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)
11 changes in sbcl-1.0.21 relative to 1.0.20:
12 * new feature: the compiler is able to track the effective type of a
13 generic function across method addition and removal.
14 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
15 appropriately typed structure slots without locking.
16 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
17 information like those gathered by TIME using a programming-friendly
19 * new feature: TIME reports time taken even if the form performs a
20 non-local transfer of control.
21 * enhancement: reduced conservativism on GENCGC platforms: on
22 average 45% less pages pinned (measured from SBCL self build).
23 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
24 mutate constant symbols or violate declaimed type of the symbol.
25 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
26 sleeping on the semaphore in heavily contested semaphores.
27 * bug fix: semaphores and condition variables were not interrupt
29 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
30 owned by other threads anymore.
31 * bug fix: FIND on lists called KEY outside the specified
32 subsequence. (reported by budden)
33 * bug fix: LOG doesn't use single-float intermediate results when
34 given mixed integer and double-float arguments, leading to better
35 precision. (reported by Bob Felts)
36 * bug fix: LOG with base zero returned values of inconsistent type.
38 changes in sbcl-1.0.20 relative to 1.0.19:
39 * minor incompatible change: OPTIMIZE qualities
40 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
41 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
42 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
43 * documentation: some slot access efficiency guidelines have been
44 added to the user manual.
45 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
46 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
48 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
49 to more efficient EQ-comparison versions more often.
50 * optimization: enhanced derivation of DOLIST iteration variable type
52 * optimization: constant folding of simple (LIST ...) forms as DOLIST
54 * optimization: runtime lookup of function definitions can be
55 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
56 * optimization: compiler is able to derive the return type of
57 (AREF (THE STRING X) Y) as being CHARACTER.
58 * optimization: CLRHASH on empty hash-tables no longer does pointless
59 work. (thanks to Alec Berryman)
60 * optimization: REPLACE deftransforms don't punt when :START1 or
62 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
63 no longer emits a code deletion note. (reported by Andrew
65 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
66 functions with an odd number of &OPTIONAL arguments, a &REST
67 argument, and one or more &KEY arguments at the call site.
68 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
69 arguments appeared at call sites as well.
70 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
71 READ-LINE &co can no longer read from them afterwards. (reported
73 * bug fix: fixed #427: unused local aliens no longer cause compiler
74 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
76 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
77 alien-stack leakage. (reported by Andy Hefner)
78 * bug fix: PROGV signals an error when an attempt to violate declared
79 type of a variable or bind a constant is made.
80 * bug fix: SET signals an error when an attempt to violate declared
81 type of a variable is made.
82 * bug fix: restart computation during the execution of a restart
83 test function no longer causes infinite recursion. (reported by
85 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
86 signals an error. (thanks to Pierre Mai)
87 * bug fix: EXPORT left symbol unexported in conflict situations.
88 (thanks to Michael Weber)
89 * bug fix: correctly handle name conflicts involving CL:NIL.
90 (thanks to Michael Weber)
91 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
92 conflicts arising from USEing package with conflicting symbols
93 correctly. (thanks to Michael Weber)
95 changes in sbcl-1.0.19 relative to 1.0.18:
96 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
97 warnings that go otherwise unhandled will be muffled if they are
98 of the type that's the value of this variable.
99 * optimization: stack allocation is slightly more efficient on x86
101 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
102 where a raw slot always is initialized using the initform whose
103 type is not know sufficiently well a compile-time are now compiled
104 correctly. (reported by John Morrison)
105 * bug fix: compiler no longer makes erronous assumptions in the
106 presense of non-foldable SATISFIES types.
107 * bug fix: stack analysis missed cleanups of dynamic-extent
108 arguments in non-let-converted calls to local functions.
109 * improvements to the Windows port:
110 ** adjusted address spaces for building on both Win32 and
111 Win64. (thanks for John Connors)
112 * fixed some bugs revealed by Paul Dietz' test suite:
113 ** interval arithmetic during type derivation used inexact integer
114 to single-float coercions.
115 ** arithmetic operations involving large integers and single
116 floats give the same results in compiled and interpreted code.
117 ** deriving the result type of COERCE no longer signals an error
118 if the derived type of the second argument is a MEMBER type
119 containing invalid type specifiers.
120 ** ADJOIN with constant NIL as second argument works correctly.
121 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
123 changes in sbcl-1.0.18 relative to 1.0.17:
124 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
125 profiles only the current thread.
126 * minor incompatible change: changes to SYMBOL-VALUE of constants
127 defined with DEFCONSTANT now signal an error.
128 * enhancement: SB-SPROF now has support for wallclock profiling,
129 and is also able to profile specific threads. REPORT output
130 has also additional sorting options.
131 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
133 * optimization: structure allocation has been improved
134 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
135 ** out of line constructors are ~10% faster.
136 ** inline constructors are ~15% faster.
137 ** inline constructors are capable of dynamic extent allocation
138 (generally on x86 and x86-64, in some cases on other platforms
140 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
142 * optimization: file compiler is now able to coalesce non-circular
143 lists, non-base strings, and bit-vectors. Additionally, constants
144 are never referenced through SYMBOL-VALUE at runtime.
145 * optimization: code defining methods on PRINT-OBJECT (and other
146 generic functions in the COMMON-LISP package) now loads faster.
147 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
149 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
150 fasl is now deleted. (reported by Attila Lendvai)
151 * bug fix: READ-LINE always returned NIL for the last line in files.
152 (reported by Yoshinori Tahara)
153 * bug fix: more accurate disassembly annotations of foreign function
154 calls. (thanks to Andy Hefner)
155 * bug fix: trimming non-simple strings and non-string string
156 designators when the there is nothing to trim works properly.
157 (thanks to James Knight)
158 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
159 (thanks to Travis Cross)
160 * fixed some bugs revealed by Paul Dietz' test suite:
161 ** NIL is a valid function name (regression at 1.0.13.38)
162 ** FILL on lists was missing its return value (regression at 1.0.12.27)
163 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
164 fill pointers properly (regression at 1.0.12.23)
165 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
166 displacement indices properly (regression at 1.0.12.23)
168 changes in sbcl-1.0.17 relative to 1.0.16:
169 * temporary regression: user code can no longer allocate closure
170 variable storage on stack, due to bug 419 without explicitly
171 requesting it. Please consult sbcl-devel for advice if you need to
172 use this feature in the meanwhile.
173 * new feature: runtime argument --control-stack-size can be used to
174 adjust thread default control stack size.
175 * enhancement: improved TIME output
176 ** all times are reported using the measured accuracy (milliseconds
177 for real and GC times, microseconds for everything else.)
178 ** processor cycle counts on x86 and x86-64.
179 ** interpreted forms are counted for both evaluator modes.
180 ** number of lambdas converted by the compiler is reported.
181 ** CPU percentage report (computed from real and total run time.)
182 ** more comprehensive run time reporting, using a condenced format
183 ** interperted form, lambda, and page fault counts are omitted
185 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
187 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
188 * optimization: two argument forms of LAST are upto ~10% faster
189 in normal SPEED policies.
190 * optimization: NCONC no longer needs to heap cons its &REST list
191 in normal SPEED policies.
192 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
194 * bug fix: UNION and NUNION work with :TEST-NOT once more,
195 regression since 1.0.9.1. (thanks to Eric Marsden)
196 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
197 since 1.0.15.36. (thanks to Paul Khuong)
198 * bug fix: LAST when always returned the whole list when given a bignum
199 as the second argument.
200 * bug fix: dynamic extent allocation of nested lists and vectors
201 could leak to otherwise accessible parts.
202 * bug fix: invalid optimization of heap-allocated alien variable
204 * bug fix: fasl header checking is less vulnerable to different
205 platform word lengths.
206 * bug fix: more correct assembler syntax for GNU binutils
207 2.18.50.0.4 support. (thanks to Marijn Schouten)
208 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
210 * bug fix: the fopcompiler can handle LOCALLY forms (with no
211 declarations) successfully. (reported by Attila Lendvai)
213 changes in sbcl-1.0.16 relative to 1.0.15:
214 * minor incompatible change: revert the changes to sb-posix's error
215 signaling added in 1.0.14.
216 * minor incompatible change: change PROBE-FILE back to returning
217 NIL whenever we can't get a truename, as was the case before 1.0.14.
218 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
219 inherits from ERROR instead of just CONDITION.
220 * new feature: SB-INTROSPECT can provide source locations for instances
221 as well. (thanks to Tobian Ritterweiler)
222 * optimization: binding special variables now generates smaller code
223 on threaded platforms.
224 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
225 and cases where no :TEST is given but the compiler can infer that
226 the element to search is of type (OR FIXNUM (NOT NUMBER)).
227 * optimization: better LOGNOT on fixnums.
228 * optimization: modular arithmetic for a particular requested width
229 is implemented using a tagged representation unless a better
230 representation is available.
231 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
232 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
233 for wellformedness and returns a specialized vector. (reported by
234 Francois-Rene Rideau)
235 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
236 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
237 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
238 unscheduling timer at the same time as another timer fires.
239 * bug fix: don't reschedule timers for dead threads.
240 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
241 * bug fix: copying output from RUN-PROGRAM to a stream signalled
242 bogus errors if select() was interrupted.
243 * enhancement: add support for fcntl's struct flock to SB-POSIX.
245 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
246 * enhancement: cleaner backtraces for interactive interrupts, as
247 well as other cases where the interesting frames used to be
248 obscured by interrupt handling frames.
249 * enhancement: untracing a whole package using (UNTRACE "FOO") is
250 now supported, and tracing a whole package using (TRACE "FOO") now
251 traces SETF-functions as well.
252 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
253 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
254 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
255 * unreadably printed representation of hash-tables now includes
257 * bug fix: partially fixed #188: type propagation from assignments
258 is now more efficient.
259 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
260 'DOUBLE-FLOAT) are not flushable.
261 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
262 full address of the object, and none of the tag bits.
263 * bug fix: readably printing hash-tables now respects other printer
264 control variables. (reported by Cedric St-Jean)
265 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
266 keyword with MAKE-HASH-TABLE.
267 * bug fix: export SB-POSIX:MKSTEMP.
268 * bug fix: SORT was not interrupt safe.
269 * bug fix: XREF accounts for the last node of each basic-block as
271 * bug fix: MAKE-INSTANCE optimizations interacted badly with
272 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
273 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
274 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
275 create obsolete instances in the case of redefinition or
276 obsoletion of a superclass. (thanks to Andy Hefner)
277 * bug fix: Support for the Alpha architecture has been revived; it had
278 suffered somewhat from lack of maintenance since sbcl-1.0.
279 * improvements to the Windows port:
280 ** The system detects the codepage to use at startup, instead of
281 using the value from the saved core. (thanks to Kei Suzuki)
283 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
284 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
285 (see documentation for details.)
286 * revived support for OpenBSD (contributed by Josh Elsasser)
287 * partially fixed bug #108: ROOM no longer suffers from occasional
288 (AVER (SAP= CURRENT END)) failures.
289 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
290 bodies no longer cause a WARNING to be signalled. (reported by
292 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
293 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
294 the bug in TRUNCATE.) (reported by Michael Weber)
295 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
296 no samples. (reported by Andy Hefner)
297 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
298 no longer appear as (NIL ...) frames in backtraces.
299 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
300 is now actually exported from SB-EXT as documented. (reported by
302 * bug fix: sb-aclrepl now correctly understands how to inspect
303 single-floats on 64-bit platforms where single-floats are not boxed.
304 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
305 yet been finalized. (reported by Levente Meszaros)
306 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
307 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
308 SBCL-specific optimize qualities.
310 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
311 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
312 an executable in the search path, and does so in the child
313 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
314 been removed; it can be found in the CVS history, for somebody who
315 needs that search behavior (see the manual).
316 * minor incompatible change: compiler policy re. weakening type
317 checks has changed: now type checks are weakened only if SAFETY < 2
319 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
320 unparsing of directory pathnames as files. Analogously,
321 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
322 filename to parse into a directory pathname.
323 * enhancement: implicit generic function creation now signals a
324 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
325 users can bind handlers for and muffle around calls to LOAD.
326 * enhancement: RUN-PROGRAM allows unicode arguments and environments
327 to be used (using the default stream external format), and allows
328 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
329 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
330 strings and vectors whose element-type or simplicity is not fully
331 known at compile-time.
332 * optimization: STRING-TRIM and related functions no longer allocate
333 a new string when no trimming needs to be performed. These functions
334 are also faster than before when the input string has been declared
336 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
337 * optimization: READ-LINE is significantly faster for files containing
339 * optimization: non-open coded uses of character comparison operators
340 (e.g. char=) no longer cons when called with more than one parameter
341 on platforms supporting dynamic-extent allocation.
342 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
343 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
344 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
346 * bug fix: some sequence functions elided bounds checking when
348 * bug fix: too liberal weakening of union-type checks when SPEED >
350 * bug fix: more bogus fixnum declarations in ROOM implementation
353 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
354 * new feature: MAKE-HASH-TABLE now experimentally accepts a
355 :SYNCHRONIZED argument, which makes the hash-table safe for
356 concurrent accesses (but not iteration.) See also:
357 SB-EXT:WITH-LOCKED-HASH-TABLE, and
358 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
359 * optimization: CONCATENATE on strings is an order of magnitude faster
360 in code compiled with (> SPEED SPACE).
361 * optimization: SUBSEQ is ~50% faster on lists.
362 * bug fix: bug 417 fixed -- source location reporting is now more robust.
363 * bug fix: SUBSEQ on a list will now correctly signal an error if if
364 END is smaller then START.
365 * bug fix: SB-PROFILE will no longer report extra consing for nested
366 calls to profiled functions.
367 * bug fix: ROOM implementation had bogus fixnum declarations which
368 could cause type-errors when calling ROOM on large images.
369 * bug fix: if file compilation is aborted, the partial fasl is now
370 deleted, and COMPILE-FILE returns NIL as the primary value.
371 * bug fix: number of thread safety issues relating to SBCL's internal
372 hash-table usage have been fixed.
373 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
374 be returned from its body when the values were being returned
375 using unknown-values return convection and the W-P-O was wrapped
376 inside an UNWIND-PROTECT.
377 * bug fix: sb-posix should now compile again under Windows, enabling
380 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
381 * incompatible change: hash-table accessor functions are no longer
382 automatically protected by locks. Concurrent accesses on the same hash-table
383 from multiple threads can give inconsistent results or even corrupt the
384 hash-table completely. Multi-threaded applications should do their own
385 locking at the correct granularity. In the current implementation it is
386 still safe to have multiple readers access the same table, but it's not
387 guaranteed that this property will be maintained in future releases.
388 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
389 and will signal an error at runtime.
390 * enhancement: SB-THREAD package now exports a semaphore interface.
391 * enhancement: CONS can now stack-allocate on x86 and
392 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
394 * enhancement: nested lists can now be stack allocated on
395 platforms providing stack allocation support.
396 * enhancement: dynamic-extent support has been extended to support
397 cases where there are multiple possible sources for the stack
399 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
400 if the mutex is uncontested on Linux.
401 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
402 as specified by the CLHS. (thanks to Tobias Rittweiler)
403 * bug fix: NaN comparison now works on x86-64.
404 * bug fix: CLOSE :ABORT T on a stream with pending output now
406 * bug fix: instances of non-standard metaclasses using standard
407 instance structure protocol sometimes missed the slot type checks
409 * bug fix: known functions can be inlined in high-debug code.
411 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
412 * minor incompatible change: the MSI installer on Windows no longer
413 associates .lisp and .fasl files with the installed SBCL.
414 * minor incompatible change: :UNIX is no longer present in *FEATURES*
415 on Windows. (thanks to Luis Oliviera)
416 * new feature: SB-CLTL2 contrib module now implements
417 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
418 * optimization: scavenging weak pointers is now more efficient,
419 requiring O(1) instead of O(N) per weak pointer to identify
420 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
421 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
422 method bodies using constant slot-names when the first argument is
423 a specializer parameter for the method.
424 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
425 names, when the class of the instance is a direct instance
426 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
427 as slow as the constant slot-name case.)
428 * optimization: member type construction is now O(N) instead
430 * optimization: UNION and NUNION are now O(N+M) for large
431 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
433 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
434 now more readable in environments like Slime which display it.
435 (thanks to Tobias C. Rittweiler)
436 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
437 was able to infer the correct argument type for the object on which
438 the CAS operation was being performed.
439 * bug fix: copy propagation interfered with parallel assignment
440 semantics in local calls. (reported by Paul Khuong)
441 * bug fix: the signed modular fixnum shift compiled to wrong code on
442 x86 and x86-64. (spotted by a slight modification to some of
445 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
446 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
447 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
448 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
450 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
451 and generic function now print the object identity for anonymous
452 instances. (thanks to Attila Lendvai)
453 * enhancement: as an extension to MOP, SBCL now supports SETF of
454 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
455 (thanks to Attila Lendvai)
456 * enhancement: sb-cover has an alternate annotation mode which puts
457 the form annotations at the car of the form, instead of annotating
458 it completely. The benefit of this mode is that it shows better
459 which forms were instrumented by the compiler.
460 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
461 by printer control variables. (reported by Dan Corkill)
462 * bug fix: system leaked memory when delayed output was performed by
463 the OS in smaller chunks then expected. (thanks to David Smith)
464 * bug fix: system leaked memory when file streams were not closed
466 * bug fix: large objects written to slow streams that were modified
467 after the write could end up with the modified state written to
468 the underlying file descriptor.
469 * bug fix: multiple threads operating in parallel on the same stream
470 could cause buffer-overflows.
471 * bug fix: source location information is stored correctly
472 (broken since 1.0.6). This bug would generally show up as the
473 Slime debugger highlighting the wrong form when the sldb-show-source
475 * bug fix: Branch forms are again annotated as branches in the sb-cover
477 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
478 on x86-64. (reported by Gregory Vanuxem)
479 * bug fix: EQUALP could return wrong results for structures with raw
480 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
481 word). (reported by Vjacheslav Fyodorov)
482 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
485 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
486 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
487 atomic compare-and-swap operations on threaded platforms.
488 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
489 allows assigning a global minimum value to optimization qualities
490 (overriding proclamations and declarations).
491 * enhancement: closed over variables can be stack-allocated on x86
493 * performance bug fix: GETHASH and (SETF GETHASH) are once again
495 * optimization: slot definition lookup is now O(1). This speeds up
496 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
497 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
498 * optimization: ASSOC and MEMBER can now be open-coded for all
499 combinations of keyword arguments when second argument is constant
500 and SPEED >= SPACE. In other cases a specialized version is
502 * bug fix: using obsoleted structure instances with TYPEP and
503 generic functions now signals a sensible error.
504 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
505 (reported by Kristoffer Kvello)
506 * bug fix: backtrace construction is now more careful when making
507 lisp-objects from pointers on the stack, to avoid creating bogus
508 objects that can be seen by the GC.
509 * bug fix: defaulting of values in contexts expecting more than 7
510 variables now works on x86-64. (reported by Christopher Laux)
511 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
513 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
514 as the property-list of a symbol.
515 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
516 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
517 situations requested, are once again file-compileable. (reported
520 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
521 * MOP improvement: support for user-defined subclasses of
522 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
523 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
524 the expansion of DEFMETHOD, is responsible for generating a form
525 which creates a list of specializers when evaluated. Additional
526 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
527 debugging and introspective support.
528 * minor incompatible change: the (unsupported) spinlock interface
529 has changed: free spinlock now has the value NIL, and a held spinlock
530 has the owning thread as its value.
531 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
532 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
534 * enhancement: name of a socket-stream is now "a socket" instead of
536 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
537 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
538 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
539 for generic functions. (Thanks to Tobias C. Rittweiler)
540 * optimization: bignum printing speed has been improved by 20-40%
541 (depending on the bignum size.)
542 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
544 * bug fix: the cache used by the CLOS to store precomputed effective
545 methods, slot offsets, and constant return values is now thread and
547 * bug fix: generic function dispatch function updating is now thread
548 and interrupt safe (in the sense that the known issues have been
550 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
551 * bug fix: interning EQL-specializers is now thread and interrupt safe.
552 * bug fix: asdf systems with dependencies to the SB-POSIX or
553 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
554 * bug fix: interrupt safety of applicable method computation has been
557 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
558 * new contrib: sb-cover, an experimental code coverage tool, is included
560 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
561 significantly faster.
562 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
563 produce more efficient code on x86-64 if the compiler can determine
564 that (+ INDEX <constant>) does not require a bounds check and FOO
565 has an element type at least 8 bits wide.
566 * enhancement: a new, experimental synchronous timeout facility is
567 provided. Refer to SB-SYS:WITH-DEADLINE for details.
568 * enhancement: when a symbol name conflict error arises, the
569 conflicting symbols are always printed with a package prefix.
570 (thanks to Kevin Reid)
571 * enhancement: stepping is now once again supported on the SPARC and
572 MIPS platforms. (It is also now more likely to work on CheneyGC
574 * enhancement: sb-sprof can now also track and report accurate call
576 * bug fixes: the treatment of non-standard subclasses of
577 SB-MOP:SPECIALIZER is more correct.
578 * incompatible change: PURIFY no longer copies the data from the
579 dynamic space into the static and read-only spaces on platforms
580 that use the generational garbage collector
581 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
583 * bug fix: GC race condition occasionally resulting in crashes with
584 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
586 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
588 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
589 a STYLE-WARNING to be signalled (regression from 1.0.4.)
590 * bug fix: an asynchronous interrupt could previously leave the
591 system running with GC inhibited.
592 * bug fix: a DECLARE form evaluated at top-level now causes an error
593 rather than silently (or verbosely) returning NIL.
594 * bug fix: trying to dissassemble functions compiled at high DEBUG could
595 result in a error being signalled due to source form lookup errors.
596 (reported by Peter Graves)
598 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
599 * incompatible change: removed writer methods for host-ent-name,
600 host-ent-addresses -- changing the values did not update the DNS
602 * minor incompatible change: changed experimental JOIN-THREAD interface
603 * documentation: the manual now lists reader and writer methods
604 in class slot documentation sections. (thanks to Richard M Kreuter)
605 * documentation: unwinding from asyncronous events has been
606 documented as unsafe.
607 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
608 in multithreaded application code.
609 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
610 platforms. (thanks to James Anderson for the optimization hint)
611 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
613 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
614 variants no longer cons.
615 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
616 their NOT- variants no longer cons.
617 * optimization: EQUAL hash tables no longer use SXHASH for objects
618 of all data types, but instead use an EQL hash for types for which
619 EQUAL is the same as EQL.
620 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
621 are significantly faster.
622 * optimization: new STRING-OUTPUT-STREAM implementation is much
623 faster for large outputs and conses 30% less on average.
624 * enhancement: XREF information is now collected for references made
625 to global variables using SYMBOL-VALUE and a constant argument.
626 * enhancement: SIGINT now causes a specific condition
627 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
628 * bug fix: FILE-POSITION used to signal an error in some cases where
629 ANSI requires it to return NIL.
630 * bug fix: ADJUST-ARRAY is now interrupt-safe.
631 * bug fix: adding and removing fd-handlers is now interrupt-safe.
632 * bug fix: inlined calls to C now ensure 16byte stack alignment on
634 * bug fix: bad type declaration in the CLOS implementation has
635 been fixed. (reported by James Anderson)
636 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
637 (thanks to Lutz Euler)
638 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
639 required by the CLHS. (reported by Samium Gromoff)
640 * bug fix: dead unbound variable references now signal an error.
641 (reported by Marco Monteiro)
642 * bug fix: / with an unused value was being deleted in safe code.
643 (thanks to Marco Monteiro and Kevin Reid)
644 * bug fix: number of characters that can be written onto a single
645 line in a file is unlimited.
646 * bug fix: some GC deadlocks caused by asynchronous interrupts have
647 been fixed by inhibiting interrupts for when GC is disabled.
648 * bug fix: some interrupt safety issues with GETHASH, (SETF
649 GETHASH), CLRHASH and REMHASH have been fixed.
650 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
651 specifier no longer causes infinite recursion.
652 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
653 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
654 * bug fix: an &environment argument with the correct variable information
655 is passed to macros that are expanded during byte compilation.
656 (reported by Samium Gromoff)
657 * bug fix: a logic error was causing the hash function for bignums to
658 have excessive amounts of collisions. (reported by Faré Rideau)
659 * bug fix: modifying the contents of an array could change the return
660 value of SXHASH on that array, which is only allowed for strings
661 and bit vectors. (bug introduced in 0.9.16)
662 * improvement: the x86-64/darwin port now passes all tests (except
663 for the debugger tests) but should still be considered
664 experimental until this is fixed.
665 * improvement: a style-warning is signaled for CASE (etc) clauses with
666 duplicate keys. (thanks to Kevin Reid)
667 * improvement: macos/x86 and macos/x86-64 now use mach exception
668 handlers for dealing with illegal instructions (for trapping and
669 error handling) and memory protection violations (for GC).
671 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
672 * new platform: experimental support for x86-64/darwin (MacOS).
673 * incompatible change: the thread-safe (on most platforms) getaddrinfo
674 and getnameinfo sockets functions are used instead of gethostbyaddr
675 and gethostbyname, on platforms where the newer functions are available.
676 As a result, the ALIASES field of HOST-ENT will always be NIL on these
678 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
679 don't print the startup banner, but behave as if --noinform was passed
680 as a command line argument. (thanks to Kevin Reid)
681 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
682 * optimization: code using alien values with undeclared types is much faster.
683 * optimization: the compiler is now able to open code SEARCH in more cases.
684 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
685 * bug fix: using standardized COMMON-LISP special variables as loop
686 variables no longer signals bogus package lock violations. (reported
688 * bug fix: declaring local loop variables to be of a range-limited type
689 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
690 (reported by Andras Simon)
691 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
692 bugs remain on x86-64.)
693 * bug fix: the #= and ## reader macros now interact reasonably with
694 funcallable instances.
695 * bug fix: type-checks for function arguments were compiled using the
696 compiler policy settings of the wrong lexical environment (in 1.0.2
698 * bug fix: SHADOW accepts characters as string designators, as required
699 by the spec (thanks to Eric Marsden)
700 * bug fix: fixed GC safety issues when foreign functions are called with
701 non-base strings as arguments
702 * bug fix: more consistent error output for fatal-compiler-warnings, like
704 * bug fix: functions with proper names were showing up as NIL in
705 backtraces (reported by Edi Weitz, regression in 1.0.2)
707 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
708 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
709 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
710 produce more efficient code on the x86 if the compiler can determine
711 that (+ INDEX <constant>) does not require a bounds check and FOO
712 has an element type at least 8 bits wide.
713 * bug fix: references to undefined variables are handled the same way
714 in toplevel forms as in the normal compiler
715 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
717 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
718 (thanks to Jon Buller)
719 * improvement: faster compilation times for complex functions
720 * improvement: added readlink support to SB-POSIX (thanks to Richard
723 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
724 * improvement: experimental support for mach exception handling on
725 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
727 * improvement: support for GBK external format.
728 (thanks to Chun Tian (binghe))
729 * improvement: the debugger now displays variables that have been closed
730 over, in code compiled with (DEBUG 2) or higher
731 * improvement: support for executable cores on NetBSD (thanks to
733 * new feature: added a RESTART-FRAME debugger command
734 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
735 be used to provide an implementation for FILE-POSITION on Gray streams
736 (thanks to Eric Marsden)
737 * new feature: add syslog, openlog and closelog support to SB-POSIX
738 (thanks to Richard Kreuter)
739 * optimization: the function call overhead in code compiled with
740 a high DEBUG optimization setting is significantly
741 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
742 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
743 (thanks to Tony Martinez)
744 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
745 evaluated once (reported by Kevin Reid)
746 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
747 works again (thanks to Kilian Sprotte)
748 * bug fix: an error is signaled for tagbodies with duplicate tags
749 (thanks to Stephen Wilson)
750 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
751 * bug fix: Win32 port can now handle foreign code unwinding Lisp
752 stack frames from alien callbacks.
753 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
754 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
755 65k pages (thanks to David Woodhouse)
756 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
758 changes in sbcl-1.0.1 relative to sbcl-1.0:
759 * new platform: FreeBSD/x86-64, including support for threading.
760 * new feature: the compiler stores cross-referencing information
761 about function calls (who-calls), macroexpansion (who-macroexpands)
762 and special variables (who-binds, who-sets, who-references) for code
763 compiled with (< SPACE 3). This information is available through the
764 sb-introspect contrib.
765 * new feature: users may subclass SEQUENCE, and have instances of
766 these classes interoperate with standard Common Lisp functions if
767 a number of methods are defined. (This feature is experimental
768 and the interface subject to change based on feedback from SBCL
769 users and the general community)
770 * improvement: sb-sprof traces call stacks to an arbitrary depth on
771 x86/x86-64, rather than the previous fixed depth of 8
772 * improvement: another pthread back-end of mutex "pthread-futex".
773 * bug fix: non-ascii command-line arguments are processed correctly
774 (thanks to Yaroslav Kavenchuk)
775 * bug fix: non-required arguments were not passed correctly when a method
776 defined using DEFMETHOD was called from a mop-generated method using
777 CALL-NEXT-METHOD (reported by Pascal Costanza)
778 * bug fix: recursion is now permitted in accessors through
779 SLOT-UNBOUND. (reported by Pascal Costanza)
780 * bug fix: an error was signaled at startup if the HOME environment
781 variable was defined, but had an empty value (reported by Peter Van Eynde)
782 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
783 signal a PROGRAM-ERROR, not a BUG.
784 * bug fix: SB-POSIX:READDIR works when built with large file support.
785 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
787 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
788 for better automatic indentation support. (thanks to Matt Pillsbury)
789 * optimization: loading generic functions no longer takes O(n^2) time,
790 proportional to the amount of methods in the generic function
791 (reported by Todd Sabin and Jeremy Brown)
792 * optimization: the FIND and POSITION family of sequence functions
793 are significantly faster on arrays whose element types have been
795 * improvements to the Windows port:
796 ** Intermittent heap corruption problems have been fixed. (thanks
797 to Alastair Bridgewater)
798 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
800 ** Lisp is able to unwind foreign exception frames from alien
801 callbacks. (thanks to Alastair Bridgewater)
803 changes in sbcl-1.0 relative to sbcl-0.9.18:
804 * improvement: experimental support for threading on FreeBSD/x86.
805 (thanks to NIIMI Satoshi)
806 * improvement: runtime option --dynamic-space-size can be used
807 to set the size of the dynamic space reserved on startup.
808 * improvement: floating point modes in effect are now saved in
809 core, and restored on startup.
810 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
811 startup, not time since first call to GET-INTERNAL-REAL-TIME.
812 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
813 threads are not running after *SAVE-HOOKS* have run.
814 * improvement: writes to CLOS instance slots are type-checked in code
815 compiled with (SAFETY 3)
816 * improvement: floating-point exception handling on FreeBSD (thanks to
818 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
819 (thanks to Zach Beane)
820 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
822 * improvement: added support for the Shift-JIS external format.
823 (contributed by NIIMI Satoshi)
824 * improvement: callbacks are supported on Linux/PPC. (thanks to
826 * bug fix: compiler bug triggered by a (non-standard) VALUES
827 declaration in a LET* was fixed. (reported by Kaersten Poeck)
828 * bug fix: file compiler no longer confuses validated and already
829 dumped structurres. (reported by Kaersten Poeck)
830 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
831 fill-pointer signals a type-error as required. (thanks to
833 * bug fix: disassemly of funcallable instances works.
834 * bug fix: single stepping on PPC.
835 * bug fix: fix thread-safety problems in the type system (generally
836 manifesting as nonsensical errors like "STRING is a bad type specifier
837 for sequences" or "The value 1 is not of type FIXNUM")
838 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
839 (reported by Josip Gracin)
840 * bug fix: an error is signaled for attempts to displace arrays with
841 incompatible element types (thanks to Mario Mommer)
842 * bug fix: more correct handling of wide characters in debug info
843 (bug reported by Attila Lendvai and fixed by Juho Snellman)
844 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
845 and don't cause extra consing
846 * optimization: MAP and MAP-INTO are significantly faster on vectors
847 whose elements types have been declared.
848 * Improvements to SB-SPROF:
849 ** Support for allocation profiling
850 ** Reduced profiling overhead, especially for long profiling runs
851 * Improvements to the Windows port:
852 ** floating point exceptions are now reported correctly.
853 ** stack exhaustion detection works partially.
854 ** more accurate GET-INTERNAL-REAL-TIME.
855 ** better RUN-PROGRAM behaviour in case of errors in the child
857 ** PROBE-FILE now simplifies pathnames correctly.
858 ** DIRECTORY now works correctly with :WILD-INFERIORS.
859 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
860 1st 1970 (although time timezone data will be unreliable).
861 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
863 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
864 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
865 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
867 * bug fix: two potential GC deadlocks affecting threaded builds.
868 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
869 returns T (reported by Anton Kazennikov)
870 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
871 with non-variable places
872 * bug fix: remove a race condition in the setting of
873 funcallable-instance functions, this should make threaded CLOS
874 code more stable against memory faults.
875 * bug fix: corruption of specials when unbinding is interrupted by an
876 asynchronous unwind (reported by Hannu Koivisto)
877 * improvement: the debugger will now also display local variables that
878 are only used once, for code compiled with a DEBUG optimization quality
881 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
882 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
883 * incompatible change: External-format support for FFI calls. The
884 SB-ALIEN:C-STRING no longer implies an ASCII
885 external-format. Instead the string is subject to external-format
886 conversion. Additionally return values of type SB-ALIEN:C-STRING
887 are by default of type (SIMPLE-ARRAY CHARACTER), not
888 SIMPLE-BASE-STRING. For an alien type that behaves like the old
889 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
890 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
891 * incompatible change: SB-EXT package no longer contains the
892 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
893 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
895 * incompatible change: the single-stepper is no longer available
896 on Alpha, Mips and Sparc platforms.
897 * minor incompatible change: the direct superclasses of
898 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
899 not (STANDARD-OBJECT FUNCTION). This makes the
900 class-precedence-lists of GENERIC-FUNCTION and
901 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
903 * minor incompatible change: the default stream external format on
904 non-unicode SBCL is no longer detected from the locale, but is always
906 * new feature: Add a version of evaluator that uses an interpreter instead
907 of the compiler. EVAL still uses the compiler by default, to switch it
908 to use the interpreter, set the value of the variable
909 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
910 * minor incompatible change: the single-stepper REPL has been merged
911 with the normal debugger (see the "Stepping" heading of the debugger help
912 for more details). The debugger command STEP will no longer switch
913 to the single-stepper REPL.
914 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
915 for a type now works.
916 * bug fix: loading of default sysinit file works. (thanks to Leonid
918 * bug fix: better detection of circularities in the file-compiler.
919 (reported by Marco Monteiro)
920 * bug fix: the CL pathname functions now work with files that have
921 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
922 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
923 systems with Unix98 pty semantics.
924 * bug fix: ASDF-INSTALL will now work with bsd tar.
925 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
927 * bug fix: timers expiring in dead threads no longer cause a
928 type-error (reported by Paul "Nonny Mouse").
929 * bug fix: thanks to more lightweight single-stepper instrumentation,
930 code compiled with (DEBUG 3) will compile and execute significantly faster,
931 and will have more accurate type-inferencing than before
932 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
933 whose bindings are modified
934 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
935 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
936 ** replace dummy implementations of CL:MACHINE-INSTANCE and
937 CL:SOFTWARE-VERSION with proper version
939 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
940 * feature: implemented the READER-METHOD-CLASS and
941 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
942 as specified by AMOP.
943 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
945 * optimization: faster LOGCOUNT implementation on x86 and x86-64
946 (thanks to Lutz Euler)
947 * optimization: hashing of general arrays and vectors has been
948 improved. (reported by Any Fingerhut)
949 * enhancement: SB-INTROSPECT is now able to find definitions of
950 profiled functions. (thanks to Troels Henriksen)
951 * enhancement: compiler-macro expansion applies now to FUNCALL forms
952 as well, allowing compiler-macros for SETF-functions to expand.
953 * enhancement: step-instrumentation no longer wraps known
954 single-value functions with multiple-value context, allowing
955 better type inference.
956 * fixed bug #337: use of MAKE-METHOD in method combination now works
957 even in the presence of user-defined method classes. (reported by
958 Bruno Haible and Pascal Costanza)
959 * fixed bug #339(c): if there are applicable methods not part of any
960 long-form method-combination group, call INVALID-METHOD-ERROR.
961 (reported by Bruno Haible)
962 * fixed bug #361: the :FUNCTION initarg in the protocol for
963 initialization of methods can now be used to override
964 internally-produced optimized functions. (reported by Bruno
966 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
967 system-provided lambda expression no longer cause warnings about
968 unbound #:|pv-table| symbols.
969 * bug fix: improved the handling of type declarations and the
970 detection of violations for keyword arguments with non-constant
972 * bug fix: potentially erronous calls to PATHNAME and
973 MERGE-PATHNAMES were being flushed in some cases.
974 (reported by Richard Kreuter)
975 * bug fix: compiled calls to TYPEP were mishandling obsolete
976 instances. (reported by James Bielman and Attila Lendvai)
977 * bug fix: format strings with ~> without matching ~< no longer
978 trigger an AVER, but signal an understandable error instead.
979 (reported by Antonio Martinez)
980 * bug fix: specifying an output-file without a pathname-type for
981 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
982 of input-file instead of "fasl". (reported by Robert Dodier)
983 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
984 (reported by James Y Knight).
985 * bug fix: compiler-macros-function did not consider the environment
986 argument for shadowing by local functions.
987 * bug fix: compiler-macros expansion was inhibited by local INLINE
989 * bug fix: inline expansions of known functions were subject to
990 step-instrumentation in high debug policies, leading to problems
992 * bug fix: compiler failed to differentiate between different CONS
994 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
995 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
996 * bug fix: #368: incorrect use of expressed vs. upgraded array
998 * bug fix: #306a: more precise unions of array types.
999 * thread-safety improvements:
1000 ** CONDITION-WAIT could return early on Linux, if the thread was
1001 interrupted and subsequently continued with SIGCONT.
1002 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1004 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1005 * added support for the ucs-2 external format. (contributed by Ivan
1007 * minor incompatible change: pretty printing of objects of type
1008 (cons symbol) is, in the default pprint-dispatch-table, now
1009 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1011 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1012 called later in a class's lifetime, possibly as late as when the
1013 first instance of the class is created. Previously,
1014 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1015 class became finalizeable.
1016 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1017 for multibyte external-format streams. (thanks to "vbzoli")
1018 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1019 with :ALLOCATION :CLASS inherited from superclasses of the
1021 * fixed bug: anonymous classes can now be created using the :NAME
1022 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1023 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1024 * fixed bug: core-files saved with :EXECUTABLE T can again be
1025 executed when SBCL_HOME isn't set. (reported by James Knight)
1026 * fixed bug: toplevel LOCALLY forms with declarations could
1027 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1028 * fixed bug: printing from several different threads using different
1029 values of *print-case* could cause invalid output, due to
1030 some internal special variables of the printer not being bound
1031 thread-locally (reported by Max Mikhanosha)
1032 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1033 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1034 (reported by Pascal Costanza)
1035 * fixed bug: FINALIZE-INHERITANCE is called from
1036 REINITIALIZE-INSTANCE on classes when the class has previously
1037 been finalized, as required by AMOP.
1038 * minor code generation optimizations:
1039 ** better register allocation in CLOS dispatching functions
1040 ** overflow detection when coercing signed bytes to fixnums on x86-64
1041 ** is now implemented with one IMUL instruction instead of three shifts
1042 ** more efficient bit-vector access on x86 and x86-64
1043 ** more efficient access to raw structure slots on x86 and x86-64
1044 * fixed some bugs revealed by Paul Dietz' test suite:
1045 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1046 return its argument.
1048 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1049 * feature: thread support on Solaris/x86, and experimental thread support
1051 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1053 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1054 default method for SLOT-UNBOUND.
1055 * minor incompatible change: prevent the user from specializing the
1056 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1057 somewhat counter to the intent of the protocol, I (CSR) think, and
1058 additionally it just doesn't work in SBCL as currently
1059 implemented, thanks to optimizations (that are always valid for
1060 the other three SLOT-VALUEish functions, but not for the setter).
1061 * minor incompatible change: the :SB-LDB feature is now enabled by
1062 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1063 the low-level debugger.
1064 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1065 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
1066 down to an SB-SYS:FD-STREAM, allowing direct communication in
1067 more cases. Temporary files and pipes are still used as a fallback
1069 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1070 on the non-reentrant C functions localtime(3) and gmtime(3).
1071 * bug fix: global optimization policy was not visible in LOCALLY and
1073 * bug fix: class objects can be used as specializers in methods.
1074 (reported by Pascal Costanza)
1075 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1076 with a type error. (reported by blitz_ on #lisp)
1077 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1078 failed with a type error. (reported by Pascal Bourguignon)
1079 * bug fix: merging pathnames against defaults with :DIRECTORY
1080 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1081 (reported by James Y Knight)
1082 * bug fix: saving large (>2GB) cores on x86-64 now works
1083 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1084 constant index and a (simple-array (signed-byte 32)) array
1085 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1086 error (patch by Robert J. Macomber)
1087 * feature: TIME now displays the amount of run-time spent in GC
1088 * bug fix: The debugger now does a better job of respecting
1089 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1090 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1091 workaround for bug 403.)
1092 * bug fix: floating point exception handling now works on FreeBSD
1093 (thanks to NIIMI Satoshi)
1094 * fixed some bugs revealed by Paul Dietz' test suite:
1095 ** MISC.641: LET-conversion were not supposed to work in late
1097 * improvements to the Win32/x86 port:
1098 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1099 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1101 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1102 * new feature: source path information is generated for macro-expansion
1103 errors for use in IDE's like Slime (thanks to Helmut Eller)
1104 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1105 * bug fix: compiler does not loop forever on an invalid type in
1107 * improvement: compilation of most CLOS applications is significantly
1109 * optimization: added a limited bytecode compiler for simple toplevel
1110 forms, speeding up compilation and FASL loading
1111 * bug fix: the statistical profiler now properly distinguishes anonymous
1114 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1115 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1116 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1117 into subclasses of the finalized class. Now user calls to
1118 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1119 system return before any subclasses are finalized.
1120 * minor incompatible change: The reader no longer ignores errors
1121 regarding non-existent packages in #+ and #- feature tests.
1122 * new feature: command line options --no-sysinit, --no-userinit to
1123 inhibit loading the corresponding init files
1124 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1125 for sending data through UDP sockets (thanks to François-René Rideau)
1126 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1127 error is signalled instead (thanks to François-René Rideau)
1128 * minor incompatible change: Error signalling behaviour of lexical
1129 operations violating package locks has changed slightly. Refer to
1130 documentation on package locks for details.
1131 * enhancement: EVAL can process IF-expressions without resorting to the
1133 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1134 constant-folded properly due to missing out-of-line functions.
1135 (reported by Mika Pihlajamäki)
1136 * bug fix: LISTEN sometimes returned T even in cases where no data was
1137 immediately available from the stream
1138 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1139 were reversed. (reported by Levente Mészáros)
1140 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1141 when given a vector displaced to an adjustable array. (reported
1143 * bug fix: derivation of float boundaries from numbers outside the
1144 appropriate float range (reported by John Wiseman)
1145 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1147 * bug fix: precomputing cache entries for generic functions with
1148 some subclasses of specializers as yet invalid does not attempt to
1149 fill a cache line with a negative offset. (reported by Levente
1151 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1152 allow condition class objects as arguments to DOCUMENTATION and
1153 (SETF DOCUMENTATION); only find and set documentation for
1154 structure names for the STRUCTURE doc-type. (suggested by Gary
1156 * improvements to the Win32/x86 port:
1157 ** (user-homedir-pathname) and default initialization file
1158 locations now know about the user's "Documents and Settings"
1159 directory (thanks to Yaroslav Kavenchuk)
1160 ** run-program is implemented (thanks to Mike Thomas)
1161 ** sockets support (thanks to Timothy Ritchey)
1162 ** better backtrace support (thanks to Alastair Bridgewater)
1163 ** sb-grovel supported
1164 ** asdf-install and sb-posix work somewhat
1165 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1166 * improvements to the Solaris/x86 port:
1167 ** works on Solaris 11/Solaris Express
1168 ** floating-point exception handling support
1169 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1170 * fixed some bugs revealed by Paul Dietz' test suite:
1171 ** REMOVE-METHOD returns its generic function argument even when
1172 the method is not one of the generic functions' methods.
1173 ** objects declared as MEMBER types can be admissible arguments to
1174 structure accessors.
1175 ** printing characters should simply be printed by the FORMAT ~:C
1177 ** compiler failure when compiling functions with hairy constant
1178 defaults for optional parameters.
1179 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1180 function, which is already optimized.
1182 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1183 * new platform: experimental support for SBCL x86/Darwin, including
1184 MacOS X 10.4.5 on Intel.
1185 * new feature: Unicode character names are now known to the system
1186 (through CHAR-NAME and NAME-CHAR).
1187 * new feature: the filesystem location of SBCL's core file is
1188 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1189 * minor incompatible change: the contrib modules SB-POSIX and
1190 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1191 this change is to make it easier to distribute
1192 location-independent binaries.
1193 * bug fix: as implied by AMOP, standardized classes no longer have
1194 slots named by external symbols of public packages. (reported by
1196 * optimization: calling functions via constant symbols -- as in
1197 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1198 via the function object as in (FUNCALL #'FOO).
1199 * optimization: on x86 and x86-64, the calling convention for the
1200 case of dealing with an unknown number of values has been altered
1201 to be friendlier to the prediction heuristics implemented,
1202 particularly with reference to CALL and RET pairing. (thanks to
1203 Alastair Bridgewater)
1204 * optimization: on x86, the code for access to thread-local symbol
1205 values and binding/unbinding of thread-local symbols is smaller.
1206 (thanks to Alastair Bridgewater)
1207 * enhancement: CONSTANTP is now able to determine constantness of
1208 more complex forms, including calls to constant-foldable standardized
1209 functions and some special forms beyond QUOTE.
1210 * fixed bug: occasional GC crashes on Solaris/x86
1211 * optimization: x86-64 supports stack allocation of results of simple
1212 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1213 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1214 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1215 (thanks to James Knight)
1216 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1217 by the standard (thanks to Stephen Compall)
1219 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1220 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1221 be used for bundling the runtime and the core file into one
1222 executable binary. This feature is not currently supported on all SBCL
1223 platforms. (thanks to James Bielman and NIIMI Satoshi)
1224 * new feature: a generational or ephemeral garbage collector is now
1225 the default on the PowerPC platform (both Linux and Darwin). The
1226 old Cheney (stop and copy) collector is a build-time option.
1227 (thanks to Cyrus Harmon, after Raymond Toy)
1228 * minor incompatible change: the method by which SBCL finds its
1229 contributed modules has changed; it no longer relies on symbolic
1230 links from an $SBCL_HOME/systems directory, but searches directly
1231 in the subdirectories of $SBCL_HOME.
1232 * enhancement: the dynamic heap size on the Linux/PPC platform is
1233 markedly larger, even using the older Cheney garbage collector.
1234 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1235 certain complicated string types. (reported by Gary King)
1236 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1237 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1238 * fixed bug: DOCUMENTATION on structures no longer signals an
1239 error if no structure type of the right name exists. (reported by
1241 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1242 applicable non-standard primary or :AROUND method on
1243 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1244 no longer cause unbound variable errors. (reported by Kevin Reid)
1245 * fixed bug: metacircle resolution in cases where methods have slots
1246 added before the slots from STANDARD-METHOD. (reported by Jean
1248 * fixed bug: the Power PC assembly code for calling into lisp
1249 sometimes computed its return address wrongly (depending on the
1250 15th bit of the address, fixed at link time). This used to
1251 manifest itself in a segmentation violation while building PCL.
1252 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1253 many others over the years)
1254 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1255 the stack pointer (treated as a signed integer) is negative.
1256 (thanks to Peter van Eynde)
1258 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1259 * new platform: experimental support for the Windows operating
1260 system has been added. (thanks to Alastair Bridgewater)
1261 * fixed several bugs in and robustified the PPC FFI (including
1262 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1263 * bug fix: allow non-simple string symbol names (reported by Paul
1265 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1267 * bug fix: saving a core corrupted callbacks on x86/x86-64
1268 * bug fix: closed a loophole in metacircularity detection and
1269 grounding in the PCL implementation of CLOS.
1270 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1271 (reported by tomppa on #lisp)
1272 * bug fix: clear the direction flag on Lisp -> C transitions, as
1273 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1274 (reported by Andrej Grozin and Hendrik Maryns)
1275 * optimization: major improvements to GC efficiency on GENCGC platforms
1276 * optimization: faster implementation of EQUAL
1277 * optimization: emit more efficient opcodes for some common
1278 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1279 * optimization: several other minor code-generation improvements on x86-64
1280 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1282 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1283 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1284 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1285 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1286 by ANSI as a generic function, and for consistency (SETF
1287 GENERIC-FUNCTION-NAME) is treated likewise.
1288 * fixed bug #233.b: make constraint propagation notice when a variable
1289 value is changed after it is referenced but before it is used
1290 * fixed bug #296: no more arbitrary behaviour from filenames with
1291 odd characters as --load arguments. (reported by Adam Warner)
1292 * fixed bug #390: :CHARACTER-SET pathname components now work as
1293 expected. (reported by Tim Daly Jr)
1294 * fixed bug #391: complicated :TYPE intersections in slot
1295 definitions no longer cause an error in PCL internals.
1296 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1297 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1298 returning the number of octets which would be written to the
1299 file-stream. (thanks to Robert J. Macomber)
1300 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1301 arguments to the internal stream functions. (thanks to David
1303 * bug fix: stack exhaustion is now handled on the x86/SunOS
1304 platform. (thanks to Daisuke Homma)
1305 * bug fix: more accurate ROOM results on GENCGC platforms
1306 * optimization: improved type inference for arithmetic-for
1307 index variables in LOOP
1308 * optimization: faster floating-point SQRT on x86-64
1309 * fixed some bugs revealed by Paul Dietz' test suite:
1310 ** DOCUMENTATION returns NIL instead of "" for method combinations
1311 that don't have a docstring
1313 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1314 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1315 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1316 therefore are not customizeable by user code (as seems to be at
1317 least permitted and maybe required by AMOP). As a consolation,
1318 however, the SBCL implementation of these functions now calls
1319 REINITIALIZE-INSTANCE as specified by AMOP.
1320 * bug fix: slot-definition documentation is propagated as per ANSI
1321 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1322 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1323 argument, as do eslotd initialization methods. (from Pascal
1324 Costanza's "Closer" project)
1325 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1326 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1328 * bug fix: it is now possible to have more than one subclass of
1329 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1330 (reported by Bruno Haible, Pascal Costanza and others)
1331 * bug fix: the dependent update protocol now works for generic
1332 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1333 and Pascal Costanza)
1334 * bug fix: condition-class instances corresponding to
1335 DEFINE-CONDITION forms are now created eagerly. (reported by
1336 Kalle Olavi Niemitalo on comp.lang.lisp)
1337 * bug fix: floating point printing is more accurate in some
1338 circumstances. (thanks to Simon Alexander)
1339 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1340 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1341 * bug fix: callbacks on OS X now preserve stack-alignment.
1342 (thanks to Cyrus Harmon)
1343 * enhancement: the x86-64 disassembler is much better at
1344 disassembling SSE instructions. (thanks to Lutz Euler)
1345 * enhancement: CHECK-TYPE on a variable now allows type-inference
1346 (thanks to Vincent Arkesteijn)
1347 * optimization: improved performance of EUC-JP external format.
1348 (thanks to NIIMI Satoshi)
1349 * optimization: performance improvements to IO on file streams of
1350 :ELEMENT-TYPE CHARACTER
1351 * optimization: much faster memory allocation on x86-64
1352 * optimization: faster garbage collections (latency and throughput)
1354 * optimization: faster bignum subtraction and division on x86 and x86-64
1355 (thanks to Lutz Euler)
1356 * optimization: more accurate type-inference for loops that use a
1357 floating point index variable or a negative step.
1359 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1360 * bug fix: add a workaround to SBCL looping infinitely at startup on
1361 Linux kernels with apparently buggy implementations of personality().
1362 (thanks to Svein Ove Aas)
1363 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1364 (thanks to David Lichteblau)
1365 * bug fix: local bindings shadow global symbol macros for
1366 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1367 * optimization: non-open coded uses of numeric comparison operators
1368 (e.g. >) no longer cons when called with more than one parameter
1369 on platforms supporting dynamic-extent allocation.
1370 * enhancement: saving cores with foreign code loaded is now supported
1371 on MIPS/Linux in addition to the previously supported platforms.
1372 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1373 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1374 * bug fix: division by zero in sb-sprof when no samples were collected
1375 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1376 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1377 (thanks to Alexander Kjeldaas)
1378 * new feature: ignoring the return values of destructive functions
1379 like NREVERSE signals a compile-time style-warning.
1380 (thanks to Kevin Reid)
1381 * new feature: the alignment of alien structure fields can be
1382 explicitly specified. (thanks to Cyrus Harmon)
1383 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1384 and dump core on SIGQUIT
1386 ** incompatible change: threads do not inherit values of specials
1387 from their parents (see manual)
1388 ** bug fix: threads stacks belonging to dead threads are freed by the
1389 next exiting thread, no need to gc to collect thread stacks anymore
1390 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1391 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1392 ** bug fix: dynamic variable and thread start related gc lossage
1393 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1395 * fixed some bugs revealed by Paul Dietz' test suite:
1396 ** SUBTYPEP is slightly more accurate on heinously complicated
1397 CONS types where some of the members have uncertain (in the
1398 NIL, NIL sense) type relationships to each other.
1400 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1401 no longer purified unless :PURIFY T is explicitly specified.
1402 ** Non-purified cores are significantly smaller than before
1404 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1405 * new feature: timers based on Zach Beane's excellent timer package
1406 * added support for the following external formats: koi8-u,
1407 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1408 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1409 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1410 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1411 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1412 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1413 * incompatible change: a threaded SBCL will no longer revert to
1414 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1415 * bug fix: interrupts are disabled until startup is complete; no
1416 more sigsegvs when receiving a signal to soon
1417 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1419 * bug fix: add a workaround for the memory randomization features in
1420 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1421 workaround will only be in effect on systems with the proc filesystem
1423 * bug fix: printing objects of type HASH-TABLE signals a
1424 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1426 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1427 have been running for over 50 days. (reported by Gilbert Baumann)
1428 * bug fix: the logic for getting names of functions gets less
1429 confused when confronded with alternate-metaclass
1430 funcallable-instances. (reported by Cyrus Harmon)
1431 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1432 consistent, even on internal alternate-metaclass objects.
1433 * bug fix: flush closure information collected by physical
1434 environment analysis prepass before the main pass. (bug reported
1436 * bug fix: compiler pack phase does not modify a hashtable, which is
1437 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1438 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1439 may be expanded. (reported by Alexander <malishev> on c.l.l)
1440 * bug fix: no more style-warnings when compiling code using LAST
1441 under high-SPEED low-SPACE settings. (reported by David Wallin
1443 * fixed some incompatibilities between SBCL's MOP and the MOP
1445 ** the METAOBJECT class is now implemented;
1446 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1447 STANDARD-OBJECT, as required;
1448 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1449 now compatible (as required by VALIDATE-SUPERCLASS); there
1450 remains a consistency requirement over the presence or absence
1451 of the FUNCTION class in the superclass list of finalized
1452 classes; see the manual for more details;
1453 ** the :AROUND method for COMPUTE-SLOTS on
1454 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1455 requested slot ordering.
1457 ** bug fix: parent thread now can be gc'ed even with a live
1459 ** bug fix: binding a special with PROGV to no value is not
1460 the same as that symbol not having been bound (thanks to
1462 ** bug fix: binding specials is thread safe (thanks to
1464 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1465 and SB-SPROF do not die with 'no handler for signal XX in
1466 interrupt_handle_now(..)' anymore
1467 ** bug fix: WITH-TIMEOUT works with multiple running threads
1468 * fixed some bugs revealed by Paul Dietz' test suite:
1469 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1470 the :method-class keyword argument.
1472 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1473 * new port: the Solaris operating system on x86 processors is now
1474 mostly supported, though some rough edges in the environment
1475 remain. (thanks to Daisuke Homma)
1476 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1477 space for its heap. (thanks to Thiemo Seufer)
1478 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1479 closures. (thanks to Thiemo Seufer)
1480 * minor incompatible change: eof selects abort in the debugger.
1481 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1482 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1484 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1485 deprecated, and will go away in a future revision of SBCL.
1486 * minor incompatible change: GC-ON and GC-OFF are no longer
1487 implemented with a counter, it does not matter how many times gc
1488 is switched on or off
1489 * bug fix: discriminating functions for generic function classes
1490 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1491 make invalid assumptions about method precedence order. (reported
1493 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1494 platforms. (thanks to Thiemo Seufer)
1495 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1496 * bug fix: correct pathname printing with printer escaping is on.
1497 (thanks to Kevin Reid)
1498 * bug fix: complex VOP definitions in "user-space" no longer trigger
1499 package locks. (reported by Zach Beane)
1500 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1501 for funcallable-instance functions, and is no different from
1503 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1504 characters in the current readtable. (reported by Nicholas Neuss)
1505 * bug fix: SUBTYPEP on various CONS types returns more a more
1506 accurate acknowledgment of its certainty.
1507 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1508 lists in some cases. This partially fixes bug 384.
1509 * flush all standard streams before prompting in the REPL and the
1511 * bug fix: signal handling and triggering gc do not conflict
1512 directly anymore, in particular a high frequency sb-sprof does
1513 not prevent gc from running
1514 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1515 approximation for timezone and DST information between the
1516 universal time epoch and the smallest negative 32-bit time_t.
1517 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1518 year 1899 so long as the encoded time is non-negative.
1519 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1520 the correct answer for non-constant rotate amounts when performing
1521 an inline 32-bit rotation.
1523 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1524 there is only one thread in the session
1525 ** bug fix: memory leak for streams created in one thread and
1526 written to in another
1527 ** bug fix: lockup when compiled with gcc4
1528 ** bug fix: race that allows the gc to be triggered when gc is
1530 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1531 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1533 ** bug fix: finished threads are gc'ed properly
1534 * fixed some bugs revealed by Paul Dietz' test suite:
1535 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1536 has arguments with a different set of applicable methods from
1537 the orignal arguments.
1538 ** The type error thrown by MAP now has a correct expected-type
1540 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1541 name a compiled function.
1542 ** Three MISC tests where a large bignum was improperly coerced to
1543 a float that couldn't represent that bignum during type
1544 derivation were fixed.
1545 ** SUBTYPEP can now handle types involving the negation of a
1546 list-form FUNCTION type.
1547 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1548 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1549 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1551 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1552 * New feature: Experimental support for bivalent streams: streams
1553 opened with :element-type :default now allow character and binary
1554 (unsigned-byte 8) I/O
1555 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1556 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1558 * bug fix: no more highly sporadic "couldn't check whether ~S is
1559 readable" when reading a stream and an interrupt hits in the middle
1560 of a select system call
1561 * compiler better recognizes complex arrays (reported by Eduardo
1563 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1564 for variables on SBCL built with linkage-tables. (reported by Luis
1566 * various error reporting improvements.
1567 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1568 (Thanks to Thiemo Seufer)
1569 * bug fix: nested reader invokations work correctly
1570 * bug fix: it is possible to have simultaneous references to foreign
1571 code and foreign data with the same name.
1573 ** added x86-64 support
1574 ** incompatible change: the threading api now works with thread
1575 objects instead of thread ids
1576 ** bug fix: threads are protected from signals and interruption when
1577 starting up or going down
1578 ** bug fix: a race where an exiting thread could lose its stack to gc
1579 ** bug fix: don't halt on infinite error in threads if possible
1580 ** fixed numerous gc deadlocks introduced in the pthread merge
1581 ** bug fix: fixed thread safety issues in read and print
1582 ** bug fix: debugger doesn't hang on session lock if interrupted at
1583 an inappropriate moment
1584 ** bug fix: run-program is now thread safe(r)
1585 ** bug fix: inner with-recursive-lock no longer releases the mutex
1586 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1587 * fixed some bugs revealed by Paul Dietz' test suite:
1588 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1589 EXPECTED-TYPE slots filled.
1590 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1592 ** MAKE-CONDITION accepts classes as type-designators.
1593 ** COMPILE may never return NIL.
1594 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1595 range before calling Unix time functions
1597 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1598 * numerous signal handling fixes to increase stability
1599 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1600 * minor incompatible change: we now correctly canonize default
1601 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1602 as per the MOP, rather than the historical (INITARG INITFUNCTION
1603 INITFORM). (reported by Bruno Haible)
1604 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1605 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1606 for more information.
1607 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1608 pathname is a directory pathname.
1609 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1610 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1612 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1613 :METACLASS argument in addition to a class name. (reported by
1614 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1615 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1616 (reported by Bruno Haible; patch from Gerd Moellmann)
1617 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1619 * bug fix: more cleanups to the floating point exception handling on
1620 x86-64 (thanks to James Knight)
1621 * bug fix: the compiler does not try to constant fold calls of
1622 COERCE and BIT-* functions when they return freshly constructed
1623 objects. (reported by Nikodemus Siivola and Paul Dietz)
1624 * optimization: improved the MIPS versions of generic (in the
1625 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1626 * optimization: direct conversion of (unsigned-byte 32) to floats on
1627 the PowerPC platform.
1628 * optimization: structure instances with raw slots now use less
1629 memory, and probably show better memory locality. (thanks to
1631 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1632 are treated as requests for stack allocation on the x86-64,
1633 Alpha-32, PPC and SPARC platforms.
1634 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1635 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1636 (thanks to Tony Martinez)
1638 ** gcing a dead thread can no longer lead to lockups
1639 ** threads block signals until they are set up properly
1640 ** errno is no longer shared by threads
1641 ** interrupt-thread restores the eflags register on x86
1642 ** fixed some lockups due to gc/thread interaction
1643 * fixed some bugs revealed by Paul Dietz' test suite:
1644 ** invalid dotted lists no longer raise a read error when
1645 *READ-SUPPRESS* is T
1646 ** signal an error if a symbol that names a declaration is used
1647 as the name of a type, or vice versa
1648 ** allow using the (declare (typespec var*)) abbreviation for
1649 (declare (type typespec var*)) with all type specifiers
1650 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1651 ** always return NIL from PPRINT-POP when OBJECT is NIL
1652 ** don't signal errors when pretty-printing malformed LABELS,
1653 FLET or MACROLET forms
1654 ** declarations in a DOLIST body are also in scope for the
1656 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1658 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1661 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1662 * fixed cross-compiler leakages that prevented building a 32-bit
1663 target with a 64-bit host compiler.
1664 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1665 opened with :IF-EXISTS :OVERWRITE.
1666 * fixed bug 281: error for an invalid qualifier in a short-form method
1667 combination method is not signalled until the faulty method is called.
1668 * bug fix: iteration variable type inferrer failed to deal with open
1669 intervals. (reported by Alan Shields)
1670 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1671 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1672 or more runtime options were provided to the sbcl binary.
1673 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1674 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1675 * optimization: there's now a fast-path for fixnum arguments in the
1676 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1677 * optimization: the code generated on x86-64 is more compact thanks
1678 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1679 * bug fix: floating-point exception handling now partly works on
1680 x86-64. (Thanks to James Knight)
1681 * improvement to the MIPS backend from Thiemo Seufer: C-style
1682 64-bit long long arguments and return values to alien functions
1684 * fixed some bugs revealed by Paul Dietz' test suite:
1685 ** the type-error signalled from WARN has a filled-in DATUM slot.
1686 ** the type-error required when a stream is not associated with
1687 a file has the stream as its datum.
1688 ** type-errors on single-floats on x86-64 no longer have
1689 :INVALID-OBJECT as the datum
1690 ** the type-errors signalled for invalid function names now have
1691 a correct expected type
1692 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1693 for structures defined with a :TYPE. Documentation strings for
1694 typed structures are no longer immediately discarded
1695 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1696 broadcast streams and synonym streams. FILE-LENGTH now also works
1697 on broadcast streams.
1699 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1700 * incompatible change: the --noprogrammer option, deprecated since
1701 version 0.7.5, has been removed. Please use the equivalent
1702 --disable-debugger option instead.
1703 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1705 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1706 inoperational for a while now) has been completely removed.
1707 * null lexical environments are now printed as #<NULL-LEXENV>,
1708 significantly reducing the amount of clutter in typical backtraces.
1709 * documentation on weak pointers, finalization, and after GC hooks
1710 has been added to the manual.
1711 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1712 as other specialized array types, is much faster. SUBSEQ and
1713 COPY-SEQ on such arrays have also been sped up.
1714 * optimization: EQL is now more efficient when at least other argument
1715 is known to be of type (OR FIXNUM (NOT NUMBER)).
1716 * fixed bug: the runtime is now less vulnerable to changes in the
1717 size of the SBCL object on OS X, and virtual memory is reserved for
1718 all spaces that need to be at a fixed address.
1719 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1720 * fixed bug: finalizers and after GC hooks that cause consing are now
1722 * fixed bug: compiler error messages and summaries are now printed to
1723 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1724 * fixed inference of the upper bound of an iteration variable.
1725 (reported by Rajat Datta).
1726 * fixed bug 211e: calling local functions with duplicated constant
1727 keyword argument no longer causes a bogus style warning about an
1729 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1730 ftype declaration to be disregarded. (reported by Dave Roberts)
1731 * fixed bug 373: caused by erronous compilation of references to alien
1732 variables in the runtime on ppc/darwin.
1733 * fixed bug 376: CONJUGATE type deriver.
1734 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1735 (reported by Baughn on #lisp)
1736 * a cross-compiler bug on non-x86 platforms has been identified and
1737 fixed. (thanks to Bruno Haible)
1738 * improvements to the MIPS runtime code for increased stability.
1739 (thanks to Thiemo Seufer)
1740 * increased the maximimum compact environment size to allow
1741 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1742 * partial workaround for bug 135: don't name the function we're
1743 calling for hairy cases of EVAL, so as not to accumulate one environment
1744 entry per call to EVAL. (reported by Kevin Reid)
1745 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1746 * optimization: type testing for non-vector arrays should be faster.
1747 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1749 * bug fix: setting 31st element of a bit vector to zero did not work
1751 * bug fix: redefining a class definition which failed due to a
1752 previous accessor / function clash now works (but see BUGS entry
1753 #380 for more problems in this area). (thanks to Zach Beane)
1754 * the long form of DEFINE-METHOD-COMBINATION disables method group
1755 checking when given a single method group with pattern *. (thanks
1757 * on x86 compiler supports stack allocation of results of simple
1758 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1759 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1760 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1761 * fixed some bugs related to Unicode integration:
1762 ** the restarts for recovering from input and output encoding
1763 errors only appear when there is in fact such an error to
1765 * fixed some bugs revealed by Paul Dietz' test suite:
1766 ** MISC.549 and similar: late transformation of unsafe type
1767 assertions into derived types caused unexpected code
1769 ** SCALE-FLOAT type deriver is less wrong.
1770 ** type derivers for EXP, LOG and similar functions compute result
1771 types for complex arguments better.
1772 ** (MISC.563) CONJUGATE type deriver works for very restricted
1774 ** out-of-line type testers for character strings are available.
1775 ** EQUAL compiler transform understands specialness of objects
1777 ** accessing double-floats stored on the stack now works on x86-64.
1778 ** debugger internals could sometimes create invalid lispobjs,
1779 resulting in GC crashes.
1780 ** MISC.548: type check weakening can convert required type into
1782 ** initialization forms for bindings are not in scope of free special
1785 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1786 * incompatible change: thread support for non-NPTL systems has
1787 been removed - locking is buggy and unreliable. A threaded
1788 SBCL build will now warn at startup and refuse to create
1789 new threads, unless futex support is detected in the kernel
1790 * incompatible change: the top level REPL now has only an ABORT
1791 restart associated with it, not TOPLEVEL and ABORT as it used to.
1792 TOP and TOPLEVEL are now available as debugger commands for
1793 returning to the top level.
1794 * incompatible change: forms evaluated in the REPL now use the
1795 global optimization policy.
1796 * incompatible change: user- and system-initialization files are
1797 no longer processed with LOAD, but by READ and EVAL; hence the
1798 global optimization policy, startup package, readtable, etc,
1800 * The .fasl file format number has been incremented because of
1801 various incompatible changes.
1802 * internal entry point details and argument counts no longer appear
1803 in backtraces unless explicitly requested by setting
1804 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1805 * built-in and standard functions no longer have names like "top
1806 level local call to FOO".
1807 * fixed bug 32: functions defined in non-null lexical environments
1808 now have more legible printed representation
1809 * fixed bug 33: functions defined in non-null lexical environemnts
1810 are now more amenable to inspection by INSPECT.
1811 * workaround for bug 354: XEPs no longer appear in backtraces unless
1812 explicitly requested.
1813 * fixed bug: receiving the signal which results from attempting to
1814 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1815 SIGBUS on other *BSDs) on architectures where the C stack is also
1816 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1817 to debugger confusion or wild instability. (reported by Cheuksan
1819 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1820 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1821 (reported by Lutz Euler)
1822 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1823 compiled code. (reported by Paul Dietz)
1824 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1825 specialized parameter to a method to something that is not TYPEP
1826 the specializer is now possible.
1827 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1828 face of package deletion.
1829 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1830 pathname. (reported by David Tolpin on c.l.l)
1831 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1832 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1833 STRING-STREAM and FILE-STREAM, respectively.
1834 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1835 than previously; in addition, its test suite is now run on build.
1836 (thanks to Luke Gorrie)
1837 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1838 * optimization: added a immediate representation for single-floats
1840 * fixed some bugs revealed by Paul Dietz' test suite:
1841 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1842 ** attempting to create a package with a colliding nickname causes
1843 correctable errors to be signalled.
1844 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1845 ** several x86-64 backend bugs related to sign-extension of immediate
1848 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1849 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1850 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1852 * fixed loading of multiply forward-referenced layouts.
1853 (thanks to Cheuksan Wang)
1854 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1855 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1856 (see COMPILE-FILE documentation for details of :PRINT option)
1857 * fixed bugs 19 and 317: fixed-format floating point printing is
1858 more accurate. This also fixes a bug reported by Adam Warner
1859 related to the ~@F format directive.
1860 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1862 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1863 dispatch table if the from-char is a dispatch macro character.
1864 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1865 by the wordsize no longer ignore the last word. (reported by Lutz
1867 * fixed bug in type checking of dynamic-extent variables. (reported
1869 * optimization: sequence traversal functions use their freedom to
1870 coerce function designators to functions.
1871 * optimization: code with many calls to CLOS methods specialized on
1872 CLOS classes has had redundant type checks removed.
1873 * optimization: added declarations to speed up operations that access
1874 the internal character database (for example STRING-UPCASE)
1875 * optimization: comparison operations between floats and sufficiently small
1876 fixnums no longer create extra rationals
1877 * fixed some bugs related to Unicode integration:
1878 ** portions of multibyte characters at the end of buffers for
1879 character-based file input are correctly transferred to the
1880 start of the buffer at the next read.
1881 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1882 passing it through to OPEN.
1883 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1884 argument given to internal calls to OPEN.
1885 ** the built-in comment readers (introduced by character sequences
1886 ";" and "#|") are more forgiving to encoding errors; they will
1887 STYLE-WARN and then attempt to resync the stream at a character
1888 boundary. (thanks to Teemu Kalvas)
1889 * fixed some bugs in the x86-64 port:
1890 ** Negative short int return values from c-calls are sign-extended
1892 ** The stack is aligned to 16-bytes for c-calls, as required by
1893 the ABI. (thanks to Cheuksan Wang)
1894 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1895 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1896 for enums. (thanks to Vincent Arkesteijn)
1897 ** Multiple small optimizations and bugfixes for floating point
1899 * fixed some bugs revealed by Paul Dietz' test suite:
1900 ** Space, Tab, Linefeed, Return and Page have the invalid
1901 secondary constituent character trait.
1902 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1904 ** WITH-INPUT-FROM-STRING should only update the index place on
1906 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1907 works more reliably.
1908 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1909 with one-dimensional arrays with fill pointers.
1910 ** TYPE-OF failed on a complex with an integer realpart and a
1912 ** compiler failure during type inference for the code of form
1913 (IF (EQL X (THE ...)) ...) (MISC.535).
1915 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1916 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1917 platform. The port supports 61-bit fixnums, large memory spaces
1918 and reloading shared object files.
1919 * enhancement: saving cores with foreign code loaded is now
1920 supported on ppc/Darwin in addition to the previously supported
1922 * enhancement: the statistical profiler now walks deeper into the
1923 call stack for more meaningful call-graphs and accrued time
1924 reports (x86/x86-64 only). It also now reports time spent in
1926 * enhancement: it is now possible to trace most individual methods
1927 of a generic function in addition to tracing the generic function
1929 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1931 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1932 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1933 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1934 produces an error. (thanks to Vincent Arkesteijn)
1935 * bug fix: NAMESTRING on pathnames with :WILD components in their
1936 directories works correctly. (thanks to Artem V. Andreev)
1937 * fixed bug 125: compiler preserves identity of closures. (reported
1939 * bug fixed: functions with &REST arguments sometimes failed with
1940 "Undefined function" when compiled with (DEBUG 3). (reported by
1942 * bug fix: overflow during compiling of setting element of a bit
1943 vector with constant index and value. (reported by Timmy Douglas)
1944 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1945 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1946 * on x86 compiler partially supports stack allocation of dynamic-extent
1948 * GO and RETURN-FROM do not check the extent of their exit points
1949 when compiled with SAFETY 0.
1950 * fixed some bugs related to Unicode integration:
1951 ** encoding and decoding errors are now much more robustly
1952 handled; it should now be possible to recover even from invalid
1953 input or output to the terminal. (thanks to Teemu Kalvas)
1954 ** provided a first cut at implementing STRING-TO-OCTETS and
1955 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1956 ** altered the SB-MD5 contributed module slightly, changing the
1957 interface just enough for it to be supportable for builds where
1958 lisp characters are not eight bits.
1959 * fixed some bugs revealed by Paul Dietz' test suite:
1960 ** the FORMATTER-generated functions for ~V[ conditionals require
1961 the correct number of arguments.
1962 ** READ-FROM-STRING returns the mandated second value when applied
1963 to displaced strings.
1964 ** the #\Rubout and #\Backspace characters are treated as invalid
1965 constituent characters by the tokenizer.
1967 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1968 * new feature: reloading changed shared object files with
1969 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1970 * new feature: references to foreign variables and functions
1971 can now be compiled and loaded before the corresponding shared
1972 object file is loaded, as long as the foreign definitions are
1973 available at runtime.
1974 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1975 supported. (thanks to Dan Debertin)
1976 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1977 just keywords. (thanks to Vincent Arkesteijn)
1978 * fixed bug #331: structure-class instances corresponding to
1979 DEFSTRUCT forms are now created eagerly.
1980 * fixed bug #345: backtraces from calls to undefined functions work
1981 on x86 as well. Related bug #61 is now also partially fixed on x86
1982 (backtraces from throws to unknown catch tags.)
1983 * bug fix: lambda-list parsing is now stricter vrt. order and number
1984 of lambda-list keywords.
1985 * bug fix: as specified by AMOP, an error is signalled if a
1986 class-option appears multiple times in a DEFCLASS form. (reported
1988 * bug fix: RANDOM can be compiled when the compiler derives the type
1989 of its numeric argument as a disjoint set of small integers.
1990 (reported by Paul Dietz)
1991 * bug fix: starting a core saved with shared objects loaded when
1992 those objects are not available no longer causes threaded SBCL to
1993 hang. (reported by Sean Ross)
1994 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1996 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1997 argument, which must always be NIL. (reported by Kalle Niemitalo)
1998 * bug fix: printing 1.0d+23 no longer results in an error.
1999 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2000 * bug fix: structure slot setters preserve evaluation order. (thanks
2002 * fixed some bugs related to Unicode integration:
2003 ** RUN-PROGRAM can allow its child to take input from a Lisp
2004 stream. (reported by Stefan Scholl)
2005 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2006 locales. (reported by Ken Causey)
2007 * fixed some bugs revealed by Paul Dietz' test suite:
2008 ** INCF, DECF and REMF evaluate their place form as specified in
2010 ** FORMATTER expands ~{ ~} iteration directives with V or #
2011 parameters correctly.
2012 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2013 consequent uses no arguments correctly.
2014 ** the system has a partial understanding of the (COMPLEX RATIO)
2016 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2019 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2020 * new feature: a build-time option (controlled by the :SB-UNICODE
2021 keyword feature, enabled by default) for building the system with
2022 support for the entire 21-bit character space defined by the
2024 * new feature: the system now has rudimentary external-format
2025 support; the primary user-visible change at this time is that
2026 characters with the high bit set (such as lower-case-e-acute) will
2027 print correctly to a terminal in a UTF-8 environment.
2028 * minor incompatible change: BASE-CHAR no longer names a class;
2029 however, CHARACTER continues to do so, as required by ANSI.
2030 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2031 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2032 has been moved to the SB-EXT package (temporarily re-exported from
2034 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2035 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2037 * fixed bug #351: better error-handlind and reporting for malformed
2039 * fixed bug #350: bignum-printing is now more memory-efficient,
2040 allowing printing of very large bignums, eg. (expt 2 10000000).
2041 (reported by Bruno Haible)
2042 * fixed bug #302: better primitive-type selection for intersection
2044 * fixed bug #308: non-graphic characters now all have names, as
2045 required. (reported by Bruno Haible)
2046 * bug fix: redefining a class with different superclasses now correctly
2047 removes it from the direct-subclasses of its previous superclasses.
2048 (reported by David Morse)
2049 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2050 new value now works. (reported by Bruno Haible)
2051 * bug fix: correct canonicalization of multiple non-standard slot
2052 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2053 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2054 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2055 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2056 now exists, an signals an error.
2057 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2058 during macroexpansion for non-iteration variables as well. (reported
2059 by Bruno Haible for CMUCL)
2060 * bug fix: Cyclic structures and unprintable objects in compiler
2061 messages no longer cause errors. (reported by Bruno Haible)
2062 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2063 and READ-FROM-STRING all now return a primary value of NIL if
2064 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2065 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2066 (reported by Bruno Haible for CMUCL)
2067 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2068 do not designate a condition. (reported by Bruno Haible for
2070 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2071 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2072 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2073 permitting resolution in favour of any of the conflicting symbols.
2074 (reported by Bruno Haible for CMUCL)
2075 * bug fix: EQUAL compiler optimizations is less aggressive on
2076 strings which can potentially compare true despite having distinct
2077 specialized array element types.
2078 * bug fix: unit enumerations can be defined without dividing by
2079 zero. (thanks to Vincent Arkesteijn)
2080 * FORMAT compile-time argument count checking has been enhanced.
2081 (report from Bruno Haible for CMUCL)
2082 * a partial workaround for the bug 262: the compiler does not try to
2083 inline-expand a local function doing RETURN-FROM from a deleted
2084 BLOCK. (thanks to Peter Denno for the bug report and to David
2085 Wragg for the simple test case)
2086 * fixed some bugs revealed by Paul Dietz' test suite:
2087 ** RENAME-PACKAGE allows all package designators as new package
2089 ** constraint propagation and lambda variable substitution are
2090 more cautious in dealing with partially deleted code.
2091 ** compiler sometimes lost reoptimization passes.
2092 ** CERROR, given a condition as condition designator, passes its
2093 remaining arguments to the continue format control without
2095 ** Case-altering FORMAT directives work correctly on non-ASCII
2097 ** The REAL type specifier handles bounds outside the floating
2098 point ranges without signalling FLOATING-POINT-OVERFLOW.
2099 ** Functions with IR1-transformations can create intercomponent
2100 references to global functions.
2101 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2103 ** FORMAT ~R treats a nil value for its first parameter correctly.
2105 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2106 * enhancement: saving cores with foreign code loaded is now
2107 supported on x86/NetBSD and sparc/Linux in addition to the previously
2108 supported platforms.
2109 * bug fix: on some platforms repeated installations caused multiple
2110 copies of HTML documentation to be installed -- should not happen
2111 any more. (reported by Stefan Scholl)
2112 * bug fix: parsing self-recursive alien record types multiple times
2113 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2114 original patch by Helmut Eller for CMUCL)
2115 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2116 (thanks to Richard Kreuter)
2117 * bug fix: defining classes whose accessors are methods on existing
2118 generic functions in other (locked) packages no longer signals
2119 bogus package lock violations. (reported by François-René Rideau)
2120 * bug fix: special variables as DEFMETHOD parameters no longer have
2121 associated bogus type declarations. (reported by David Wragg and
2123 * bug fix: read-write consistency on streams of element-type
2124 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2125 * bug fix: redefiniton of the only method of a generic function with
2126 no DEFGENERIC no longer emits a full WARNING. In addition,
2127 redefinition of generic functions with no DEFGENERIC to an
2128 incompatible lambda list now signals an error. (thanks to Zach
2130 * bug fix: DEFGENERIC now works even when there's a function of the
2131 same name in an enclosing lexical environment. (thanks to Zach
2133 * fixed compiler failure, caused by instrumenting code during
2134 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2135 * optimization: added loop analysis and improved register allocation
2136 * fixed some bugs revealed by Paul Dietz' test suite:
2137 ** POSITION on displaced vectors with non-zero displacement
2138 returns the right answer.
2139 ** (SIMPLE-STRING) is a valid type specifier for sequence
2141 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2143 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2144 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2146 ** The justification version of the FORMAT ~< directive treats
2147 non-zero minpad parameter correctly.
2149 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2150 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2151 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2152 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2153 the supported interface.
2154 * new feature: Single-stepping of code compiled with DEBUG 2 or
2155 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2156 * new feature: saving cores with foreign code loaded is now
2157 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2158 Timothy Moore's work for CMUCL)
2159 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2160 parameters to * instead of NIL if no initform is supplied.
2161 (reported by Johan Bockgaard)
2162 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2163 correctly when FUNCALL appears as the car of the form. Note:
2164 despite this FUNCALL forms are not currently subject to
2165 compiler-macro expansion. (port of Raymond Toy's fix for the
2166 same from CMUCL, reported by Johan Bockgaard)
2167 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2168 (thanks for Teemu Kalvas)
2169 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2170 case instead of terminating the iteration (thanks for Julian
2171 Squires, Sean Champ and Raymond Toy)
2172 * bug fix: incorrect expansion of defgeneric that caused a style
2173 warning. (thanks for Zach Beane)
2174 * bug fix: slot accessor effective method computation works properly
2175 for classes with multiple non-standard applicable methods on
2176 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2177 * on x86 compiler supports stack allocation of results of LIST and
2178 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2179 CMUCL implementation by Gerd Moellmann)
2180 * fixed some bugs revealed by Paul Dietz' test suite:
2181 ** FORMAT strings with both the ~<~:;~> form of the justification
2182 directive and pretty-printing directives cause an error.
2184 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2185 * incompatible change: the internal functions
2186 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2187 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2188 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2189 instead of the old functions.
2190 * new feature: on platforms where "dladdr" is available foreign
2191 function names now appear in backtraces. (based on Helmut Eller's
2193 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2194 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2196 * documentation: ASDF manual is now distributed with SBCL.
2197 * bug fix: SBCL can now load its contributed modules with REQUIRE
2198 even if the system-provided entries have been removed from
2199 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2201 * bug fix: backtraces involving undefined functions or assembly
2202 routines are more informative. (thanks to Brian Downing)
2203 * bug fix: mutually referent alien structures now work correctly.
2204 (reported by Rick Taube)
2205 * bug fix: structures defined by WITH-ALIEN can be referred to
2206 within other definitions in the same WITH-ALIEN.
2207 * bug fix: division operators (MOD, TRUNCATE and the like) with
2208 constant zero divisors and integer dividends no longer generate
2210 * bug fix: provide default methods for INPUT-STREAM-P and
2211 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2212 * bug fix: improve the detection and resolution of MOP metacycles
2213 (where effective method computation depends on the generic function
2214 having its effective method computed). (reported by Bruno Haible)
2215 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2216 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2217 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2218 * bug fix: pathnames with relative directory components can be
2219 represented relative to default pathnames.
2220 * optimization: in taking the GCD of bignums, reduce the two bignums
2221 to approximately the same size (using Euclid's algorithm) before
2222 applying the more sophisticated binary GCD. (thanks to Juho
2224 * optimization: COUNT on bitvectors now operates word-at-a-time.
2225 * optimization: ASH with a positive, but not necessarily constant,
2226 (leftwards) shift, when in a modular context, is compiled to a
2228 * fixed some bugs revealed by Paul Dietz' test suite:
2229 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2230 if the corresponding argument is NIL.
2231 ** FORMAT directives accepting a minpad parameter treat negative
2233 ** literal commas as character parameters to FORMAT directives are
2235 ** literal spaces directly after ~<Newline> directives within a
2236 format-logical-block (~:< ~@:>) do not induce :FILL-style
2237 conditional newlines.
2238 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2239 as specified: it no longer includes conditional newlines).
2240 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2242 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2243 :PER-LINE-PREFIX argument does not evaluate to a string.
2245 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2246 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2247 the manual for details; for now, package locks can be disabled by
2248 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2249 if no major problems are found then it is likely that they will be
2250 compiled in unconditionally.
2251 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2252 unsupported operators on all platforms. To load a shared library
2253 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2254 object file, link it into a shared library outside of SBCL and
2255 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2257 * bug workaround: ROOM T and the GC were not working together
2258 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2259 reporting facility aren't preserved. That reporting has been
2260 disabled, so now until and unless someone figures out how to make it
2261 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2262 * minor incompatible change: as threatened around sbcl-0.8.0, the
2263 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2264 The symbols are also exported from SB-PCL for backwards
2265 compatibility, but more so than before SB-PCL should be treated as
2266 an implementation-internal package.
2267 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2269 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2270 It is known as of this release not to work on the Alpha, however.
2271 * fixed bug #167: errors signalled due to illegal syntax in method
2272 bodies are now more legible.
2273 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2274 designators and can hence be used with TYPEP.
2275 * fixed bug #333: CHECK-TYPE now ensures that the type error
2276 signalled, if any, has the right object to be accessed by
2277 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2278 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2279 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2281 * fixed bug #334: programmatic addition of slots using specialized
2282 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2283 and :ALLOCATION :CLASS slots.
2284 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2285 float exponents. (rereported by Peter Seibel)
2286 * fixed a bug: #\Space (and other whitespace characters) are no
2287 longer considered to be macro characters in standard syntax by
2288 GET-MACRO-CHARACTER.
2289 * fixed bug: initialization of condition class metaobjects no longer
2290 causes an instance of the condition to be created. (reported by
2292 * fixed bug: it is now possible to have slots such that
2293 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2294 description is neither :INSTANCE nor :CLASS.
2295 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2296 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2297 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2298 non-standard SLOT-DEFINITION-ALLOCATION values.
2299 * fixed bugs: various race conditions handling exiting threads.
2300 CL-PPCRE's thread torture test now passes.
2301 * fixed arguably-a-bug: GC time proportional to number of threads in
2302 system even when most of them are idle
2303 * optimization: improved performance of BIT and SBIT on bit-vectors.
2304 * .fasl file incompatibility: The fasl file version number has
2305 been incremented because of changes associated with package locks.
2307 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2308 * minor incompatible change: the system no longer provides
2309 optimization hints (conditions which are TYPEP
2310 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2311 mode; these hints are still emitted when higher SPEED optimization
2313 * new contrib module: a sampling profiler (profiling by statistical
2314 sampling, rather than by instrumenting functions) is available as
2315 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2316 * the behaviour of the standard function ED is now customizeable by
2317 third parties through a hook variable: see ED's documentation
2318 string for information on the protocol.
2319 * the compiler no longer emits efficiency notes for (FUNCALL X)
2320 when the type of X is uncertain under default optimization
2322 * fixed bug 276: mutating a binding of a specialized parameter to a
2323 method to something that is not TYPEP the specializer is now
2325 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2326 exhibit the required behaviour on the x86 platform. (thanks to
2327 Peter van Eynde, Eric Marsden and Bruno Haible)
2328 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2329 even for difficult arguments. (reported by Peter Graves)
2330 * fixed bug 141a: the backquote printer now descends quoted
2332 * fixed another bug in backquote printing: no more destructive
2333 modification of the form's list structure. (reported by Brian
2335 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2336 move between its address being taken and the call to
2337 interrupt_thread, fixing a crashing race condition.
2338 * the SB-POSIX contrib implementation has been adjusted so that it
2339 no longer exhibits ridiculously poor performance when constructing
2340 instances corresponding to C structs.
2342 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2343 * minor incompatible change: the sb-grovel contrib now treats C
2344 structures as alien (in the sense of SB-ALIEN) objects rather than
2345 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2346 has implications for memory management of client code
2347 (sb-grovel-returned objects must now be manually managed) and for
2348 type safety (alien objects now have full types).
2349 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2350 used to control emission of compiler diagnostics, rather than the
2351 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2352 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2353 quality should be considered deprecated.
2354 * install.sh now installs the user manual as well
2355 * (not quite a new documentable feature, but worth considering in
2356 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2357 beginnings of a semantically meaningful condition hierarchy is
2358 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2360 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2361 designator as the defaults argument.
2362 * fixed bug: Displaced arrays whose displaced-to array has become
2363 too small now cause ARRAY-DIMENSION to signal an error, providing
2364 for safer bounds-checking. (reported by Bruno Haible)
2365 * fixed bug: DEFCLASS slot definitions with identical :READER and
2366 :WRITER names now signal a reasonable error. (reported by Thomas
2368 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2370 * fixed bug: Invalid slot specification errors now print correctly.
2371 (thanks to Zach Beane)
2372 * fixed bug 320: Shared to local slot value transfers in class
2373 redefinitions now happen corrently from superclasses as
2374 well. (reported by Bruno Haible)
2375 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2377 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2378 improper lists correctly. (reported by Bruno Haible)
2379 * fixed bug 313: source-transform for <fun-name> was erroneously
2380 applied to a call of a value of a variable with name <fun-name>.
2381 (reported by Antonio Menezes Leitao)
2382 * fixed bug 307: The obsolete instance protocol ensures that
2383 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2384 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2385 remove dead unknown-values globs from the middle of the stack.
2386 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2387 easier to resume long computations after using *BREAK-ON-SIGNALS*
2388 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2389 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2390 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2391 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2392 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2393 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2395 * fixed a (fairly theoretical) bug in string printing: if
2396 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2397 does not have array-element-type equal to the most general string
2399 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2400 function name. (thanks to Nikodemus Siivola)
2401 * fixed bug: compile-time format string checker failed on
2402 non-closed ~{. (reported by Thomas F Burdick)
2403 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2404 don't warn on legal loop code involving more than one aggregate
2405 boolean. (thanks to Nikodemus Siivola)
2406 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2407 in keyword position to MAKE-PACKAGE was not being reported as
2409 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2410 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2412 * fixed bugs 280 and 312: the checking for multiple definitions in a
2413 file is less likely to become confused by uses of inline
2415 * fixed bug: the #S reader macro performs the keyword coercion
2416 specified for slot names. (reported by Kalle Niemitalo)
2417 * fixed bug: lambda lists may contain symbols whose names start with
2418 & but are not lambda-list-keywords; their occurrence triggers a
2420 * fixed bug 321: define-method-combination argument lambda lists do
2421 not cause invalid code to be generated when &optional or &aux
2422 variables have default values. (reported by Bruno Haible)
2423 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2424 have CLOS classes; this appears also to have fixed reported
2425 instability in stack exhaustion detection.
2426 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2427 MUFFLE-WARNING when no associated restart is present is now
2429 * optimization: rearranged the expansion of various defining macros
2430 so that each expands into only one top-level form in a
2431 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2433 * optimization: used a previously little-used slot in symbols to
2434 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2436 * fixed some bugs revealed by Paul Dietz' test suite:
2437 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2439 ** Fixed an optimization bug related to inheritance of initforms
2440 from local to shared slots.
2441 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2442 ** CAST optimizer forgot to flush argument derived type.
2443 ** print/read consistency on floats is now orders of magnitude
2444 more likely. (thanks also to Bruno Haible for a similar report
2446 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2447 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2448 for objects of type REAL. Make it so.
2449 ** SXHASH is specified (infelicitously) to respect similarity,
2450 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2451 it so. (thanks to Markus Ziegler)
2452 ** on the Alpha, the compiler succeeds in compiling functions
2453 returning a known number of arguments greater than 63.
2454 ** fixed handling of invalid NIL arguments in keyword position
2456 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2458 ** provided a readably-printable representation for RANDOM-STATE
2460 ** ensured that pathnames, where they have a namestring, always
2461 print using #P"..." syntax.
2463 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2464 * Support for the forthcoming 2.0 version of the NetBSD kernel
2465 running on x86 hardware has been added. (thanks to Perry
2466 E. Metzger most immediately, and others for their past work)
2467 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2468 current release version; SBCL's OpenBSD support had been broken
2469 since about the time of OpenBSD's switch to ELF binary format.)
2470 * [placeholder for DX summary]
2471 ** user code with &REST lists declared dynamic-extent, under high
2472 speed or space and low safety and debug optimization policy.
2473 * The manual has been converted to Texinfo format and the debugger
2474 chapter from the cmucl manual has been added.
2475 * A facility has been added to extract documentation strings from
2476 sbcl and store them as Texinfo-formatted snippets for inclusion in
2477 the manual (via Texinfo's @include directive)
2478 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2479 the test case to Dave Roberts)
2480 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2481 pointers. (reported by Sean Ross)
2482 * bug fix: PROFILE output is printed nicely even for large numerical
2483 values. (thanks to Zach Beane)
2484 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2485 greater than 32 handle EOF correctly.
2486 * bug fix: on X86 an immediate argument of the IMUL instruction is
2487 correctly printed in disassembly. (thanks to Lutz Euler)
2488 * bug fix: class slots in redefined classes preserve their old
2489 values. (thanks to Bruno Haible and Nikodemus Siivola)
2490 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2491 succeeds. (reported by Marco Baringer)
2492 * fixed some bugs revealed by Paul Dietz' test suite:
2493 ** READ-SEQUENCE now works on ECHO-STREAMs.
2494 ** RATIONALIZE works more according to its specification. (thanks
2497 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2498 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2499 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2500 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2501 should matter to you only if you rebind the printer control
2502 variables and then find you want different bindings in the
2503 debugger than in the ordinary execution of your program.)
2504 * The runtime build system has been tweaked to support building
2505 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2506 assembler and linker. (thanks to Nikodemus Siivola)
2507 * Unbound, undefined, undeclared variables now trigger full
2508 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2509 more often programmer error than deliberate exploitation of undefined
2511 * optimization: the hash algorithm for strings has changed to one
2512 that is less vulnerable to spurious collisions. (thanks to Juho
2514 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2515 needless bounds checking. (thanks to Juho Snellman)
2516 * optimization: implemented multiplication as a modular
2517 (UNSIGNED-BYTE 32) operation on the PPC backend.
2518 * fixed some bugs revealed by Paul Dietz' test suite:
2519 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2521 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2522 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2523 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2525 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2527 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2529 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2530 their output stream on EOF from read.
2531 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2532 have been read to end-of-file.
2533 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2535 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2536 description of determination of which consecutive characters
2538 ** Printing the "Space" character with escaping on now yields "#\\ ",
2539 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2540 ** Reading floating-point numbers with *READ-BASE* set to a number
2541 less than 10 works correctly.
2542 ** Reading floating-point numbers with *READ-BASE* set to a number
2543 more than 10 works correctly.
2544 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2545 the readtable currently in effect.
2547 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2548 * minor incompatible change: parsing of namestrings on a physical
2549 (Unix) host has changed; numbers after the final #\. in a
2550 namestring are no longer interpreted as a version field. This is
2551 intented to be largely invisible to the user, except that the
2552 meaning of the namestring "*.*.*" has changed: it now refers to a
2553 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2554 should usually be replaced by
2555 (make-pathname :name :wild :type :wild :version :wild)
2556 with the added benefit that this is more likely to be portable.
2557 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2558 option to OPEN now signals an error if the file being opened
2559 exists; this may have an impact on existing code.
2560 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2561 signals received, on the PowerPC platforms (both Linux and
2562 Darwin). (thanks to Pierre Mai for pointing out the location of
2564 * several fixes on OS X: The system now builds and runs cleanly on
2565 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2566 (thanks to Brian Mastenbrook)
2567 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2568 second-resolution: integer multiples of 1/3600 between -24 and 24.
2569 (thanks to Vincent Arkesteijn)
2570 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2571 type when called with 1 argument; PEEK-CHAR checked type of
2572 PEEK-TYPE only after having read first character from a
2573 stream. (reported by Peter Graves)
2574 * bug fix: the garbage collector now has much better locality
2575 behaviour, and in particular no longer treats all memory as being
2576 exhausted when many small objects point to each other in a deeply
2578 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2579 recognized as being TYPEP their class.
2580 * bug fix: the PUSHNEW documentation string has been corrected.
2581 (thanks to Vincent Arkesteijn)
2582 * bug fix: defaulting of the value for the last of an atypically
2583 large number of multiple values being bound was not being
2584 performed correctly on the Alpha or PPC platforms
2585 * optimization: implemented multiplication as a modular
2586 (UNSIGNED-BYTE 32) operation on the x86 backend.
2587 * optimization: SEARCH on simple-base-strings can now be open-coded.
2588 (see also contrib/compiler-extras.lisp for inspiration for
2589 teaching the compiler about the Boyer-Moore algorithm).
2590 * value, returned by MAX (and MIN) called with several EQUALP, but
2591 not EQL, arguments now does not depend on compiler settings.
2592 * fixed some bugs revealed by Paul Dietz' test suite:
2593 ** in stack analysis liveness information is propagated from
2594 non-local entry points.
2595 ** pathwise CAST removing failed when the CAST node did not start
2597 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2598 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2600 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2601 host is already defined.
2602 ** RENAME-FILE works on streams instead of signalling an internal
2604 ** PEEK-CHAR uses the current readtable when determining whether
2605 or not a character is whitespace.
2606 ** MERGE-PATHNAMES handles the case when the pathname does not
2607 specify a name while the default-pathname specifies a version.
2608 ** Pathnames now stand a better chance of respecting print/read
2610 ** Attempting to use standardized file system operators with a
2611 pathname with invalid :DIRECTORY components signals a
2613 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2615 ** DIRECTORY on logical pathnames is more correct.
2616 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2617 signal a TYPE-ERROR if their argument is not a stream.
2618 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2619 designator argument does not designate a stream.
2620 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2621 examining the synonym.
2622 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2624 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2625 element-type larger than ([UN]SIGNED-BYTE 32).
2627 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2628 * When built with the :SB-FUTEX feature, threaded builds now take
2629 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2630 for faster/more reliable mutex and condition variable support.
2631 * Incompatible change (but one you probably shouldn't have been using
2632 anyway): the interface and code for arbitrating between multiple
2633 threads in the same user session has been redesigned.
2634 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2635 error when called without an explicit environment argument.
2636 (thanks to Nikodemus Siivola)
2637 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2638 confused about their position. (thanks to Adam Warner and Gerd
2640 * bug fix: Pretty printing backquoted forms with unquotations in the
2641 argument list position of various code constructs such as LAMBDA
2642 now works correctly. (reported by Paul Dietz)
2643 * bug fix: Pretty printing unquotations no longer loses all
2644 stream position information.
2645 * optimization: performance of string output streams is now less
2646 poor for multiple small sequence writes.
2647 * optimization: performance of CSUBTYPEP in the presence of complex
2648 expressions involving CONS and NOT many times has been improved.
2649 (reported by Paul Dietz)
2650 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2652 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2653 streams for the REPL, for the debugger, and so on. (thanks to
2655 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2656 * fixed some bugs revealed by Paul Dietz' test suite:
2657 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2658 is ignored if it is too small, rather than propagating through
2659 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2660 ** extremely complex negations of CONS types were not being
2661 sufficiently canonicalized, leading to inconsistencies in
2663 ** VALUES tranformer lost derived type.
2665 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2666 * fixed a bootstrapping bug: the build process no longer assumes
2667 that the various BOOLE-related constants have the same value in
2668 host and target lisps. (noted by Paul Dietz' test suite on an
2669 SBCL binary built from CLISP)
2670 * The system can now be dynamically linked on the MIPS platform,
2671 which enables dynamic loading of foreign code from Lisp. (thanks
2672 to Ralf Baechle for discussions on the MIPS ABI)
2673 * The system now records debugging information for its own source
2674 files in a filesystem-position-independent manner, relative to
2675 the "SYS" logical host.
2676 * fixed a compiler bug: MV-LET convertion did not check references
2677 to the "max args" entry point. (reported by Brian Downing)
2678 * tweaked disassembly notes to be less confident about proclaiming
2679 some instruction as an LRA. (thanks to Brian Downing)
2680 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2681 now each have their own history, command character, and other
2682 characteristics. (thanks to David Lichteblau)
2683 * fixed some bugs revealed by Paul Dietz' test suite:
2684 ** compiler failure in compiling LOGAND expressions including a
2686 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2687 shift greater than 32.
2688 ** FLUSH-DEST did not mark blocks for type check regeneration.
2689 ** HANDLER-CASE failed to accept declarations in handler clauses
2690 in some circumstances.
2692 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2693 * New code in contrib/sb-introspect (still probably not entirely
2694 stable yet) provides some support for smart Lisp development
2695 environments like SLIME.
2696 * The conditions signalled for errors occurring when loading .fasl
2697 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2698 in a way which should help ASDF recover gracefully.
2699 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2700 stuff has been cleaned up. If you code contrib/ stuff, this might
2701 affect you, and you can look at contrib/README, contrib/STANDARDS,
2702 and/or the 0.8.4.27 diff to check.
2703 * In full calls the compiler now does not generate checks for declared
2704 argument types for all arguments.
2705 * various threading fixes
2706 ** and some experimental patches which didn't make it into
2707 the main tree for this release, but which are shipped in
2708 contrib/experimental-thread.patch as a possible fix for some
2709 failures (deadlock, spinning...) in GC-intensive multithreaded
2711 * fixed PPC build problem (source code incompatibility of different
2712 library versions): added offsetof() hackery which attempts to divine
2713 where glibc maintainers put uc_mcontext today
2714 * fixed bug 282: compiler does not trust type assertions while passing
2715 arguments to a full call.
2716 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2717 &OPTIONAL VALUES type parameter.
2718 * fix bug 214: algorithm for noting rejected templates is now more
2719 similar to that of template seletion. (also reported by rydis on
2721 * fixed bug 141b: printing backquoted information readably and prettily
2722 inserts a space where necessary.
2723 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2724 (CHARACTER 10) are now reported as errors, rather than propagated
2725 as unknown types. (reported by piso on #lisp)
2726 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2727 with values NIL and :ERROR. (thanks to Milan Zamazal)
2728 * fixed bug 191c: CLOS now does proper keyword argument checking as
2729 described in CLHS 7.6.5 and 7.6.5.1.
2730 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2731 longer raise an error; further, using a list as a for-as-arithmetic
2732 counter now raises a meaningful error.
2733 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2734 now understood by sequence creation functions such as MAKE-SEQUENCE
2736 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2737 a STRING-INPUT-STREAM.
2738 * compiler enhancement: SIGNUM is now better able to derive the type
2740 * type declarations inside WITH-SLOTS are checked. (reported by
2742 * fixed some bugs revealed by Paul Dietz' test suite:
2743 ** incorrect optimization of TRUNCATE for a positive first
2744 argument and negative second.
2745 ** compiler failure in let-convertion during flushing dead code.
2746 ** compiler failure while deriving type of TRUNCATE on an
2747 interval, containing 0.
2748 ** ASH of a negative bignum by a negative bignum count now returns
2750 ** intersection of CONS types now canonicalizes properly, fixing
2751 inconsistencies in SUBTYPEP.
2753 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2754 * incompatible change: The --disable-debugger command line
2755 option now clobbers the debugger at a more fundamental
2756 level, by redefining #'INVOKE-DEBUGGER instead of by
2757 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2758 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2759 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2760 enter the debugger REPL and then suffer recursive errors
2761 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2762 while under the new system, BREAK in --disable-debugger mode
2763 terminates the system just as an unhandled error would.
2764 * fixed compiler performance when processing loops with a step >1;
2765 * bug fix: DOCUMENTATION now retrieves generic function
2766 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2767 support has been systematized, and now supports the methods
2768 specified by ANSI, along with a default method and a method for
2769 slot documentation. (reported by Nathan Froyd)
2770 * bug fix: effective methods associated with a generic function are
2771 no longer cached over a change of that generic function's method
2772 combination. (reported by Andreas Fuchs)
2773 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2774 names a non-existent directory. (thanks to Andreas Fuchs)
2775 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2776 the PPC platform, lead to stack corruption; this has been fixed.
2777 (reported by Rainer Joswig)
2778 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2779 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2780 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2781 arithmetic in general, is now much more reliable on the PPC
2783 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2784 platform now returns the right answer.
2785 * optimization: restored some effective method precomputation in
2786 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2787 precomputation is now tunable.
2788 * optimization: compiler-internal data structure use has been
2789 reviewed, and changes have been made that should improve the
2790 performance of the compiler by about 20%.
2791 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2792 simple-base-strings and simple-bit-vectors is improved.
2793 * optimization: the optimization of 32-bit logical and arithmetic
2794 functions introduced in version 0.8.3 on the x86 has been
2795 implemented on the mips, ppc and sparc platforms; an
2796 implementation of the same facility, but for 64-bit arithmetic,
2797 has been added for the alpha.
2798 * microoptimization: the compiler is better able to make use of the
2799 x86 LEA instruction for multiplication by constants.
2800 * bug fix: in some situations compiler did not report usage of
2801 generic arithmetic in (SPEED 3) policy.
2802 * bug 145b fix: compiler used wrong type specifier while converting
2803 MEMBER-types to numeric.
2804 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2806 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2807 index argument now works properly on the Alpha platform.
2808 * bug fix: floating point exception treatment on the Alpha platform
2810 * bug fix: FILE-POSITION works much better on string input and
2811 output streams. (thanks to Nikodemus Siivola)
2812 * bug fix: many threading/garbage collection symptoms sorted.
2813 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2814 might be pseudo-atomic.
2815 * internal change: Stopping for GC is now done with signals not
2816 ptrace. GC is now done in whichever thread wanted it, instead of
2818 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2820 * bug fix: result form in DO is not contained in the implicit
2822 * incompatible change: ICR structure is changed; the value part of
2823 CONTINUATION is now called LVAR; corresponding functions are
2824 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2825 * added type deriver for ISQRT (thanks to Robert E. Brown).
2826 * bug fix: better support for loading from the command line when an
2827 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2829 * fixed some bugs revealed by Paul Dietz' test suite:
2830 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2831 ** ROUND and FROUND now give the right answer when given very
2832 small float arguments.
2833 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2835 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2836 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2837 ** LCM should return a non-negative integer.
2838 ** PARSE-INTEGER returned the index of a terminator instead of the
2839 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2840 ** PARSE-INTEGER returned an incorrect index being applied to a
2842 ** LCM with two arguments of 0 returns 0 rather than signalling
2844 ** unsigned addition of a 32-bit constant with the high bit set no
2845 longer causes an internal compiler error on the x86.
2846 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2847 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2848 with negative last argument.
2849 ** byte specifiers with zero size and position no longer cause
2850 an error during type derivation.
2851 ** bignum multiplication on the Alpha platform now returns the
2853 * porting: The system now builds on SuSE AMD64, although it still
2854 generates a 32-bit binary.
2855 * .fasl file incompatibility: The fasl file version number has
2856 been incremented (because of the changes to internal compiler
2857 data structures referred to above).
2859 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2860 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2861 more accurately, on the Darwin kernel running on PowerPC hardware.
2862 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2863 * Compiler code deletion notes now signal a condition of type
2864 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2865 an associated MUFFLE-WARNING restart.
2866 * The compiler now performs limited argument count validation of
2867 constant format strings in FORMAT, and where appropriate in ERROR,
2868 CERROR and WARN. (thanks to Gerd Moellmann)
2869 * New ASDF-INSTALL contrib can be used for automatic download and
2870 installation of third-party Lisp code from CCLAN or other sites
2872 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2873 which forces another thread to execute a function supplied by the
2875 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2876 now accept and act upon their :ELEMENT-TYPE keyword argument.
2877 (reported by Martin Atzmueller, Edi Weitz)
2878 * bug fix: FILE-POSITION now accepts position designators up to
2879 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2880 is the greater. (thanks to Patrik Nordebo)
2881 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2882 Kalle Olavi Niemitalo)
2883 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2884 printable. (reported by Eric Marsden)
2885 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2886 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2887 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2888 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2889 (thanks to Stig Sandoe)
2890 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2891 UNREAD-CHAR now correctly includes the unread character in the
2892 target sequence. (thanks to Gerd Moellmann)
2893 * bug fix in threaded builds: the system can now be suspended and
2894 resumed by shell job control with minimal disruption.
2895 * bug fixes in times and timezones >2038AD
2896 * better handling of "where is GNU make?" problem in build scripts
2897 (thanks to Nikodemus Siivola)
2898 * new optimization: inside a named function any reference to a
2899 function with the same name is considered to be a self-reference;
2900 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2901 optimization quality.
2902 * new optimization on x86: logical functions and + now have
2903 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2904 used when the result is truncated to 32 bits.
2905 * VALUES declaration is partially enabled.
2906 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2907 * fixed some bugs revealed by Paul Dietz' test suite:
2908 ** The system now obeys the constraint imposed by
2909 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2910 types form a lattice under type intersection.
2911 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2912 ** ASSOC now ignores NIL elements in an alist.
2913 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2914 and (1+ MOST-POSITIVE-FIXNUM) answers.
2915 ** The addition of a method with invalid qualifiers to a generic
2916 function does not cause an error to be signalled immediately;
2917 a warning is signalled, and the error is generated only on
2918 calling the generic function.
2919 * changed .fasl file version number, in part to add type codes for
2920 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2921 obscure ANSI requirements
2923 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2924 * fixed bug 148: failure to inline-expand a local function left
2925 garbage, confusing the compiler.
2926 * fixed bugs 3cd: structure slot readers perform type check if the
2927 slot can have an invalid value (i.e. it is either not initialized
2928 or can be written with a less specific slot writer).
2929 * bug fix: the compiler now traps array references to elements off
2930 the end of an array; previously, the bounds checking in some
2931 circumstances could go off-by-one.
2932 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2934 * type declarations for array element types now obey the description
2935 on the CLHS page "Declaration TYPE", as per discussions on
2936 sbcl-help around 2003-05-08. This means that a declaration
2937 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2938 declaration, all references to BAR will be asserted or assumed
2939 (with THE, so dependent on compiler policy) to involve objects of
2940 type FOO. Note that no such declaration is implied in
2941 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2942 * declared types of functions from the "Conditions"
2943 chapter. (reported by Paul Dietz)
2944 * bug fix: CERROR accepts a function as its first argument.
2945 * bug fix: NTH an NTHCDR accept a bignum as index
2946 arguments. (reported by Adam Warner)
2947 * optimization: character compare routines now optimize comparing
2948 against a constant character. (reported by Gilbert Baumann)
2949 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2950 argument now works properly on the MIPS platform.
2951 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2952 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2953 declared to have a functional type, does not cause an error
2955 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2956 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2957 * fixed bug: if last continuation of a deleted block has a
2958 destination, this destination should be deleted too. (reported by
2960 * fixed a bug in the bootstrap process: the host compiler's values
2961 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2962 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2963 test case from Patrik Nordebo)
2964 * improved the ability of the disassembler on the PPC platform to
2965 provide helpful disassembly notes.
2966 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2967 the class in more cases than previously.
2968 * bug fix: FILE-POSITION now understands :START and :END for
2969 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2970 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2971 8) data. (thanks to David Lichteblau)
2972 * bug fix: it is possible to add a method to a generic function
2973 without lambda list.
2974 * bug fix: reader failed to signal END-OF-FILE inside an
2975 object representation. (reported by Nikodemus Siivola)
2976 * fixed some bugs revealed by Paul Dietz' test suite:
2977 ** LAST and [N]BUTLAST should accept a bignum.
2978 ** condition slot accessors are methods.
2979 ** (VECTOR NIL) is a subtype of STRING.
2981 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2982 * minor incompatible change: some nonsensical specialized lambda
2983 lists (used in DEFMETHOD) which were previously ignored now signal
2985 * minor incompatible change: the system is now aware of the types of
2986 variables in the COMMON-LISP package, and will signal errors for
2987 most violations of these type constraints (where previously they
2988 were silently accepted).
2989 * minor incompatible change: COMPILE-FILE now uses the freedom
2990 afforded (ANSI 3.2.2.3) to use derived function types for
2991 functions defined in the same file. This also permits the system
2992 to warn on static type mismatches and function
2993 redefinition. (Currently it does not work with high DEBUG level.)
2994 * minor incompatible change: VALUES declaration is disabled.
2995 * When issuing notes, the compiler now signals a condition of type
2996 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2997 restart for use in user handlers. It is expected that the
2998 COMPILER-NOTE condition will eventually become a condition
2999 supertype to a hierarchy of note types, which will then be
3000 handleable in a similar fashion. However, at the moment, no such
3001 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3002 but it's an implementation detail, not a classification for the
3004 * Changes in type checking closed the following bugs:
3005 ** type checking of unused values (192b, 194d, 203);
3006 ** template selection based on unsafe type assertions (192c, 236);
3007 ** type checking in branches (194bc).
3008 * A short form of VALUES type specifier has ANSI meaning (it has
3009 increased the number of situations when SBCL cannot perform type
3011 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3012 keywords or constants is permissible.
3013 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3014 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3015 classes. (thanks to Antonio Martinez)
3016 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3017 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3018 argument to a pretty-print stream. Instead, it leaves any such
3019 operations to DESCRIBE-OBJECT methods.
3020 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3022 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3023 lambda lists. (thanks to David Lichteblau)
3024 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3025 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3026 (thanks to Gerd Moellmann)
3027 * fixed bug 246: increased compilation speed of long
3028 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3030 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3032 * DEFCONSTANT now throws a condition of type
3033 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3034 constant to a non-EQL value; CONTINUE and ABORT restarts
3035 respectively change and preserve the value.
3036 * fixed bug 63: The code walker, part of the implementation of CLOS,
3037 is now better at handling symbol macros.
3038 * bug fix: There is no longer an internal implementation type named
3039 CL:LENGTH. (reported by Raymond Toy)
3040 * bug fix: In macro-like defining macros/special operators the
3041 implicit block does not enclose the lambda list.
3042 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3043 enclosed in parentheses) are not suitable as type specifiers, and
3044 their use properly signals an error now.
3045 * bug fix: An argument count mismatch for a type specifier in code
3046 being compiled no longer causes an unhandled error at compile
3047 time, but signals a compile-time warning.
3048 * fixed simple vector readable printing
3049 * bug fix: DESCRIBE takes more care over whether the class
3050 precedence list slot of a class is bound before accessing it.
3051 (reported by Markus Krummenacker)
3052 * bug fix: FORMATTER can successfully compile pretty-printer format
3053 strings which use variants of the ~* directive inside.
3054 * bug fix: SEARCH now applies its TEST predicate to the elements of
3055 the arguments in the correct order. (thanks to Wolfhard Buss)
3056 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3057 in type inference. (thanks to Robert E. Brown)
3058 * bug fix: Reading in symbols with an explicit package name of ""
3059 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3060 * fixed some bugs revealed by Paul Dietz' test suite:
3061 ** NIL is now allowed as a structure slot name.
3062 ** Arbitrary numbers, not just REALs, are allowed in certain
3063 circumstances in LOOP for-as-arithmetic clauses.
3064 ** Multiple class redefinitions before slot access no longer
3065 causes a type error.
3066 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3067 association between the name and a class.
3068 ** Generic functions with non-standard method-combination and over
3069 five methods all of which return constants no longer return NIL
3070 after the first few invocations. (thanks to Gerd Moellmann)
3071 ** CALL-NEXT-METHOD with no arguments now passes the original
3072 values of the arguments, even in the presence of assignment.
3073 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3075 ** DISASSEMBLE works with closures and funcallable instances.
3076 ** ADD-METHOD now returns the generic function, not the new method.
3077 ** FIND-METHOD signals an error if the lengths of the specializers
3078 is incompatible with the generic function, even if the ERRORP
3080 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3081 which its argument is a member.
3082 ** DEFCLASS only redefines the class named by its class-name
3083 argument if that name is the proper name of the class;
3084 otherwise, it creates a new class.
3085 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3086 of the UNBOUND-SLOT condition to the name of the slot.
3087 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3088 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3089 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3090 SLOT-MAKUNBOUND in the specified fashion.
3092 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3093 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3094 cross-compilation host. As a consequence, we can now bootstrap our
3095 way up to SBCL starting with a bare gcc toolchain and human-readable
3096 source code (first the source to CLISP, then the source to SBCL).
3097 * A contributed module containing a partial implementation of the
3098 simple-streams interface has been included. (thanks to Rudi
3100 * A contributed module implementing the RFC1321 Message Digest
3101 Algorithm, known as MD5, has been included.
3102 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3103 no longer has any effect, as the code controlled by this feature
3104 has been deleted. (As far as we know, no-one has ever built using
3105 this feature, and its semantics were confused in any case).
3106 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3107 work on conditions (as required by the ANSI specification),
3108 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3109 expected behaviour on conditions. Users should note, however,
3110 that such behaviour is not required by the ANSI specification,
3111 and so use of this behaviour may render their code unportable.
3112 * fixed some bugs revealed by Paul Dietz' test suite:
3113 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3115 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3116 ** SXHASH on condition objects no longer returns NIL.
3117 ** :ALLOCATION :CLASS slots are better treated; their values are
3118 updated on class redefinition, and initforms inherited from
3119 superclasses are applied.
3120 ** REMOVE-METHOD returns its generic function argument even when
3121 no method was removed.
3122 ** SHARED-INITIALIZE now initializes the values of the requested
3123 slots, including those with :ALLOCATION :CLASS.
3124 ** ALLOCATE-INSTANCE now works on structure classes defined via
3125 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3127 ** SLOT-EXISTS-P now works on conditions, as well as structures
3129 ** MAKE-LOAD-FORM now has the required methods on
3130 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3131 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3132 keyword, and now implements the SLOT-NAMES argument.
3133 ** methods with &OPTIONAL arguments no longer allow too many
3134 arguments to be passed in the call without error.
3135 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3136 option is consistent with the required arguments of the generic
3137 function lambda list.
3138 * bug fix: REQUIRE accepts a string designator. (Thanks to
3140 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3141 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3142 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3143 * bug fix: The compiler now checks for duplicated variables in macro
3145 * bug fix: SETQ on globals returns the correct value.
3146 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3147 not a COMPILER-ERROR (followed by some other strange error on
3148 choosing the CONTINUE restart).
3149 * bug fix: make.sh and friends are now more consistent in the way that
3150 they look for GNU "make".
3152 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3153 * experimental native threads support (on x86 Linux >=2.4 only).
3154 This is not compiled in by default: you need to add :SB-THREAD to
3155 the target features. See the "Beyond ANSI" chapter of the manual
3157 * fix for longstanding nonANSIism: The old distinction between
3158 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3159 The return value from CL:FIND-CLASS is now a CLOS class, and
3160 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3161 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3162 * An interface to the MetaObject Protocol, as described in Kiczales,
3163 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3164 MIT Press, 1991, is available from the SB-MOP package.
3165 * incompatible change: the SB-PCL package should now be considered
3166 a private implementation detail, and no longer a semi-private MOP
3168 * minor incompatible change: due to rearrangement for threads, the
3169 control stack and binding stack are now allocated at arbitrary
3170 addresses instead of being hardcoded per-port. Users affected by
3171 this probably have to be doing advanced things with shared
3172 libraries, and will know who they are.
3173 * minor incompatible change: Previously, all --eval forms used were
3174 processed with READ before any of them were processed with EVAL.
3175 Now each --eval form is processed with both READ and EVAL before
3176 the next --eval form is processed. (Thus package operations like
3177 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3178 work as the user might reasonably expect.)
3179 * minor incompatible change: *STANDARD-INPUT* is now only an
3180 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3182 * minor incompatible change: Y-OR-N-P is now character-oriented,
3183 not line oriented. Also, YES-OR-NO-P now works without errors.
3184 (thanks to Antonio Martinez)
3185 * sb-aclrepl module improvements: an integrated inspector, added
3186 repl features, and a bug fix to :trace command.
3187 * Known functions, which cannot be open coded by the backend, are
3188 considered to be able to check types of their arguments. (fixing
3189 a bug report by Nathan J. Froyd)
3190 * fixed a bug in computing method discriminating functions: It is
3191 now possible to define methods specialized on classes which have
3192 forward-referenced superclasses. (thanks to Gerd Moellmann)
3193 * fixed evaluation order in optional entries (reported by Gilbert
3195 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3196 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3197 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3198 required-by-AMOP NAME argument, as well as CLASS and
3199 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3200 * fixed bug 20: DEFMETHOD can define methods using names that are
3201 not the proper names of classes to designate class specializers.
3202 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3203 Unix file descriptors, instead of blowing up. (thanks to Antonio
3205 * Garbage collection refactoring: user-visible change is that a
3206 call to the GC function during WITHOUT-GCING will not do garbage
3207 collection until the end of the WITHOUT-GCING. If you were doing
3208 this you were probably losing anyway.
3209 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3210 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3211 TYPEP the latter but not the former.
3212 * The compiler issues a full WARNING for calls to undefined functions
3213 with names from the CL package.
3214 * MAP-INTO for a vector destination is open coded. (reported by
3215 Brian Downing on c.l.l)
3216 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3217 documentation string.
3218 * fixed some bugs revealed by Paul Dietz' test suite:
3219 ** COPY-ALIST now signals an error if its argument is a dotted
3221 ** Condition slots are now accessed more correctly in the presence
3222 of multiple initargs for a given slot.
3223 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3224 exclude restarts of the same name associated with a different
3226 ** DEFCLASS of forward-referenced classes with another
3227 forward-referenced class in the superclasses list no longer
3229 ** Condition slots are now initialized once each, not multiple
3230 times. (thanks to Gerd Moellmann)
3231 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3233 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3234 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3235 arguments contain duplicated elements.
3236 ** RESTART-CASE understands local macros.
3237 ** RESTART-CASE associates exactly its own restarts with a condition.
3238 ** ENDP in safe mode checks its argument to be of type LIST.
3239 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3240 in question is unbound.
3241 ** Optimized MAKE-INSTANCE functions no longer cause internal
3242 assertion failures in the presence of duplicate initargs.
3243 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3244 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3245 propagated even in the ctor optimized implementation of
3247 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3249 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3250 * a better implementation of SXHASH on (simple) bit vectors,
3251 measured both in execution speed and in distribution of results
3252 over the positive fixnums, has been installed. Likewise, a better
3253 implementation of EQUAL for simple bit vectors is now available.
3254 * fixed CEILING optimization for a divisor of form 2^k.
3255 * fixed bug 240 (emitting extra style warnings "using the lexical
3256 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3257 by Antonio Martinez)
3258 * fixed SXHASH, giving different results for NIL depending on type
3259 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3260 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3261 variable. (found by Rolf Wester)
3262 * SBCL does not ignore type declarations for special
3263 variables. (reported by rif on c.l.l 2003-03-05)
3264 * some bug fixes in contrib/sb-aclrepl/
3265 * fixed some bugs revealed by Paul Dietz' test suite:
3266 ** a bug in the CONS type specifier, whereby the CAR and CDR
3267 types got intertwined, has been fixed;
3268 ** the type system is now able to reason about the interaction
3269 between INTEGER and RATIO types more completely;
3270 ** APPEND, [N]REVERSE and NRECONC check that those their
3271 arguments, which must be proper lists, are really so;
3272 ** An array specialized to be unable to hold elements has been
3273 implemented, as required -- yes, really -- by ANSI;
3274 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3275 on malformed property lists;
3277 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3278 * incompatible packaging change: in line with Unix convention,
3279 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3280 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3281 * REQUIRE and PROVIDE are now optionally capable of doing something
3282 useful. See the documentation string for REQUIRE.
3283 * infrastructure for a managed SBCL contrib system: contributed
3284 modules in this release include:
3285 ** the ASDF system definition facility;
3286 ** an interface to the BSD Sockets API;
3287 ** an ACL-like convenience interface to the repl;
3288 (thanks to Kevin Rosenberg)
3289 ** an implementation of ROTATE-BYTE, with efficient implementation
3291 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3292 gives a closer approximation to the right answer than previously.
3293 (thanks to Raymond Toy)
3294 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3295 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3296 optional environment arguments, as required by ANSI.
3297 * fixed bugs in other functions taking environment objects, allowing
3298 calls with an explicit NIL environment argument to be compiled
3300 * fixed bug 228: primary return values from
3301 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3302 COMPILE or FUNCTION.
3303 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3304 :TYPE will no longer signal errors on innocuous objects.
3305 * fixed bug 231b: SETQ is better at respecting type declarations in
3306 the lexical environment.
3307 * fixed a bug in DEFCLASS: classes named by symbols with no or
3308 unprintable packages can now be defined.
3309 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3310 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3311 * fixed compiler failure related to checking types of functions.
3312 (reported by Robert E. Brown)
3313 * the compiler is now much more consistent in its error-checking
3314 treatment of bounding index arguments to sequence functions: in
3315 (SAFETY 3) code, errors will be signalled in almost all cases if
3316 invalid sequence bounding indices are passed to functions defined
3317 by ANSI to operate on sequences.
3318 * fixed a bug in the build procedure: documentation of SBCL-specific
3319 packages is now preserved and available in the final Lisp image.
3320 * lifted FDEFINITION lookup out of loops in the implementation of
3321 many list operations. (thanks to Robert E. Brown)
3322 * fixed a bug in the reader: the #n# reader macro now works for
3323 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3324 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3325 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3326 E. Brown and Tony Martinez respectively)
3327 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3328 SLOT-BOUNDP: in particular, it is now able to optimize them much
3329 better, and is now not vulnerable to having packages renamed.
3330 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3331 always called when a slot is not present in an instance. (thanks
3333 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3334 longer signal an error on unreading a character following EOF on
3335 the previous constituent stream. (thanks to Tony Martinez)
3336 * fixed some bugs revealed by Paul Dietz' test suite:
3337 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3338 not just nonnegative fixnums;
3339 ** the logical bit-array operators such as BIT-AND now accept an
3340 explicit NIL for their "opt-arg" argument (to indicate a
3341 freshly-consed result bit-array);
3342 ** ELT now signals an error on an invalid sequence index in safe
3344 ** the type system is now cleverer about negations of numeric
3345 types, and consequently understands the BIGNUM and RATIO types
3347 ** the type system is now cleverer about the interaction between
3348 INTEGER and RATIO types: while bugs still remain, many more
3349 cases are accurately computed;
3350 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3351 if it is in the last clause;
3352 ** CONSTANTLY now correctly returns a side-effect-free function in
3354 ** DECLARE is no longer treated as a special-operator; in
3355 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3356 * incremented fasl file version number due to the change in the
3357 DEFSTRUCT-SLOT-DESCRIPTION structure.
3359 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3360 * minor incompatible change: code processed by the "interpreter" or
3361 EVAL now has a compilation optimization policy of (DEBUG 2)
3362 (changed from (DEBUG 1)) to improve debuggability of interactive
3363 development, and to allow the use of the debug RETURN command in
3365 * an experimental implementation of the RETURN command for the
3366 debugger has been included. (thanks to Frederik Kuivinen)
3367 * fixed bug 62: constraints were not propagated into a loop.
3368 * fixed bug in embedded calls of SORT (reported and investigated by
3370 * fixed some bugs revealed by Paul Dietz' test suite:
3371 ** printing and reading of arrays with some dimensions having
3372 length 0 (thanks to Gerd Moellmann);
3373 ** BOA constructor with &AUX argument without a default value does
3374 not cause a type error;
3375 ** CONSTANTP now returns true for all self-evaluating objects.
3377 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3378 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3379 accessors that are related by inheritance, as specified in the
3380 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3381 Valtteri Vuorikoski)
3382 * The compiler is now able to inline functions that were defined in
3383 a complex lexical environment (e.g. inside a MACROLET).
3384 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3386 * Support for the upcoming FreeBSD-5.0 release has been included.
3387 (thanks to Dag-Erling Smorgrav)
3388 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3389 effect when it is not in a toplevel context.
3390 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3391 stand a better chance of being correct. (thanks to Gerd
3393 * fixed bug in COERCE, which now signals an error on coercing a
3394 rational to a bounded real type which excludes the expected
3396 * The compiler is now able to derive types more accurately from the
3397 COERCE and COMPILE functions.
3398 * fixed bug 223: functional binding is considered to be constant
3399 only for symbols in the CL package.
3400 * fixed bug 231: SETQ did not check the type of a variable being set
3401 (reported by Robert E. Brown)
3402 * A new optimization for MAKE-INSTANCE has been included, fixing
3403 various bugs (including relating to :ALLOCATION :CLASS slots and
3404 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3406 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3407 ** As required by ANSI, LOOP now disallows anonymous collection
3408 clauses such as COLLECT I in conjunction with aggregate boolean
3409 clauses such as THEREIS (= I 1);
3410 ** LOOP now signals an error when any variable is reused in the
3411 same loop (including the potentially useful construct analogous
3412 to WITH A = 1 WITH A = (1+ A);
3413 ** IT is only a special loop symbol within the first clause of a
3414 conditional loop clause;
3415 ** LOOP with a typed iteration variable over a hashtable now
3416 signals a type error iff it should.
3417 * fixed some other bugs revealed by Paul Dietz' test suite:
3418 ** FILE-STREAM now names the class previously known as FD-STREAM;
3419 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3420 argument) no longer signals an error;
3421 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3422 of the slot symbol, rather than using the current package
3423 ((:CONC-NAME "") continues to intern the slot's name in the
3425 * incremented fasl file version number, because of the incompatible
3426 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3427 of the new implementation of DEFINE-COMPILER-MACRO.
3429 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3430 * Support for building SBCL for MIPS platforms running in
3431 little-endian mode has now been checked in, and basic
3432 functionality on said platforms verified.
3433 * minor incompatible change: PCL now records the pathname of a file
3434 in which methods and the like are defined, rather than its
3436 * minor incompatible change: TRUENAME now considers the truename of
3437 a file naming a directory to be the pathname with :DIRECTORY
3438 component indicating that directory.
3439 * minor incompatible change: a NAMED clause in the extended form of
3440 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3441 reason for the previous behaviour is unclear.
3442 * more systematization and improvement of CLOS and MOP conformance
3443 in PCL (thanks to Gerd Moellman and Pierre Mai):
3444 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3446 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3447 primary methods with no specializers;
3448 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3450 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3451 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3452 FORWARD-REFERENCED-CLASSes; error reporting on
3453 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3455 ** SXHASH on CLOS instances now uses a slot internal to the
3456 instance to return different numbers on distinct instances,
3457 while preserving the same return value through invocations of
3459 ** DEFMETHOD signals errors when methods with longer incongruent
3460 lambda lists are added to generic functions;
3461 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3462 CLASS, as specified in AMOP;
3463 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3464 on the order returned by the primary method for classes of
3465 class STANDARD-CLASS;
3466 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3467 * fixed some bugs shown by Paul Dietz' test suite:
3468 ** DOLIST puts its body in TAGBODY;
3469 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3471 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3472 value producing form;
3473 ** if more variables are given to PROGV than values, extra
3474 variables are bound and made to have no value;
3475 ** NSUBSTITUTE on list arguments gets the right answer with
3477 ** ELT signals an error of type TYPE-ERROR when the index argument
3478 is not a valid sequence index;
3479 ** LOOP signals (at macroexpansion time) an error of type
3480 PROGRAM-ERROR when duplicate variable names are found;
3481 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3482 ** FUNCALL of special-operators now cause an error of type
3484 ** PSETQ now works as required in the presence of side-effecting
3485 symbol-macro places;
3486 ** NCONC accepts any object as its last argument;
3487 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3489 ** loop-for-as-package does not require a package to be explicitely
3491 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3493 * fixed bug 166: compiler preserves "there is a way to go"
3494 invariant when deleting code.
3495 * fixed bug 172: macro lambda lists with required arguments after
3496 &REST arguments now cause an error to be signalled. (thanks to
3498 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3499 bug. (thanks to Matthew Danish)
3500 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3502 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3503 arguments are lexically rebound. (thanks to Gerd Moellmann and
3505 * fixed bug 194: error messages are now more informative when there
3506 is no primary method applicable in a call to a generic
3507 function. (thanks to Gerd Moellmann)
3508 * fixed bug in command line argument checking (thanks to Julian
3510 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3511 (thanks to Matthew Danish)
3512 * incremented fasl file version number, because of the
3513 SXHASH-related changes in the layout of CLOS data structures
3515 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3516 * minor incompatible change: The runtime (the Unix executable named
3517 "sbcl") is now much pickier about the .core files it will load.
3518 Essentially it now requires .core files to descend from the same
3519 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3520 as the runtime does. (The intent is to prevent the crashes which
3521 can occur, and which can even be reported as mysterious failures,
3522 when people patch the sources or change the build parameters
3523 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3524 sbcl and .core files.)
3525 * fixed bug: VALUES-LIST is no longer optimized away.
3526 * fixed bug 142: The FFI conversion of C string values to Lisp
3527 string values no longer conses excessively. (thanks to Nathan
3528 Froyd porting Raymond Toy's fix to CMU CL)
3529 * began to systematize and improve MOP conformance in PCL (thanks to
3530 Nathan Froyd, Gerd Moellman and Pierre Mai):
3531 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3533 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3534 ** FINALIZE-INHERITANCE is now called on class finalization;
3535 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3536 argument precedence order.
3537 * fixed bug 202: The compiler no longer fails on functions whose
3538 derived types contradict their declared type.
3539 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3540 so it can be non-toplevel.
3541 * The fasl file version number has changed (because of the new
3542 implementation of DEFMACRO).
3543 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3544 safe code, that any length requirement by their type-specifier
3545 argument is valid. The exceptions to this are described in bug
3547 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3548 functions now check the types of their inputs as required by ANSI.
3549 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3550 introduced symbol is DECLAREd to be SPECIAL.
3551 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3552 * fixed bug: :COUNT argument to sequence functions may be negative
3553 * fixed bug: body of DO-SYMBOLS may contain declarations
3554 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3555 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3556 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3557 symbol macro only once
3558 * fixed printing of call frame when argument list is unavailable
3559 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3560 * compiler no longer signals WARNING on unknown keyword
3563 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3564 * A beta-quality port to the mips architecture running Linux,
3565 based on the old CMUCL backend, has been made. It has been tested
3566 on a big-endian kernel, and works sufficiently well to be able to
3567 rebuild itself; it has not been tested in little-endian mode.
3568 * fixed an inconsistency between gencgc.c and purify.c which made
3569 dumping/loading .core files unreliable
3570 * fixed bug 120a: The compiler now deals correctly with IFs where
3571 the consequent is the same as the alternative, instead of
3572 misderiving the return type. (thanks to Alexey Dejneka)
3573 * fixed bug 113: Logical pathnames are now dumpable (the logical
3574 host is resolved at load-time, throwing an error if it is not
3576 * fixed bug 174: FORMAT's error message is slightly clearer when a
3577 non-printing character is used in a format directive.
3578 * fixed several bugs in compiler checking of type declarations, i.e.
3579 violations of the Python "declarations are assertions" principle
3580 (thanks to Alexey Dejneka)
3581 * fixed several bugs in PCL's error checking (thanks to Gerd
3583 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3585 * fixed bug in compilation of functions as first class values
3586 (thanks to Antonio Martinez-Shotton)
3587 * The compiler's handling TYPE-ERRORs which it can prove will
3588 inevitably happen at runtime has been cleaned up and corrected
3589 in several ways. (thanks to Alexey Dejneka)
3590 * improved argument type checking for various basic arithmetic
3591 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3592 lavished on them in the past that they can be compiled in many
3593 ways in different special cases
3594 * fixed bug 181: compiler checks validity of user supplied type
3596 * cleaned up code flushing in optimization: Function calls which
3597 should signal errors for safety purposes (e.g. which ANSI says
3598 should signal errors when their arguments are of incorrect type)
3599 are no longer optimized away.
3600 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3601 * incremented fasl file version number, because changes in the
3602 implementation of sequence functions like COERCE caused
3603 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3604 in old inline expansions) to become undefined. (Actually these
3605 changes were later undone, so we might very well be binary
3606 compatible with 0.7.7 after all, but leaving the version number
3607 incremented seemed like the simplest and most conservative
3610 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3611 * An alpha-quality port to the parisc architecture running Linux,
3612 based on the old CMUCL backend, has been made. This, even more so
3613 than the other backends, should be considered still a work in
3614 progress; known problems include that the Linux kernel in 64-bit
3615 mode does not propagate the correct sigcontext structure to
3616 userspace, and consequently SBCL on a parisc64 kernel will not
3618 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3619 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3620 suppressed.) Also now that the compiler is looking at declarations
3621 in the environment, it checks optimization declarations as well,
3622 and suppresses inlining when (> DEBUG SPEED).
3623 * More fixes have been made to treatment of floating point exception
3624 treatment and other Unix signals. In particular, floating point
3625 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3626 * The detection and handling of control stack exhaustion (infinite
3627 or very deeply nested recursion) has changed. Stack exhaustion
3628 detection is now done by write-protecting pages at the OS level
3629 and applies at all optimization settings; when found, a
3630 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3631 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3632 be caught using IGNORE-ERRORS.
3633 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3634 correctly when transforming recursive LABELS functions to LETs.
3635 (thanks to Alexey Dejneka)
3636 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3637 that are names of constants or global variables.
3638 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3639 alien routines with docstrings.
3640 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3641 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3643 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3644 (thanks to Eric Marsden)
3645 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3646 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3647 * Bug fix: The SPARC backend can now compile functions involving
3648 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3649 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3650 to a functions expecting a PATHNAME-DESIGNATOR.
3651 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3652 lambda lists. (thanks to Alexey Dejneka)
3653 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3654 functions so that EXPORTing the name of the class doesn't cause
3655 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3656 to Antonio Martinez for reporting this)
3657 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3658 constant-folded without failing an assertion. (thanks to Einar
3659 Floystad Dorum for reporting this)
3660 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3661 types are now dealt with more correctly by the compiler.
3662 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3663 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3664 behaviour analogous to RENAME-FILE. This puts its behaviour more
3665 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3667 * The fasl file version number has changed again. (because of the
3668 bug fix involving the names of PCL MAKE-INSTANCE functions)
3670 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3671 * bug fix: Floating point exceptions are treated much more
3672 consistently on the x86/Linux and PPC/Linux platforms.
3673 * Array initialization with :INITIAL-ELEMENT is now much faster for
3674 cases when the compiler cannot open code the array creation, but
3675 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3676 array accesses have also seen a speed increase.
3677 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3678 specified. (This was at the root of some bad interactions between
3679 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3681 * bug fix: Internal error arguments for undefined functions are now
3682 computed correctly on the PPC/Linux platform.
3683 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3684 Raymond Toy's patch for CMU CL)
3685 * Support for the Solaris 9 operating environment has been included
3686 (thanks to Daniel Merritt)
3687 * A very ugly but hopefully complete draft of the missing FFI chapter
3688 of the manual has been created by reformatting the corresponding
3689 CMU CL manual chapter into (currently very ugly and incoherent)
3690 DocBook and bringing it up to date for SBCL behavior. Thus, the
3691 manual is now essentially complete, at least by my extreme
3692 once-and-only-once standards, whereby it's acceptable to refer to
3693 the doc strings of SB-EXT functions as the primary documentation.
3694 * The fasl file version number has changed again, due to cleanup of
3695 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3696 is no longer a static symbol.)
3698 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3699 * SBCL now builds with OpenMCL (version 0.12) as the
3700 cross-compilation host; also, more progress has been made toward
3701 bootstrapping under CLISP.
3702 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3704 * bug 158 fixed: The compiler can now deal with integer loop
3705 increments different from 1; fixing this turned out also to fix
3707 * bug 169 fixed: no more bogus warnings about using lexical bindings
3708 despite the presence of perfectly good SPECIAL declarations (thanks
3709 to David Lichteblau)
3710 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3711 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3712 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3714 * bug 180 fixed: Method combination specifications no longer ignore
3715 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3716 * bug fix: Structure type predicate functions now check their argument
3717 count as they should.
3718 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3719 correctly. (thanks to Pierre Mai)
3720 * minor incompatible change: The --noprogrammer option is deprecated
3721 in favor of the new --disable-debugger option, which is very similar.
3722 (The major difference is that it takes effect at a slightly different
3723 time at startup, causing handling of errors in --sysinit and
3724 --userinit files will be affected differently.) The
3725 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3726 been added to allow this functionality to be controlled from ordinary
3727 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3728 maintainers, who might want to run non-interactive scripts to
3729 build SBCL cores which will later be used interactively.)
3730 * minor incompatible change: The LOAD function no longer, when given
3731 a wild pathname to load, loads all files matching that pathname.
3732 Instead, an error of type FILE-ERROR is signalled.
3734 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3735 * bug 147 fixed: The compiler preserves its block link/count
3736 invariants more correctly now so that it doesn't crash. (thanks
3738 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3740 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3741 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3742 no-value-for-this-character as NIL (as specified by ANSI).
3743 * HOST-NAMESTRING on physical pathnames now returns a string that is
3744 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3745 (thanks to Christophe Rhodes)
3746 * The Alpha port handles icache flushing more correctly. (thanks to
3748 * More progress has been made toward bootstrapping under CLISP. (thanks
3749 to Christophe Rhodes)
3750 * The fasl file format has changed again, because dynamic loading
3751 on OpenBSD (which has non-ELF object files) motivated some cleanups
3752 in the way that foreign symbols are transformed and passed around.
3753 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3754 is no longer treated as whitespace by the reader, but instead as
3755 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3756 |A
\7fB|, instead of A as it used to.
3758 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3759 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3760 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3761 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3762 this as of 0.7.1.45, but was left out of the previous news section
3763 (thanks to Dan Barlow)
3764 * SBCL now runs on the Solaris operating system on SPARC architectures
3765 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3766 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3767 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3768 * SPARC backend cleanups, allowing builds of cores optimized for V8
3769 and V9 SPARCS, and also emission of code targeted to a particular
3770 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3772 * SBCL is closer to bootstrapping under CLISP, thanks to various
3773 fixes by Christophe Rhodes.
3774 * The fasl file format has changed again, to allow the compiler's
3775 INFO database to support symbol macros.
3776 * The user manual (in doc/) is formatted into HTML more nicely.
3777 (thanks to coreythomas)
3778 * The system is smarter about SUBTYPEP relationships, especially
3779 those involving NOT types (including types such as ATOM which are
3780 represented internally using NOT types). Thus SUBTYPEP is less
3781 likely to return (VALUES NIL NIL) in general, and in particular
3782 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3784 * The fasl file format has changed again, because the internal
3785 representation of types now includes a new slot to support the new
3786 SUBTYPEP-of-NOT-types logic.
3787 * (not a change in the main branch of SBCL, but a related prototype
3788 which can hopefully be merged into the main branch of SBCL in the
3789 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3790 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3791 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3792 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3793 dispatch macro character combinations. (thanks to Alexey Dejneka)
3794 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3795 that superficially look like logical namestrings correctly.
3796 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3798 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3799 default supplied-p) for &optional and &key arguments. (thanks to
3802 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3803 * incompatible change: The compiler is now less aggressive about
3804 tail call optimization, doing it only when (> SPACE DEBUG) or
3805 (> SPEED DEBUG). (This is an incompatible change because there are
3806 programs which relied on the old CMU-CL-style behavior to optimize
3807 away their unbounded recursion which will now die of stack overflow.)
3808 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3809 for non-GENCGC systems has been increased to 20M (since that
3810 seems much closer to the likely performance optimum for modern
3811 systems than the old 4M value was)
3812 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3813 and *DEBUG-PRINT-LEVEL*
3814 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3815 of CMU CL's support for SPARC, and various endianness and other
3816 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3817 * new syntactic sugar for the Unix command line: --load foo.bar is now
3818 an alternate notation for --eval '(load "foo.bar")'.
3820 ** The system now detects stack overflow and handles it gracefully,
3821 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3822 optimization settings. (This is a good thing in general, and
3823 its introduction in this version should be particularly timely
3824 for anyone whose code fails because of suppression of tail
3826 ** The system now hunts for the C variable "environ" in a more
3827 devious way, to avoid segfaults when the C library version differs
3828 between compile time and run time. (thanks to Christophe Rhodes)
3829 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3830 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3831 ** The compiler no longer issues bogus style warnings for undefined
3832 classes in the same source file as the DEFCLASSes which defined
3833 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3835 ** fixes in CONDITION class precedence list for undefined function
3836 errors (thanks to Alexei Dejneka)
3837 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3838 correctly. (thanks to Dan Barlow)
3839 ** portability fixes aiming at bootstrapping under CLISP (thanks
3840 to Dave McDonald and Christophe Rhodes)
3841 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3842 ** fixes in type translation and and type inference (thanks to
3844 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3845 ** various fixes in the new ports (thanks to Dan Barlow)
3846 * several changes related to debugging:
3847 ** suppression of tail recursion, as noted above
3848 ** stack overflow detection, as noted above
3849 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3850 is now the default. (For some time encapsulation has been more
3851 reliable than the breakpoint-based :ENCAPSULATE NIL
3852 implementation, at least on X86 systems; and I just noticed that
3853 encapsulation also seems closer to the spirit of the ANSI
3856 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3858 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3859 up properly again. (There was a packaging bug in 0.7.0 which
3860 left their definitions in SB-SYS::LOAD-FOREIGN and
3861 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3862 vital for most things which interface to C-level interfaces,
3863 like extensions working with sockets or databases or
3864 Perl-compatible regexes or whatever, and the need to fix
3865 this bug is the main reason that 0.7.1 was released so
3867 ** DEFGENERIC is now choosier about the methods it redefines, so that
3868 reLOADing a previously-LOADed file containing DEFGENERICs does
3869 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3870 cycle now works as it should. (thanks to Alexey Dejneka)
3871 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3872 fixed by Christophe Rhodes. (He actually submitted this patch
3873 months ago, and I delayed until after 0.7.0.)
3874 ** Bug 111 (internal compiler confusion about runtime checks on
3875 FUNCTION types) was fixed by Alexey Dejneka.
3876 * Some internal cleanups (getting rid of variables which aren't
3877 needed now that the byte interpreter is gone) caused the fasl
3878 file format number to change again.
3880 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3881 * major incompatible change: The default fasl file extension, i.e. the
3882 default extension for files produced by COMPILE-FILE, has changed
3883 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3885 ** There are many changes in the implementation of the compiler.
3886 SBCL is now essentially a compiler-only implementation of ANSI
3887 Common Lisp. EVAL still "interprets" a few special cases, but
3888 almost all the interesting cases are handled by creating
3889 a LAMBDA expression, calling COMPILE on it, then calling
3890 FUNCALL on the result.
3891 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3892 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3893 Since the code is newer, there might still be some new bugs
3894 (though not as many as before Martin Atzmueller's fixes:-). But
3895 the new code is substantially simpler and clearer, and hopefully
3896 any remaining bugs will be simpler, less fundamental, and more
3897 fixable then the bugs in the old code.
3898 ** The revised compiler is still a little unsteady on its feet.
3900 *** The debugging information it produces (particularly the names
3901 of FUNCTION objects) is sometimes much less useful than what
3902 the old compiler produced.
3903 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3904 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3905 within a MACROLET) has been temporarily weakened.
3906 ** There are new compiler optimizations for various functions:
3907 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3908 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3909 *** the math functions TRUNCATE, FLOOR, and CEILING
3910 *** the function-of-all-trades COERCE
3911 Mostly these should be transparent, but there's one
3912 potentially-annoying problem (bug 117): when the compiler
3913 inline-expands a function and does type analysis on the result,
3914 it can create control paths which have type mismatches, and
3915 when it can't prove that those control paths aren't taken,
3916 it will issue WARNINGs about the type mismatches. This is
3917 a particular problem in practice for the new sequence functions.
3918 It's not clear how this should be fixed, and for now, a
3919 workaround is given in the entry for 117 in the BUGS file.
3920 ** (Because of the interaction between the two previous items --
3921 occasional inlining problems and new inline expansions -- some
3922 of the new sequence function optimizations won't really kick in
3923 completely until debugging information, and then inlining, are
3924 straightened out in some future version.)
3925 * minor incompatible changes:
3926 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3927 DIRECTORY no longer implicitly promotes NIL slots of its
3928 pathname argument to :WILD. In particular, when you ask for the
3929 contents of a directory (which you used to be able to do without
3930 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3931 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3932 ** changes in behavior that ANSI explicitly defines to be
3933 implementation dependent:
3934 *** The new compiler-only implementation still conforms with ANSI,
3935 but acts a little different than before. Besides the obvious
3936 changes in performance tradeoffs (that the cost per form passed
3937 to EVAL has gone up, and the cost per form executed by EVAL
3938 has gone down), the behavior of the system changes a little
3939 because there are no longer any interpreted function objects.
3940 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3941 e.g. doing COMPILE on the output of interactive DEFUN is
3943 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3945 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3946 has changed from (:CL) to NIL.
3947 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3948 e.g. #\Newline and #\Tab, don't have names specified in the
3949 ANSI Common Lisp standard, is now based on their ASCII symbolic
3950 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3951 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3952 are no longer used for output.
3953 ** changes in internal implementation constants:
3954 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3955 4 million. (If your application spends a lot of time GCing and
3956 you have a lot of RAM, you might want to experiment with
3957 increasing it even more.)
3958 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3959 However, almost all old code should still continue to work without
3960 immediate update, as SB-C-CALL is now a (deprecated) nickname
3962 ** Old operator names in the style DEF-FOO are now deprecated in
3963 favor of new corresponding names DEFINE-FOO, for consistency with
3964 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3965 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3966 internal symbols, but a few supported extensions like
3967 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3968 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3969 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3970 etc. as you get deeper into recursive calls to the debugger
3971 command loop, instead of the old "5]", "5]]", "5]]]"
3972 sequence. (I was motivated to do this when squabbles between
3973 ILISP and SBCL left me very deeply nested in the debugger. In the
3974 short term, this change will probably provoke more ILISP/SBCL
3975 squabbles, but hopefully it will be an improvement in the long run.)
3976 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3977 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3978 function, since declaiming return types involving aliens is
3979 (1) annoyingly messy to do by hand and (2) vital to efficient
3980 compilation of code which calls such functions.
3981 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3982 longer reexported by the SB-EXT package. They're solely useful
3983 for alien code, so it seems more logical that you should get
3984 them from the SB-ALIEN package, not in SB-EXT.
3985 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3986 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3987 features. Instead, the code that they used to control is always
3988 built into the system.
3989 * many other bug fixes
3990 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3991 advantage of the new EVAL-WHEN stuff and to clean them up in
3992 general, and they are now more ANSI-compliant in a number of
3993 ways. Martin Atzmueller is responsible for a lot of this.
3994 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3996 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3997 *** correct ERROR type for various file operations
3998 *** some fixes for Lisp streams
3999 *** DEFMETHOD syntax checking
4000 *** changing old weird representation of debug information as
4001 strings (which, among their other deficiencies, don't transform
4002 correctly when you rename packages, and don't change their
4003 print representation when you change things like *PACKAGE*
4004 and *PRINT-LENGTH*) to symbols and lists of symbols
4005 He also made several improvements and fixed several bugs in DESCRIBE.
4006 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4008 *** misbehavior of WRITE-STRING/WRITE-LINE
4009 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4010 and several other LOOP problems as well
4011 *** DIRECTORY when similar filenames are present
4012 *** DEFGENERIC with :METHOD options
4013 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4014 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4015 *** argument ordering in FIND with :TEST option
4016 *** mishandled package designator argument in APROPOS-LIST
4017 *** various problems in the backquote readmacro
4018 *** a bug in APROPOS
4019 *** probably some others that I'm not describing very well here,
4020 since the CVS log documents them by reference to sbcl-devel
4021 messages, and the SourceForge archives aren't working well.:-(
4022 ** Dan Barlow improved the Alpha port (and is making progress on the
4023 PPC port, for those of you who think different).
4024 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4025 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4026 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4027 continues to work on the SPARC port (for those of you in a position
4028 to look down upon our little PC-compatible boxes from a great height).
4029 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4030 and uses that copy, rather than the current dynamic value, when
4031 it's trying to decide whether to truncate output. Thus e.g.
4032 (let ((*print-lines* 50))
4033 (pprint-logical-block (stream nil)
4035 (let ((*print-lines* 8))
4036 (print (aref possiblybigthings i) stream)))))
4037 should now truncate the logical block only at 50 lines, instead of
4038 often truncating it at 8 lines, as it did before.
4039 * The doc/cmucl/ directory, containing old CMU CL documentation
4040 from the time of the fork, is no longer part of the base system.
4041 SourceForge has shut down its anonymous FTP service, and with it
4042 my original plan for distributing the old CMU CL documentation
4043 there. For now, if you need these files you can download an old
4044 SBCL source release and extract them from it.
4045 * The fasl file version number changed again, for dozens of reasons,
4046 some of which are apparent above.
4048 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4049 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4050 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4051 patch, so that DISASSEMBLE output is much nicer.
4052 * The code in the SB-PROFILE package now seems reasonably stable.
4053 I still haven't decided what the final interface should look like
4054 (I'd like PROFILE to interact cleanly with TRACE, since both
4055 facilities use function encapsulation) but if you have a need
4056 for profiling now, you can probably use it successfully with
4057 the current CMU-CL-style interface.
4058 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4059 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4060 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4061 link now returns the dangling link itself, and for similar
4062 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4063 link itself. (In these cases the old code signalled an error and
4064 looped endlessly, respectively.) Thus, DIRECTORY now works even
4065 in the presence of dangling and cyclic symbolic links.
4066 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4067 is now a supported extension again, since the consensus on
4068 sbcl-devel was that it can be useful for ordinary development
4069 work, not just for debugging SBCL itself.
4070 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4071 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4072 that currently-defined functions might be redefined later with
4073 different return types.
4074 * Hash tables can be printed readably, as inspired by CMU CL code
4075 of Eric Marsden and SBCL code of Martin Atzmueller.
4076 * better error handling in CLOS method combination, thanks to
4077 Martin Atzmueller porting Pierre Mai's CMU CL patches
4078 * more overflow fixes for >16Mbyte I/O buffers
4079 * A bug in READ has been fixed, so that now a single Ctrl-D
4080 character suffices to cause end-of-file on character streams.
4081 In particular, now you only need one Ctrl-D at the command
4082 line (not two) to exit SBCL.
4083 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4085 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4086 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4087 does the right thing.
4088 * The implementation of some type tests, especially for CONDITION
4089 types, is now tidier and maybe faster, due to CMU CL code
4090 originally by Douglas Crosher, ported by Martin Atzmueller.
4091 * Some math functions have been fixed, and there are new
4092 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4093 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4094 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4095 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4096 likely to be built into sbcl-0.7.0.)
4097 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4098 CL, ported to SBCL by Martin Atzmueller)
4099 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4100 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4101 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4102 * There's a new slam.sh hack to shorten the edit/compile/debug
4103 cycle for low-level changes to SBCL itself, and a new
4104 :SB-AFTER-XC-CORE target feature to control the generation of
4105 the after-xc.core file needed by slam.sh.
4106 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4107 COMPILE-FILE is no longer supported, so that now every function
4108 gets an entry point, so that block compilation looks a little
4109 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4110 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4111 returns the number of bytes consed since the system started,
4112 rather than the number consed since the first time the function
4113 was called. (The new definition parallels ANSI functions like
4114 CL:GET-INTERNAL-RUN-TIME.)
4115 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4116 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4117 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4118 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4119 * Fasl file version numbers are now independent of the target CPU,
4120 since historically most system changes which required version
4121 number changes have affected all CPUs equally. Similarly,
4122 the byte fasl file version is now equal to the ordinary
4125 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4126 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4127 is no longer recognized. I apologize for this, because it was
4128 listed in SB-EXT as a supported extension, but I found that
4129 its existing behavior was poorly specified, as well as incorrectly
4130 specified, and it looked like too much of a mess to straighten it
4131 out. I have enough on my hands trying to get ANSI stuff to work..
4132 * many patches ported from CMU CL by Martin Atzmueller, with
4133 half a dozen bug fixes in pretty-printing and the debugger, and
4134 half a dozen others elsewhere
4135 * fixed bug 13: Floating point infinities are now supported again.
4136 They might still be a little bit flaky, but thanks to bug reports
4137 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4138 as flaky as they were.
4139 * The --noprogrammer command line option is now supported. (Its
4140 behavior is slightly different in detail from what the old man
4141 page claimed it would do, but it's still appropriate under the
4142 same circumstances that the man page talks about.)
4143 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4144 are now supported, and enabled by default. Thus, the compiler can
4145 handle many floating point and complex operations much less
4146 inefficiently. (Thus e.g. you can implement a complex FFT
4148 * The compiler now detects type mismatches between DECLAIM FTYPE
4149 and DEFUN better, and implements CHECK-TYPE more correctly, and
4150 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4151 thanks to patches from Martin Atzmueller.
4152 * various fixes to make the cross-compiler more portable to
4153 ANSI-conforming-but-different cross-compilation hosts (notably
4154 Lispworks for Windows, following bug reports from Arthur Lemmens)
4155 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4156 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4157 fixed thanks to Pierre Mai's CMU CL patches.
4158 * fixes to keep the system from overflowing internal counters when
4159 it tries to use i/o buffers larger than 16M bytes
4160 * fixed bug 45a: Various internal functions required to support
4161 complex special functions have been merged from CMU CL sources.
4162 (When I was first setting up SBCL, I misunderstood a compile-time
4163 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4164 * improved support for type intersection and union, fixing bug 12
4165 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4166 more obscure bugs as well
4167 * some steps toward byte-compiling non-performance-critical
4168 parts of the system, courtesy of patches from Martin Atzmueller
4169 * Christophe Rhodes has made some debian packages of sbcl at
4170 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4171 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4172 stable, but are nonetheless usable. When he's ready, I'd be happy
4173 to add them to the SourceForge "File Releases" section. (And if
4174 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4175 * new fasl file format version number (because of changes in
4176 internal representation of (OR ..) types to accommodate the new
4177 support for (AND ..) types, among other things)
4179 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4180 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4182 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4184 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4185 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4186 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4187 * DESCRIBE now works on structure objects again.
4188 * Most function call argument type mismatches are now handled as
4189 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4190 know whether the function will be redefined before the call is
4191 executed. (The compiler could flag local calls with full WARNINGs,
4192 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4193 it doesn't keep track of enough information to know whether calls
4194 are local in this sense.)
4195 * Compiler output is now more verbose, with messages truncated
4196 later than before. (There should be some supported way for users
4197 to override the default verbosity, but I haven't decided how to
4198 provide it yet, so this behavior is still controlled by the internal
4199 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4200 src/compiler/ir1util.lisp.)
4201 * Fasl file format version numbers have increased again, because
4202 support for the Gray streams extension changes the layout of the
4203 system's STREAM objects.
4204 * The Gray subclassable streams extension now works, thanks to a
4205 patch from Martin Atzmueller.
4206 * The full LOAD-FOREIGN extension (not just the primitive
4207 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4208 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4209 but like most other programs, it defaults to copying the Unix
4210 environment from the original process instead of starting the
4211 new process in an empty environment.
4212 * Extensions which manipulate the Unix environment now support
4213 an :ENVIRONMENT keyword option which doesn't smash case or
4214 do other bad things. The CMU-CL-style :ENV option is retained
4215 for porting convenience.
4216 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4217 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4219 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4221 * A patch from Martin Atzmueller seems to have solved the SIGINT
4222 problem, and as far as we know, signal-handling now works cleanly.
4223 (If you find any new bugs, please report them!)
4224 * The system no longer defaults Lisp source file names to types
4225 ".l", ".cl", or ".lsp", but only to ".lisp".
4226 * The compiler no longer uses special default file extensions for
4227 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4228 seems to expect a single default extension for all compiled code,
4229 and there's no compelling reason to try to stretch the standard
4230 to allow two different extensions.) Instead, byte-compiled files
4231 default to the same extension as native-compiled files.
4232 * Fasl file format version numbers have increased again, because
4233 a rearrangement of internal implementation packages made some
4234 dumped symbols in old fasl files unreadable in new cores.
4235 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4236 many fewer weird special cases.
4237 * Bug #17 (differing COMPILE-FILE behavior between logical and
4238 physical pathnames) has been fixed, and some related misbehavior too,
4239 thanks to a patch from Martin Atzmueller.
4240 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4241 by Tim Moore, ported to SBCL by Martin Atzmueller.
4242 * Martin Atzmueller fixed several filesystem-related problems,
4243 including bug #36, in part by porting CMU CL patches, which were
4244 written in part by Paul Werkowski.
4245 * More compiler warnings in src/runtime/ are gone, thanks to
4246 more patches from Martin Atzmueller.
4247 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4250 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4252 * DESCRIBE now works on CONDITION objects.
4253 * The debugger now handles errors which arise when trying to print
4254 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4256 * The build system now uses an additional file, customize-target-features.lisp,
4257 to allow local modifications to the target *FEATURES* list. (The point of
4258 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4259 debugging features enabled, without having to worry about propagating it
4260 into everyone's system when I do a "cvs update".) When no
4261 customize-target-features.lisp file exists, the target *FEATURES* list
4262 should be constructed the same way as before.
4263 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4264 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4265 warn about behavior which is undefined under the ANSI spec. Note: This
4266 is specified by ANSI, but it's not very popular with programmers.
4267 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4268 macro in the SBCL sources for an example of a workaround which you
4269 might use to make portable ANSI-standard code which does what you want.
4270 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4271 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4272 when it's used as a non-toplevel form. (This is required in order
4273 to implement the DEFCONSTANT-EQX macro.)
4274 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4275 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4276 funny things, due to bugs in the implementation of EVAL-WHEN
4277 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4278 non-toplevel DEFCONSTANTs are uncommon.)
4279 * The core file version number and fasl file version number have been
4280 incremented, because the old noncompliant DEFCONSTANT behavior involved
4281 calling functions which no longer exist, and because I also took the
4282 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4283 * fixed bug 1 (error handling before read-eval-print loop starts), and
4284 redid debugger restarts and related debugger commands somewhat while
4286 ** The QUIT debugger command is gone, since it did something
4287 rather different than the SB-EXT:QUIT command, and since it never
4288 worked properly outside the main toplevel read/eval/print loop.
4289 Invoking the new TOPLEVEL restart provides the same functionality.
4290 ** The GO debugger command is also gone, since you can just invoke
4291 the CONTINUE restart directly instead.
4292 ** The TOP debugger command is also gone, since it's redundant with the
4293 FRAME 0 command, and since it interfered with abbreviations for the
4295 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4297 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4298 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4299 some porting work by Martin Atzmueller)
4300 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4301 with a boatload of patches which he ported to SBCL. Now that those
4303 ** The system tries to make sure that its low-priority messages
4304 are prefixed by semicolons, to help people who like to use
4305 syntax highlighting in their ILISP buffer. (This patch
4306 was originally due to Raymond Toy.)
4307 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4308 patches originally written by Raymond Toy.
4309 ** The compiler understands coercion between single-value and
4310 multiple-VALUES type expressions better, getting rid of some very
4311 weird behavior, thanks to patches originally by Robert MacLachlan
4312 and Douglas Crosher.
4313 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4314 lambda lists, thanks to a patch originally by Pierre Mai.
4315 ** The system no longer bogusly warns about "abbreviated type
4317 ** The compiler gets less confused by inlining and RETURN-FROM,
4318 thanks to some patches originally by Tim Moore.
4319 ** The system no longer hangs when dumping circular lists to fasl
4320 files, thanks to a patch originally from Douglas Crosher.
4321 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4322 undefined function error.
4323 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4324 in structure slot initforms) for now, documented workaround instead:-|
4325 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4326 * fixed bug 5: added stubs for various Gray stream functions called
4327 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4328 installed, at least appropriate type errors are generated
4329 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4330 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4331 consistently in DEFMETHOD forms.
4332 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4333 it doesn't seem to affect SBCL after all
4334 * The C runtime system now builds with better optimization and many
4335 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4337 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4339 * The system is now under CVS at SourceForge (instead of the
4340 CVS repository on my home machine).
4341 * The new signal handling code has been tweaked to treat register
4342 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4343 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4344 downstream, e.g. in debug-int.lisp, has implicit dependencies
4345 on the unsignedness of integer representation of machine words,
4346 and that caused the system to bomb out with infinite regress
4347 when trying to recover from type errors involving signed values,
4348 e.g. (BUTLAST '(1 2 3) -1).)
4349 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4350 (This was one of the bugs Peter Van Eynde reported back in July.)
4351 * The system now uses code inspired by Colin Walters' O(N)
4352 implementation of MAP (from the cmucl-imp@cons.org mailing
4353 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4354 inline the MAP operation, and there is more than one
4355 sequence argument to the MAP call (so that it can't just
4356 do ETYPECASE once and for all based on the type of the
4357 single sequence argument). (The old non-inline implementation
4358 of the general M-argument sequence-of-length-N case required
4359 O(M*N*N) time when any of the sequence arguments were LISTs.)
4360 * The QUIT :UNIX-CODE keyword argument has been renamed to
4361 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4362 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4364 * Raymond Wiker's patches to port dynamic loading from Linux to
4365 FreeBSD have been added.
4366 * The BUGS file is now more nearly up to date, thanks in large part
4367 to Martin Atzmueller's review of it.
4368 * The debugger now flushes standard output streams before it begins
4369 its output ("debugger invoked" and so forth).
4370 * The core version number and fasl file version number have both
4371 been incremented, because of incompatible changes in the layout
4373 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4374 used to not be called for a saved Lisp image.)
4375 * Martin Atzmueller's version of a patch to fix a compiler crash,
4376 as posted on sbcl-devel 13 September 2000, has been installed.
4377 * Instead of installing Martin Atzmueller's patch for the
4378 compiler transform for SUBSEQ, I deleted the compiler transform,
4379 and transforms for some similar consing operations.
4380 * A bug in signal handling which kept TRACE from working on OpenBSD
4382 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4383 away by constant folding
4384 * The system now defines its address space constants in one place
4385 (in the Lisp sources), and propagates them automatically elsewhere
4386 (through GENESIS and the sbcl.h file). Therefore, patching the
4387 address map is less unnecessarily tedious and error-prone. The
4388 Lisp names of address space constants have also been systematized.
4389 * CVS tags like dollar-Header-dollar have been removed from
4390 the sources, because they have never saved me trouble and
4391 they've been source of trouble working with patches and other
4392 diff-related operations.
4393 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4394 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4396 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4398 * The system has been ported to OpenBSD.
4399 * The system now compiles with a simple "sh make.sh" on the systems
4400 that it's supported on. I.e., now you no longer need to tweak
4401 text in the target-features.lisp-expr and symlinks in src/runtime/
4402 by hand, the make.sh takes care of it for you.
4403 * The system is no longer so grossly inefficient when compiling code
4404 involving vectors implemented as general (not simple) vectors (VECTOR T),
4405 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4406 which dares to use the various sequence functions on non-simple
4407 vectors, takes less of a performance hit.
4408 * There is now a primitive type predicate VECTOR-T-P
4409 to test for the (VECTOR T) type, so that e.g.
4410 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4411 can now be compiled with some semblance of efficiency. (The old code
4412 turned the type declaration into a full call to %TYPEP at runtime!)
4413 * AREF on (VECTOR T) is still not fast, since it's still compiled
4414 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4415 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4416 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4417 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4418 without making full calls to SUBTYPEP at runtime.
4419 (Some analogous efficiency issues for non-simple vectors specialized to
4420 element types other than T, or for non-simple multidimensional arrays,
4421 have not been addressed. They could almost certainly be handled the
4422 same way if anyone is motivated to do so.)
4423 * The changes in array handling break binary compatibility, so
4424 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4425 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4426 instead of (VALUES T T).
4427 * By following the instructions that Dan Barlow posted to sbcl-devel
4428 on 2 July 2000, I was able to enable primitive dynamic object
4429 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4430 functionality is not implemented (since it calls ld to resolve
4431 library references automatically, requiring RUN-PROGRAM for its
4432 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4433 try to resolve library references) is now supported.
4434 * The system now flushes the standard output streams when it terminates,
4435 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4436 them at several other probably-convenient times, e.g. in each pass of
4437 the toplevel read-eval-print loop, and after evaluating a form given
4438 as an "--eval" command-line option. (These changes were motivated by a
4439 discussion of stream flushing issues on cmucl-imp in August 2000.)
4440 * The source transform for TYPEP of array types no longer assumes
4441 that an array whose element type is a not-yet-defined type
4442 is implemented as an array of T, but instead punts, so that the
4443 type will be interpreted at runtime.
4444 * There is now some support for cross-compiling in make.sh: each of
4445 the phases of make.sh has its own script. (This should be transparent
4446 to people doing ordinary, non-cross-compile builds.)
4447 * Since my laptop doesn't have hundreds of megabytes of memory like
4448 my desktop machine, I became more motivated to do some items on
4449 my to-do list in order to reduce the size of the system a little:
4450 ** Arrange for various needed-only-at-cold-init things to be
4451 uninterned after cold init. To support this, those things have
4452 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4453 symbols with such names are now uninterned after cold init).
4454 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4455 fasl files for cold load.
4456 ** Remove the old compiler structure pooling code (which used to
4457 be conditional on the target feature :SB-ALLOC) completely.
4458 ** Redo the representation of some data in cold init to be more compact.
4459 (I also looked into supporting byte compiled code at bootstrap time,
4460 which would probably reduce the size of the system a lot, but that
4461 looked too complicated, so I punted for now.)
4462 * The maximum signal nesting depth in the src/runtime/ support code has
4463 been reduced from 4096 to 256. (I don't know any reason for the very
4464 large old value. If the new smaller value turns out to break something,
4465 I'll probably just bump it back up.)
4466 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4468 * Many, many bugs reported by Peter Van Eynde have been added to
4469 the BUGS list; some have even been fixed.
4470 * While enabling dynamic object file loading, I tried to make the
4471 code easier to understand, renaming various functions and variables
4472 with less ambiguous names, and changing some function calling
4473 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4474 * While trying to figure out how to do the OpenBSD port, I tried to
4475 clean up some of the code in src/runtime/. In particular, I dropped
4476 support for non-POSIX signal handling, added various comments,
4477 tweaked the code to reduce the number of compilation warnings, and
4478 renamed some files to increase consistency.
4479 * To support the new automatic configuration functionality in make.sh,
4480 the source file target-features.lisp-expr has been replaced with the
4481 source file base-target-features.lisp-expr and the machine-generated
4482 file local-target-features.lisp-expr.
4483 * fixed a stupid quoting error in make.sh so that using CMU CL
4484 "lisp -batch" as cross-compilation host works again
4486 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4488 * DESCRIBE no longer tries to call itself recursively to describe
4489 bound/fbound values, so that it no longer fails on symbols which are
4490 bound to themselves (like keywords, T, and NIL).
4491 * DESCRIBE now works on generic functions.
4492 * The printer now prints less-screwed-up representations of closures
4493 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4494 * A private symbol is used instead of the :EMPTY keyword previously
4495 used to mark empty slots in hash tables. Thus
4496 (DEFVAR *HT* (MAKE-HASH-TABLE))
4497 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4498 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4499 now does what ANSI says that it should. (You can still get
4500 similar noncompliant behavior if bang on the hash table
4501 implementation with all the symbols you get back from
4502 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4503 This breaks binary compatibility, since tests for equality to
4504 :EMPTY are wired into things like the macroexpansion of
4505 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4507 * There's now a minimal placeholder implementation for CL:STEP,
4508 as required by ANSI.
4509 * An obscure bug in the interaction of the normal compiler, the byte
4510 compiler, inlining, and structure predicates has been patched
4511 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4512 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4514 * Missing ordinary arguments in a macro call are now detected even
4515 when the macro lambda list contains &KEY or &REST.
4516 * The debugger no longer complains about encountering the top of the
4517 stack when you type "FRAME 0" to explicitly instruct it to go to
4518 the top of the stack. And it now prints the frame you request even
4519 if it's the current frame (instead of saying "You are here.").
4520 * As specified by ANSI, the system now always prints keywords
4521 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4523 * The default initial SIZE of HASH-TABLEs is now smaller.
4524 * Type information from CLOS class dispatch is now propagated
4525 into DEFMETHOD bodies, so that e.g.
4526 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4528 is now basically equivalent to
4529 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4530 (DECLARE (TYPE SINGLE-FLOAT X))
4532 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4533 operation, without having to do run-time type dispatch.
4534 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4535 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4536 * Since I don't seem to be making big file reorganizations very often
4537 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4538 is overflowing my ability to conveniently back them up), I've finally
4539 checked the system into CVS. (The CVS repository is on my home system,
4540 not at SourceForge -- putting it on SourceForge might come later.)
4541 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4542 high-level GC-NOTIFY-FOO functions send their output. (There's
4543 still very little control of where low-level verbose GC functions
4544 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4545 less than it used to -- the GC-NOTIFY-FOO functions are now under
4546 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4547 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4548 in only one place in the source code, and propagates it automatically
4549 everywhere that it's needed. Thus e.g. when I bump the version from
4550 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4551 * The C source files now include boilerplate legalese and documentation
4552 at the head of each file (just as the Lisp source files already did).
4553 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4554 to his page will be replaced with a link to his new CLiki service.
4556 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4558 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4559 * The build process now looks for GNU make under the default name "gmake",
4560 instead of "make" as it used to. If GNU make is not available as "gmake"
4561 on your system, you can change this default behavior by setting the
4562 GNUMAKE environment variable.
4563 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4564 can build without error under CMU CL.
4566 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4568 * There is now a partial SBCL user manual (with some new text and some
4569 text cribbed from the CMU CL manual).
4570 * The beginnings of a profiler have been added (starting with the
4571 CMU CL profiler and simplifying and cleaning up). Eventually the
4572 main interface should be through the TRACE macro, but for now,
4573 it's still accessed through vaguely CMU-CL-style functions and macros
4574 exported from the package SB-PROFILE.
4575 * Some problems left over from porting CMU CL to the new
4576 cross-compilation bootstrap process have been cleaned up:
4577 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4578 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4579 part when getting rid of the compiler *BACKEND* stuff.)
4580 ** The value of *NULL-TYPE* was screwed up, because it was
4581 being initialized before the type system knew the final
4582 definition of the 'NULL type. This screwed up several key
4583 optimizations in the compiler, causing inefficiency in all sorts
4584 of places. (I found it because I wanted to understand why
4585 GET-INTERNAL-RUN-TIME was consing.)
4586 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4587 PROCLAIM FTYPE information. Unfortunately this broke binary
4588 compatibility again, since now the forms output by DEFGENERIC
4589 to refer to functions which didn't exist in 0.6.3.
4590 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4591 can use the new (as of 0.6.3) transform for SOME into MAP into
4593 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4594 values to (INTEGER 0 1000000), so that the time code will no longer
4595 occasionally get blown up by Linux returning 1000000 microseconds
4596 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4597 its output conform to the ANSI spec. (Alas, this makes its output
4598 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4599 * A full call to MAP NIL with a single sequence argument no longer conses.
4600 * fixes to problems pointed out by Martin Atzmueller:
4601 * The manual page no longer talks about multiprocessing as though
4602 it were currently supported.
4603 * The ILISP support patches have been removed from the distribution,
4604 because as of version 5.10.1, ILISP now supports SBCL without us
4605 having to maintain patches.
4606 * added a modified version of Raymond Toy's recent CMU CL patch for
4607 EQUALP comparison of HASH-TABLE
4609 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4611 * The system still can't cross-compile itself with
4612 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4613 entails), but at least it can compile more of itself that way
4614 than it used to be able to, and various buglets which were uncovered
4615 by trying to cross-compile itself that way have now been fixed.
4616 * This release breaks binary compatibility again. This time
4617 at least I've incremented the FASL file format version to 2, so that the
4618 problem can be detected reliably instead of just causing weird errors.
4619 * various new style warnings:
4620 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4621 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4622 ** using the lexical binding of a variable named in the *FOO* style
4623 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4624 as specified by ANSI.
4625 * *RANDOM-STATE* is no longer automatically initialized from
4626 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4627 default behavior of the system is to repeat its behavior every time
4628 it's run. If you'd like to change this behavior, you can always
4629 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4630 old convention there was no comparably easy way to get the system to
4631 repeat its behavior every time it was run.
4632 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4633 FTYPE declarations has been removed, in favor of their ANSI
4634 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4635 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4636 the types of their sequence arguments can be determined at compile time.
4637 This is done through a new open code expansion for MAP which eliminates
4638 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4639 argument types can be determined at compile time.
4640 * The optimizer now transforms COERCE into an identity operation when it
4641 can prove that the coerced object is already of the correct type. (This
4642 can be a win for machine generated code, including the output of other
4643 optimization transforms, such as the MAP transform above.)
4644 * Credit information has been moved from source file headers into CREDITS.
4645 * Source file headers have been made more standard.
4646 * The CASE macro now compiles without complaining even when it has
4649 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4651 * (Note that the way that the PCL macroexpansions were rewritten
4652 to accommodate the change in DEFGENERIC below breaks binary
4653 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4654 not run under sbcl-0.6.2. Once we get out of alpha releases,
4655 i.e. hit release 1.0.0, we'll probably try to maintain binary
4656 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4657 and sbcl-1.4.4. Until then, however, it might be fairly common
4658 for maintenance releases to break binary compatibility.)
4659 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4661 * The reporting of unhandled signals has been changed to print some
4662 explanatory text as well as the report form. (Previously only
4663 the report form was printed.)
4664 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4665 it defines, so that the compiler no longer issues undefined function
4666 warnings for compiled-but-not-yet-loaded generic functions.
4667 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4668 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4669 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4670 section of the ANSI spec are supported.
4671 * The "" nickname for the "KEYWORD" package has been removed.
4672 The reader still handles symbol tokens which begin with a package marker
4673 as keywords, but it doesn't expose its mechanism for doing so in the
4674 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4675 * The system now issues STYLE-WARNINGs for contradictory TYPE
4676 proclamations. (Warnings for contradictory FTYPE proclamations would
4677 be nice too, but those can't be done usefully unless the type system
4678 is made smarter about FUNCTION types.)
4679 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4680 systematized, so that "*target-*.lisp is supposed to exist only on the
4681 target and imply that there's a related file which exists on the
4682 host, and *host-*.lisp is supposed to exist only on the host and imply
4683 that there's a related file which exists on the target. This involves a
4684 lot of renaming. Hopefully the acute confusion caused by the renaming
4685 will be justified by the reduction in chronic confusion..
4686 ** runtime-type.lisp -> early-target-type.lisp
4687 ** target-type.lisp -> late-target-type.lisp
4688 ** early-host-format.lisp -> early-format.lisp
4689 ** late-host-format.lisp -> late-format.lisp
4690 ** host-error.lisp -> misc-error.lisp
4691 ** early-error.lisp -> early-target-error.lisp
4692 ** late-error.lisp -> late-target-error.lisp
4693 ** host-defboot.lisp -> early-defboot.lisp
4694 ** code/misc.lisp -> code/target-misc.lisp
4695 ** code/host-misc.lisp -> code/misc.lisp
4696 ** code/numbers.lisp -> code/target-numbers.lisp
4697 ** code/early-numbers.lisp -> numbers.lisp
4698 ** early-host-type.lisp -> early-type.lisp
4699 ** late-host-type.lisp -> late-type.lisp
4700 ** host-typep.lisp -> typep.lisp
4701 ** load.lisp -> target-load.lisp
4702 ** host-load.lisp -> load.lisp
4703 ** host-disassem.lisp -> disassem.lisp
4704 ** host-insts.lisp -> insts.lisp
4705 ** byte-comp.lisp -> target-byte-comp.lisp
4706 ** host-byte-comp.lisp -> byte-comp.lisp
4707 ** host-signal.lisp -> signal.lisp
4708 ** host-defstruct.lisp -> defstruct.lisp
4709 ** late-target-type.lisp -> deftypes-for-target.lisp
4710 Furthermore, several other previously target-only files foo.lisp (e.g.
4711 hash-table.lisp and random.lisp) have been split into a target-and-host
4712 foo.lisp file and a target-only target-foo.lisp file, with their key type
4713 definitions in the target-and-host part, so that the cross-compiler will
4714 know more about target types.
4715 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4716 gone away. In their place are various *BACKEND-FOO* variables
4717 corresponding to the slots of the old structure.
4718 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4719 into the target SBCL has been fixed.
4720 * The chill.lisp system for loading cold code into a running SBCL
4722 * Support for the CMU CL "scavenger hook" extension has been removed.
4723 (It was undocumented and unused in the CMU CL sources that SBCL was
4724 derived from, and stale in sbcl-0.6.1.)
4725 * Various errors in the cross-compiler type system were detected
4726 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4727 (enabling various consistency checks). Many of them were fixed,
4728 but some hard problems remain, so the compiler is back to
4729 running without *TYPE-SYSTEM-INITIALIZED* for now.
4730 * As part of the cross-compiler type system cleanup, I implemented
4731 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4732 * I have started adding UNCROSS calls throughout the type system
4733 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4734 on cross-compiler input files will be able to go away, and various
4736 * CONSTANTP now returns true for quoted forms (as explicitly required
4739 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4741 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4742 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4743 (MAKE-LIST -1). In the longer term, it ought to become true
4744 that declarations are assertions even at SAFETY 1. For now, it's not
4745 quite true even at SAFETY 3, but it's at least more nearly true..
4746 (Note that this change seems to increases the size of the system by
4747 O(5%) and to decrease the speed of the compiler by 20% or more.)
4748 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4749 for the problem of printing dozens of lines of distracting information
4750 about low-level system machinery as part of the top stack frame
4751 on entry to the debugger when an undefined function was called.
4752 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4753 is not reset to COMMON-LISP-USER.
4754 * Compilation of stuff related to dyncount.lisp has been made conditional
4755 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4756 smaller. The various dyncount-related symbols have been moved into
4757 a new "SB-DYNCOUNT" package.
4758 * tty-inspect.lisp has been renamed to inspect.lisp.
4759 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4760 feature has gone away. (When we eventually port to other flavors of
4761 libc and/or Unix, we'll try to make the differences between flavors
4762 invisible at the user level.)
4763 * Various other *FEATURES* tags, and/or their associated conditionals,
4764 have been removed if obsolescent, or given better documentation, or
4765 sometimes given more-mnemonic names.
4767 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4769 * tidied up "make.sh" script
4770 * tidied up system directory structure
4771 * better "clean.sh" behavior
4772 * added doc/FOR-CMUCL-DEVELOPERS
4773 * many many small tweaks to output format, e.g. removing possibly-confusing
4774 trailing #\. character in DESCRIBE-INSTANCE
4775 * (EQUALP #\A 'A) no longer signals an error.
4776 * new hashing code, including EQUALP hashing
4777 * tidied up Lisp initialization and toplevel
4778 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4779 * command line argument processing
4780 * added POSIX-GETENV function to deal with Unix-ish environment variables
4781 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4782 terminating SBCL on EOF
4783 * non-verbose GC by default
4784 * There is no more "sbcl" shell script; the sbcl file is now the C
4785 runtime executable (just like CMU CL).
4786 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4788 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4789 symbol and package information as Lisp native symbol and package objects
4790 instead of strings naming symbols and strings naming packages. This way,
4791 whenever packages are renamed (as in warm init), debug information is
4792 transformed along with everything else.
4793 * tweaked the optimization policy declarations which control the building
4794 of SBCL itself. Now, among other things, the system no longer saves
4795 source location debugging information. (This helps two problems at once
4796 by reducing SBCL size and by keeping SBCL from trying to look for its
4797 sources -- which may not exist -- when reporting errors.)
4798 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4799 debugging and testing purposes
4800 * cleaned up printing, making the printer call PRINT-OBJECT for
4801 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4802 methods, giving nearly-ANSI behavior
4803 * converted almost all special variables to use *FOO* naming convention
4804 * deleted PARSE-TIME functionality, since it can be done portably
4805 * moved some files out of cold init into warm init
4806 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4808 * regularized formatting of source files
4809 * added an install.sh script
4810 * fixed ridiculous memory usage of cross-compiler by making
4811 compiler/alloc.lisp not try to do pooling unless it can hook
4812 itself into the GC of the cross-compilation host. Now the system
4813 builds nicely on my old laptop.
4814 * added :SB-ALLOC in target-features.lisp-expr
4815 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4816 was not implemented)
4817 * re-did condition handling and note reporting in the compiler. Notes
4818 are no longer handled by signalling conditions. Style warnings
4819 and warnings are handled more correctly and reported in such a way
4820 that it's easy to find one or the other in your output (so that you
4821 can e.g. figure out which of many problems caused COMPILE-FILE to
4823 * changed the severity of several compiler warnings from full WARNING
4824 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4825 compiler note reporting so that it doesn't use the condition system
4826 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4827 COMPILE-FILE command)
4828 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4829 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4830 no longer supported, since their implementation was deeply intertwingled
4831 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4832 * removed bogus "support" for compiler macros named (SETF FOO), and
4833 removed the compiler macro for SETF INFO (but only after making a fool
4834 of myself on the cmucl-imp mailing list by posting a bogus patch for
4835 DEFINE-COMPILER-MACRO..)
4836 * Compiled files containing forms which have side effects on the Lisp
4837 reader (such as DEFPACKAGE forms) are now handled more correctly.
4838 (Compiler queuing of top level lambdas has been suppressed by setting
4839 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4840 * deleted various currently-unused source files, e.g. gengc.lisp. They
4841 may be added back at some point e.g. when porting to other architectures,
4842 but until they are it's distracting to distribute them and to try to
4844 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4845 there were so many of them they're just distractions, and UNCROSS is
4846 known to be able to handle the current sources
4847 * moved PROFILE functionality into TRACE, so that it will be clear
4848 how the wrapping and unwrapping of functions when you profile them
4849 interacts with the wrapping and unwrapping of functions when you
4850 trace them. (Actually, the functionality isn't there yet, but at least
4851 the interface specification is there. Hopefully, the functionality will
4852 arrive with some future maintenance release.)
4853 * removed host-oops.lisp
4854 * changed signature of QUIT function to allow UNIX-CODE argument
4855 * fixed READ-SEQUENCE bug
4856 * tweaked verbose GC output so that it looks more like the progress
4857 output that ANSI specifies for functions like LOAD
4858 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4859 * added <http://sbcl.sourceforge.com> to the banner information printed by