream is now "a socket" instead of
538 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
539 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
540 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
541 for generic functions. (Thanks to Tobias C. Rittweiler)
542 * optimization: bignum printing speed has been improved by 20-40%
543 (depending on the bignum size.)
544 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
546 * bug fix: the cache used by the CLOS to store precomputed effective
547 methods, slot offsets, and constant return values is now thread and
549 * bug fix: generic function dispatch function updating is now thread
550 and interrupt safe (in the sense that the known issues have been
552 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
553 * bug fix: interning EQL-specializers is now thread and interrupt safe.
554 * bug fix: asdf systems with dependencies to the SB-POSIX or
555 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
556 * bug fix: interrupt safety of applicable method computation has been
559 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
560 * new contrib: sb-cover, an experimental code coverage tool, is included
562 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
563 significantly faster.
564 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
565 produce more efficient code on x86-64 if the compiler can determine
566 that (+ INDEX <constant>) does not require a bounds check and FOO
567 has an element type at least 8 bits wide.
568 * enhancement: a new, experimental synchronous timeout facility is
569 provided. Refer to SB-SYS:WITH-DEADLINE for details.
570 * enhancement: when a symbol name conflict error arises, the
571 conflicting symbols are always printed with a package prefix.
572 (thanks to Kevin Reid)
573 * enhancement: stepping is now once again supported on the SPARC and
574 MIPS platforms. (It is also now more likely to work on CheneyGC
576 * enhancement: sb-sprof can now also track and report accurate call
578 * bug fixes: the treatment of non-standard subclasses of
579 SB-MOP:SPECIALIZER is more correct.
580 * incompatible change: PURIFY no longer copies the data from the
581 dynamic space into the static and read-only spaces on platforms
582 that use the generational garbage collector
583 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
585 * bug fix: GC race condition occasionally resulting in crashes with
586 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
588 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
590 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
591 a STYLE-WARNING to be signalled (regression from 1.0.4.)
592 * bug fix: an asynchronous interrupt could previously leave the
593 system running with GC inhibited.
594 * bug fix: a DECLARE form evaluated at top-level now causes an error
595 rather than silently (or verbosely) returning NIL.
596 * bug fix: trying to dissassemble functions compiled at high DEBUG could
597 result in a error being signalled due to source form lookup errors.
598 (reported by Peter Graves)
600 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
601 * incompatible change: removed writer methods for host-ent-name,
602 host-ent-addresses -- changing the values did not update the DNS
604 * minor incompatible change: changed experimental JOIN-THREAD interface
605 * documentation: the manual now lists reader and writer methods
606 in class slot documentation sections. (thanks to Richard M Kreuter)
607 * documentation: unwinding from asyncronous events has been
608 documented as unsafe.
609 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
610 in multithreaded application code.
611 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
612 platforms. (thanks to James Anderson for the optimization hint)
613 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
615 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
616 variants no longer cons.
617 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
618 their NOT- variants no longer cons.
619 * optimization: EQUAL hash tables no longer use SXHASH for objects
620 of all data types, but instead use an EQL hash for types for which
621 EQUAL is the same as EQL.
622 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
623 are significantly faster.
624 * optimization: new STRING-OUTPUT-STREAM implementation is much
625 faster for large outputs and conses 30% less on average.
626 * enhancement: XREF information is now collected for references made
627 to global variables using SYMBOL-VALUE and a constant argument.
628 * enhancement: SIGINT now causes a specific condition
629 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
630 * bug fix: FILE-POSITION used to signal an error in some cases where
631 ANSI requires it to return NIL.
632 * bug fix: ADJUST-ARRAY is now interrupt-safe.
633 * bug fix: adding and removing fd-handlers is now interrupt-safe.
634 * bug fix: inlined calls to C now ensure 16byte stack alignment on
636 * bug fix: bad type declaration in the CLOS implementation has
637 been fixed. (reported by James Anderson)
638 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
639 (thanks to Lutz Euler)
640 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
641 required by the CLHS. (reported by Samium Gromoff)
642 * bug fix: dead unbound variable references now signal an error.
643 (reported by Marco Monteiro)
644 * bug fix: / with an unused value was being deleted in safe code.
645 (thanks to Marco Monteiro and Kevin Reid)
646 * bug fix: number of characters that can be written onto a single
647 line in a file is unlimited.
648 * bug fix: some GC deadlocks caused by asynchronous interrupts have
649 been fixed by inhibiting interrupts for when GC is disabled.
650 * bug fix: some interrupt safety issues with GETHASH, (SETF
651 GETHASH), CLRHASH and REMHASH have been fixed.
652 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
653 specifier no longer causes infinite recursion.
654 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
655 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
656 * bug fix: an &environment argument with the correct variable information
657 is passed to macros that are expanded during byte compilation.
658 (reported by Samium Gromoff)
659 * bug fix: a logic error was causing the hash function for bignums to
660 have excessive amounts of collisions. (reported by Faré Rideau)
661 * bug fix: modifying the contents of an array could change the return
662 value of SXHASH on that array, which is only allowed for strings
663 and bit vectors. (bug introduced in 0.9.16)
664 * improvement: the x86-64/darwin port now passes all tests (except
665 for the debugger tests) but should still be considered
666 experimental until this is fixed.
667 * improvement: a style-warning is signaled for CASE (etc) clauses with
668 duplicate keys. (thanks to Kevin Reid)
669 * improvement: macos/x86 and macos/x86-64 now use mach exception
670 handlers for dealing with illegal instructions (for trapping and
671 error handling) and memory protection violations (for GC).
673 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
674 * new platform: experimental support for x86-64/darwin (MacOS).
675 * incompatible change: the thread-safe (on most platforms) getaddrinfo
676 and getnameinfo sockets functions are used instead of gethostbyaddr
677 and gethostbyname, on platforms where the newer functions are available.
678 As a result, the ALIASES field of HOST-ENT will always be NIL on these
680 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
681 don't print the startup banner, but behave as if --noinform was passed
682 as a command line argument. (thanks to Kevin Reid)
683 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
684 * optimization: code using alien values with undeclared types is much faster.
685 * optimization: the compiler is now able to open code SEARCH in more cases.
686 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
687 * bug fix: using standardized COMMON-LISP special variables as loop
688 variables no longer signals bogus package lock violations. (reported
690 * bug fix: declaring local loop variables to be of a range-limited type
691 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
692 (reported by Andras Simon)
693 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
694 bugs remain on x86-64.)
695 * bug fix: the #= and ## reader macros now interact reasonably with
696 funcallable instances.
697 * bug fix: type-checks for function arguments were compiled using the
698 compiler policy settings of the wrong lexical environment (in 1.0.2
700 * bug fix: SHADOW accepts characters as string designators, as required
701 by the spec (thanks to Eric Marsden)
702 * bug fix: fixed GC safety issues when foreign functions are called with
703 non-base strings as arguments
704 * bug fix: more consistent error output for fatal-compiler-warnings, like
706 * bug fix: functions with proper names were showing up as NIL in
707 backtraces (reported by Edi Weitz, regression in 1.0.2)
709 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
710 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
711 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
712 produce more efficient code on the x86 if the compiler can determine
713 that (+ INDEX <constant>) does not require a bounds check and FOO
714 has an element type at least 8 bits wide.
715 * bug fix: references to undefined variables are handled the same way
716 in toplevel forms as in the normal compiler
717 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
719 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
720 (thanks to Jon Buller)
721 * improvement: faster compilation times for complex functions
722 * improvement: added readlink support to SB-POSIX (thanks to Richard
725 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
726 * improvement: experimental support for mach exception handling on
727 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
729 * improvement: support for GBK external format.
730 (thanks to Chun Tian (binghe))
731 * improvement: the debugger now displays variables that have been closed
732 over, in code compiled with (DEBUG 2) or higher
733 * improvement: support for executable cores on NetBSD (thanks to
735 * new feature: added a RESTART-FRAME debugger command
736 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
737 be used to provide an implementation for FILE-POSITION on Gray streams
738 (thanks to Eric Marsden)
739 * new feature: add syslog, openlog and closelog support to SB-POSIX
740 (thanks to Richard Kreuter)
741 * optimization: the function call overhead in code compiled with
742 a high DEBUG optimization setting is significantly
743 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
744 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
745 (thanks to Tony Martinez)
746 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
747 evaluated once (reported by Kevin Reid)
748 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
749 works again (thanks to Kilian Sprotte)
750 * bug fix: an error is signaled for tagbodies with duplicate tags
751 (thanks to Stephen Wilson)
752 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
753 * bug fix: Win32 port can now handle foreign code unwinding Lisp
754 stack frames from alien callbacks.
755 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
756 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
757 65k pages (thanks to David Woodhouse)
758 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
760 changes in sbcl-1.0.1 relative to sbcl-1.0:
761 * new platform: FreeBSD/x86-64, including support for threading.
762 * new feature: the compiler stores cross-referencing information
763 about function calls (who-calls), macroexpansion (who-macroexpands)
764 and special variables (who-binds, who-sets, who-references) for code
765 compiled with (< SPACE 3). This information is available through the
766 sb-introspect contrib.
767 * new feature: users may subclass SEQUENCE, and have instances of
768 these classes interoperate with standard Common Lisp functions if
769 a number of methods are defined. (This feature is experimental
770 and the interface subject to change based on feedback from SBCL
771 users and the general community)
772 * improvement: sb-sprof traces call stacks to an arbitrary depth on
773 x86/x86-64, rather than the previous fixed depth of 8
774 * improvement: another pthread back-end of mutex "pthread-futex".
775 * bug fix: non-ascii command-line arguments are processed correctly
776 (thanks to Yaroslav Kavenchuk)
777 * bug fix: non-required arguments were not passed correctly when a method
778 defined using DEFMETHOD was called from a mop-generated method using
779 CALL-NEXT-METHOD (reported by Pascal Costanza)
780 * bug fix: recursion is now permitted in accessors through
781 SLOT-UNBOUND. (reported by Pascal Costanza)
782 * bug fix: an error was signaled at startup if the HOME environment
783 variable was defined, but had an empty value (reported by Peter Van Eynde)
784 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
785 signal a PROGRAM-ERROR, not a BUG.
786 * bug fix: SB-POSIX:READDIR works when built with large file support.
787 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
789 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
790 for better automatic indentation support. (thanks to Matt Pillsbury)
791 * optimization: loading generic functions no longer takes O(n^2) time,
792 proportional to the amount of methods in the generic function
793 (reported by Todd Sabin and Jeremy Brown)
794 * optimization: the FIND and POSITION family of sequence functions
795 are significantly faster on arrays whose element types have been
797 * improvements to the Windows port:
798 ** Intermittent heap corruption problems have been fixed. (thanks
799 to Alastair Bridgewater)
800 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
802 ** Lisp is able to unwind foreign exception frames from alien
803 callbacks. (thanks to Alastair Bridgewater)
805 changes in sbcl-1.0 relative to sbcl-0.9.18:
806 * improvement: experimental support for threading on FreeBSD/x86.
807 (thanks to NIIMI Satoshi)
808 * improvement: runtime option --dynamic-space-size can be used
809 to set the size of the dynamic space reserved on startup.
810 * improvement: floating point modes in effect are now saved in
811 core, and restored on startup.
812 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
813 startup, not time since first call to GET-INTERNAL-REAL-TIME.
814 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
815 threads are not running after *SAVE-HOOKS* have run.
816 * improvement: writes to CLOS instance slots are type-checked in code
817 compiled with (SAFETY 3)
818 * improvement: floating-point exception handling on FreeBSD (thanks to
820 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
821 (thanks to Zach Beane)
822 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
824 * improvement: added support for the Shift-JIS external format.
825 (contributed by NIIMI Satoshi)
826 * improvement: callbacks are supported on Linux/PPC. (thanks to
828 * bug fix: compiler bug triggered by a (non-standard) VALUES
829 declaration in a LET* was fixed. (reported by Kaersten Poeck)
830 * bug fix: file compiler no longer confuses validated and already
831 dumped structurres. (reported by Kaersten Poeck)
832 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
833 fill-pointer signals a type-error as required. (thanks to
835 * bug fix: disassemly of funcallable instances works.
836 * bug fix: single stepping on PPC.
837 * bug fix: fix thread-safety problems in the type system (generally
838 manifesting as nonsensical errors like "STRING is a bad type specifier
839 for sequences" or "The value 1 is not of type FIXNUM")
840 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
841 (reported by Josip Gracin)
842 * bug fix: an error is signaled for attempts to displace arrays with
843 incompatible element types (thanks to Mario Mommer)
844 * bug fix: more correct handling of wide characters in debug info
845 (bug reported by Attila Lendvai and fixed by Juho Snellman)
846 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
847 and don't cause extra consing
848 * optimization: MAP and MAP-INTO are significantly faster on vectors
849 whose elements types have been declared.
850 * Improvements to SB-SPROF:
851 ** Support for allocation profiling
852 ** Reduced profiling overhead, especially for long profiling runs
853 * Improvements to the Windows port:
854 ** floating point exceptions are now reported correctly.
855 ** stack exhaustion detection works partially.
856 ** more accurate GET-INTERNAL-REAL-TIME.
857 ** better RUN-PROGRAM behaviour in case of errors in the child
859 ** PROBE-FILE now simplifies pathnames correctly.
860 ** DIRECTORY now works correctly with :WILD-INFERIORS.
861 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
862 1st 1970 (although time timezone data will be unreliable).
863 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
865 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
866 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
867 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
869 * bug fix: two potential GC deadlocks affecting threaded builds.
870 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
871 returns T (reported by Anton Kazennikov)
872 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
873 with non-variable places
874 * bug fix: remove a race condition in the setting of
875 funcallable-instance functions, this should make threaded CLOS
876 code more stable against memory faults.
877 * bug fix: corruption of specials when unbinding is interrupted by an
878 asynchronous unwind (reported by Hannu Koivisto)
879 * improvement: the debugger will now also display local variables that
880 are only used once, for code compiled with a DEBUG optimization quality
883 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
884 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
885 * incompatible change: External-format support for FFI calls. The
886 SB-ALIEN:C-STRING no longer implies an ASCII
887 external-format. Instead the string is subject to external-format
888 conversion. Additionally return values of type SB-ALIEN:C-STRING
889 are by default of type (SIMPLE-ARRAY CHARACTER), not
890 SIMPLE-BASE-STRING. For an alien type that behaves like the old
891 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
892 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
893 * incompatible change: SB-EXT package no longer contains the
894 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
895 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
897 * incompatible change: the single-stepper is no longer available
898 on Alpha, Mips and Sparc platforms.
899 * minor incompatible change: the direct superclasses of
900 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
901 not (STANDARD-OBJECT FUNCTION). This makes the
902 class-precedence-lists of GENERIC-FUNCTION and
903 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
905 * minor incompatible change: the default stream external format on
906 non-unicode SBCL is no longer detected from the locale, but is always
908 * new feature: Add a version of evaluator that uses an interpreter instead
909 of the compiler. EVAL still uses the compiler by default, to switch it
910 to use the interpreter, set the value of the variable
911 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
912 * minor incompatible change: the single-stepper REPL has been merged
913 with the normal debugger (see the "Stepping" heading of the debugger help
914 for more details). The debugger command STEP will no longer switch
915 to the single-stepper REPL.
916 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
917 for a type now works.
918 * bug fix: loading of default sysinit file works. (thanks to Leonid
920 * bug fix: better detection of circularities in the file-compiler.
921 (reported by Marco Monteiro)
922 * bug fix: the CL pathname functions now work with files that have
923 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
924 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
925 systems with Unix98 pty semantics.
926 * bug fix: ASDF-INSTALL will now work with bsd tar.
927 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
929 * bug fix: timers expiring in dead threads no longer cause a
930 type-error (reported by Paul "Nonny Mouse").
931 * bug fix: thanks to more lightweight single-stepper instrumentation,
932 code compiled with (DEBUG 3) will compile and execute significantly faster,
933 and will have more accurate type-inferencing than before
934 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
935 whose bindings are modified
936 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
937 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
938 ** replace dummy implementations of CL:MACHINE-INSTANCE and
939 CL:SOFTWARE-VERSION with proper version
941 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
942 * feature: implemented the READER-METHOD-CLASS and
943 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
944 as specified by AMOP.
945 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
947 * optimization: faster LOGCOUNT implementation on x86 and x86-64
948 (thanks to Lutz Euler)
949 * optimization: hashing of general arrays and vectors has been
950 improved. (reported by Any Fingerhut)
951 * enhancement: SB-INTROSPECT is now able to find definitions of
952 profiled functions. (thanks to Troels Henriksen)
953 * enhancement: compiler-macro expansion applies now to FUNCALL forms
954 as well, allowing compiler-macros for SETF-functions to expand.
955 * enhancement: step-instrumentation no longer wraps known
956 single-value functions with multiple-value context, allowing
957 better type inference.
958 * fixed bug #337: use of MAKE-METHOD in method combination now works
959 even in the presence of user-defined method classes. (reported by
960 Bruno Haible and Pascal Costanza)
961 * fixed bug #339(c): if there are applicable methods not part of any
962 long-form method-combination group, call INVALID-METHOD-ERROR.
963 (reported by Bruno Haible)
964 * fixed bug #361: the :FUNCTION initarg in the protocol for
965 initialization of methods can now be used to override
966 internally-produced optimized functions. (reported by Bruno
968 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
969 system-provided lambda expression no longer cause warnings about
970 unbound #:|pv-table| symbols.
971 * bug fix: improved the handling of type declarations and the
972 detection of violations for keyword arguments with non-constant
974 * bug fix: potentially erronous calls to PATHNAME and
975 MERGE-PATHNAMES were being flushed in some cases.
976 (reported by Richard Kreuter)
977 * bug fix: compiled calls to TYPEP were mishandling obsolete
978 instances. (reported by James Bielman and Attila Lendvai)
979 * bug fix: format strings with ~> without matching ~< no longer
980 trigger an AVER, but signal an understandable error instead.
981 (reported by Antonio Martinez)
982 * bug fix: specifying an output-file without a pathname-type for
983 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
984 of input-file instead of "fasl". (reported by Robert Dodier)
985 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
986 (reported by James Y Knight).
987 * bug fix: compiler-macros-function did not consider the environment
988 argument for shadowing by local functions.
989 * bug fix: compiler-macros expansion was inhibited by local INLINE
991 * bug fix: inline expansions of known functions were subject to
992 step-instrumentation in high debug policies, leading to problems
994 * bug fix: compiler failed to differentiate between different CONS
996 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
997 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
998 * bug fix: #368: incorrect use of expressed vs. upgraded array
1000 * bug fix: #306a: more precise unions of array types.
1001 * thread-safety improvements:
1002 ** CONDITION-WAIT could return early on Linux, if the thread was
1003 interrupted and subsequently continued with SIGCONT.
1004 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1006 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1007 * added support for the ucs-2 external format. (contributed by Ivan
1009 * minor incompatible change: pretty printing of objects of type
1010 (cons symbol) is, in the default pprint-dispatch-table, now
1011 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1013 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1014 called later in a class's lifetime, possibly as late as when the
1015 first instance of the class is created. Previously,
1016 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1017 class became finalizeable.
1018 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1019 for multibyte external-format streams. (thanks to "vbzoli")
1020 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1021 with :ALLOCATION :CLASS inherited from superclasses of the
1023 * fixed bug: anonymous classes can now be created using the :NAME
1024 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1025 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1026 * fixed bug: core-files saved with :EXECUTABLE T can again be
1027 executed when SBCL_HOME isn't set. (reported by James Knight)
1028 * fixed bug: toplevel LOCALLY forms with declarations could
1029 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1030 * fixed bug: printing from several different threads using different
1031 values of *print-case* could cause invalid output, due to
1032 some internal special variables of the printer not being bound
1033 thread-locally (reported by Max Mikhanosha)
1034 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1035 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1036 (reported by Pascal Costanza)
1037 * fixed bug: FINALIZE-INHERITANCE is called from
1038 REINITIALIZE-INSTANCE on classes when the class has previously
1039 been finalized, as required by AMOP.
1040 * minor code generation optimizations:
1041 ** better register allocation in CLOS dispatching functions
1042 ** overflow detection when coercing signed bytes to fixnums on x86-64
1043 ** is now implemented with one IMUL instruction instead of three shifts
1044 ** more efficient bit-vector access on x86 and x86-64
1045 ** more efficient access to raw structure slots on x86 and x86-64
1046 * fixed some bugs revealed by Paul Dietz' test suite:
1047 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1048 return its argument.
1050 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1051 * feature: thread support on Solaris/x86, and experimental thread support
1053 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1055 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1056 default method for SLOT-UNBOUND.
1057 * minor incompatible change: prevent the user from specializing the
1058 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1059 somewhat counter to the intent of the protocol, I (CSR) think, and
1060 additionally it just doesn't work in SBCL as currently
1061 implemented, thanks to optimizations (that are always valid for
1062 the other three SLOT-VALUEish functions, but not for the setter).
1063 * minor incompatible change: the :SB-LDB feature is now enabled by
1064 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1065 the low-level debugger.
1066 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1067 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1068 down to an SB-SYS:FD-STREAM, allowing direct communication in
1069 more cases. Temporary files and pipes are still used as a fallback
1071 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1072 on the non-reentrant C functions localtime(3) and gmtime(3).
1073 * bug fix: global optimization policy was not visible in LOCALLY and
1075 * bug fix: class objects can be used as specializers in methods.
1076 (reported by Pascal Costanza)
1077 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1078 with a type error. (reported by blitz_ on #lisp)
1079 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1080 failed with a type error. (reported by Pascal Bourguignon)
1081 * bug fix: merging pathnames against defaults with :DIRECTORY
1082 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1083 (reported by James Y Knight)
1084 * bug fix: saving large (>2GB) cores on x86-64 now works
1085 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1086 constant index and a (simple-array (signed-byte 32)) array
1087 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1088 error (patch by Robert J. Macomber)
1089 * feature: TIME now displays the amount of run-time spent in GC
1090 * bug fix: The debugger now does a better job of respecting
1091 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1092 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1093 workaround for bug 403.)
1094 * bug fix: floating point exception handling now works on FreeBSD
1095 (thanks to NIIMI Satoshi)
1096 * fixed some bugs revealed by Paul Dietz' test suite:
1097 ** MISC.641: LET-conversion were not supposed to work in late
1099 * improvements to the Win32/x86 port:
1100 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1101 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1103 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1104 * new feature: source path information is generated for macro-expansion
1105 errors for use in IDE's like Slime (thanks to Helmut Eller)
1106 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1107 * bug fix: compiler does not loop forever on an invalid type in
1109 * improvement: compilation of most CLOS applications is significantly
1111 * optimization: added a limited bytecode compiler for simple toplevel
1112 forms, speeding up compilation and FASL loading
1113 * bug fix: the statistical profiler now properly distinguishes anonymous
1116 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1117 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1118 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1119 into subclasses of the finalized class. Now user calls to
1120 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1121 system return before any subclasses are finalized.
1122 * minor incompatible change: The reader no longer ignores errors
1123 regarding non-existent packages in #+ and #- feature tests.
1124 * new feature: command line options --no-sysinit, --no-userinit to
1125 inhibit loading the corresponding init files
1126 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1127 for sending data through UDP sockets (thanks to François-René Rideau)
1128 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1129 error is signalled instead (thanks to François-René Rideau)
1130 * minor incompatible change: Error signalling behaviour of lexical
1131 operations violating package locks has changed slightly. Refer to
1132 documentation on package locks for details.
1133 * enhancement: EVAL can process IF-expressions without resorting to the
1135 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1136 constant-folded properly due to missing out-of-line functions.
1137 (reported by Mika Pihlajamäki)
1138 * bug fix: LISTEN sometimes returned T even in cases where no data was
1139 immediately available from the stream
1140 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1141 were reversed. (reported by Levente Mészáros)
1142 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1143 when given a vector displaced to an adjustable array. (reported
1145 * bug fix: derivation of float boundaries from numbers outside the
1146 appropriate float range (reported by John Wiseman)
1147 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1149 * bug fix: precomputing cache entries for generic functions with
1150 some subclasses of specializers as yet invalid does not attempt to
1151 fill a cache line with a negative offset. (reported by Levente
1153 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1154 allow condition class objects as arguments to DOCUMENTATION and
1155 (SETF DOCUMENTATION); only find and set documentation for
1156 structure names for the STRUCTURE doc-type. (suggested by Gary
1158 * improvements to the Win32/x86 port:
1159 ** (user-homedir-pathname) and default initialization file
1160 locations now know about the user's "Documents and Settings"
1161 directory (thanks to Yaroslav Kavenchuk)
1162 ** run-program is implemented (thanks to Mike Thomas)
1163 ** sockets support (thanks to Timothy Ritchey)
1164 ** better backtrace support (thanks to Alastair Bridgewater)
1165 ** sb-grovel supported
1166 ** asdf-install and sb-posix work somewhat
1167 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1168 * improvements to the Solaris/x86 port:
1169 ** works on Solaris 11/Solaris Express
1170 ** floating-point exception handling support
1171 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1172 * fixed some bugs revealed by Paul Dietz' test suite:
1173 ** REMOVE-METHOD returns its generic function argument even when
1174 the method is not one of the generic functions' methods.
1175 ** objects declared as MEMBER types can be admissible arguments to
1176 structure accessors.
1177 ** printing characters should simply be printed by the FORMAT ~:C
1179 ** compiler failure when compiling functions with hairy constant
1180 defaults for optional parameters.
1181 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1182 function, which is already optimized.
1184 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1185 * new platform: experimental support for SBCL x86/Darwin, including
1186 MacOS X 10.4.5 on Intel.
1187 * new feature: Unicode character names are now known to the system
1188 (through CHAR-NAME and NAME-CHAR).
1189 * new feature: the filesystem location of SBCL's core file is
1190 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1191 * minor incompatible change: the contrib modules SB-POSIX and
1192 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1193 this change is to make it easier to distribute
1194 location-independent binaries.
1195 * bug fix: as implied by AMOP, standardized classes no longer have
1196 slots named by external symbols of public packages. (reported by
1198 * optimization: calling functions via constant symbols -- as in
1199 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1200 via the function object as in (FUNCALL #'FOO).
1201 * optimization: on x86 and x86-64, the calling convention for the
1202 case of dealing with an unknown number of values has been altered
1203 to be friendlier to the prediction heuristics implemented,
1204 particularly with reference to CALL and RET pairing. (thanks to
1205 Alastair Bridgewater)
1206 * optimization: on x86, the code for access to thread-local symbol
1207 values and binding/unbinding of thread-local symbols is smaller.
1208 (thanks to Alastair Bridgewater)
1209 * enhancement: CONSTANTP is now able to determine constantness of
1210 more complex forms, including calls to constant-foldable standardized
1211 functions and some special forms beyond QUOTE.
1212 * fixed bug: occasional GC crashes on Solaris/x86
1213 * optimization: x86-64 supports stack allocation of results of simple
1214 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1215 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1216 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1217 (thanks to James Knight)
1218 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1219 by the standard (thanks to Stephen Compall)
1221 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1222 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1223 be used for bundling the runtime and the core file into one
1224 executable binary. This feature is not currently supported on all SBCL
1225 platforms. (thanks to James Bielman and NIIMI Satoshi)
1226 * new feature: a generational or ephemeral garbage collector is now
1227 the default on the PowerPC platform (both Linux and Darwin). The
1228 old Cheney (stop and copy) collector is a build-time option.
1229 (thanks to Cyrus Harmon, after Raymond Toy)
1230 * minor incompatible change: the method by which SBCL finds its
1231 contributed modules has changed; it no longer relies on symbolic
1232 links from an $SBCL_HOME/systems directory, but searches directly
1233 in the subdirectories of $SBCL_HOME.
1234 * enhancement: the dynamic heap size on the Linux/PPC platform is
1235 markedly larger, even using the older Cheney garbage collector.
1236 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1237 certain complicated string types. (reported by Gary King)
1238 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1239 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1240 * fixed bug: DOCUMENTATION on structures no longer signals an
1241 error if no structure type of the right name exists. (reported by
1243 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1244 applicable non-standard primary or :AROUND method on
1245 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1246 no longer cause unbound variable errors. (reported by Kevin Reid)
1247 * fixed bug: metacircle resolution in cases where methods have slots
1248 added before the slots from STANDARD-METHOD. (reported by Jean
1250 * fixed bug: the Power PC assembly code for calling into lisp
1251 sometimes computed its return address wrongly (depending on the
1252 15th bit of the address, fixed at link time). This used to
1253 manifest itself in a segmentation violation while building PCL.
1254 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1255 many others over the years)
1256 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1257 the stack pointer (treated as a signed integer) is negative.
1258 (thanks to Peter van Eynde)
1260 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1261 * new platform: experimental support for the Windows operating
1262 system has been added. (thanks to Alastair Bridgewater)
1263 * fixed several bugs in and robustified the PPC FFI (including
1264 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1265 * bug fix: allow non-simple string symbol names (reported by Paul
1267 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1269 * bug fix: saving a core corrupted callbacks on x86/x86-64
1270 * bug fix: closed a loophole in metacircularity detection and
1271 grounding in the PCL implementation of CLOS.
1272 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1273 (reported by tomppa on #lisp)
1274 * bug fix: clear the direction flag on Lisp -> C transitions, as
1275 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1276 (reported by Andrej Grozin and Hendrik Maryns)
1277 * optimization: major improvements to GC efficiency on GENCGC platforms
1278 * optimization: faster implementation of EQUAL
1279 * optimization: emit more efficient opcodes for some common
1280 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1281 * optimization: several other minor code-generation improvements on x86-64
1282 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1284 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1285 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1286 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1287 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1288 by ANSI as a generic function, and for consistency (SETF
1289 GENERIC-FUNCTION-NAME) is treated likewise.
1290 * fixed bug #233.b: make constraint propagation notice when a variable
1291 value is changed after it is referenced but before it is used
1292 * fixed bug #296: no more arbitrary behaviour from filenames with
1293 odd characters as --load arguments. (reported by Adam Warner)
1294 * fixed bug #390: :CHARACTER-SET pathname components now work as
1295 expected. (reported by Tim Daly Jr)
1296 * fixed bug #391: complicated :TYPE intersections in slot
1297 definitions no longer cause an error in PCL internals.
1298 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1299 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1300 returning the number of octets which would be written to the
1301 file-stream. (thanks to Robert J. Macomber)
1302 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1303 arguments to the internal stream functions. (thanks to David
1305 * bug fix: stack exhaustion is now handled on the x86/SunOS
1306 platform. (thanks to Daisuke Homma)
1307 * bug fix: more accurate ROOM results on GENCGC platforms
1308 * optimization: improved type inference for arithmetic-for
1309 index variables in LOOP
1310 * optimization: faster floating-point SQRT on x86-64
1311 * fixed some bugs revealed by Paul Dietz' test suite:
1312 ** DOCUMENTATION returns NIL instead of "" for method combinations
1313 that don't have a docstring
1315 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1316 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1317 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1318 therefore are not customizeable by user code (as seems to be at
1319 least permitted and maybe required by AMOP). As a consolation,
1320 however, the SBCL implementation of these functions now calls
1321 REINITIALIZE-INSTANCE as specified by AMOP.
1322 * bug fix: slot-definition documentation is propagated as per ANSI
1323 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1324 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1325 argument, as do eslotd initialization methods. (from Pascal
1326 Costanza's "Closer" project)
1327 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1328 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1330 * bug fix: it is now possible to have more than one subclass of
1331 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1332 (reported by Bruno Haible, Pascal Costanza and others)
1333 * bug fix: the dependent update protocol now works for generic
1334 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1335 and Pascal Costanza)
1336 * bug fix: condition-class instances corresponding to
1337 DEFINE-CONDITION forms are now created eagerly. (reported by
1338 Kalle Olavi Niemitalo on comp.lang.lisp)
1339 * bug fix: floating point printing is more accurate in some
1340 circumstances. (thanks to Simon Alexander)
1341 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1342 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1343 * bug fix: callbacks on OS X now preserve stack-alignment.
1344 (thanks to Cyrus Harmon)
1345 * enhancement: the x86-64 disassembler is much better at
1346 disassembling SSE instructions. (thanks to Lutz Euler)
1347 * enhancement: CHECK-TYPE on a variable now allows type-inference
1348 (thanks to Vincent Arkesteijn)
1349 * optimization: improved performance of EUC-JP external format.
1350 (thanks to NIIMI Satoshi)
1351 * optimization: performance improvements to IO on file streams of
1352 :ELEMENT-TYPE CHARACTER
1353 * optimization: much faster memory allocation on x86-64
1354 * optimization: faster garbage collections (latency and throughput)
1356 * optimization: faster bignum subtraction and division on x86 and x86-64
1357 (thanks to Lutz Euler)
1358 * optimization: more accurate type-inference for loops that use a
1359 floating point index variable or a negative step.
1361 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1362 * bug fix: add a workaround to SBCL looping infinitely at startup on
1363 Linux kernels with apparently buggy implementations of personality().
1364 (thanks to Svein Ove Aas)
1365 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1366 (thanks to David Lichteblau)
1367 * bug fix: local bindings shadow global symbol macros for
1368 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1369 * optimization: non-open coded uses of numeric comparison operators
1370 (e.g. >) no longer cons when called with more than one parameter
1371 on platforms supporting dynamic-extent allocation.
1372 * enhancement: saving cores with foreign code loaded is now supported
1373 on MIPS/Linux in addition to the previously supported platforms.
1374 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1375 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1376 * bug fix: division by zero in sb-sprof when no samples were collected
1377 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1378 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1379 (thanks to Alexander Kjeldaas)
1380 * new feature: ignoring the return values of destructive functions
1381 like NREVERSE signals a compile-time style-warning.
1382 (thanks to Kevin Reid)
1383 * new feature: the alignment of alien structure fields can be
1384 explicitly specified. (thanks to Cyrus Harmon)
1385 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1386 and dump core on SIGQUIT
1388 ** incompatible change: threads do not inherit values of specials
1389 from their parents (see manual)
1390 ** bug fix: threads stacks belonging to dead threads are freed by the
1391 next exiting thread, no need to gc to collect thread stacks anymore
1392 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1393 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1394 ** bug fix: dynamic variable and thread start related gc lossage
1395 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1397 * fixed some bugs revealed by Paul Dietz' test suite:
1398 ** SUBTYPEP is slightly more accurate on heinously complicated
1399 CONS types where some of the members have uncertain (in the
1400 NIL, NIL sense) type relationships to each other.
1402 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1403 no longer purified unless :PURIFY T is explicitly specified.
1404 ** Non-purified cores are significantly smaller than before
1406 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1407 * new feature: timers based on Zach Beane's excellent timer package
1408 * added support for the following external formats: koi8-u,
1409 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1410 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1411 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1412 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1413 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1414 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1415 * incompatible change: a threaded SBCL will no longer revert to
1416 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1417 * bug fix: interrupts are disabled until startup is complete; no
1418 more sigsegvs when receiving a signal to soon
1419 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1421 * bug fix: add a workaround for the memory randomization features in
1422 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1423 workaround will only be in effect on systems with the proc filesystem
1425 * bug fix: printing objects of type HASH-TABLE signals a
1426 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1428 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1429 have been running for over 50 days. (reported by Gilbert Baumann)
1430 * bug fix: the logic for getting names of functions gets less
1431 confused when confronded with alternate-metaclass
1432 funcallable-instances. (reported by Cyrus Harmon)
1433 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1434 consistent, even on internal alternate-metaclass objects.
1435 * bug fix: flush closure information collected by physical
1436 environment analysis prepass before the main pass. (bug reported
1438 * bug fix: compiler pack phase does not modify a hashtable, which is
1439 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1440 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1441 may be expanded. (reported by Alexander <malishev> on c.l.l)
1442 * bug fix: no more style-warnings when compiling code using LAST
1443 under high-SPEED low-SPACE settings. (reported by David Wallin
1445 * fixed some incompatibilities between SBCL's MOP and the MOP
1447 ** the METAOBJECT class is now implemented;
1448 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1449 STANDARD-OBJECT, as required;
1450 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1451 now compatible (as required by VALIDATE-SUPERCLASS); there
1452 remains a consistency requirement over the presence or absence
1453 of the FUNCTION class in the superclass list of finalized
1454 classes; see the manual for more details;
1455 ** the :AROUND method for COMPUTE-SLOTS on
1456 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1457 requested slot ordering.
1459 ** bug fix: parent thread now can be gc'ed even with a live
1461 ** bug fix: binding a special with PROGV to no value is not
1462 the same as that symbol not having been bound (thanks to
1464 ** bug fix: binding specials is thread safe (thanks to
1466 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1467 and SB-SPROF do not die with 'no handler for signal XX in
1468 interrupt_handle_now(..)' anymore
1469 ** bug fix: WITH-TIMEOUT works with multiple running threads
1470 * fixed some bugs revealed by Paul Dietz' test suite:
1471 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1472 the :method-class keyword argument.
1474 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1475 * new port: the Solaris operating system on x86 processors is now
1476 mostly supported, though some rough edges in the environment
1477 remain. (thanks to Daisuke Homma)
1478 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1479 space for its heap. (thanks to Thiemo Seufer)
1480 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1481 closures. (thanks to Thiemo Seufer)
1482 * minor incompatible change: eof selects abort in the debugger.
1483 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1484 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1486 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1487 deprecated, and will go away in a future revision of SBCL.
1488 * minor incompatible change: GC-ON and GC-OFF are no longer
1489 implemented with a counter, it does not matter how many times gc
1490 is switched on or off
1491 * bug fix: discriminating functions for generic function classes
1492 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1493 make invalid assumptions about method precedence order. (reported
1495 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1496 platforms. (thanks to Thiemo Seufer)
1497 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1498 * bug fix: correct pathname printing with printer escaping is on.
1499 (thanks to Kevin Reid)
1500 * bug fix: complex VOP definitions in "user-space" no longer trigger
1501 package locks. (reported by Zach Beane)
1502 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1503 for funcallable-instance functions, and is no different from
1505 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1506 characters in the current readtable. (reported by Nicholas Neuss)
1507 * bug fix: SUBTYPEP on various CONS types returns more a more
1508 accurate acknowledgment of its certainty.
1509 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1510 lists in some cases. This partially fixes bug 384.
1511 * flush all standard streams before prompting in the REPL and the
1513 * bug fix: signal handling and triggering gc do not conflict
1514 directly anymore, in particular a high frequency sb-sprof does
1515 not prevent gc from running
1516 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1517 approximation for timezone and DST information between the
1518 universal time epoch and the smallest negative 32-bit time_t.
1519 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1520 year 1899 so long as the encoded time is non-negative.
1521 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1522 the correct answer for non-constant rotate amounts when performing
1523 an inline 32-bit rotation.
1525 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1526 there is only one thread in the session
1527 ** bug fix: memory leak for streams created in one thread and
1528 written to in another
1529 ** bug fix: lockup when compiled with gcc4
1530 ** bug fix: race that allows the gc to be triggered when gc is
1532 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1533 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1535 ** bug fix: finished threads are gc'ed properly
1536 * fixed some bugs revealed by Paul Dietz' test suite:
1537 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1538 has arguments with a different set of applicable methods from
1539 the orignal arguments.
1540 ** The type error thrown by MAP now has a correct expected-type
1542 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1543 name a compiled function.
1544 ** Three MISC tests where a large bignum was improperly coerced to
1545 a float that couldn't represent that bignum during type
1546 derivation were fixed.
1547 ** SUBTYPEP can now handle types involving the negation of a
1548 list-form FUNCTION type.
1549 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1550 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1551 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1553 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1554 * New feature: Experimental support for bivalent streams: streams
1555 opened with :element-type :default now allow character and binary
1556 (unsigned-byte 8) I/O
1557 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1558 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1560 * bug fix: no more highly sporadic "couldn't check whether ~S is
1561 readable" when reading a stream and an interrupt hits in the middle
1562 of a select system call
1563 * compiler better recognizes complex arrays (reported by Eduardo
1565 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1566 for variables on SBCL built with linkage-tables. (reported by Luis
1568 * various error reporting improvements.
1569 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1570 (Thanks to Thiemo Seufer)
1571 * bug fix: nested reader invokations work correctly
1572 * bug fix: it is possible to have simultaneous references to foreign
1573 code and foreign data with the same name.
1575 ** added x86-64 support
1576 ** incompatible change: the threading api now works with thread
1577 objects instead of thread ids
1578 ** bug fix: threads are protected from signals and interruption when
1579 starting up or going down
1580 ** bug fix: a race where an exiting thread could lose its stack to gc
1581 ** bug fix: don't halt on infinite error in threads if possible
1582 ** fixed numerous gc deadlocks introduced in the pthread merge
1583 ** bug fix: fixed thread safety issues in read and print
1584 ** bug fix: debugger doesn't hang on session lock if interrupted at
1585 an inappropriate moment
1586 ** bug fix: run-program is now thread safe(r)
1587 ** bug fix: inner with-recursive-lock no longer releases the mutex
1588 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1589 * fixed some bugs revealed by Paul Dietz' test suite:
1590 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1591 EXPECTED-TYPE slots filled.
1592 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1594 ** MAKE-CONDITION accepts classes as type-designators.
1595 ** COMPILE may never return NIL.
1596 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1597 range before calling Unix time functions
1599 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1600 * numerous signal handling fixes to increase stability
1601 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1602 * minor incompatible change: we now correctly canonize default
1603 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1604 as per the MOP, rather than the historical (INITARG INITFUNCTION
1605 INITFORM). (reported by Bruno Haible)
1606 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1607 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1608 for more information.
1609 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1610 pathname is a directory pathname.
1611 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1612 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1614 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1615 :METACLASS argument in addition to a class name. (reported by
1616 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1617 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1618 (reported by Bruno Haible; patch from Gerd Moellmann)
1619 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1621 * bug fix: more cleanups to the floating point exception handling on
1622 x86-64 (thanks to James Knight)
1623 * bug fix: the compiler does not try to constant fold calls of
1624 COERCE and BIT-* functions when they return freshly constructed
1625 objects. (reported by Nikodemus Siivola and Paul Dietz)
1626 * optimization: improved the MIPS versions of generic (in the
1627 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1628 * optimization: direct conversion of (unsigned-byte 32) to floats on
1629 the PowerPC platform.
1630 * optimization: structure instances with raw slots now use less
1631 memory, and probably show better memory locality. (thanks to
1633 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1634 are treated as requests for stack allocation on the x86-64,
1635 Alpha-32, PPC and SPARC platforms.
1636 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1637 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1638 (thanks to Tony Martinez)
1640 ** gcing a dead thread can no longer lead to lockups
1641 ** threads block signals until they are set up properly
1642 ** errno is no longer shared by threads
1643 ** interrupt-thread restores the eflags register on x86
1644 ** fixed some lockups due to gc/thread interaction
1645 * fixed some bugs revealed by Paul Dietz' test suite:
1646 ** invalid dotted lists no longer raise a read error when
1647 *READ-SUPPRESS* is T
1648 ** signal an error if a symbol that names a declaration is used
1649 as the name of a type, or vice versa
1650 ** allow using the (declare (typespec var*)) abbreviation for
1651 (declare (type typespec var*)) with all type specifiers
1652 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1653 ** always return NIL from PPRINT-POP when OBJECT is NIL
1654 ** don't signal errors when pretty-printing malformed LABELS,
1655 FLET or MACROLET forms
1656 ** declarations in a DOLIST body are also in scope for the
1658 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1660 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1663 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1664 * fixed cross-compiler leakages that prevented building a 32-bit
1665 target with a 64-bit host compiler.
1666 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1667 opened with :IF-EXISTS :OVERWRITE.
1668 * fixed bug 281: error for an invalid qualifier in a short-form method
1669 combination method is not signalled until the faulty method is called.
1670 * bug fix: iteration variable type inferrer failed to deal with open
1671 intervals. (reported by Alan Shields)
1672 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1673 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1674 or more runtime options were provided to the sbcl binary.
1675 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1676 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1677 * optimization: there's now a fast-path for fixnum arguments in the
1678 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1679 * optimization: the code generated on x86-64 is more compact thanks
1680 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1681 * bug fix: floating-point exception handling now partly works on
1682 x86-64. (Thanks to James Knight)
1683 * improvement to the MIPS backend from Thiemo Seufer: C-style
1684 64-bit long long arguments and return values to alien functions
1686 * fixed some bugs revealed by Paul Dietz' test suite:
1687 ** the type-error signalled from WARN has a filled-in DATUM slot.
1688 ** the type-error required when a stream is not associated with
1689 a file has the stream as its datum.
1690 ** type-errors on single-floats on x86-64 no longer have
1691 :INVALID-OBJECT as the datum
1692 ** the type-errors signalled for invalid function names now have
1693 a correct expected type
1694 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1695 for structures defined with a :TYPE. Documentation strings for
1696 typed structures are no longer immediately discarded
1697 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1698 broadcast streams and synonym streams. FILE-LENGTH now also works
1699 on broadcast streams.
1701 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1702 * incompatible change: the --noprogrammer option, deprecated since
1703 version 0.7.5, has been removed. Please use the equivalent
1704 --disable-debugger option instead.
1705 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1707 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1708 inoperational for a while now) has been completely removed.
1709 * null lexical environments are now printed as #<NULL-LEXENV>,
1710 significantly reducing the amount of clutter in typical backtraces.
1711 * documentation on weak pointers, finalization, and after GC hooks
1712 has been added to the manual.
1713 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1714 as other specialized array types, is much faster. SUBSEQ and
1715 COPY-SEQ on such arrays have also been sped up.
1716 * optimization: EQL is now more efficient when at least other argument
1717 is known to be of type (OR FIXNUM (NOT NUMBER)).
1718 * fixed bug: the runtime is now less vulnerable to changes in the
1719 size of the SBCL object on OS X, and virtual memory is reserved for
1720 all spaces that need to be at a fixed address.
1721 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1722 * fixed bug: finalizers and after GC hooks that cause consing are now
1724 * fixed bug: compiler error messages and summaries are now printed to
1725 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1726 * fixed inference of the upper bound of an iteration variable.
1727 (reported by Rajat Datta).
1728 * fixed bug 211e: calling local functions with duplicated constant
1729 keyword argument no longer causes a bogus style warning about an
1731 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1732 ftype declaration to be disregarded. (reported by Dave Roberts)
1733 * fixed bug 373: caused by erronous compilation of references to alien
1734 variables in the runtime on ppc/darwin.
1735 * fixed bug 376: CONJUGATE type deriver.
1736 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1737 (reported by Baughn on #lisp)
1738 * a cross-compiler bug on non-x86 platforms has been identified and
1739 fixed. (thanks to Bruno Haible)
1740 * improvements to the MIPS runtime code for increased stability.
1741 (thanks to Thiemo Seufer)
1742 * increased the maximimum compact environment size to allow
1743 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1744 * partial workaround for bug 135: don't name the function we're
1745 calling for hairy cases of EVAL, so as not to accumulate one environment
1746 entry per call to EVAL. (reported by Kevin Reid)
1747 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1748 * optimization: type testing for non-vector arrays should be faster.
1749 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1751 * bug fix: setting 31st element of a bit vector to zero did not work
1753 * bug fix: redefining a class definition which failed due to a
1754 previous accessor / function clash now works (but see BUGS entry
1755 #380 for more problems in this area). (thanks to Zach Beane)
1756 * the long form of DEFINE-METHOD-COMBINATION disables method group
1757 checking when given a single method group with pattern *. (thanks
1759 * on x86 compiler supports stack allocation of results of simple
1760 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1761 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1762 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1763 * fixed some bugs related to Unicode integration:
1764 ** the restarts for recovering from input and output encoding
1765 errors only appear when there is in fact such an error to
1767 * fixed some bugs revealed by Paul Dietz' test suite:
1768 ** MISC.549 and similar: late transformation of unsafe type
1769 assertions into derived types caused unexpected code
1771 ** SCALE-FLOAT type deriver is less wrong.
1772 ** type derivers for EXP, LOG and similar functions compute result
1773 types for complex arguments better.
1774 ** (MISC.563) CONJUGATE type deriver works for very restricted
1776 ** out-of-line type testers for character strings are available.
1777 ** EQUAL compiler transform understands specialness of objects
1779 ** accessing double-floats stored on the stack now works on x86-64.
1780 ** debugger internals could sometimes create invalid lispobjs,
1781 resulting in GC crashes.
1782 ** MISC.548: type check weakening can convert required type into
1784 ** initialization forms for bindings are not in scope of free special
1787 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1788 * incompatible change: thread support for non-NPTL systems has
1789 been removed - locking is buggy and unreliable. A threaded
1790 SBCL build will now warn at startup and refuse to create
1791 new threads, unless futex support is detected in the kernel
1792 * incompatible change: the top level REPL now has only an ABORT
1793 restart associated with it, not TOPLEVEL and ABORT as it used to.
1794 TOP and TOPLEVEL are now available as debugger commands for
1795 returning to the top level.
1796 * incompatible change: forms evaluated in the REPL now use the
1797 global optimization policy.
1798 * incompatible change: user- and system-initialization files are
1799 no longer processed with LOAD, but by READ and EVAL; hence the
1800 global optimization policy, startup package, readtable, etc,
1802 * The .fasl file format number has been incremented because of
1803 various incompatible changes.
1804 * internal entry point details and argument counts no longer appear
1805 in backtraces unless explicitly requested by setting
1806 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1807 * built-in and standard functions no longer have names like "top
1808 level local call to FOO".
1809 * fixed bug 32: functions defined in non-null lexical environments
1810 now have more legible printed representation
1811 * fixed bug 33: functions defined in non-null lexical environemnts
1812 are now more amenable to inspection by INSPECT.
1813 * workaround for bug 354: XEPs no longer appear in backtraces unless
1814 explicitly requested.
1815 * fixed bug: receiving the signal which results from attempting to
1816 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1817 SIGBUS on other *BSDs) on architectures where the C stack is also
1818 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1819 to debugger confusion or wild instability. (reported by Cheuksan
1821 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1822 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1823 (reported by Lutz Euler)
1824 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1825 compiled code. (reported by Paul Dietz)
1826 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1827 specialized parameter to a method to something that is not TYPEP
1828 the specializer is now possible.
1829 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1830 face of package deletion.
1831 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1832 pathname. (reported by David Tolpin on c.l.l)
1833 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1834 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1835 STRING-STREAM and FILE-STREAM, respectively.
1836 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1837 than previously; in addition, its test suite is now run on build.
1838 (thanks to Luke Gorrie)
1839 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1840 * optimization: added a immediate representation for single-floats
1842 * fixed some bugs revealed by Paul Dietz' test suite:
1843 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1844 ** attempting to create a package with a colliding nickname causes
1845 correctable errors to be signalled.
1846 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1847 ** several x86-64 backend bugs related to sign-extension of immediate
1850 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1851 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1852 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1854 * fixed loading of multiply forward-referenced layouts.
1855 (thanks to Cheuksan Wang)
1856 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1857 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1858 (see COMPILE-FILE documentation for details of :PRINT option)
1859 * fixed bugs 19 and 317: fixed-format floating point printing is
1860 more accurate. This also fixes a bug reported by Adam Warner
1861 related to the ~@F format directive.
1862 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1864 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1865 dispatch table if the from-char is a dispatch macro character.
1866 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1867 by the wordsize no longer ignore the last word. (reported by Lutz
1869 * fixed bug in type checking of dynamic-extent variables. (reported
1871 * optimization: sequence traversal functions use their freedom to
1872 coerce function designators to functions.
1873 * optimization: code with many calls to CLOS methods specialized on
1874 CLOS classes has had redundant type checks removed.
1875 * optimization: added declarations to speed up operations that access
1876 the internal character database (for example STRING-UPCASE)
1877 * optimization: comparison operations between floats and sufficiently small
1878 fixnums no longer create extra rationals
1879 * fixed some bugs related to Unicode integration:
1880 ** portions of multibyte characters at the end of buffers for
1881 character-based file input are correctly transferred to the
1882 start of the buffer at the next read.
1883 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1884 passing it through to OPEN.
1885 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1886 argument given to internal calls to OPEN.
1887 ** the built-in comment readers (introduced by character sequences
1888 ";" and "#|") are more forgiving to encoding errors; they will
1889 STYLE-WARN and then attempt to resync the stream at a character
1890 boundary. (thanks to Teemu Kalvas)
1891 * fixed some bugs in the x86-64 port:
1892 ** Negative short int return values from c-calls are sign-extended
1894 ** The stack is aligned to 16-bytes for c-calls, as required by
1895 the ABI. (thanks to Cheuksan Wang)
1896 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1897 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1898 for enums. (thanks to Vincent Arkesteijn)
1899 ** Multiple small optimizations and bugfixes for floating point
1901 * fixed some bugs revealed by Paul Dietz' test suite:
1902 ** Space, Tab, Linefeed, Return and Page have the invalid
1903 secondary constituent character trait.
1904 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1906 ** WITH-INPUT-FROM-STRING should only update the index place on
1908 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1909 works more reliably.
1910 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1911 with one-dimensional arrays with fill pointers.
1912 ** TYPE-OF failed on a complex with an integer realpart and a
1914 ** compiler failure during type inference for the code of form
1915 (IF (EQL X (THE ...)) ...) (MISC.535).
1917 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1918 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1919 platform. The port supports 61-bit fixnums, large memory spaces
1920 and reloading shared object files.
1921 * enhancement: saving cores with foreign code loaded is now
1922 supported on ppc/Darwin in addition to the previously supported
1924 * enhancement: the statistical profiler now walks deeper into the
1925 call stack for more meaningful call-graphs and accrued time
1926 reports (x86/x86-64 only). It also now reports time spent in
1928 * enhancement: it is now possible to trace most individual methods
1929 of a generic function in addition to tracing the generic function
1931 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1933 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1934 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1935 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1936 produces an error. (thanks to Vincent Arkesteijn)
1937 * bug fix: NAMESTRING on pathnames with :WILD components in their
1938 directories works correctly. (thanks to Artem V. Andreev)
1939 * fixed bug 125: compiler preserves identity of closures. (reported
1941 * bug fixed: functions with &REST arguments sometimes failed with
1942 "Undefined function" when compiled with (DEBUG 3). (reported by
1944 * bug fix: overflow during compiling of setting element of a bit
1945 vector with constant index and value. (reported by Timmy Douglas)
1946 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1947 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1948 * on x86 compiler partially supports stack allocation of dynamic-extent
1950 * GO and RETURN-FROM do not check the extent of their exit points
1951 when compiled with SAFETY 0.
1952 * fixed some bugs related to Unicode integration:
1953 ** encoding and decoding errors are now much more robustly
1954 handled; it should now be possible to recover even from invalid
1955 input or output to the terminal. (thanks to Teemu Kalvas)
1956 ** provided a first cut at implementing STRING-TO-OCTETS and
1957 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1958 ** altered the SB-MD5 contributed module slightly, changing the
1959 interface just enough for it to be supportable for builds where
1960 lisp characters are not eight bits.
1961 * fixed some bugs revealed by Paul Dietz' test suite:
1962 ** the FORMATTER-generated functions for ~V[ conditionals require
1963 the correct number of arguments.
1964 ** READ-FROM-STRING returns the mandated second value when applied
1965 to displaced strings.
1966 ** the #\Rubout and #\Backspace characters are treated as invalid
1967 constituent characters by the tokenizer.
1969 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1970 * new feature: reloading changed shared object files with
1971 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1972 * new feature: references to foreign variables and functions
1973 can now be compiled and loaded before the corresponding shared
1974 object file is loaded, as long as the foreign definitions are
1975 available at runtime.
1976 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1977 supported. (thanks to Dan Debertin)
1978 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1979 just keywords. (thanks to Vincent Arkesteijn)
1980 * fixed bug #331: structure-class instances corresponding to
1981 DEFSTRUCT forms are now created eagerly.
1982 * fixed bug #345: backtraces from calls to undefined functions work
1983 on x86 as well. Related bug #61 is now also partially fixed on x86
1984 (backtraces from throws to unknown catch tags.)
1985 * bug fix: lambda-list parsing is now stricter vrt. order and number
1986 of lambda-list keywords.
1987 * bug fix: as specified by AMOP, an error is signalled if a
1988 class-option appears multiple times in a DEFCLASS form. (reported
1990 * bug fix: RANDOM can be compiled when the compiler derives the type
1991 of its numeric argument as a disjoint set of small integers.
1992 (reported by Paul Dietz)
1993 * bug fix: starting a core saved with shared objects loaded when
1994 those objects are not available no longer causes threaded SBCL to
1995 hang. (reported by Sean Ross)
1996 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1998 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1999 argument, which must always be NIL. (reported by Kalle Niemitalo)
2000 * bug fix: printing 1.0d+23 no longer results in an error.
2001 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2002 * bug fix: structure slot setters preserve evaluation order. (thanks
2004 * fixed some bugs related to Unicode integration:
2005 ** RUN-PROGRAM can allow its child to take input from a Lisp
2006 stream. (reported by Stefan Scholl)
2007 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2008 locales. (reported by Ken Causey)
2009 * fixed some bugs revealed by Paul Dietz' test suite:
2010 ** INCF, DECF and REMF evaluate their place form as specified in
2012 ** FORMATTER expands ~{ ~} iteration directives with V or #
2013 parameters correctly.
2014 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2015 consequent uses no arguments correctly.
2016 ** the system has a partial understanding of the (COMPLEX RATIO)
2018 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2021 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2022 * new feature: a build-time option (controlled by the :SB-UNICODE
2023 keyword feature, enabled by default) for building the system with
2024 support for the entire 21-bit character space defined by the
2026 * new feature: the system now has rudimentary external-format
2027 support; the primary user-visible change at this time is that
2028 characters with the high bit set (such as lower-case-e-acute) will
2029 print correctly to a terminal in a UTF-8 environment.
2030 * minor incompatible change: BASE-CHAR no longer names a class;
2031 however, CHARACTER continues to do so, as required by ANSI.
2032 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2033 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2034 has been moved to the SB-EXT package (temporarily re-exported from
2036 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2037 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2039 * fixed bug #351: better error-handlind and reporting for malformed
2041 * fixed bug #350: bignum-printing is now more memory-efficient,
2042 allowing printing of very large bignums, eg. (expt 2 10000000).
2043 (reported by Bruno Haible)
2044 * fixed bug #302: better primitive-type selection for intersection
2046 * fixed bug #308: non-graphic characters now all have names, as
2047 required. (reported by Bruno Haible)
2048 * bug fix: redefining a class with different superclasses now correctly
2049 removes it from the direct-subclasses of its previous superclasses.
2050 (reported by David Morse)
2051 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2052 new value now works. (reported by Bruno Haible)
2053 * bug fix: correct canonicalization of multiple non-standard slot
2054 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2055 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2056 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2057 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2058 now exists, an signals an error.
2059 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2060 during macroexpansion for non-iteration variables as well. (reported
2061 by Bruno Haible for CMUCL)
2062 * bug fix: Cyclic structures and unprintable objects in compiler
2063 messages no longer cause errors. (reported by Bruno Haible)
2064 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2065 and READ-FROM-STRING all now return a primary value of NIL if
2066 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2067 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2068 (reported by Bruno Haible for CMUCL)
2069 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2070 do not designate a condition. (reported by Bruno Haible for
2072 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2073 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2074 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2075 permitting resolution in favour of any of the conflicting symbols.
2076 (reported by Bruno Haible for CMUCL)
2077 * bug fix: EQUAL compiler optimizations is less aggressive on
2078 strings which can potentially compare true despite having distinct
2079 specialized array element types.
2080 * bug fix: unit enumerations can be defined without dividing by
2081 zero. (thanks to Vincent Arkesteijn)
2082 * FORMAT compile-time argument count checking has been enhanced.
2083 (report from Bruno Haible for CMUCL)
2084 * a partial workaround for the bug 262: the compiler does not try to
2085 inline-expand a local function doing RETURN-FROM from a deleted
2086 BLOCK. (thanks to Peter Denno for the bug report and to David
2087 Wragg for the simple test case)
2088 * fixed some bugs revealed by Paul Dietz' test suite:
2089 ** RENAME-PACKAGE allows all package designators as new package
2091 ** constraint propagation and lambda variable substitution are
2092 more cautious in dealing with partially deleted code.
2093 ** compiler sometimes lost reoptimization passes.
2094 ** CERROR, given a condition as condition designator, passes its
2095 remaining arguments to the continue format control without
2097 ** Case-altering FORMAT directives work correctly on non-ASCII
2099 ** The REAL type specifier handles bounds outside the floating
2100 point ranges without signalling FLOATING-POINT-OVERFLOW.
2101 ** Functions with IR1-transformations can create intercomponent
2102 references to global functions.
2103 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2105 ** FORMAT ~R treats a nil value for its first parameter correctly.
2107 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2108 * enhancement: saving cores with foreign code loaded is now
2109 supported on x86/NetBSD and sparc/Linux in addition to the previously
2110 supported platforms.
2111 * bug fix: on some platforms repeated installations caused multiple
2112 copies of HTML documentation to be installed -- should not happen
2113 any more. (reported by Stefan Scholl)
2114 * bug fix: parsing self-recursive alien record types multiple times
2115 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2116 original patch by Helmut Eller for CMUCL)
2117 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2118 (thanks to Richard Kreuter)
2119 * bug fix: defining classes whose accessors are methods on existing
2120 generic functions in other (locked) packages no longer signals
2121 bogus package lock violations. (reported by François-René Rideau)
2122 * bug fix: special variables as DEFMETHOD parameters no longer have
2123 associated bogus type declarations. (reported by David Wragg and
2125 * bug fix: read-write consistency on streams of element-type
2126 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2127 * bug fix: redefiniton of the only method of a generic function with
2128 no DEFGENERIC no longer emits a full WARNING. In addition,
2129 redefinition of generic functions with no DEFGENERIC to an
2130 incompatible lambda list now signals an error. (thanks to Zach
2132 * bug fix: DEFGENERIC now works even when there's a function of the
2133 same name in an enclosing lexical environment. (thanks to Zach
2135 * fixed compiler failure, caused by instrumenting code during
2136 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2137 * optimization: added loop analysis and improved register allocation
2138 * fixed some bugs revealed by Paul Dietz' test suite:
2139 ** POSITION on displaced vectors with non-zero displacement
2140 returns the right answer.
2141 ** (SIMPLE-STRING) is a valid type specifier for sequence
2143 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2145 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2146 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2148 ** The justification version of the FORMAT ~< directive treats
2149 non-zero minpad parameter correctly.
2151 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2152 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2153 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2154 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2155 the supported interface.
2156 * new feature: Single-stepping of code compiled with DEBUG 2 or
2157 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2158 * new feature: saving cores with foreign code loaded is now
2159 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2160 Timothy Moore's work for CMUCL)
2161 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2162 parameters to * instead of NIL if no initform is supplied.
2163 (reported by Johan Bockgaard)
2164 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2165 correctly when FUNCALL appears as the car of the form. Note:
2166 despite this FUNCALL forms are not currently subject to
2167 compiler-macro expansion. (port of Raymond Toy's fix for the
2168 same from CMUCL, reported by Johan Bockgaard)
2169 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2170 (thanks for Teemu Kalvas)
2171 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2172 case instead of terminating the iteration (thanks for Julian
2173 Squires, Sean Champ and Raymond Toy)
2174 * bug fix: incorrect expansion of defgeneric that caused a style
2175 warning. (thanks for Zach Beane)
2176 * bug fix: slot accessor effective method computation works properly
2177 for classes with multiple non-standard applicable methods on
2178 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2179 * on x86 compiler supports stack allocation of results of LIST and
2180 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2181 CMUCL implementation by Gerd Moellmann)
2182 * fixed some bugs revealed by Paul Dietz' test suite:
2183 ** FORMAT strings with both the ~<~:;~> form of the justification
2184 directive and pretty-printing directives cause an error.
2186 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2187 * incompatible change: the internal functions
2188 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2189 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2190 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2191 instead of the old functions.
2192 * new feature: on platforms where "dladdr" is available foreign
2193 function names now appear in backtraces. (based on Helmut Eller's
2195 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2196 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2198 * documentation: ASDF manual is now distributed with SBCL.
2199 * bug fix: SBCL can now load its contributed modules with REQUIRE
2200 even if the system-provided entries have been removed from
2201 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2203 * bug fix: backtraces involving undefined functions or assembly
2204 routines are more informative. (thanks to Brian Downing)
2205 * bug fix: mutually referent alien structures now work correctly.
2206 (reported by Rick Taube)
2207 * bug fix: structures defined by WITH-ALIEN can be referred to
2208 within other definitions in the same WITH-ALIEN.
2209 * bug fix: division operators (MOD, TRUNCATE and the like) with
2210 constant zero divisors and integer dividends no longer generate
2212 * bug fix: provide default methods for INPUT-STREAM-P and
2213 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2214 * bug fix: improve the detection and resolution of MOP metacycles
2215 (where effective method computation depends on the generic function
2216 having its effective method computed). (reported by Bruno Haible)
2217 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2218 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2219 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2220 * bug fix: pathnames with relative directory components can be
2221 represented relative to default pathnames.
2222 * optimization: in taking the GCD of bignums, reduce the two bignums
2223 to approximately the same size (using Euclid's algorithm) before
2224 applying the more sophisticated binary GCD. (thanks to Juho
2226 * optimization: COUNT on bitvectors now operates word-at-a-time.
2227 * optimization: ASH with a positive, but not necessarily constant,
2228 (leftwards) shift, when in a modular context, is compiled to a
2230 * fixed some bugs revealed by Paul Dietz' test suite:
2231 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2232 if the corresponding argument is NIL.
2233 ** FORMAT directives accepting a minpad parameter treat negative
2235 ** literal commas as character parameters to FORMAT directives are
2237 ** literal spaces directly after ~<Newline> directives within a
2238 format-logical-block (~:< ~@:>) do not induce :FILL-style
2239 conditional newlines.
2240 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2241 as specified: it no longer includes conditional newlines).
2242 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2244 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2245 :PER-LINE-PREFIX argument does not evaluate to a string.
2247 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2248 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2249 the manual for details; for now, package locks can be disabled by
2250 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2251 if no major problems are found then it is likely that they will be
2252 compiled in unconditionally.
2253 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2254 unsupported operators on all platforms. To load a shared library
2255 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2256 object file, link it into a shared library outside of SBCL and
2257 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2259 * bug workaround: ROOM T and the GC were not working together
2260 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2261 reporting facility aren't preserved. That reporting has been
2262 disabled, so now until and unless someone figures out how to make it
2263 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2264 * minor incompatible change: as threatened around sbcl-0.8.0, the
2265 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2266 The symbols are also exported from SB-PCL for backwards
2267 compatibility, but more so than before SB-PCL should be treated as
2268 an implementation-internal package.
2269 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2271 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2272 It is known as of this release not to work on the Alpha, however.
2273 * fixed bug #167: errors signalled due to illegal syntax in method
2274 bodies are now more legible.
2275 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2276 designators and can hence be used with TYPEP.
2277 * fixed bug #333: CHECK-TYPE now ensures that the type error
2278 signalled, if any, has the right object to be accessed by
2279 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2280 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2281 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2283 * fixed bug #334: programmatic addition of slots using specialized
2284 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2285 and :ALLOCATION :CLASS slots.
2286 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2287 float exponents. (rereported by Peter Seibel)
2288 * fixed a bug: #\Space (and other whitespace characters) are no
2289 longer considered to be macro characters in standard syntax by
2290 GET-MACRO-CHARACTER.
2291 * fixed bug: initialization of condition class metaobjects no longer
2292 causes an instance of the condition to be created. (reported by
2294 * fixed bug: it is now possible to have slots such that
2295 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2296 description is neither :INSTANCE nor :CLASS.
2297 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2298 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2299 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2300 non-standard SLOT-DEFINITION-ALLOCATION values.
2301 * fixed bugs: various race conditions handling exiting threads.
2302 CL-PPCRE's thread torture test now passes.
2303 * fixed arguably-a-bug: GC time proportional to number of threads in
2304 system even when most of them are idle
2305 * optimization: improved performance of BIT and SBIT on bit-vectors.
2306 * .fasl file incompatibility: The fasl file version number has
2307 been incremented because of changes associated with package locks.
2309 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2310 * minor incompatible change: the system no longer provides
2311 optimization hints (conditions which are TYPEP
2312 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2313 mode; these hints are still emitted when higher SPEED optimization
2315 * new contrib module: a sampling profiler (profiling by statistical
2316 sampling, rather than by instrumenting functions) is available as
2317 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2318 * the behaviour of the standard function ED is now customizeable by
2319 third parties through a hook variable: see ED's documentation
2320 string for information on the protocol.
2321 * the compiler no longer emits efficiency notes for (FUNCALL X)
2322 when the type of X is uncertain under default optimization
2324 * fixed bug 276: mutating a binding of a specialized parameter to a
2325 method to something that is not TYPEP the specializer is now
2327 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2328 exhibit the required behaviour on the x86 platform. (thanks to
2329 Peter van Eynde, Eric Marsden and Bruno Haible)
2330 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2331 even for difficult arguments. (reported by Peter Graves)
2332 * fixed bug 141a: the backquote printer now descends quoted
2334 * fixed another bug in backquote printing: no more destructive
2335 modification of the form's list structure. (reported by Brian
2337 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2338 move between its address being taken and the call to
2339 interrupt_thread, fixing a crashing race condition.
2340 * the SB-POSIX contrib implementation has been adjusted so that it
2341 no longer exhibits ridiculously poor performance when constructing
2342 instances corresponding to C structs.
2344 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2345 * minor incompatible change: the sb-grovel contrib now treats C
2346 structures as alien (in the sense of SB-ALIEN) objects rather than
2347 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2348 has implications for memory management of client code
2349 (sb-grovel-returned objects must now be manually managed) and for
2350 type safety (alien objects now have full types).
2351 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2352 used to control emission of compiler diagnostics, rather than the
2353 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2354 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2355 quality should be considered deprecated.
2356 * install.sh now installs the user manual as well
2357 * (not quite a new documentable feature, but worth considering in
2358 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2359 beginnings of a semantically meaningful condition hierarchy is
2360 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2362 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2363 designator as the defaults argument.
2364 * fixed bug: Displaced arrays whose displaced-to array has become
2365 too small now cause ARRAY-DIMENSION to signal an error, providing
2366 for safer bounds-checking. (reported by Bruno Haible)
2367 * fixed bug: DEFCLASS slot definitions with identical :READER and
2368 :WRITER names now signal a reasonable error. (reported by Thomas
2370 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2372 * fixed bug: Invalid slot specification errors now print correctly.
2373 (thanks to Zach Beane)
2374 * fixed bug 320: Shared to local slot value transfers in class
2375 redefinitions now happen corrently from superclasses as
2376 well. (reported by Bruno Haible)
2377 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2379 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2380 improper lists correctly. (reported by Bruno Haible)
2381 * fixed bug 313: source-transform for <fun-name> was erroneously
2382 applied to a call of a value of a variable with name <fun-name>.
2383 (reported by Antonio Menezes Leitao)
2384 * fixed bug 307: The obsolete instance protocol ensures that
2385 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2386 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2387 remove dead unknown-values globs from the middle of the stack.
2388 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2389 easier to resume long computations after using *BREAK-ON-SIGNALS*
2390 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2391 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2392 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2393 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2394 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2395 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2397 * fixed a (fairly theoretical) bug in string printing: if
2398 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2399 does not have array-element-type equal to the most general string
2401 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2402 function name. (thanks to Nikodemus Siivola)
2403 * fixed bug: compile-time format string checker failed on
2404 non-closed ~{. (reported by Thomas F Burdick)
2405 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2406 don't warn on legal loop code involving more than one aggregate
2407 boolean. (thanks to Nikodemus Siivola)
2408 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2409 in keyword position to MAKE-PACKAGE was not being reported as
2411 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2412 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2414 * fixed bugs 280 and 312: the checking for multiple definitions in a
2415 file is less likely to become confused by uses of inline
2417 * fixed bug: the #S reader macro performs the keyword coercion
2418 specified for slot names. (reported by Kalle Niemitalo)
2419 * fixed bug: lambda lists may contain symbols whose names start with
2420 & but are not lambda-list-keywords; their occurrence triggers a
2422 * fixed bug 321: define-method-combination argument lambda lists do
2423 not cause invalid code to be generated when &optional or &aux
2424 variables have default values. (reported by Bruno Haible)
2425 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2426 have CLOS classes; this appears also to have fixed reported
2427 instability in stack exhaustion detection.
2428 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2429 MUFFLE-WARNING when no associated restart is present is now
2431 * optimization: rearranged the expansion of various defining macros
2432 so that each expands into only one top-level form in a
2433 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2435 * optimization: used a previously little-used slot in symbols to
2436 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2438 * fixed some bugs revealed by Paul Dietz' test suite:
2439 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2441 ** Fixed an optimization bug related to inheritance of initforms
2442 from local to shared slots.
2443 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2444 ** CAST optimizer forgot to flush argument derived type.
2445 ** print/read consistency on floats is now orders of magnitude
2446 more likely. (thanks also to Bruno Haible for a similar report
2448 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2449 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2450 for objects of type REAL. Make it so.
2451 ** SXHASH is specified (infelicitously) to respect similarity,
2452 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2453 it so. (thanks to Markus Ziegler)
2454 ** on the Alpha, the compiler succeeds in compiling functions
2455 returning a known number of arguments greater than 63.
2456 ** fixed handling of invalid NIL arguments in keyword position
2458 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2460 ** provided a readably-printable representation for RANDOM-STATE
2462 ** ensured that pathnames, where they have a namestring, always
2463 print using #P"..." syntax.
2465 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2466 * Support for the forthcoming 2.0 version of the NetBSD kernel
2467 running on x86 hardware has been added. (thanks to Perry
2468 E. Metzger most immediately, and others for their past work)
2469 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2470 current release version; SBCL's OpenBSD support had been broken
2471 since about the time of OpenBSD's switch to ELF binary format.)
2472 * [placeholder for DX summary]
2473 ** user code with &REST lists declared dynamic-extent, under high
2474 speed or space and low safety and debug optimization policy.
2475 * The manual has been converted to Texinfo format and the debugger
2476 chapter from the cmucl manual has been added.
2477 * A facility has been added to extract documentation strings from
2478 sbcl and store them as Texinfo-formatted snippets for inclusion in
2479 the manual (via Texinfo's @include directive)
2480 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2481 the test case to Dave Roberts)
2482 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2483 pointers. (reported by Sean Ross)
2484 * bug fix: PROFILE output is printed nicely even for large numerical
2485 values. (thanks to Zach Beane)
2486 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2487 greater than 32 handle EOF correctly.
2488 * bug fix: on X86 an immediate argument of the IMUL instruction is
2489 correctly printed in disassembly. (thanks to Lutz Euler)
2490 * bug fix: class slots in redefined classes preserve their old
2491 values. (thanks to Bruno Haible and Nikodemus Siivola)
2492 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2493 succeeds. (reported by Marco Baringer)
2494 * fixed some bugs revealed by Paul Dietz' test suite:
2495 ** READ-SEQUENCE now works on ECHO-STREAMs.
2496 ** RATIONALIZE works more according to its specification. (thanks
2499 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2500 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2501 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2502 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2503 should matter to you only if you rebind the printer control
2504 variables and then find you want different bindings in the
2505 debugger than in the ordinary execution of your program.)
2506 * The runtime build system has been tweaked to support building
2507 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2508 assembler and linker. (thanks to Nikodemus Siivola)
2509 * Unbound, undefined, undeclared variables now trigger full
2510 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2511 more often programmer error than deliberate exploitation of undefined
2513 * optimization: the hash algorithm for strings has changed to one
2514 that is less vulnerable to spurious collisions. (thanks to Juho
2516 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2517 needless bounds checking. (thanks to Juho Snellman)
2518 * optimization: implemented multiplication as a modular
2519 (UNSIGNED-BYTE 32) operation on the PPC backend.
2520 * fixed some bugs revealed by Paul Dietz' test suite:
2521 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2523 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2524 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2525 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2527 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2529 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2531 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2532 their output stream on EOF from read.
2533 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2534 have been read to end-of-file.
2535 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2537 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2538 description of determination of which consecutive characters
2540 ** Printing the "Space" character with escaping on now yields "#\\ ",
2541 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2542 ** Reading floating-point numbers with *READ-BASE* set to a number
2543 less than 10 works correctly.
2544 ** Reading floating-point numbers with *READ-BASE* set to a number
2545 more than 10 works correctly.
2546 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2547 the readtable currently in effect.
2549 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2550 * minor incompatible change: parsing of namestrings on a physical
2551 (Unix) host has changed; numbers after the final #\. in a
2552 namestring are no longer interpreted as a version field. This is
2553 intented to be largely invisible to the user, except that the
2554 meaning of the namestring "*.*.*" has changed: it now refers to a
2555 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2556 should usually be replaced by
2557 (make-pathname :name :wild :type :wild :version :wild)
2558 with the added benefit that this is more likely to be portable.
2559 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2560 option to OPEN now signals an error if the file being opened
2561 exists; this may have an impact on existing code.
2562 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2563 signals received, on the PowerPC platforms (both Linux and
2564 Darwin). (thanks to Pierre Mai for pointing out the location of
2566 * several fixes on OS X: The system now builds and runs cleanly on
2567 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2568 (thanks to Brian Mastenbrook)
2569 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2570 second-resolution: integer multiples of 1/3600 between -24 and 24.
2571 (thanks to Vincent Arkesteijn)
2572 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2573 type when called with 1 argument; PEEK-CHAR checked type of
2574 PEEK-TYPE only after having read first character from a
2575 stream. (reported by Peter Graves)
2576 * bug fix: the garbage collector now has much better locality
2577 behaviour, and in particular no longer treats all memory as being
2578 exhausted when many small objects point to each other in a deeply
2580 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2581 recognized as being TYPEP their class.
2582 * bug fix: the PUSHNEW documentation string has been corrected.
2583 (thanks to Vincent Arkesteijn)
2584 * bug fix: defaulting of the value for the last of an atypically
2585 large number of multiple values being bound was not being
2586 performed correctly on the Alpha or PPC platforms
2587 * optimization: implemented multiplication as a modular
2588 (UNSIGNED-BYTE 32) operation on the x86 backend.
2589 * optimization: SEARCH on simple-base-strings can now be open-coded.
2590 (see also contrib/compiler-extras.lisp for inspiration for
2591 teaching the compiler about the Boyer-Moore algorithm).
2592 * value, returned by MAX (and MIN) called with several EQUALP, but
2593 not EQL, arguments now does not depend on compiler settings.
2594 * fixed some bugs revealed by Paul Dietz' test suite:
2595 ** in stack analysis liveness information is propagated from
2596 non-local entry points.
2597 ** pathwise CAST removing failed when the CAST node did not start
2599 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2600 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2602 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2603 host is already defined.
2604 ** RENAME-FILE works on streams instead of signalling an internal
2606 ** PEEK-CHAR uses the current readtable when determining whether
2607 or not a character is whitespace.
2608 ** MERGE-PATHNAMES handles the case when the pathname does not
2609 specify a name while the default-pathname specifies a version.
2610 ** Pathnames now stand a better chance of respecting print/read
2612 ** Attempting to use standardized file system operators with a
2613 pathname with invalid :DIRECTORY components signals a
2615 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2617 ** DIRECTORY on logical pathnames is more correct.
2618 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2619 signal a TYPE-ERROR if their argument is not a stream.
2620 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2621 designator argument does not designate a stream.
2622 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2623 examining the synonym.
2624 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2626 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2627 element-type larger than ([UN]SIGNED-BYTE 32).
2629 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2630 * When built with the :SB-FUTEX feature, threaded builds now take
2631 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2632 for faster/more reliable mutex and condition variable support.
2633 * Incompatible change (but one you probably shouldn't have been using
2634 anyway): the interface and code for arbitrating between multiple
2635 threads in the same user session has been redesigned.
2636 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2637 error when called without an explicit environment argument.
2638 (thanks to Nikodemus Siivola)
2639 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2640 confused about their position. (thanks to Adam Warner and Gerd
2642 * bug fix: Pretty printing backquoted forms with unquotations in the
2643 argument list position of various code constructs such as LAMBDA
2644 now works correctly. (reported by Paul Dietz)
2645 * bug fix: Pretty printing unquotations no longer loses all
2646 stream position information.
2647 * optimization: performance of string output streams is now less
2648 poor for multiple small sequence writes.
2649 * optimization: performance of CSUBTYPEP in the presence of complex
2650 expressions involving CONS and NOT many times has been improved.
2651 (reported by Paul Dietz)
2652 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2654 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2655 streams for the REPL, for the debugger, and so on. (thanks to
2657 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2658 * fixed some bugs revealed by Paul Dietz' test suite:
2659 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2660 is ignored if it is too small, rather than propagating through
2661 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2662 ** extremely complex negations of CONS types were not being
2663 sufficiently canonicalized, leading to inconsistencies in
2665 ** VALUES tranformer lost derived type.
2667 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2668 * fixed a bootstrapping bug: the build process no longer assumes
2669 that the various BOOLE-related constants have the same value in
2670 host and target lisps. (noted by Paul Dietz' test suite on an
2671 SBCL binary built from CLISP)
2672 * The system can now be dynamically linked on the MIPS platform,
2673 which enables dynamic loading of foreign code from Lisp. (thanks
2674 to Ralf Baechle for discussions on the MIPS ABI)
2675 * The system now records debugging information for its own source
2676 files in a filesystem-position-independent manner, relative to
2677 the "SYS" logical host.
2678 * fixed a compiler bug: MV-LET convertion did not check references
2679 to the "max args" entry point. (reported by Brian Downing)
2680 * tweaked disassembly notes to be less confident about proclaiming
2681 some instruction as an LRA. (thanks to Brian Downing)
2682 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2683 now each have their own history, command character, and other
2684 characteristics. (thanks to David Lichteblau)
2685 * fixed some bugs revealed by Paul Dietz' test suite:
2686 ** compiler failure in compiling LOGAND expressions including a
2688 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2689 shift greater than 32.
2690 ** FLUSH-DEST did not mark blocks for type check regeneration.
2691 ** HANDLER-CASE failed to accept declarations in handler clauses
2692 in some circumstances.
2694 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2695 * New code in contrib/sb-introspect (still probably not entirely
2696 stable yet) provides some support for smart Lisp development
2697 environments like SLIME.
2698 * The conditions signalled for errors occurring when loading .fasl
2699 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2700 in a way which should help ASDF recover gracefully.
2701 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2702 stuff has been cleaned up. If you code contrib/ stuff, this might
2703 affect you, and you can look at contrib/README, contrib/STANDARDS,
2704 and/or the 0.8.4.27 diff to check.
2705 * In full calls the compiler now does not generate checks for declared
2706 argument types for all arguments.
2707 * various threading fixes
2708 ** and some experimental patches which didn't make it into
2709 the main tree for this release, but which are shipped in
2710 contrib/experimental-thread.patch as a possible fix for some
2711 failures (deadlock, spinning...) in GC-intensive multithreaded
2713 * fixed PPC build problem (source code incompatibility of different
2714 library versions): added offsetof() hackery which attempts to divine
2715 where glibc maintainers put uc_mcontext today
2716 * fixed bug 282: compiler does not trust type assertions while passing
2717 arguments to a full call.
2718 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2719 &OPTIONAL VALUES type parameter.
2720 * fix bug 214: algorithm for noting rejected templates is now more
2721 similar to that of template seletion. (also reported by rydis on
2723 * fixed bug 141b: printing backquoted information readably and prettily
2724 inserts a space where necessary.
2725 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2726 (CHARACTER 10) are now reported as errors, rather than propagated
2727 as unknown types. (reported by piso on #lisp)
2728 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2729 with values NIL and :ERROR. (thanks to Milan Zamazal)
2730 * fixed bug 191c: CLOS now does proper keyword argument checking as
2731 described in CLHS 7.6.5 and 7.6.5.1.
2732 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2733 longer raise an error; further, using a list as a for-as-arithmetic
2734 counter now raises a meaningful error.
2735 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2736 now understood by sequence creation functions such as MAKE-SEQUENCE
2738 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2739 a STRING-INPUT-STREAM.
2740 * compiler enhancement: SIGNUM is now better able to derive the type
2742 * type declarations inside WITH-SLOTS are checked. (reported by
2744 * fixed some bugs revealed by Paul Dietz' test suite:
2745 ** incorrect optimization of TRUNCATE for a positive first
2746 argument and negative second.
2747 ** compiler failure in let-convertion during flushing dead code.
2748 ** compiler failure while deriving type of TRUNCATE on an
2749 interval, containing 0.
2750 ** ASH of a negative bignum by a negative bignum count now returns
2752 ** intersection of CONS types now canonicalizes properly, fixing
2753 inconsistencies in SUBTYPEP.
2755 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2756 * incompatible change: The --disable-debugger command line
2757 option now clobbers the debugger at a more fundamental
2758 level, by redefining #'INVOKE-DEBUGGER instead of by
2759 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2760 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2761 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2762 enter the debugger REPL and then suffer recursive errors
2763 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2764 while under the new system, BREAK in --disable-debugger mode
2765 terminates the system just as an unhandled error would.
2766 * fixed compiler performance when processing loops with a step >1;
2767 * bug fix: DOCUMENTATION now retrieves generic function
2768 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2769 support has been systematized, and now supports the methods
2770 specified by ANSI, along with a default method and a method for
2771 slot documentation. (reported by Nathan Froyd)
2772 * bug fix: effective methods associated with a generic function are
2773 no longer cached over a change of that generic function's method
2774 combination. (reported by Andreas Fuchs)
2775 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2776 names a non-existent directory. (thanks to Andreas Fuchs)
2777 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2778 the PPC platform, lead to stack corruption; this has been fixed.
2779 (reported by Rainer Joswig)
2780 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2781 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2782 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2783 arithmetic in general, is now much more reliable on the PPC
2785 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2786 platform now returns the right answer.
2787 * optimization: restored some effective method precomputation in
2788 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2789 precomputation is now tunable.
2790 * optimization: compiler-internal data structure use has been
2791 reviewed, and changes have been made that should improve the
2792 performance of the compiler by about 20%.
2793 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2794 simple-base-strings and simple-bit-vectors is improved.
2795 * optimization: the optimization of 32-bit logical and arithmetic
2796 functions introduced in version 0.8.3 on the x86 has been
2797 implemented on the mips, ppc and sparc platforms; an
2798 implementation of the same facility, but for 64-bit arithmetic,
2799 has been added for the alpha.
2800 * microoptimization: the compiler is better able to make use of the
2801 x86 LEA instruction for multiplication by constants.
2802 * bug fix: in some situations compiler did not report usage of
2803 generic arithmetic in (SPEED 3) policy.
2804 * bug 145b fix: compiler used wrong type specifier while converting
2805 MEMBER-types to numeric.
2806 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2808 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2809 index argument now works properly on the Alpha platform.
2810 * bug fix: floating point exception treatment on the Alpha platform
2812 * bug fix: FILE-POSITION works much better on string input and
2813 output streams. (thanks to Nikodemus Siivola)
2814 * bug fix: many threading/garbage collection symptoms sorted.
2815 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2816 might be pseudo-atomic.
2817 * internal change: Stopping for GC is now done with signals not
2818 ptrace. GC is now done in whichever thread wanted it, instead of
2820 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2822 * bug fix: result form in DO is not contained in the implicit
2824 * incompatible change: ICR structure is changed; the value part of
2825 CONTINUATION is now called LVAR; corresponding functions are
2826 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2827 * added type deriver for ISQRT (thanks to Robert E. Brown).
2828 * bug fix: better support for loading from the command line when an
2829 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2831 * fixed some bugs revealed by Paul Dietz' test suite:
2832 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2833 ** ROUND and FROUND now give the right answer when given very
2834 small float arguments.
2835 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2837 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2838 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2839 ** LCM should return a non-negative integer.
2840 ** PARSE-INTEGER returned the index of a terminator instead of the
2841 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2842 ** PARSE-INTEGER returned an incorrect index being applied to a
2844 ** LCM with two arguments of 0 returns 0 rather than signalling
2846 ** unsigned addition of a 32-bit constant with the high bit set no
2847 longer causes an internal compiler error on the x86.
2848 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2849 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2850 with negative last argument.
2851 ** byte specifiers with zero size and position no longer cause
2852 an error during type derivation.
2853 ** bignum multiplication on the Alpha platform now returns the
2855 * porting: The system now builds on SuSE AMD64, although it still
2856 generates a 32-bit binary.
2857 * .fasl file incompatibility: The fasl file version number has
2858 been incremented (because of the changes to internal compiler
2859 data structures referred to above).
2861 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2862 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2863 more accurately, on the Darwin kernel running on PowerPC hardware.
2864 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2865 * Compiler code deletion notes now signal a condition of type
2866 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2867 an associated MUFFLE-WARNING restart.
2868 * The compiler now performs limited argument count validation of
2869 constant format strings in FORMAT, and where appropriate in ERROR,
2870 CERROR and WARN. (thanks to Gerd Moellmann)
2871 * New ASDF-INSTALL contrib can be used for automatic download and
2872 installation of third-party Lisp code from CCLAN or other sites
2874 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2875 which forces another thread to execute a function supplied by the
2877 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2878 now accept and act upon their :ELEMENT-TYPE keyword argument.
2879 (reported by Martin Atzmueller, Edi Weitz)
2880 * bug fix: FILE-POSITION now accepts position designators up to
2881 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2882 is the greater. (thanks to Patrik Nordebo)
2883 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2884 Kalle Olavi Niemitalo)
2885 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2886 printable. (reported by Eric Marsden)
2887 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2888 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2889 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2890 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2891 (thanks to Stig Sandoe)
2892 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2893 UNREAD-CHAR now correctly includes the unread character in the
2894 target sequence. (thanks to Gerd Moellmann)
2895 * bug fix in threaded builds: the system can now be suspended and
2896 resumed by shell job control with minimal disruption.
2897 * bug fixes in times and timezones >2038AD
2898 * better handling of "where is GNU make?" problem in build scripts
2899 (thanks to Nikodemus Siivola)
2900 * new optimization: inside a named function any reference to a
2901 function with the same name is considered to be a self-reference;
2902 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2903 optimization quality.
2904 * new optimization on x86: logical functions and + now have
2905 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2906 used when the result is truncated to 32 bits.
2907 * VALUES declaration is partially enabled.
2908 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2909 * fixed some bugs revealed by Paul Dietz' test suite:
2910 ** The system now obeys the constraint imposed by
2911 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2912 types form a lattice under type intersection.
2913 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2914 ** ASSOC now ignores NIL elements in an alist.
2915 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2916 and (1+ MOST-POSITIVE-FIXNUM) answers.
2917 ** The addition of a method with invalid qualifiers to a generic
2918 function does not cause an error to be signalled immediately;
2919 a warning is signalled, and the error is generated only on
2920 calling the generic function.
2921 * changed .fasl file version number, in part to add type codes for
2922 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2923 obscure ANSI requirements
2925 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2926 * fixed bug 148: failure to inline-expand a local function left
2927 garbage, confusing the compiler.
2928 * fixed bugs 3cd: structure slot readers perform type check if the
2929 slot can have an invalid value (i.e. it is either not initialized
2930 or can be written with a less specific slot writer).
2931 * bug fix: the compiler now traps array references to elements off
2932 the end of an array; previously, the bounds checking in some
2933 circumstances could go off-by-one.
2934 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2936 * type declarations for array element types now obey the description
2937 on the CLHS page "Declaration TYPE", as per discussions on
2938 sbcl-help around 2003-05-08. This means that a declaration
2939 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2940 declaration, all references to BAR will be asserted or assumed
2941 (with THE, so dependent on compiler policy) to involve objects of
2942 type FOO. Note that no such declaration is implied in
2943 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2944 * declared types of functions from the "Conditions"
2945 chapter. (reported by Paul Dietz)
2946 * bug fix: CERROR accepts a function as its first argument.
2947 * bug fix: NTH an NTHCDR accept a bignum as index
2948 arguments. (reported by Adam Warner)
2949 * optimization: character compare routines now optimize comparing
2950 against a constant character. (reported by Gilbert Baumann)
2951 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2952 argument now works properly on the MIPS platform.
2953 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2954 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2955 declared to have a functional type, does not cause an error
2957 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2958 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2959 * fixed bug: if last continuation of a deleted block has a
2960 destination, this destination should be deleted too. (reported by
2962 * fixed a bug in the bootstrap process: the host compiler's values
2963 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2964 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2965 test case from Patrik Nordebo)
2966 * improved the ability of the disassembler on the PPC platform to
2967 provide helpful disassembly notes.
2968 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2969 the class in more cases than previously.
2970 * bug fix: FILE-POSITION now understands :START and :END for
2971 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2972 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2973 8) data. (thanks to David Lichteblau)
2974 * bug fix: it is possible to add a method to a generic function
2975 without lambda list.
2976 * bug fix: reader failed to signal END-OF-FILE inside an
2977 object representation. (reported by Nikodemus Siivola)
2978 * fixed some bugs revealed by Paul Dietz' test suite:
2979 ** LAST and [N]BUTLAST should accept a bignum.
2980 ** condition slot accessors are methods.
2981 ** (VECTOR NIL) is a subtype of STRING.
2983 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2984 * minor incompatible change: some nonsensical specialized lambda
2985 lists (used in DEFMETHOD) which were previously ignored now signal
2987 * minor incompatible change: the system is now aware of the types of
2988 variables in the COMMON-LISP package, and will signal errors for
2989 most violations of these type constraints (where previously they
2990 were silently accepted).
2991 * minor incompatible change: COMPILE-FILE now uses the freedom
2992 afforded (ANSI 3.2.2.3) to use derived function types for
2993 functions defined in the same file. This also permits the system
2994 to warn on static type mismatches and function
2995 redefinition. (Currently it does not work with high DEBUG level.)
2996 * minor incompatible change: VALUES declaration is disabled.
2997 * When issuing notes, the compiler now signals a condition of type
2998 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2999 restart for use in user handlers. It is expected that the
3000 COMPILER-NOTE condition will eventually become a condition
3001 supertype to a hierarchy of note types, which will then be
3002 handleable in a similar fashion. However, at the moment, no such
3003 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3004 but it's an implementation detail, not a classification for the
3006 * Changes in type checking closed the following bugs:
3007 ** type checking of unused values (192b, 194d, 203);
3008 ** template selection based on unsafe type assertions (192c, 236);
3009 ** type checking in branches (194bc).
3010 * A short form of VALUES type specifier has ANSI meaning (it has
3011 increased the number of situations when SBCL cannot perform type
3013 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3014 keywords or constants is permissible.
3015 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3016 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3017 classes. (thanks to Antonio Martinez)
3018 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3019 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3020 argument to a pretty-print stream. Instead, it leaves any such
3021 operations to DESCRIBE-OBJECT methods.
3022 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3024 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3025 lambda lists. (thanks to David Lichteblau)
3026 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3027 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3028 (thanks to Gerd Moellmann)
3029 * fixed bug 246: increased compilation speed of long
3030 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3032 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3034 * DEFCONSTANT now throws a condition of type
3035 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3036 constant to a non-EQL value; CONTINUE and ABORT restarts
3037 respectively change and preserve the value.
3038 * fixed bug 63: The code walker, part of the implementation of CLOS,
3039 is now better at handling symbol macros.
3040 * bug fix: There is no longer an internal implementation type named
3041 CL:LENGTH. (reported by Raymond Toy)
3042 * bug fix: In macro-like defining macros/special operators the
3043 implicit block does not enclose the lambda list.
3044 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3045 enclosed in parentheses) are not suitable as type specifiers, and
3046 their use properly signals an error now.
3047 * bug fix: An argument count mismatch for a type specifier in code
3048 being compiled no longer causes an unhandled error at compile
3049 time, but signals a compile-time warning.
3050 * fixed simple vector readable printing
3051 * bug fix: DESCRIBE takes more care over whether the class
3052 precedence list slot of a class is bound before accessing it.
3053 (reported by Markus Krummenacker)
3054 * bug fix: FORMATTER can successfully compile pretty-printer format
3055 strings which use variants of the ~* directive inside.
3056 * bug fix: SEARCH now applies its TEST predicate to the elements of
3057 the arguments in the correct order. (thanks to Wolfhard Buss)
3058 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3059 in type inference. (thanks to Robert E. Brown)
3060 * bug fix: Reading in symbols with an explicit package name of ""
3061 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3062 * fixed some bugs revealed by Paul Dietz' test suite:
3063 ** NIL is now allowed as a structure slot name.
3064 ** Arbitrary numbers, not just REALs, are allowed in certain
3065 circumstances in LOOP for-as-arithmetic clauses.
3066 ** Multiple class redefinitions before slot access no longer
3067 causes a type error.
3068 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3069 association between the name and a class.
3070 ** Generic functions with non-standard method-combination and over
3071 five methods all of which return constants no longer return NIL
3072 after the first few invocations. (thanks to Gerd Moellmann)
3073 ** CALL-NEXT-METHOD with no arguments now passes the original
3074 values of the arguments, even in the presence of assignment.
3075 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3077 ** DISASSEMBLE works with closures and funcallable instances.
3078 ** ADD-METHOD now returns the generic function, not the new method.
3079 ** FIND-METHOD signals an error if the lengths of the specializers
3080 is incompatible with the generic function, even if the ERRORP
3082 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3083 which its argument is a member.
3084 ** DEFCLASS only redefines the class named by its class-name
3085 argument if that name is the proper name of the class;
3086 otherwise, it creates a new class.
3087 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3088 of the UNBOUND-SLOT condition to the name of the slot.
3089 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3090 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3091 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3092 SLOT-MAKUNBOUND in the specified fashion.
3094 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3095 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3096 cross-compilation host. As a consequence, we can now bootstrap our
3097 way up to SBCL starting with a bare gcc toolchain and human-readable
3098 source code (first the source to CLISP, then the source to SBCL).
3099 * A contributed module containing a partial implementation of the
3100 simple-streams interface has been included. (thanks to Rudi
3102 * A contributed module implementing the RFC1321 Message Digest
3103 Algorithm, known as MD5, has been included.
3104 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3105 no longer has any effect, as the code controlled by this feature
3106 has been deleted. (As far as we know, no-one has ever built using
3107 this feature, and its semantics were confused in any case).
3108 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3109 work on conditions (as required by the ANSI specification),
3110 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3111 expected behaviour on conditions. Users should note, however,
3112 that such behaviour is not required by the ANSI specification,
3113 and so use of this behaviour may render their code unportable.
3114 * fixed some bugs revealed by Paul Dietz' test suite:
3115 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3117 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3118 ** SXHASH on condition objects no longer returns NIL.
3119 ** :ALLOCATION :CLASS slots are better treated; their values are
3120 updated on class redefinition, and initforms inherited from
3121 superclasses are applied.
3122 ** REMOVE-METHOD returns its generic function argument even when
3123 no method was removed.
3124 ** SHARED-INITIALIZE now initializes the values of the requested
3125 slots, including those with :ALLOCATION :CLASS.
3126 ** ALLOCATE-INSTANCE now works on structure classes defined via
3127 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3129 ** SLOT-EXISTS-P now works on conditions, as well as structures
3131 ** MAKE-LOAD-FORM now has the required methods on
3132 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3133 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3134 keyword, and now implements the SLOT-NAMES argument.
3135 ** methods with &OPTIONAL arguments no longer allow too many
3136 arguments to be passed in the call without error.
3137 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3138 option is consistent with the required arguments of the generic
3139 function lambda list.
3140 * bug fix: REQUIRE accepts a string designator. (Thanks to
3142 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3143 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3144 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3145 * bug fix: The compiler now checks for duplicated variables in macro
3147 * bug fix: SETQ on globals returns the correct value.
3148 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3149 not a COMPILER-ERROR (followed by some other strange error on
3150 choosing the CONTINUE restart).
3151 * bug fix: make.sh and friends are now more consistent in the way that
3152 they look for GNU "make".
3154 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3155 * experimental native threads support (on x86 Linux >=2.4 only).
3156 This is not compiled in by default: you need to add :SB-THREAD to
3157 the target features. See the "Beyond ANSI" chapter of the manual
3159 * fix for longstanding nonANSIism: The old distinction between
3160 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3161 The return value from CL:FIND-CLASS is now a CLOS class, and
3162 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3163 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3164 * An interface to the MetaObject Protocol, as described in Kiczales,
3165 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3166 MIT Press, 1991, is available from the SB-MOP package.
3167 * incompatible change: the SB-PCL package should now be considered
3168 a private implementation detail, and no longer a semi-private MOP
3170 * minor incompatible change: due to rearrangement for threads, the
3171 control stack and binding stack are now allocated at arbitrary
3172 addresses instead of being hardcoded per-port. Users affected by
3173 this probably have to be doing advanced things with shared
3174 libraries, and will know who they are.
3175 * minor incompatible change: Previously, all --eval forms used were
3176 processed with READ before any of them were processed with EVAL.
3177 Now each --eval form is processed with both READ and EVAL before
3178 the next --eval form is processed. (Thus package operations like
3179 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3180 work as the user might reasonably expect.)
3181 * minor incompatible change: *STANDARD-INPUT* is now only an
3182 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3184 * minor incompatible change: Y-OR-N-P is now character-oriented,
3185 not line oriented. Also, YES-OR-NO-P now works without errors.
3186 (thanks to Antonio Martinez)
3187 * sb-aclrepl module improvements: an integrated inspector, added
3188 repl features, and a bug fix to :trace command.
3189 * Known functions, which cannot be open coded by the backend, are
3190 considered to be able to check types of their arguments. (fixing
3191 a bug report by Nathan J. Froyd)
3192 * fixed a bug in computing method discriminating functions: It is
3193 now possible to define methods specialized on classes which have
3194 forward-referenced superclasses. (thanks to Gerd Moellmann)
3195 * fixed evaluation order in optional entries (reported by Gilbert
3197 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3198 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3199 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3200 required-by-AMOP NAME argument, as well as CLASS and
3201 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3202 * fixed bug 20: DEFMETHOD can define methods using names that are
3203 not the proper names of classes to designate class specializers.
3204 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3205 Unix file descriptors, instead of blowing up. (thanks to Antonio
3207 * Garbage collection refactoring: user-visible change is that a
3208 call to the GC function during WITHOUT-GCING will not do garbage
3209 collection until the end of the WITHOUT-GCING. If you were doing
3210 this you were probably losing anyway.
3211 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3212 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3213 TYPEP the latter but not the former.
3214 * The compiler issues a full WARNING for calls to undefined functions
3215 with names from the CL package.
3216 * MAP-INTO for a vector destination is open coded. (reported by
3217 Brian Downing on c.l.l)
3218 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3219 documentation string.
3220 * fixed some bugs revealed by Paul Dietz' test suite:
3221 ** COPY-ALIST now signals an error if its argument is a dotted
3223 ** Condition slots are now accessed more correctly in the presence
3224 of multiple initargs for a given slot.
3225 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3226 exclude restarts of the same name associated with a different
3228 ** DEFCLASS of forward-referenced classes with another
3229 forward-referenced class in the superclasses list no longer
3231 ** Condition slots are now initialized once each, not multiple
3232 times. (thanks to Gerd Moellmann)
3233 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3235 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3236 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3237 arguments contain duplicated elements.
3238 ** RESTART-CASE understands local macros.
3239 ** RESTART-CASE associates exactly its own restarts with a condition.
3240 ** ENDP in safe mode checks its argument to be of type LIST.
3241 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3242 in question is unbound.
3243 ** Optimized MAKE-INSTANCE functions no longer cause internal
3244 assertion failures in the presence of duplicate initargs.
3245 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3246 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3247 propagated even in the ctor optimized implementation of
3249 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3251 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3252 * a better implementation of SXHASH on (simple) bit vectors,
3253 measured both in execution speed and in distribution of results
3254 over the positive fixnums, has been installed. Likewise, a better
3255 implementation of EQUAL for simple bit vectors is now available.
3256 * fixed CEILING optimization for a divisor of form 2^k.
3257 * fixed bug 240 (emitting extra style warnings "using the lexical
3258 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3259 by Antonio Martinez)
3260 * fixed SXHASH, giving different results for NIL depending on type
3261 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3262 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3263 variable. (found by Rolf Wester)
3264 * SBCL does not ignore type declarations for special
3265 variables. (reported by rif on c.l.l 2003-03-05)
3266 * some bug fixes in contrib/sb-aclrepl/
3267 * fixed some bugs revealed by Paul Dietz' test suite:
3268 ** a bug in the CONS type specifier, whereby the CAR and CDR
3269 types got intertwined, has been fixed;
3270 ** the type system is now able to reason about the interaction
3271 between INTEGER and RATIO types more completely;
3272 ** APPEND, [N]REVERSE and NRECONC check that those their
3273 arguments, which must be proper lists, are really so;
3274 ** An array specialized to be unable to hold elements has been
3275 implemented, as required -- yes, really -- by ANSI;
3276 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3277 on malformed property lists;
3279 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3280 * incompatible packaging change: in line with Unix convention,
3281 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3282 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3283 * REQUIRE and PROVIDE are now optionally capable of doing something
3284 useful. See the documentation string for REQUIRE.
3285 * infrastructure for a managed SBCL contrib system: contributed
3286 modules in this release include:
3287 ** the ASDF system definition facility;
3288 ** an interface to the BSD Sockets API;
3289 ** an ACL-like convenience interface to the repl;
3290 (thanks to Kevin Rosenberg)
3291 ** an implementation of ROTATE-BYTE, with efficient implementation
3293 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3294 gives a closer approximation to the right answer than previously.
3295 (thanks to Raymond Toy)
3296 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3297 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3298 optional environment arguments, as required by ANSI.
3299 * fixed bugs in other functions taking environment objects, allowing
3300 calls with an explicit NIL environment argument to be compiled
3302 * fixed bug 228: primary return values from
3303 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3304 COMPILE or FUNCTION.
3305 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3306 :TYPE will no longer signal errors on innocuous objects.
3307 * fixed bug 231b: SETQ is better at respecting type declarations in
3308 the lexical environment.
3309 * fixed a bug in DEFCLASS: classes named by symbols with no or
3310 unprintable packages can now be defined.
3311 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3312 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3313 * fixed compiler failure related to checking types of functions.
3314 (reported by Robert E. Brown)
3315 * the compiler is now much more consistent in its error-checking
3316 treatment of bounding index arguments to sequence functions: in
3317 (SAFETY 3) code, errors will be signalled in almost all cases if
3318 invalid sequence bounding indices are passed to functions defined
3319 by ANSI to operate on sequences.
3320 * fixed a bug in the build procedure: documentation of SBCL-specific
3321 packages is now preserved and available in the final Lisp image.
3322 * lifted FDEFINITION lookup out of loops in the implementation of
3323 many list operations. (thanks to Robert E. Brown)
3324 * fixed a bug in the reader: the #n# reader macro now works for
3325 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3326 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3327 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3328 E. Brown and Tony Martinez respectively)
3329 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3330 SLOT-BOUNDP: in particular, it is now able to optimize them much
3331 better, and is now not vulnerable to having packages renamed.
3332 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3333 always called when a slot is not present in an instance. (thanks
3335 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3336 longer signal an error on unreading a character following EOF on
3337 the previous constituent stream. (thanks to Tony Martinez)
3338 * fixed some bugs revealed by Paul Dietz' test suite:
3339 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3340 not just nonnegative fixnums;
3341 ** the logical bit-array operators such as BIT-AND now accept an
3342 explicit NIL for their "opt-arg" argument (to indicate a
3343 freshly-consed result bit-array);
3344 ** ELT now signals an error on an invalid sequence index in safe
3346 ** the type system is now cleverer about negations of numeric
3347 types, and consequently understands the BIGNUM and RATIO types
3349 ** the type system is now cleverer about the interaction between
3350 INTEGER and RATIO types: while bugs still remain, many more
3351 cases are accurately computed;
3352 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3353 if it is in the last clause;
3354 ** CONSTANTLY now correctly returns a side-effect-free function in
3356 ** DECLARE is no longer treated as a special-operator; in
3357 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3358 * incremented fasl file version number due to the change in the
3359 DEFSTRUCT-SLOT-DESCRIPTION structure.
3361 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3362 * minor incompatible change: code processed by the "interpreter" or
3363 EVAL now has a compilation optimization policy of (DEBUG 2)
3364 (changed from (DEBUG 1)) to improve debuggability of interactive
3365 development, and to allow the use of the debug RETURN command in
3367 * an experimental implementation of the RETURN command for the
3368 debugger has been included. (thanks to Frederik Kuivinen)
3369 * fixed bug 62: constraints were not propagated into a loop.
3370 * fixed bug in embedded calls of SORT (reported and investigated by
3372 * fixed some bugs revealed by Paul Dietz' test suite:
3373 ** printing and reading of arrays with some dimensions having
3374 length 0 (thanks to Gerd Moellmann);
3375 ** BOA constructor with &AUX argument without a default value does
3376 not cause a type error;
3377 ** CONSTANTP now returns true for all self-evaluating objects.
3379 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3380 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3381 accessors that are related by inheritance, as specified in the
3382 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3383 Valtteri Vuorikoski)
3384 * The compiler is now able to inline functions that were defined in
3385 a complex lexical environment (e.g. inside a MACROLET).
3386 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3388 * Support for the upcoming FreeBSD-5.0 release has been included.
3389 (thanks to Dag-Erling Smorgrav)
3390 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3391 effect when it is not in a toplevel context.
3392 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3393 stand a better chance of being correct. (thanks to Gerd
3395 * fixed bug in COERCE, which now signals an error on coercing a
3396 rational to a bounded real type which excludes the expected
3398 * The compiler is now able to derive types more accurately from the
3399 COERCE and COMPILE functions.
3400 * fixed bug 223: functional binding is considered to be constant
3401 only for symbols in the CL package.
3402 * fixed bug 231: SETQ did not check the type of a variable being set
3403 (reported by Robert E. Brown)
3404 * A new optimization for MAKE-INSTANCE has been included, fixing
3405 various bugs (including relating to :ALLOCATION :CLASS slots and
3406 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3408 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3409 ** As required by ANSI, LOOP now disallows anonymous collection
3410 clauses such as COLLECT I in conjunction with aggregate boolean
3411 clauses such as THEREIS (= I 1);
3412 ** LOOP now signals an error when any variable is reused in the
3413 same loop (including the potentially useful construct analogous
3414 to WITH A = 1 WITH A = (1+ A);
3415 ** IT is only a special loop symbol within the first clause of a
3416 conditional loop clause;
3417 ** LOOP with a typed iteration variable over a hashtable now
3418 signals a type error iff it should.
3419 * fixed some other bugs revealed by Paul Dietz' test suite:
3420 ** FILE-STREAM now names the class previously known as FD-STREAM;
3421 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3422 argument) no longer signals an error;
3423 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3424 of the slot symbol, rather than using the current package
3425 ((:CONC-NAME "") continues to intern the slot's name in the
3427 * incremented fasl file version number, because of the incompatible
3428 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3429 of the new implementation of DEFINE-COMPILER-MACRO.
3431 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3432 * Support for building SBCL for MIPS platforms running in
3433 little-endian mode has now been checked in, and basic
3434 functionality on said platforms verified.
3435 * minor incompatible change: PCL now records the pathname of a file
3436 in which methods and the like are defined, rather than its
3438 * minor incompatible change: TRUENAME now considers the truename of
3439 a file naming a directory to be the pathname with :DIRECTORY
3440 component indicating that directory.
3441 * minor incompatible change: a NAMED clause in the extended form of
3442 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3443 reason for the previous behaviour is unclear.
3444 * more systematization and improvement of CLOS and MOP conformance
3445 in PCL (thanks to Gerd Moellman and Pierre Mai):
3446 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3448 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3449 primary methods with no specializers;
3450 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3452 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3453 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3454 FORWARD-REFERENCED-CLASSes; error reporting on
3455 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3457 ** SXHASH on CLOS instances now uses a slot internal to the
3458 instance to return different numbers on distinct instances,
3459 while preserving the same return value through invocations of
3461 ** DEFMETHOD signals errors when methods with longer incongruent
3462 lambda lists are added to generic functions;
3463 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3464 CLASS, as specified in AMOP;
3465 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3466 on the order returned by the primary method for classes of
3467 class STANDARD-CLASS;
3468 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3469 * fixed some bugs shown by Paul Dietz' test suite:
3470 ** DOLIST puts its body in TAGBODY;
3471 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3473 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3474 value producing form;
3475 ** if more variables are given to PROGV than values, extra
3476 variables are bound and made to have no value;
3477 ** NSUBSTITUTE on list arguments gets the right answer with
3479 ** ELT signals an error of type TYPE-ERROR when the index argument
3480 is not a valid sequence index;
3481 ** LOOP signals (at macroexpansion time) an error of type
3482 PROGRAM-ERROR when duplicate variable names are found;
3483 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3484 ** FUNCALL of special-operators now cause an error of type
3486 ** PSETQ now works as required in the presence of side-effecting
3487 symbol-macro places;
3488 ** NCONC accepts any object as its last argument;
3489 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3491 ** loop-for-as-package does not require a package to be explicitely
3493 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3495 * fixed bug 166: compiler preserves "there is a way to go"
3496 invariant when deleting code.
3497 * fixed bug 172: macro lambda lists with required arguments after
3498 &REST arguments now cause an error to be signalled. (thanks to
3500 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3501 bug. (thanks to Matthew Danish)
3502 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3504 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3505 arguments are lexically rebound. (thanks to Gerd Moellmann and
3507 * fixed bug 194: error messages are now more informative when there
3508 is no primary method applicable in a call to a generic
3509 function. (thanks to Gerd Moellmann)
3510 * fixed bug in command line argument checking (thanks to Julian
3512 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3513 (thanks to Matthew Danish)
3514 * incremented fasl file version number, because of the
3515 SXHASH-related changes in the layout of CLOS data structures
3517 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3518 * minor incompatible change: The runtime (the Unix executable named
3519 "sbcl") is now much pickier about the .core files it will load.
3520 Essentially it now requires .core files to descend from the same
3521 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3522 as the runtime does. (The intent is to prevent the crashes which
3523 can occur, and which can even be reported as mysterious failures,
3524 when people patch the sources or change the build parameters
3525 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3526 sbcl and .core files.)
3527 * fixed bug: VALUES-LIST is no longer optimized away.
3528 * fixed bug 142: The FFI conversion of C string values to Lisp
3529 string values no longer conses excessively. (thanks to Nathan
3530 Froyd porting Raymond Toy's fix to CMU CL)
3531 * began to systematize and improve MOP conformance in PCL (thanks to
3532 Nathan Froyd, Gerd Moellman and Pierre Mai):
3533 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3535 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3536 ** FINALIZE-INHERITANCE is now called on class finalization;
3537 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3538 argument precedence order.
3539 * fixed bug 202: The compiler no longer fails on functions whose
3540 derived types contradict their declared type.
3541 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3542 so it can be non-toplevel.
3543 * The fasl file version number has changed (because of the new
3544 implementation of DEFMACRO).
3545 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3546 safe code, that any length requirement by their type-specifier
3547 argument is valid. The exceptions to this are described in bug
3549 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3550 functions now check the types of their inputs as required by ANSI.
3551 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3552 introduced symbol is DECLAREd to be SPECIAL.
3553 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3554 * fixed bug: :COUNT argument to sequence functions may be negative
3555 * fixed bug: body of DO-SYMBOLS may contain declarations
3556 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3557 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3558 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3559 symbol macro only once
3560 * fixed printing of call frame when argument list is unavailable
3561 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3562 * compiler no longer signals WARNING on unknown keyword
3565 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3566 * A beta-quality port to the mips architecture running Linux,
3567 based on the old CMUCL backend, has been made. It has been tested
3568 on a big-endian kernel, and works sufficiently well to be able to
3569 rebuild itself; it has not been tested in little-endian mode.
3570 * fixed an inconsistency between gencgc.c and purify.c which made
3571 dumping/loading .core files unreliable
3572 * fixed bug 120a: The compiler now deals correctly with IFs where
3573 the consequent is the same as the alternative, instead of
3574 misderiving the return type. (thanks to Alexey Dejneka)
3575 * fixed bug 113: Logical pathnames are now dumpable (the logical
3576 host is resolved at load-time, throwing an error if it is not
3578 * fixed bug 174: FORMAT's error message is slightly clearer when a
3579 non-printing character is used in a format directive.
3580 * fixed several bugs in compiler checking of type declarations, i.e.
3581 violations of the Python "declarations are assertions" principle
3582 (thanks to Alexey Dejneka)
3583 * fixed several bugs in PCL's error checking (thanks to Gerd
3585 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3587 * fixed bug in compilation of functions as first class values
3588 (thanks to Antonio Martinez-Shotton)
3589 * The compiler's handling TYPE-ERRORs which it can prove will
3590 inevitably happen at runtime has been cleaned up and corrected
3591 in several ways. (thanks to Alexey Dejneka)
3592 * improved argument type checking for various basic arithmetic
3593 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3594 lavished on them in the past that they can be compiled in many
3595 ways in different special cases
3596 * fixed bug 181: compiler checks validity of user supplied type
3598 * cleaned up code flushing in optimization: Function calls which
3599 should signal errors for safety purposes (e.g. which ANSI says
3600 should signal errors when their arguments are of incorrect type)
3601 are no longer optimized away.
3602 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3603 * incremented fasl file version number, because changes in the
3604 implementation of sequence functions like COERCE caused
3605 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3606 in old inline expansions) to become undefined. (Actually these
3607 changes were later undone, so we might very well be binary
3608 compatible with 0.7.7 after all, but leaving the version number
3609 incremented seemed like the simplest and most conservative
3612 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3613 * An alpha-quality port to the parisc architecture running Linux,
3614 based on the old CMUCL backend, has been made. This, even more so
3615 than the other backends, should be considered still a work in
3616 progress; known problems include that the Linux kernel in 64-bit
3617 mode does not propagate the correct sigcontext structure to
3618 userspace, and consequently SBCL on a parisc64 kernel will not
3620 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3621 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3622 suppressed.) Also now that the compiler is looking at declarations
3623 in the environment, it checks optimization declarations as well,
3624 and suppresses inlining when (> DEBUG SPEED).
3625 * More fixes have been made to treatment of floating point exception
3626 treatment and other Unix signals. In particular, floating point
3627 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3628 * The detection and handling of control stack exhaustion (infinite
3629 or very deeply nested recursion) has changed. Stack exhaustion
3630 detection is now done by write-protecting pages at the OS level
3631 and applies at all optimization settings; when found, a
3632 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3633 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3634 be caught using IGNORE-ERRORS.
3635 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3636 correctly when transforming recursive LABELS functions to LETs.
3637 (thanks to Alexey Dejneka)
3638 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3639 that are names of constants or global variables.
3640 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3641 alien routines with docstrings.
3642 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3643 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3645 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3646 (thanks to Eric Marsden)
3647 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3648 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3649 * Bug fix: The SPARC backend can now compile functions involving
3650 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3651 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3652 to a functions expecting a PATHNAME-DESIGNATOR.
3653 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3654 lambda lists. (thanks to Alexey Dejneka)
3655 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3656 functions so that EXPORTing the name of the class doesn't cause
3657 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3658 to Antonio Martinez for reporting this)
3659 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3660 constant-folded without failing an assertion. (thanks to Einar
3661 Floystad Dorum for reporting this)
3662 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3663 types are now dealt with more correctly by the compiler.
3664 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3665 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3666 behaviour analogous to RENAME-FILE. This puts its behaviour more
3667 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3669 * The fasl file version number has changed again. (because of the
3670 bug fix involving the names of PCL MAKE-INSTANCE functions)
3672 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3673 * bug fix: Floating point exceptions are treated much more
3674 consistently on the x86/Linux and PPC/Linux platforms.
3675 * Array initialization with :INITIAL-ELEMENT is now much faster for
3676 cases when the compiler cannot open code the array creation, but
3677 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3678 array accesses have also seen a speed increase.
3679 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3680 specified. (This was at the root of some bad interactions between
3681 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3683 * bug fix: Internal error arguments for undefined functions are now
3684 computed correctly on the PPC/Linux platform.
3685 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3686 Raymond Toy's patch for CMU CL)
3687 * Support for the Solaris 9 operating environment has been included
3688 (thanks to Daniel Merritt)
3689 * A very ugly but hopefully complete draft of the missing FFI chapter
3690 of the manual has been created by reformatting the corresponding
3691 CMU CL manual chapter into (currently very ugly and incoherent)
3692 DocBook and bringing it up to date for SBCL behavior. Thus, the
3693 manual is now essentially complete, at least by my extreme
3694 once-and-only-once standards, whereby it's acceptable to refer to
3695 the doc strings of SB-EXT functions as the primary documentation.
3696 * The fasl file version number has changed again, due to cleanup of
3697 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3698 is no longer a static symbol.)
3700 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3701 * SBCL now builds with OpenMCL (version 0.12) as the
3702 cross-compilation host; also, more progress has been made toward
3703 bootstrapping under CLISP.
3704 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3706 * bug 158 fixed: The compiler can now deal with integer loop
3707 increments different from 1; fixing this turned out also to fix
3709 * bug 169 fixed: no more bogus warnings about using lexical bindings
3710 despite the presence of perfectly good SPECIAL declarations (thanks
3711 to David Lichteblau)
3712 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3713 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3714 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3716 * bug 180 fixed: Method combination specifications no longer ignore
3717 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3718 * bug fix: Structure type predicate functions now check their argument
3719 count as they should.
3720 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3721 correctly. (thanks to Pierre Mai)
3722 * minor incompatible change: The --noprogrammer option is deprecated
3723 in favor of the new --disable-debugger option, which is very similar.
3724 (The major difference is that it takes effect at a slightly different
3725 time at startup, causing handling of errors in --sysinit and
3726 --userinit files will be affected differently.) The
3727 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3728 been added to allow this functionality to be controlled from ordinary
3729 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3730 maintainers, who might want to run non-interactive scripts to
3731 build SBCL cores which will later be used interactively.)
3732 * minor incompatible change: The LOAD function no longer, when given
3733 a wild pathname to load, loads all files matching that pathname.
3734 Instead, an error of type FILE-ERROR is signalled.
3736 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3737 * bug 147 fixed: The compiler preserves its block link/count
3738 invariants more correctly now so that it doesn't crash. (thanks
3740 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3742 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3743 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3744 no-value-for-this-character as NIL (as specified by ANSI).
3745 * HOST-NAMESTRING on physical pathnames now returns a string that is
3746 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3747 (thanks to Christophe Rhodes)
3748 * The Alpha port handles icache flushing more correctly. (thanks to
3750 * More progress has been made toward bootstrapping under CLISP. (thanks
3751 to Christophe Rhodes)
3752 * The fasl file format has changed again, because dynamic loading
3753 on OpenBSD (which has non-ELF object files) motivated some cleanups
3754 in the way that foreign symbols are transformed and passed around.
3755 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3756 is no longer treated as whitespace by the reader, but instead as
3757 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3758 |A
\x7fB|, instead of A as it used to.
3760 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3761 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3762 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3763 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3764 this as of 0.7.1.45, but was left out of the previous news section
3765 (thanks to Dan Barlow)
3766 * SBCL now runs on the Solaris operating system on SPARC architectures
3767 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3768 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3769 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3770 * SPARC backend cleanups, allowing builds of cores optimized for V8
3771 and V9 SPARCS, and also emission of code targeted to a particular
3772 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3774 * SBCL is closer to bootstrapping under CLISP, thanks to various
3775 fixes by Christophe Rhodes.
3776 * The fasl file format has changed again, to allow the compiler's
3777 INFO database to support symbol macros.
3778 * The user manual (in doc/) is formatted into HTML more nicely.
3779 (thanks to coreythomas)
3780 * The system is smarter about SUBTYPEP relationships, especially
3781 those involving NOT types (including types such as ATOM which are
3782 represented internally using NOT types). Thus SUBTYPEP is less
3783 likely to return (VALUES NIL NIL) in general, and in particular
3784 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3786 * The fasl file format has changed again, because the internal
3787 representation of types now includes a new slot to support the new
3788 SUBTYPEP-of-NOT-types logic.
3789 * (not a change in the main branch of SBCL, but a related prototype
3790 which can hopefully be merged into the main branch of SBCL in the
3791 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3792 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3793 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3794 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3795 dispatch macro character combinations. (thanks to Alexey Dejneka)
3796 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3797 that superficially look like logical namestrings correctly.
3798 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3800 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3801 default supplied-p) for &optional and &key arguments. (thanks to
3804 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3805 * incompatible change: The compiler is now less aggressive about
3806 tail call optimization, doing it only when (> SPACE DEBUG) or
3807 (> SPEED DEBUG). (This is an incompatible change because there are
3808 programs which relied on the old CMU-CL-style behavior to optimize
3809 away their unbounded recursion which will now die of stack overflow.)
3810 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3811 for non-GENCGC systems has been increased to 20M (since that
3812 seems much closer to the likely performance optimum for modern
3813 systems than the old 4M value was)
3814 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3815 and *DEBUG-PRINT-LEVEL*
3816 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3817 of CMU CL's support for SPARC, and various endianness and other
3818 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3819 * new syntactic sugar for the Unix command line: --load foo.bar is now
3820 an alternate notation for --eval '(load "foo.bar")'.
3822 ** The system now detects stack overflow and handles it gracefully,
3823 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3824 optimization settings. (This is a good thing in general, and
3825 its introduction in this version should be particularly timely
3826 for anyone whose code fails because of suppression of tail
3828 ** The system now hunts for the C variable "environ" in a more
3829 devious way, to avoid segfaults when the C library version differs
3830 between compile time and run time. (thanks to Christophe Rhodes)
3831 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3832 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3833 ** The compiler no longer issues bogus style warnings for undefined
3834 classes in the same source file as the DEFCLASSes which defined
3835 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3837 ** fixes in CONDITION class precedence list for undefined function
3838 errors (thanks to Alexei Dejneka)
3839 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3840 correctly. (thanks to Dan Barlow)
3841 ** portability fixes aiming at bootstrapping under CLISP (thanks
3842 to Dave McDonald and Christophe Rhodes)
3843 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3844 ** fixes in type translation and and type inference (thanks to
3846 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3847 ** various fixes in the new ports (thanks to Dan Barlow)
3848 * several changes related to debugging:
3849 ** suppression of tail recursion, as noted above
3850 ** stack overflow detection, as noted above
3851 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3852 is now the default. (For some time encapsulation has been more
3853 reliable than the breakpoint-based :ENCAPSULATE NIL
3854 implementation, at least on X86 systems; and I just noticed that
3855 encapsulation also seems closer to the spirit of the ANSI
3858 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3860 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3861 up properly again. (There was a packaging bug in 0.7.0 which
3862 left their definitions in SB-SYS::LOAD-FOREIGN and
3863 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3864 vital for most things which interface to C-level interfaces,
3865 like extensions working with sockets or databases or
3866 Perl-compatible regexes or whatever, and the need to fix
3867 this bug is the main reason that 0.7.1 was released so
3869 ** DEFGENERIC is now choosier about the methods it redefines, so that
3870 reLOADing a previously-LOADed file containing DEFGENERICs does
3871 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3872 cycle now works as it should. (thanks to Alexey Dejneka)
3873 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3874 fixed by Christophe Rhodes. (He actually submitted this patch
3875 months ago, and I delayed until after 0.7.0.)
3876 ** Bug 111 (internal compiler confusion about runtime checks on
3877 FUNCTION types) was fixed by Alexey Dejneka.
3878 * Some internal cleanups (getting rid of variables which aren't
3879 needed now that the byte interpreter is gone) caused the fasl
3880 file format number to change again.
3882 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3883 * major incompatible change: The default fasl file extension, i.e. the
3884 default extension for files produced by COMPILE-FILE, has changed
3885 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3887 ** There are many changes in the implementation of the compiler.
3888 SBCL is now essentially a compiler-only implementation of ANSI
3889 Common Lisp. EVAL still "interprets" a few special cases, but
3890 almost all the interesting cases are handled by creating
3891 a LAMBDA expression, calling COMPILE on it, then calling
3892 FUNCALL on the result.
3893 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3894 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3895 Since the code is newer, there might still be some new bugs
3896 (though not as many as before Martin Atzmueller's fixes:-). But
3897 the new code is substantially simpler and clearer, and hopefully
3898 any remaining bugs will be simpler, less fundamental, and more
3899 fixable then the bugs in the old code.
3900 ** The revised compiler is still a little unsteady on its feet.
3902 *** The debugging information it produces (particularly the names
3903 of FUNCTION objects) is sometimes much less useful than what
3904 the old compiler produced.
3905 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3906 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3907 within a MACROLET) has been temporarily weakened.
3908 ** There are new compiler optimizations for various functions:
3909 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3910 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3911 *** the math functions TRUNCATE, FLOOR, and CEILING
3912 *** the function-of-all-trades COERCE
3913 Mostly these should be transparent, but there's one
3914 potentially-annoying problem (bug 117): when the compiler
3915 inline-expands a function and does type analysis on the result,
3916 it can create control paths which have type mismatches, and
3917 when it can't prove that those control paths aren't taken,
3918 it will issue WARNINGs about the type mismatches. This is
3919 a particular problem in practice for the new sequence functions.
3920 It's not clear how this should be fixed, and for now, a
3921 workaround is given in the entry for 117 in the BUGS file.
3922 ** (Because of the interaction between the two previous items --
3923 occasional inlining problems and new inline expansions -- some
3924 of the new sequence function optimizations won't really kick in
3925 completely until debugging information, and then inlining, are
3926 straightened out in some future version.)
3927 * minor incompatible changes:
3928 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3929 DIRECTORY no longer implicitly promotes NIL slots of its
3930 pathname argument to :WILD. In particular, when you ask for the
3931 contents of a directory (which you used to be able to do without
3932 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3933 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3934 ** changes in behavior that ANSI explicitly defines to be
3935 implementation dependent:
3936 *** The new compiler-only implementation still conforms with ANSI,
3937 but acts a little different than before. Besides the obvious
3938 changes in performance tradeoffs (that the cost per form passed
3939 to EVAL has gone up, and the cost per form executed by EVAL
3940 has gone down), the behavior of the system changes a little
3941 because there are no longer any interpreted function objects.
3942 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3943 e.g. doing COMPILE on the output of interactive DEFUN is
3945 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3947 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3948 has changed from (:CL) to NIL.
3949 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3950 e.g. #\Newline and #\Tab, don't have names specified in the
3951 ANSI Common Lisp standard, is now based on their ASCII symbolic
3952 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3953 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3954 are no longer used for output.
3955 ** changes in internal implementation constants:
3956 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3957 4 million. (If your application spends a lot of time GCing and
3958 you have a lot of RAM, you might want to experiment with
3959 increasing it even more.)
3960 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3961 However, almost all old code should still continue to work without
3962 immediate update, as SB-C-CALL is now a (deprecated) nickname
3964 ** Old operator names in the style DEF-FOO are now deprecated in
3965 favor of new corresponding names DEFINE-FOO, for consistency with
3966 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3967 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3968 internal symbols, but a few supported extensions like
3969 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3970 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3971 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3972 etc. as you get deeper into recursive calls to the debugger
3973 command loop, instead of the old "5]", "5]]", "5]]]"
3974 sequence. (I was motivated to do this when squabbles between
3975 ILISP and SBCL left me very deeply nested in the debugger. In the
3976 short term, this change will probably provoke more ILISP/SBCL
3977 squabbles, but hopefully it will be an improvement in the long run.)
3978 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3979 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3980 function, since declaiming return types involving aliens is
3981 (1) annoyingly messy to do by hand and (2) vital to efficient
3982 compilation of code which calls such functions.
3983 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3984 longer reexported by the SB-EXT package. They're solely useful
3985 for alien code, so it seems more logical that you should get
3986 them from the SB-ALIEN package, not in SB-EXT.
3987 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3988 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3989 features. Instead, the code that they used to control is always
3990 built into the system.
3991 * many other bug fixes
3992 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3993 advantage of the new EVAL-WHEN stuff and to clean them up in
3994 general, and they are now more ANSI-compliant in a number of
3995 ways. Martin Atzmueller is responsible for a lot of this.
3996 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3998 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3999 *** correct ERROR type for various file operations
4000 *** some fixes for Lisp streams
4001 *** DEFMETHOD syntax checking
4002 *** changing old weird representation of debug information as
4003 strings (which, among their other deficiencies, don't transform
4004 correctly when you rename packages, and don't change their
4005 print representation when you change things like *PACKAGE*
4006 and *PRINT-LENGTH*) to symbols and lists of symbols
4007 He also made several improvements and fixed several bugs in DESCRIBE.
4008 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4010 *** misbehavior of WRITE-STRING/WRITE-LINE
4011 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4012 and several other LOOP problems as well
4013 *** DIRECTORY when similar filenames are present
4014 *** DEFGENERIC with :METHOD options
4015 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4016 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4017 *** argument ordering in FIND with :TEST option
4018 *** mishandled package designator argument in APROPOS-LIST
4019 *** various problems in the backquote readmacro
4020 *** a bug in APROPOS
4021 *** probably some others that I'm not describing very well here,
4022 since the CVS log documents them by reference to sbcl-devel
4023 messages, and the SourceForge archives aren't working well.:-(
4024 ** Dan Barlow improved the Alpha port (and is making progress on the
4025 PPC port, for those of you who think different).
4026 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4027 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4028 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4029 continues to work on the SPARC port (for those of you in a position
4030 to look down upon our little PC-compatible boxes from a great height).
4031 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4032 and uses that copy, rather than the current dynamic value, when
4033 it's trying to decide whether to truncate output. Thus e.g.
4034 (let ((*print-lines* 50))
4035 (pprint-logical-block (stream nil)
4037 (let ((*print-lines* 8))
4038 (print (aref possiblybigthings i) stream)))))
4039 should now truncate the logical block only at 50 lines, instead of
4040 often truncating it at 8 lines, as it did before.
4041 * The doc/cmucl/ directory, containing old CMU CL documentation
4042 from the time of the fork, is no longer part of the base system.
4043 SourceForge has shut down its anonymous FTP service, and with it
4044 my original plan for distributing the old CMU CL documentation
4045 there. For now, if you need these files you can download an old
4046 SBCL source release and extract them from it.
4047 * The fasl file version number changed again, for dozens of reasons,
4048 some of which are apparent above.
4050 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4051 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4052 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4053 patch, so that DISASSEMBLE output is much nicer.
4054 * The code in the SB-PROFILE package now seems reasonably stable.
4055 I still haven't decided what the final interface should look like
4056 (I'd like PROFILE to interact cleanly with TRACE, since both
4057 facilities use function encapsulation) but if you have a need
4058 for profiling now, you can probably use it successfully with
4059 the current CMU-CL-style interface.
4060 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4061 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4062 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4063 link now returns the dangling link itself, and for similar
4064 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4065 link itself. (In these cases the old code signalled an error and
4066 looped endlessly, respectively.) Thus, DIRECTORY now works even
4067 in the presence of dangling and cyclic symbolic links.
4068 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4069 is now a supported extension again, since the consensus on
4070 sbcl-devel was that it can be useful for ordinary development
4071 work, not just for debugging SBCL itself.
4072 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4073 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4074 that currently-defined functions might be redefined later with
4075 different return types.
4076 * Hash tables can be printed readably, as inspired by CMU CL code
4077 of Eric Marsden and SBCL code of Martin Atzmueller.
4078 * better error handling in CLOS method combination, thanks to
4079 Martin Atzmueller porting Pierre Mai's CMU CL patches
4080 * more overflow fixes for >16Mbyte I/O buffers
4081 * A bug in READ has been fixed, so that now a single Ctrl-D
4082 character suffices to cause end-of-file on character streams.
4083 In particular, now you only need one Ctrl-D at the command
4084 line (not two) to exit SBCL.
4085 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4087 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4088 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4089 does the right thing.
4090 * The implementation of some type tests, especially for CONDITION
4091 types, is now tidier and maybe faster, due to CMU CL code
4092 originally by Douglas Crosher, ported by Martin Atzmueller.
4093 * Some math functions have been fixed, and there are new
4094 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4095 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4096 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4097 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4098 likely to be built into sbcl-0.7.0.)
4099 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4100 CL, ported to SBCL by Martin Atzmueller)
4101 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4102 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4103 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4104 * There's a new slam.sh hack to shorten the edit/compile/debug
4105 cycle for low-level changes to SBCL itself, and a new
4106 :SB-AFTER-XC-CORE target feature to control the generation of
4107 the after-xc.core file needed by slam.sh.
4108 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4109 COMPILE-FILE is no longer supported, so that now every function
4110 gets an entry point, so that block compilation looks a little
4111 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4112 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4113 returns the number of bytes consed since the system started,
4114 rather than the number consed since the first time the function
4115 was called. (The new definition parallels ANSI functions like
4116 CL:GET-INTERNAL-RUN-TIME.)
4117 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4118 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4119 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4120 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4121 * Fasl file version numbers are now independent of the target CPU,
4122 since historically most system changes which required version
4123 number changes have affected all CPUs equally. Similarly,
4124 the byte fasl file version is now equal to the ordinary
4127 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4128 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4129 is no longer recognized. I apologize for this, because it was
4130 listed in SB-EXT as a supported extension, but I found that
4131 its existing behavior was poorly specified, as well as incorrectly
4132 specified, and it looked like too much of a mess to straighten it
4133 out. I have enough on my hands trying to get ANSI stuff to work..
4134 * many patches ported from CMU CL by Martin Atzmueller, with
4135 half a dozen bug fixes in pretty-printing and the debugger, and
4136 half a dozen others elsewhere
4137 * fixed bug 13: Floating point infinities are now supported again.
4138 They might still be a little bit flaky, but thanks to bug reports
4139 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4140 as flaky as they were.
4141 * The --noprogrammer command line option is now supported. (Its
4142 behavior is slightly different in detail from what the old man
4143 page claimed it would do, but it's still appropriate under the
4144 same circumstances that the man page talks about.)
4145 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4146 are now supported, and enabled by default. Thus, the compiler can
4147 handle many floating point and complex operations much less
4148 inefficiently. (Thus e.g. you can implement a complex FFT
4150 * The compiler now detects type mismatches between DECLAIM FTYPE
4151 and DEFUN better, and implements CHECK-TYPE more correctly, and
4152 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4153 thanks to patches from Martin Atzmueller.
4154 * various fixes to make the cross-compiler more portable to
4155 ANSI-conforming-but-different cross-compilation hosts (notably
4156 Lispworks for Windows, following bug reports from Arthur Lemmens)
4157 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4158 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4159 fixed thanks to Pierre Mai's CMU CL patches.
4160 * fixes to keep the system from overflowing internal counters when
4161 it tries to use i/o buffers larger than 16M bytes
4162 * fixed bug 45a: Various internal functions required to support
4163 complex special functions have been merged from CMU CL sources.
4164 (When I was first setting up SBCL, I misunderstood a compile-time
4165 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4166 * improved support for type intersection and union, fixing bug 12
4167 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4168 more obscure bugs as well
4169 * some steps toward byte-compiling non-performance-critical
4170 parts of the system, courtesy of patches from Martin Atzmueller
4171 * Christophe Rhodes has made some debian packages of sbcl at
4172 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4173 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4174 stable, but are nonetheless usable. When he's ready, I'd be happy
4175 to add them to the SourceForge "File Releases" section. (And if
4176 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4177 * new fasl file format version number (because of changes in
4178 internal representation of (OR ..) types to accommodate the new
4179 support for (AND ..) types, among other things)
4181 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4182 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4184 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4186 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4187 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4188 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4189 * DESCRIBE now works on structure objects again.
4190 * Most function call argument type mismatches are now handled as
4191 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4192 know whether the function will be redefined before the call is
4193 executed. (The compiler could flag local calls with full WARNINGs,
4194 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4195 it doesn't keep track of enough information to know whether calls
4196 are local in this sense.)
4197 * Compiler output is now more verbose, with messages truncated
4198 later than before. (There should be some supported way for users
4199 to override the default verbosity, but I haven't decided how to
4200 provide it yet, so this behavior is still controlled by the internal
4201 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4202 src/compiler/ir1util.lisp.)
4203 * Fasl file format version numbers have increased again, because
4204 support for the Gray streams extension changes the layout of the
4205 system's STREAM objects.
4206 * The Gray subclassable streams extension now works, thanks to a
4207 patch from Martin Atzmueller.
4208 * The full LOAD-FOREIGN extension (not just the primitive
4209 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4210 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4211 but like most other programs, it defaults to copying the Unix
4212 environment from the original process instead of starting the
4213 new process in an empty environment.
4214 * Extensions which manipulate the Unix environment now support
4215 an :ENVIRONMENT keyword option which doesn't smash case or
4216 do other bad things. The CMU-CL-style :ENV option is retained
4217 for porting convenience.
4218 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4219 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4221 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4223 * A patch from Martin Atzmueller seems to have solved the SIGINT
4224 problem, and as far as we know, signal-handling now works cleanly.
4225 (If you find any new bugs, please report them!)
4226 * The system no longer defaults Lisp source file names to types
4227 ".l", ".cl", or ".lsp", but only to ".lisp".
4228 * The compiler no longer uses special default file extensions for
4229 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4230 seems to expect a single default extension for all compiled code,
4231 and there's no compelling reason to try to stretch the standard
4232 to allow two different extensions.) Instead, byte-compiled files
4233 default to the same extension as native-compiled files.
4234 * Fasl file format version numbers have increased again, because
4235 a rearrangement of internal implementation packages made some
4236 dumped symbols in old fasl files unreadable in new cores.
4237 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4238 many fewer weird special cases.
4239 * Bug #17 (differing COMPILE-FILE behavior between logical and
4240 physical pathnames) has been fixed, and some related misbehavior too,
4241 thanks to a patch from Martin Atzmueller.
4242 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4243 by Tim Moore, ported to SBCL by Martin Atzmueller.
4244 * Martin Atzmueller fixed several filesystem-related problems,
4245 including bug #36, in part by porting CMU CL patches, which were
4246 written in part by Paul Werkowski.
4247 * More compiler warnings in src/runtime/ are gone, thanks to
4248 more patches from Martin Atzmueller.
4249 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4252 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4254 * DESCRIBE now works on CONDITION objects.
4255 * The debugger now handles errors which arise when trying to print
4256 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4258 * The build system now uses an additional file, customize-target-features.lisp,
4259 to allow local modifications to the target *FEATURES* list. (The point of
4260 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4261 debugging features enabled, without having to worry about propagating it
4262 into everyone's system when I do a "cvs update".) When no
4263 customize-target-features.lisp file exists, the target *FEATURES* list
4264 should be constructed the same way as before.
4265 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4266 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4267 warn about behavior which is undefined under the ANSI spec. Note: This
4268 is specified by ANSI, but it's not very popular with programmers.
4269 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4270 macro in the SBCL sources for an example of a workaround which you
4271 might use to make portable ANSI-standard code which does what you want.
4272 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4273 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4274 when it's used as a non-toplevel form. (This is required in order
4275 to implement the DEFCONSTANT-EQX macro.)
4276 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4277 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4278 funny things, due to bugs in the implementation of EVAL-WHEN
4279 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4280 non-toplevel DEFCONSTANTs are uncommon.)
4281 * The core file version number and fasl file version number have been
4282 incremented, because the old noncompliant DEFCONSTANT behavior involved
4283 calling functions which no longer exist, and because I also took the
4284 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4285 * fixed bug 1 (error handling before read-eval-print loop starts), and
4286 redid debugger restarts and related debugger commands somewhat while
4288 ** The QUIT debugger command is gone, since it did something
4289 rather different than the SB-EXT:QUIT command, and since it never
4290 worked properly outside the main toplevel read/eval/print loop.
4291 Invoking the new TOPLEVEL restart provides the same functionality.
4292 ** The GO debugger command is also gone, since you can just invoke
4293 the CONTINUE restart directly instead.
4294 ** The TOP debugger command is also gone, since it's redundant with the
4295 FRAME 0 command, and since it interfered with abbreviations for the
4297 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4299 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4300 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4301 some porting work by Martin Atzmueller)
4302 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4303 with a boatload of patches which he ported to SBCL. Now that those
4305 ** The system tries to make sure that its low-priority messages
4306 are prefixed by semicolons, to help people who like to use
4307 syntax highlighting in their ILISP buffer. (This patch
4308 was originally due to Raymond Toy.)
4309 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4310 patches originally written by Raymond Toy.
4311 ** The compiler understands coercion between single-value and
4312 multiple-VALUES type expressions better, getting rid of some very
4313 weird behavior, thanks to patches originally by Robert MacLachlan
4314 and Douglas Crosher.
4315 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4316 lambda lists, thanks to a patch originally by Pierre Mai.
4317 ** The system no longer bogusly warns about "abbreviated type
4319 ** The compiler gets less confused by inlining and RETURN-FROM,
4320 thanks to some patches originally by Tim Moore.
4321 ** The system no longer hangs when dumping circular lists to fasl
4322 files, thanks to a patch originally from Douglas Crosher.
4323 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4324 undefined function error.
4325 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4326 in structure slot initforms) for now, documented workaround instead:-|
4327 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4328 * fixed bug 5: added stubs for various Gray stream functions called
4329 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4330 installed, at least appropriate type errors are generated
4331 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4332 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4333 consistently in DEFMETHOD forms.
4334 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4335 it doesn't seem to affect SBCL after all
4336 * The C runtime system now builds with better optimization and many
4337 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4339 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4341 * The system is now under CVS at SourceForge (instead of the
4342 CVS repository on my home machine).
4343 * The new signal handling code has been tweaked to treat register
4344 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4345 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4346 downstream, e.g. in debug-int.lisp, has implicit dependencies
4347 on the unsignedness of integer representation of machine words,
4348 and that caused the system to bomb out with infinite regress
4349 when trying to recover from type errors involving signed values,
4350 e.g. (BUTLAST '(1 2 3) -1).)
4351 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4352 (This was one of the bugs Peter Van Eynde reported back in July.)
4353 * The system now uses code inspired by Colin Walters' O(N)
4354 implementation of MAP (from the cmucl-imp@cons.org mailing
4355 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4356 inline the MAP operation, and there is more than one
4357 sequence argument to the MAP call (so that it can't just
4358 do ETYPECASE once and for all based on the type of the
4359 single sequence argument). (The old non-inline implementation
4360 of the general M-argument sequence-of-length-N case required
4361 O(M*N*N) time when any of the sequence arguments were LISTs.)
4362 * The QUIT :UNIX-CODE keyword argument has been renamed to
4363 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4364 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4366 * Raymond Wiker's patches to port dynamic loading from Linux to
4367 FreeBSD have been added.
4368 * The BUGS file is now more nearly up to date, thanks in large part
4369 to Martin Atzmueller's review of it.
4370 * The debugger now flushes standard output streams before it begins
4371 its output ("debugger invoked" and so forth).
4372 * The core version number and fasl file version number have both
4373 been incremented, because of incompatible changes in the layout
4375 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4376 used to not be called for a saved Lisp image.)
4377 * Martin Atzmueller's version of a patch to fix a compiler crash,
4378 as posted on sbcl-devel 13 September 2000, has been installed.
4379 * Instead of installing Martin Atzmueller's patch for the
4380 compiler transform for SUBSEQ, I deleted the compiler transform,
4381 and transforms for some similar consing operations.
4382 * A bug in signal handling which kept TRACE from working on OpenBSD
4384 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4385 away by constant folding
4386 * The system now defines its address space constants in one place
4387 (in the Lisp sources), and propagates them automatically elsewhere
4388 (through GENESIS and the sbcl.h file). Therefore, patching the
4389 address map is less unnecessarily tedious and error-prone. The
4390 Lisp names of address space constants have also been systematized.
4391 * CVS tags like dollar-Header-dollar have been removed from
4392 the sources, because they have never saved me trouble and
4393 they've been source of trouble working with patches and other
4394 diff-related operations.
4395 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4396 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4398 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4400 * The system has been ported to OpenBSD.
4401 * The system now compiles with a simple "sh make.sh" on the systems
4402 that it's supported on. I.e., now you no longer need to tweak
4403 text in the target-features.lisp-expr and symlinks in src/runtime/
4404 by hand, the make.sh takes care of it for you.
4405 * The system is no longer so grossly inefficient when compiling code
4406 involving vectors implemented as general (not simple) vectors (VECTOR T),
4407 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4408 which dares to use the various sequence functions on non-simple
4409 vectors, takes less of a performance hit.
4410 * There is now a primitive type predicate VECTOR-T-P
4411 to test for the (VECTOR T) type, so that e.g.
4412 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4413 can now be compiled with some semblance of efficiency. (The old code
4414 turned the type declaration into a full call to %TYPEP at runtime!)
4415 * AREF on (VECTOR T) is still not fast, since it's still compiled
4416 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4417 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4418 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4419 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4420 without making full calls to SUBTYPEP at runtime.
4421 (Some analogous efficiency issues for non-simple vectors specialized to
4422 element types other than T, or for non-simple multidimensional arrays,
4423 have not been addressed. They could almost certainly be handled the
4424 same way if anyone is motivated to do so.)
4425 * The changes in array handling break binary compatibility, so
4426 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4427 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4428 instead of (VALUES T T).
4429 * By following the instructions that Dan Barlow posted to sbcl-devel
4430 on 2 July 2000, I was able to enable primitive dynamic object
4431 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4432 functionality is not implemented (since it calls ld to resolve
4433 library references automatically, requiring RUN-PROGRAM for its
4434 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4435 try to resolve library references) is now supported.
4436 * The system now flushes the standard output streams when it terminates,
4437 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4438 them at several other probably-convenient times, e.g. in each pass of
4439 the toplevel read-eval-print loop, and after evaluating a form given
4440 as an "--eval" command-line option. (These changes were motivated by a
4441 discussion of stream flushing issues on cmucl-imp in August 2000.)
4442 * The source transform for TYPEP of array types no longer assumes
4443 that an array whose element type is a not-yet-defined type
4444 is implemented as an array of T, but instead punts, so that the
4445 type will be interpreted at runtime.
4446 * There is now some support for cross-compiling in make.sh: each of
4447 the phases of make.sh has its own script. (This should be transparent
4448 to people doing ordinary, non-cross-compile builds.)
4449 * Since my laptop doesn't have hundreds of megabytes of memory like
4450 my desktop machine, I became more motivated to do some items on
4451 my to-do list in order to reduce the size of the system a little:
4452 ** Arrange for various needed-only-at-cold-init things to be
4453 uninterned after cold init. To support this, those things have
4454 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4455 symbols with such names are now uninterned after cold init).
4456 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4457 fasl files for cold load.
4458 ** Remove the old compiler structure pooling code (which used to
4459 be conditional on the target feature :SB-ALLOC) completely.
4460 ** Redo the representation of some data in cold init to be more compact.
4461 (I also looked into supporting byte compiled code at bootstrap time,
4462 which would probably reduce the size of the system a lot, but that
4463 looked too complicated, so I punted for now.)
4464 * The maximum signal nesting depth in the src/runtime/ support code has
4465 been reduced from 4096 to 256. (I don't know any reason for the very
4466 large old value. If the new smaller value turns out to break something,
4467 I'll probably just bump it back up.)
4468 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4470 * Many, many bugs reported by Peter Van Eynde have been added to
4471 the BUGS list; some have even been fixed.
4472 * While enabling dynamic object file loading, I tried to make the
4473 code easier to understand, renaming various functions and variables
4474 with less ambiguous names, and changing some function calling
4475 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4476 * While trying to figure out how to do the OpenBSD port, I tried to
4477 clean up some of the code in src/runtime/. In particular, I dropped
4478 support for non-POSIX signal handling, added various comments,
4479 tweaked the code to reduce the number of compilation warnings, and
4480 renamed some files to increase consistency.
4481 * To support the new automatic configuration functionality in make.sh,
4482 the source file target-features.lisp-expr has been replaced with the
4483 source file base-target-features.lisp-expr and the machine-generated
4484 file local-target-features.lisp-expr.
4485 * fixed a stupid quoting error in make.sh so that using CMU CL
4486 "lisp -batch" as cross-compilation host works again
4488 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4490 * DESCRIBE no longer tries to call itself recursively to describe
4491 bound/fbound values, so that it no longer fails on symbols which are
4492 bound to themselves (like keywords, T, and NIL).
4493 * DESCRIBE now works on generic functions.
4494 * The printer now prints less-screwed-up representations of closures
4495 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4496 * A private symbol is used instead of the :EMPTY keyword previously
4497 used to mark empty slots in hash tables. Thus
4498 (DEFVAR *HT* (MAKE-HASH-TABLE))
4499 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4500 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4501 now does what ANSI says that it should. (You can still get
4502 similar noncompliant behavior if bang on the hash table
4503 implementation with all the symbols you get back from
4504 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4505 This breaks binary compatibility, since tests for equality to
4506 :EMPTY are wired into things like the macroexpansion of
4507 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4509 * There's now a minimal placeholder implementation for CL:STEP,
4510 as required by ANSI.
4511 * An obscure bug in the interaction of the normal compiler, the byte
4512 compiler, inlining, and structure predicates has been patched
4513 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4514 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4516 * Missing ordinary arguments in a macro call are now detected even
4517 when the macro lambda list contains &KEY or &REST.
4518 * The debugger no longer complains about encountering the top of the
4519 stack when you type "FRAME 0" to explicitly instruct it to go to
4520 the top of the stack. And it now prints the frame you request even
4521 if it's the current frame (instead of saying "You are here.").
4522 * As specified by ANSI, the system now always prints keywords
4523 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4525 * The default initial SIZE of HASH-TABLEs is now smaller.
4526 * Type information from CLOS class dispatch is now propagated
4527 into DEFMETHOD bodies, so that e.g.
4528 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4530 is now basically equivalent to
4531 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4532 (DECLARE (TYPE SINGLE-FLOAT X))
4534 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4535 operation, without having to do run-time type dispatch.
4536 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4537 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4538 * Since I don't seem to be making big file reorganizations very often
4539 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4540 is overflowing my ability to conveniently back them up), I've finally
4541 checked the system into CVS. (The CVS repository is on my home system,
4542 not at SourceForge -- putting it on SourceForge might come later.)
4543 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4544 high-level GC-NOTIFY-FOO functions send their output. (There's
4545 still very little control of where low-level verbose GC functions
4546 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4547 less than it used to -- the GC-NOTIFY-FOO functions are now under
4548 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4549 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4550 in only one place in the source code, and propagates it automatically
4551 everywhere that it's needed. Thus e.g. when I bump the version from
4552 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4553 * The C source files now include boilerplate legalese and documentation
4554 at the head of each file (just as the Lisp source files already did).
4555 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4556 to his page will be replaced with a link to his new CLiki service.
4558 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4560 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4561 * The build process now looks for GNU make under the default name "gmake",
4562 instead of "make" as it used to. If GNU make is not available as "gmake"
4563 on your system, you can change this default behavior by setting the
4564 GNUMAKE environment variable.
4565 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4566 can build without error under CMU CL.
4568 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4570 * There is now a partial SBCL user manual (with some new text and some
4571 text cribbed from the CMU CL manual).
4572 * The beginnings of a profiler have been added (starting with the
4573 CMU CL profiler and simplifying and cleaning up). Eventually the
4574 main interface should be through the TRACE macro, but for now,
4575 it's still accessed through vaguely CMU-CL-style functions and macros
4576 exported from the package SB-PROFILE.
4577 * Some problems left over from porting CMU CL to the new
4578 cross-compilation bootstrap process have been cleaned up:
4579 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4580 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4581 part when getting rid of the compiler *BACKEND* stuff.)
4582 ** The value of *NULL-TYPE* was screwed up, because it was
4583 being initialized before the type system knew the final
4584 definition of the 'NULL type. This screwed up several key
4585 optimizations in the compiler, causing inefficiency in all sorts
4586 of places. (I found it because I wanted to understand why
4587 GET-INTERNAL-RUN-TIME was consing.)
4588 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4589 PROCLAIM FTYPE information. Unfortunately this broke binary
4590 compatibility again, since now the forms output by DEFGENERIC
4591 to refer to functions which didn't exist in 0.6.3.
4592 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4593 can use the new (as of 0.6.3) transform for SOME into MAP into
4595 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4596 values to (INTEGER 0 1000000), so that the time code will no longer
4597 occasionally get blown up by Linux returning 1000000 microseconds
4598 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4599 its output conform to the ANSI spec. (Alas, this makes its output
4600 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4601 * A full call to MAP NIL with a single sequence argument no longer conses.
4602 * fixes to problems pointed out by Martin Atzmueller:
4603 * The manual page no longer talks about multiprocessing as though
4604 it were currently supported.
4605 * The ILISP support patches have been removed from the distribution,
4606 because as of version 5.10.1, ILISP now supports SBCL without us
4607 having to maintain patches.
4608 * added a modified version of Raymond Toy's recent CMU CL patch for
4609 EQUALP comparison of HASH-TABLE
4611 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4613 * The system still can't cross-compile itself with
4614 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4615 entails), but at least it can compile more of itself that way
4616 than it used to be able to, and various buglets which were uncovered
4617 by trying to cross-compile itself that way have now been fixed.
4618 * This release breaks binary compatibility again. This time
4619 at least I've incremented the FASL file format version to 2, so that the
4620 problem can be detected reliably instead of just causing weird errors.
4621 * various new style warnings:
4622 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4623 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4624 ** using the lexical binding of a variable named in the *FOO* style
4625 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4626 as specified by ANSI.
4627 * *RANDOM-STATE* is no longer automatically initialized from
4628 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4629 default behavior of the system is to repeat its behavior every time
4630 it's run. If you'd like to change this behavior, you can always
4631 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4632 old convention there was no comparably easy way to get the system to
4633 repeat its behavior every time it was run.
4634 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4635 FTYPE declarations has been removed, in favor of their ANSI
4636 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4637 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4638 the types of their sequence arguments can be determined at compile time.
4639 This is done through a new open code expansion for MAP which eliminates
4640 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4641 argument types can be determined at compile time.
4642 * The optimizer now transforms COERCE into an identity operation when it
4643 can prove that the coerced object is already of the correct type. (This
4644 can be a win for machine generated code, including the output of other
4645 optimization transforms, such as the MAP transform above.)
4646 * Credit information has been moved from source file headers into CREDITS.
4647 * Source file headers have been made more standard.
4648 * The CASE macro now compiles without complaining even when it has
4651 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4653 * (Note that the way that the PCL macroexpansions were rewritten
4654 to accommodate the change in DEFGENERIC below breaks binary
4655 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4656 not run under sbcl-0.6.2. Once we get out of alpha releases,
4657 i.e. hit release 1.0.0, we'll probably try to maintain binary
4658 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4659 and sbcl-1.4.4. Until then, however, it might be fairly common
4660 for maintenance releases to break binary compatibility.)
4661 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4663 * The reporting of unhandled signals has been changed to print some
4664 explanatory text as well as the report form. (Previously only
4665 the report form was printed.)
4666 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4667 it defines, so that the compiler no longer issues undefined function
4668 warnings for compiled-but-not-yet-loaded generic functions.
4669 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4670 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4671 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4672 section of the ANSI spec are supported.
4673 * The "" nickname for the "KEYWORD" package has been removed.
4674 The reader still handles symbol tokens which begin with a package marker
4675 as keywords, but it doesn't expose its mechanism for doing so in the
4676 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4677 * The system now issues STYLE-WARNINGs for contradictory TYPE
4678 proclamations. (Warnings for contradictory FTYPE proclamations would
4679 be nice too, but those can't be done usefully unless the type system
4680 is made smarter about FUNCTION types.)
4681 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4682 systematized, so that "*target-*.lisp is supposed to exist only on the
4683 target and imply that there's a related file which exists on the
4684 host, and *host-*.lisp is supposed to exist only on the host and imply
4685 that there's a related file which exists on the target. This involves a
4686 lot of renaming. Hopefully the acute confusion caused by the renaming
4687 will be justified by the reduction in chronic confusion..
4688 ** runtime-type.lisp -> early-target-type.lisp
4689 ** target-type.lisp -> late-target-type.lisp
4690 ** early-host-format.lisp -> early-format.lisp
4691 ** late-host-format.lisp -> late-format.lisp
4692 ** host-error.lisp -> misc-error.lisp
4693 ** early-error.lisp -> early-target-error.lisp
4694 ** late-error.lisp -> late-target-error.lisp
4695 ** host-defboot.lisp -> early-defboot.lisp
4696 ** code/misc.lisp -> code/target-misc.lisp
4697 ** code/host-misc.lisp -> code/misc.lisp
4698 ** code/numbers.lisp -> code/target-numbers.lisp
4699 ** code/early-numbers.lisp -> numbers.lisp
4700 ** early-host-type.lisp -> early-type.lisp
4701 ** late-host-type.lisp -> late-type.lisp
4702 ** host-typep.lisp -> typep.lisp
4703 ** load.lisp -> target-load.lisp
4704 ** host-load.lisp -> load.lisp
4705 ** host-disassem.lisp -> disassem.lisp
4706 ** host-insts.lisp -> insts.lisp
4707 ** byte-comp.lisp -> target-byte-comp.lisp
4708 ** host-byte-comp.lisp -> byte-comp.lisp
4709 ** host-signal.lisp -> signal.lisp
4710 ** host-defstruct.lisp -> defstruct.lisp
4711 ** late-target-type.lisp -> deftypes-for-target.lisp
4712 Furthermore, several other previously target-only files foo.lisp (e.g.
4713 hash-table.lisp and random.lisp) have been split into a target-and-host
4714 foo.lisp file and a target-only target-foo.lisp file, with their key type
4715 definitions in the target-and-host part, so that the cross-compiler will
4716 know more about target types.
4717 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4718 gone away. In their place are various *BACKEND-FOO* variables
4719 corresponding to the slots of the old structure.
4720 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4721 into the target SBCL has been fixed.
4722 * The chill.lisp system for loading cold code into a running SBCL
4724 * Support for the CMU CL "scavenger hook" extension has been removed.
4725 (It was undocumented and unused in the CMU CL sources that SBCL was
4726 derived from, and stale in sbcl-0.6.1.)
4727 * Various errors in the cross-compiler type system were detected
4728 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4729 (enabling various consistency checks). Many of them were fixed,
4730 but some hard problems remain, so the compiler is back to
4731 running without *TYPE-SYSTEM-INITIALIZED* for now.
4732 * As part of the cross-compiler type system cleanup, I implemented
4733 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4734 * I have started adding UNCROSS calls throughout the type system
4735 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4736 on cross-compiler input files will be able to go away, and various
4738 * CONSTANTP now returns true for quoted forms (as explicitly required
4741 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4743 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4744 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4745 (MAKE-LIST -1). In the longer term, it ought to become true
4746 that declarations are assertions even at SAFETY 1. For now, it's not
4747 quite true even at SAFETY 3, but it's at least more nearly true..
4748 (Note that this change seems to increases the size of the system by
4749 O(5%) and to decrease the speed of the compiler by 20% or more.)
4750 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4751 for the problem of printing dozens of lines of distracting information
4752 about low-level system machinery as part of the top stack frame
4753 on entry to the debugger when an undefined function was called.
4754 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4755 is not reset to COMMON-LISP-USER.
4756 * Compilation of stuff related to dyncount.lisp has been made conditional
4757 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4758 smaller. The various dyncount-related symbols have been moved into
4759 a new "SB-DYNCOUNT" package.
4760 * tty-inspect.lisp has been renamed to inspect.lisp.
4761 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4762 feature has gone away. (When we eventually port to other flavors of
4763 libc and/or Unix, we'll try to make the differences between flavors
4764 invisible at the user level.)
4765 * Various other *FEATURES* tags, and/or their associated conditionals,
4766 have been removed if obsolescent, or given better documentation, or
4767 sometimes given more-mnemonic names.
4769 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4771 * tidied up "make.sh" script
4772 * tidied up system directory structure
4773 * better "clean.sh" behavior
4774 * added doc/FOR-CMUCL-DEVELOPERS
4775 * many many small tweaks to output format, e.g. removing possibly-confusing
4776 trailing #\. character in DESCRIBE-INSTANCE
4777 * (EQUALP #\A 'A) no longer signals an error.
4778 * new hashing code, including EQUALP hashing
4779 * tidied up Lisp initialization and toplevel
4780 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4781 * command line argument processing
4782 * added POSIX-GETENV function to deal with Unix-ish environment variables
4783 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4784 terminating SBCL on EOF
4785 * non-verbose GC by default
4786 * There is no more "sbcl" shell script; the sbcl file is now the C
4787 runtime executable (just like CMU CL).
4788 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4790 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4791 symbol and package information as Lisp native symbol and package objects
4792 instead of strings naming symbols and strings naming packages. This way,
4793 whenever packages are renamed (as in warm init), debug information is
4794 transformed along with everything else.
4795 * tweaked the optimization policy declarations which control the building
4796 of SBCL itself. Now, among other things, the system no longer saves
4797 source location debugging information. (This helps two problems at once
4798 by reducing SBCL size and by keeping SBCL from trying to look for its
4799 sources -- which may not exist -- when reporting errors.)
4800 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4801 debugging and testing purposes
4802 * cleaned up printing, making the printer call PRINT-OBJECT for
4803 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4804 methods, giving nearly-ANSI behavior
4805 * converted almost all special variables to use *FOO* naming convention
4806 * deleted PARSE-TIME functionality, since it can be done portably
4807 * moved some files out of cold init into warm init
4808 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4810 * regularized formatting of source files
4811 * added an install.sh script
4812 * fixed ridiculous memory usage of cross-compiler by making
4813 compiler/alloc.lisp not try to do pooling unless it can hook
4814 itself into the GC of the cross-compilation host. Now the system
4815 builds nicely on my old laptop.
4816 * added :SB-ALLOC in target-features.lisp-expr
4817 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4818 was not implemented)
4819 * re-did condition handling and note reporting in the compiler. Notes
4820 are no longer handled by signalling conditions. Style warnings
4821 and warnings are handled more correctly and reported in such a way
4822 that it's easy to find one or the other in your output (so that you
4823 can e.g. figure out which of many problems caused COMPILE-FILE to
4825 * changed the severity of several compiler warnings from full WARNING
4826 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4827 compiler note reporting so that it doesn't use the condition system
4828 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4829 COMPILE-FILE command)
4830 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4831 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4832 no longer supported, since their implementation was deeply intertwingled
4833 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4834 * removed bogus "support" for compiler macros named (SETF FOO), and
4835 removed the compiler macro for SETF INFO (but only after making a fool
4836 of myself on the cmucl-imp mailing list by posting a bogus patch for
4837 DEFINE-COMPILER-MACRO..)
4838 * Compiled files containing forms which have side effects on the Lisp
4839 reader (such as DEFPACKAGE forms) are now handled more correctly.
4840 (Compiler queuing of top level lambdas has been suppressed by setting
4841 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4842 * deleted various currently-unused source files, e.g. gengc.lisp. They
4843 may be added back at some point e.g. when porting to other architectures,
4844 but until they are it's distracting to distribute them and to try to
4846 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4847 there were so many of them they're just distractions, and UNCROSS is
4848 known to be able to handle the current sources
4849 * moved PROFILE functionality into TRACE, so that it will be clear
4850 how the wrapping and unwrapping of functions when you profile them
4851 interacts with the wrapping and unwrapping of functions when you
4852 trace them. (Actually, the functionality isn't there yet, but at least
4853 the interface specification is there. Hopefully, the functionality will
4854 arrive with some future maintenance release.)
4855 * removed host-oops.lisp
4856 * changed signature of QUIT function to allow UNIX-CODE argument
4857 * fixed READ-SEQUENCE bug
4858 * tweaked verbose GC output so that it looks more like the progress
4859 output that ANSI specifies for functions like LOAD
4860 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4861 * added <http://sbcl.sourceforge.com> to the banner information printed by