1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.18 relative to 1.0.17:
3 * optimization: structure allocation has been improved
4 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
5 ** out of line constructors are ~10% faster.
6 ** inline constructors are ~15% faster.
7 ** inline constructors are capable of dynamic extent allocation
8 (generally on x86 and x86-64, in some cases on other platforms
10 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
13 changes in sbcl-1.0.17 relative to 1.0.16:
14 * temporary regression: user code can no longer allocate closure
15 variable storage on stack, due to bug 419 without explicitly
16 requesting it. Please consult sbcl-devel for advice if you need to
17 use this feature in the meanwhile.
18 * new feature: runtime argument --control-stack-size can be used to
19 adjust thread default control stack size.
20 * enhancement: improved TIME output
21 ** all times are reported using the measured accuracy (milliseconds
22 for real and GC times, microseconds for everything else.)
23 ** processor cycle counts on x86 and x86-64.
24 ** interpreted forms are counted for both evaluator modes.
25 ** number of lambdas converted by the compiler is reported.
26 ** CPU percentage report (computed from real and total run time.)
27 ** more comprehensive run time reporting, using a condenced format
28 ** interperted form, lambda, and page fault counts are omitted
30 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
32 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
33 * optimization: two argument forms of LAST are upto ~10% faster
34 in normal SPEED policies.
35 * optimization: NCONC no longer needs to heap cons its &REST list
36 in normal SPEED policies.
37 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
39 * bug fix: UNION and NUNION work with :TEST-NOT once more,
40 regression since 1.0.9.1. (thanks to Eric Marsden)
41 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
42 since 1.0.15.36. (thanks to Paul Khuong)
43 * bug fix: LAST when always returned the whole list when given a bignum
44 as the second argument.
45 * bug fix: dynamic extent allocation of nested lists and vectors
46 could leak to otherwise accessible parts.
47 * bug fix: invalid optimization of heap-allocated alien variable
49 * bug fix: fasl header checking is less vulnerable to different
50 platform word lengths.
51 * bug fix: more correct assembler syntax for GNU binutils
52 2.18.50.0.4 support. (thanks to Marijn Schouten)
53 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
55 * bug fix: the fopcompiler can handle LOCALLY forms (with no
56 declarations) successfully. (reported by Attila Lendvai)
58 changes in sbcl-1.0.16 relative to 1.0.15:
59 * minor incompatible change: revert the changes to sb-posix's error
60 signaling added in 1.0.14.
61 * minor incompatible change: change PROBE-FILE back to returning
62 NIL whenever we can't get a truename, as was the case before 1.0.14.
63 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
64 inherits from ERROR instead of just CONDITION.
65 * new feature: SB-INTROSPECT can provide source locations for instances
66 as well. (thanks to Tobian Ritterweiler)
67 * optimization: binding special variables now generates smaller code
68 on threaded platforms.
69 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
70 and cases where no :TEST is given but the compiler can infer that
71 the element to search is of type (OR FIXNUM (NOT NUMBER)).
72 * optimization: better LOGNOT on fixnums.
73 * optimization: modular arithmetic for a particular requested width
74 is implemented using a tagged representation unless a better
75 representation is available.
76 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
77 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
78 for wellformedness and returns a specialized vector. (reported by
80 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
81 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
82 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
83 unscheduling timer at the same time as another timer fires.
84 * bug fix: don't reschedule timers for dead threads.
85 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
86 * bug fix: copying output from RUN-PROGRAM to a stream signalled
87 bogus errors if select() was interrupted.
88 * enhancement: add support for fcntl's struct flock to SB-POSIX.
90 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
91 * enhancement: cleaner backtraces for interactive interrupts, as
92 well as other cases where the interesting frames used to be
93 obscured by interrupt handling frames.
94 * enhancement: untracing a whole package using (UNTRACE "FOO") is
95 now supported, and tracing a whole package using (TRACE "FOO") now
96 traces SETF-functions as well.
97 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
98 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
99 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
100 * unreadably printed representation of hash-tables now includes
102 * bug fix: partially fixed #188: type propagation from assignments
103 is now more efficient.
104 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
105 'DOUBLE-FLOAT) are not flushable.
106 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
107 full address of the object, and none of the tag bits.
108 * bug fix: readably printing hash-tables now respects other printer
109 control variables. (reported by Cedric St-Jean)
110 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
111 keyword with MAKE-HASH-TABLE.
112 * bug fix: export SB-POSIX:MKSTEMP.
113 * bug fix: SORT was not interrupt safe.
114 * bug fix: XREF accounts for the last node of each basic-block as
116 * bug fix: MAKE-INSTANCE optimizations interacted badly with
117 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
118 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
119 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
120 create obsolete instances in the case of redefinition or
121 obsoletion of a superclass. (thanks to Andy Hefner)
122 * bug fix: Support for the Alpha architecture has been revived; it had
123 suffered somewhat from lack of maintenance since sbcl-1.0.
124 * improvements to the Windows port:
125 ** The system detects the codepage to use at startup, instead of
126 using the value from the saved core. (thanks to Kei Suzuki)
128 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
129 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
130 (see documentation for details.)
131 * revived support for OpenBSD (contributed by Josh Elsasser)
132 * partially fixed bug #108: ROOM no longer suffers from occasional
133 (AVER (SAP= CURRENT END)) failures.
134 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
135 bodies no longer cause a WARNING to be signalled. (reported by
137 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
138 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
139 the bug in TRUNCATE.) (reported by Michael Weber)
140 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
141 no samples. (reported by Andy Hefner)
142 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
143 no longer appear as (NIL ...) frames in backtraces.
144 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
145 is now actually exported from SB-EXT as documented. (reported by
147 * bug fix: sb-aclrepl now correctly understands how to inspect
148 single-floats on 64-bit platforms where single-floats are not boxed.
149 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
150 yet been finalized. (reported by Levente Meszaros)
151 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
152 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
153 SBCL-specific optimize qualities.
155 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
156 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
157 an executable in the search path, and does so in the child
158 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
159 been removed; it can be found in the CVS history, for somebody who
160 needs that search behavior (see the manual).
161 * minor incompatible change: compiler policy re. weakening type
162 checks has changed: now type checks are weakened only if SAFETY < 2
164 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
165 unparsing of directory pathnames as files. Analogously,
166 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
167 filename to parse into a directory pathname.
168 * enhancement: implicit generic function creation now signals a
169 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
170 users can bind handlers for and muffle around calls to LOAD.
171 * enhancement: RUN-PROGRAM allows unicode arguments and environments
172 to be used (using the default stream external format), and allows
173 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
174 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
175 strings and vectors whose element-type or simplicity is not fully
176 known at compile-time.
177 * optimization: STRING-TRIM and related functions no longer allocate
178 a new string when no trimming needs to be performed. These functions
179 are also faster than before when the input string has been declared
181 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
182 * optimization: READ-LINE is significantly faster for files containing
184 * optimization: non-open coded uses of character comparison operators
185 (e.g. char=) no longer cons when called with more than one parameter
186 on platforms supporting dynamic-extent allocation.
187 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
188 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
189 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
191 * bug fix: some sequence functions elided bounds checking when
193 * bug fix: too liberal weakening of union-type checks when SPEED >
195 * bug fix: more bogus fixnum declarations in ROOM implementation
198 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
199 * new feature: MAKE-HASH-TABLE now experimentally accepts a
200 :SYNCHRONIZED argument, which makes the hash-table safe for
201 concurrent accesses (but not iteration.) See also:
202 SB-EXT:WITH-LOCKED-HASH-TABLE, and
203 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
204 * optimization: CONCATENATE on strings is an order of magnitude faster
205 in code compiled with (> SPEED SPACE).
206 * optimization: SUBSEQ is ~50% faster on lists.
207 * bug fix: bug 417 fixed -- source location reporting is now more robust.
208 * bug fix: SUBSEQ on a list will now correctly signal an error if if
209 END is smaller then START.
210 * bug fix: SB-PROFILE will no longer report extra consing for nested
211 calls to profiled functions.
212 * bug fix: ROOM implementation had bogus fixnum declarations which
213 could cause type-errors when calling ROOM on large images.
214 * bug fix: if file compilation is aborted, the partial fasl is now
215 deleted, and COMPILE-FILE returns NIL as the primary value.
216 * bug fix: number of thread safety issues relating to SBCL's internal
217 hash-table usage have been fixed.
218 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
219 be returned from its body when the values were being returned
220 using unknown-values return convection and the W-P-O was wrapped
221 inside an UNWIND-PROTECT.
222 * bug fix: sb-posix should now compile again under Windows, enabling
225 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
226 * incompatible change: hash-table accessor functions are no longer
227 automatically protected by locks. Concurrent accesses on the same hash-table
228 from multiple threads can give inconsistent results or even corrupt the
229 hash-table completely. Multi-threaded applications should do their own
230 locking at the correct granularity. In the current implementation it is
231 still safe to have multiple readers access the same table, but it's not
232 guaranteed that this property will be maintained in future releases.
233 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
234 and will signal an error at runtime.
235 * enhancement: SB-THREAD package now exports a semaphore interface.
236 * enhancement: CONS can now stack-allocate on x86 and
237 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
239 * enhancement: nested lists can now be stack allocated on
240 platforms providing stack allocation support.
241 * enhancement: dynamic-extent support has been extended to support
242 cases where there are multiple possible sources for the stack
244 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
245 if the mutex is uncontested on Linux.
246 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
247 as specified by the CLHS. (thanks to Tobias Rittweiler)
248 * bug fix: NaN comparison now works on x86-64.
249 * bug fix: CLOSE :ABORT T on a stream with pending output now
251 * bug fix: instances of non-standard metaclasses using standard
252 instance structure protocol sometimes missed the slot type checks
254 * bug fix: known functions can be inlined in high-debug code.
256 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
257 * minor incompatible change: the MSI installer on Windows no longer
258 associates .lisp and .fasl files with the installed SBCL.
259 * minor incompatible change: :UNIX is no longer present in *FEATURES*
260 on Windows. (thanks to Luis Oliviera)
261 * new feature: SB-CLTL2 contrib module now implements
262 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
263 * optimization: scavenging weak pointers is now more efficient,
264 requiring O(1) instead of O(N) per weak pointer to identify
265 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
266 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
267 method bodies using constant slot-names when the first argument is
268 a specializer parameter for the method.
269 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
270 names, when the class of the instance is a direct instance
271 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
272 as slow as the constant slot-name case.)
273 * optimization: member type construction is now O(N) instead
275 * optimization: UNION and NUNION are now O(N+M) for large
276 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
278 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
279 now more readable in environments like Slime which display it.
280 (thanks to Tobias C. Rittweiler)
281 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
282 was able to infer the correct argument type for the object on which
283 the CAS operation was being performed.
284 * bug fix: copy propagation interfered with parallel assignment
285 semantics in local calls. (reported by Paul Khuong)
286 * bug fix: the signed modular fixnum shift compiled to wrong code on
287 x86 and x86-64. (spotted by a slight modification to some of
290 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
291 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
292 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
293 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
295 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
296 and generic function now print the object identity for anonymous
297 instances. (thanks to Attila Lendvai)
298 * enhancement: as an extension to MOP, SBCL now supports SETF of
299 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
300 (thanks to Attila Lendvai)
301 * enhancement: sb-cover has an alternate annotation mode which puts
302 the form annotations at the car of the form, instead of annotating
303 it completely. The benefit of this mode is that it shows better
304 which forms were instrumented by the compiler.
305 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
306 by printer control variables. (reported by Dan Corkill)
307 * bug fix: system leaked memory when delayed output was performed by
308 the OS in smaller chunks then expected. (thanks to David Smith)
309 * bug fix: system leaked memory when file streams were not closed
311 * bug fix: large objects written to slow streams that were modified
312 after the write could end up with the modified state written to
313 the underlying file descriptor.
314 * bug fix: multiple threads operating in parallel on the same stream
315 could cause buffer-overflows.
316 * bug fix: source location information is stored correctly
317 (broken since 1.0.6). This bug would generally show up as the
318 Slime debugger highlighting the wrong form when the sldb-show-source
320 * bug fix: Branch forms are again annotated as branches in the sb-cover
322 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
323 on x86-64. (reported by Gregory Vanuxem)
324 * bug fix: EQUALP could return wrong results for structures with raw
325 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
326 word). (reported by Vjacheslav Fyodorov)
327 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
330 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
331 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
332 atomic compare-and-swap operations on threaded platforms.
333 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
334 allows assigning a global minimum value to optimization qualities
335 (overriding proclamations and declarations).
336 * enhancement: closed over variables can be stack-allocated on x86
338 * performance bug fix: GETHASH and (SETF GETHASH) are once again
340 * optimization: slot definition lookup is now O(1). This speeds up
341 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
342 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
343 * optimization: ASSOC and MEMBER can now be open-coded for all
344 combinations of keyword arguments when second argument is constant
345 and SPEED >= SPACE. In other cases a specialized version is
347 * bug fix: using obsoleted structure instances with TYPEP and
348 generic functions now signals a sensible error.
349 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
350 (reported by Kristoffer Kvello)
351 * bug fix: backtrace construction is now more careful when making
352 lisp-objects from pointers on the stack, to avoid creating bogus
353 objects that can be seen by the GC.
354 * bug fix: defaulting of values in contexts expecting more than 7
355 variables now works on x86-64. (reported by Christopher Laux)
356 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
358 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
359 as the property-list of a symbol.
360 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
361 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
362 situations requested, are once again file-compileable. (reported
365 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
366 * MOP improvement: support for user-defined subclasses of
367 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
368 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
369 the expansion of DEFMETHOD, is responsible for generating a form
370 which creates a list of specializers when evaluated. Additional
371 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
372 debugging and introspective support.
373 * minor incompatible change: the (unsupported) spinlock interface
374 has changed: free spinlock now has the value NIL, and a held spinlock
375 has the owning thread as its value.
376 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
377 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
379 * enhancement: name of a socket-stream is now "a socket" instead of
381 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
382 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
383 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
384 for generic functions. (Thanks to Tobias C. Rittweiler)
385 * optimization: bignum printing speed has been improved by 20-40%
386 (depending on the bignum size.)
387 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
389 * bug fix: the cache used by the CLOS to store precomputed effective
390 methods, slot offsets, and constant return values is now thread and
392 * bug fix: generic function dispatch function updating is now thread
393 and interrupt safe (in the sense that the known issues have been
395 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
396 * bug fix: interning EQL-specializers is now thread and interrupt safe.
397 * bug fix: asdf systems with dependencies to the SB-POSIX or
398 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
399 * bug fix: interrupt safety of applicable method computation has been
402 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
403 * new contrib: sb-cover, an experimental code coverage tool, is included
405 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
406 significantly faster.
407 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
408 produce more efficient code on x86-64 if the compiler can determine
409 that (+ INDEX <constant>) does not require a bounds check and FOO
410 has an element type at least 8 bits wide.
411 * enhancement: a new, experimental synchronous timeout facility is
412 provided. Refer to SB-SYS:WITH-DEADLINE for details.
413 * enhancement: when a symbol name conflict error arises, the
414 conflicting symbols are always printed with a package prefix.
415 (thanks to Kevin Reid)
416 * enhancement: stepping is now once again supported on the SPARC and
417 MIPS platforms. (It is also now more likely to work on CheneyGC
419 * enhancement: sb-sprof can now also track and report accurate call
421 * bug fixes: the treatment of non-standard subclasses of
422 SB-MOP:SPECIALIZER is more correct.
423 * incompatible change: PURIFY no longer copies the data from the
424 dynamic space into the static and read-only spaces on platforms
425 that use the generational garbage collector
426 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
428 * bug fix: GC race condition occasionally resulting in crashes with
429 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
431 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
433 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
434 a STYLE-WARNING to be signalled (regression from 1.0.4.)
435 * bug fix: an asynchronous interrupt could previously leave the
436 system running with GC inhibited.
437 * bug fix: a DECLARE form evaluated at top-level now causes an error
438 rather than silently (or verbosely) returning NIL.
439 * bug fix: trying to dissassemble functions compiled at high DEBUG could
440 result in a error being signalled due to source form lookup errors.
441 (reported by Peter Graves)
443 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
444 * incompatible change: removed writer methods for host-ent-name,
445 host-ent-addresses -- changing the values did not update the DNS
447 * minor incompatible change: changed experimental JOIN-THREAD interface
448 * documentation: the manual now lists reader and writer methods
449 in class slot documentation sections. (thanks to Richard M Kreuter)
450 * documentation: unwinding from asyncronous events has been
451 documented as unsafe.
452 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
453 in multithreaded application code.
454 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
455 platforms. (thanks to James Anderson for the optimization hint)
456 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
458 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
459 variants no longer cons.
460 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
461 their NOT- variants no longer cons.
462 * optimization: EQUAL hash tables no longer use SXHASH for objects
463 of all data types, but instead use an EQL hash for types for which
464 EQUAL is the same as EQL.
465 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
466 are significantly faster.
467 * optimization: new STRING-OUTPUT-STREAM implementation is much
468 faster for large outputs and conses 30% less on average.
469 * enhancement: XREF information is now collected for references made
470 to global variables using SYMBOL-VALUE and a constant argument.
471 * enhancement: SIGINT now causes a specific condition
472 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
473 * bug fix: FILE-POSITION used to signal an error in some cases where
474 ANSI requires it to return NIL.
475 * bug fix: ADJUST-ARRAY is now interrupt-safe.
476 * bug fix: adding and removing fd-handlers is now interrupt-safe.
477 * bug fix: inlined calls to C now ensure 16byte stack alignment on
479 * bug fix: bad type declaration in the CLOS implementation has
480 been fixed. (reported by James Anderson)
481 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
482 (thanks to Lutz Euler)
483 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
484 required by the CLHS. (reported by Samium Gromoff)
485 * bug fix: dead unbound variable references now signal an error.
486 (reported by Marco Monteiro)
487 * bug fix: / with an unused value was being deleted in safe code.
488 (thanks to Marco Monteiro and Kevin Reid)
489 * bug fix: number of characters that can be written onto a single
490 line in a file is unlimited.
491 * bug fix: some GC deadlocks caused by asynchronous interrupts have
492 been fixed by inhibiting interrupts for when GC is disabled.
493 * bug fix: some interrupt safety issues with GETHASH, (SETF
494 GETHASH), CLRHASH and REMHASH have been fixed.
495 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
496 specifier no longer causes infinite recursion.
497 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
498 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
499 * bug fix: an &environment argument with the correct variable information
500 is passed to macros that are expanded during byte compilation.
501 (reported by Samium Gromoff)
502 * bug fix: a logic error was causing the hash function for bignums to
503 have excessive amounts of collisions. (reported by Faré Rideau)
504 * bug fix: modifying the contents of an array could change the return
505 value of SXHASH on that array, which is only allowed for strings
506 and bit vectors. (bug introduced in 0.9.16)
507 * improvement: the x86-64/darwin port now passes all tests (except
508 for the debugger tests) but should still be considered
509 experimental until this is fixed.
510 * improvement: a style-warning is signaled for CASE (etc) clauses with
511 duplicate keys. (thanks to Kevin Reid)
512 * improvement: macos/x86 and macos/x86-64 now use mach exception
513 handlers for dealing with illegal instructions (for trapping and
514 error handling) and memory protection violations (for GC).
516 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
517 * new platform: experimental support for x86-64/darwin (MacOS).
518 * incompatible change: the thread-safe (on most platforms) getaddrinfo
519 and getnameinfo sockets functions are used instead of gethostbyaddr
520 and gethostbyname, on platforms where the newer functions are available.
521 As a result, the ALIASES field of HOST-ENT will always be NIL on these
523 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
524 don't print the startup banner, but behave as if --noinform was passed
525 as a command line argument. (thanks to Kevin Reid)
526 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
527 * optimization: code using alien values with undeclared types is much faster.
528 * optimization: the compiler is now able to open code SEARCH in more cases.
529 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
530 * bug fix: using standardized COMMON-LISP special variables as loop
531 variables no longer signals bogus package lock violations. (reported
533 * bug fix: declaring local loop variables to be of a range-limited type
534 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
535 (reported by Andras Simon)
536 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
537 bugs remain on x86-64.)
538 * bug fix: the #= and ## reader macros now interact reasonably with
539 funcallable instances.
540 * bug fix: type-checks for function arguments were compiled using the
541 compiler policy settings of the wrong lexical environment (in 1.0.2
543 * bug fix: SHADOW accepts characters as string designators, as required
544 by the spec (thanks to Eric Marsden)
545 * bug fix: fixed GC safety issues when foreign functions are called with
546 non-base strings as arguments
547 * bug fix: more consistent error output for fatal-compiler-warnings, like
549 * bug fix: functions with proper names were showing up as NIL in
550 backtraces (reported by Edi Weitz, regression in 1.0.2)
552 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
553 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
554 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
555 produce more efficient code on the x86 if the compiler can determine
556 that (+ INDEX <constant>) does not require a bounds check and FOO
557 has an element type at least 8 bits wide.
558 * bug fix: references to undefined variables are handled the same way
559 in toplevel forms as in the normal compiler
560 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
562 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
563 (thanks to Jon Buller)
564 * improvement: faster compilation times for complex functions
565 * improvement: added readlink support to SB-POSIX (thanks to Richard
568 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
569 * improvement: experimental support for mach exception handling on
570 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
572 * improvement: support for GBK external format.
573 (thanks to Chun Tian (binghe))
574 * improvement: the debugger now displays variables that have been closed
575 over, in code compiled with (DEBUG 2) or higher
576 * improvement: support for executable cores on NetBSD (thanks to
578 * new feature: added a RESTART-FRAME debugger command
579 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
580 be used to provide an implementation for FILE-POSITION on Gray streams
581 (thanks to Eric Marsden)
582 * new feature: add syslog, openlog and closelog support to SB-POSIX
583 (thanks to Richard Kreuter)
584 * optimization: the function call overhead in code compiled with
585 a high DEBUG optimization setting is significantly
586 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
587 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
588 (thanks to Tony Martinez)
589 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
590 evaluated once (reported by Kevin Reid)
591 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
592 works again (thanks to Kilian Sprotte)
593 * bug fix: an error is signaled for tagbodies with duplicate tags
594 (thanks to Stephen Wilson)
595 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
596 * bug fix: Win32 port can now handle foreign code unwinding Lisp
597 stack frames from alien callbacks.
598 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
599 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
600 65k pages (thanks to David Woodhouse)
601 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
603 changes in sbcl-1.0.1 relative to sbcl-1.0:
604 * new platform: FreeBSD/x86-64, including support for threading.
605 * new feature: the compiler stores cross-referencing information
606 about function calls (who-calls), macroexpansion (who-macroexpands)
607 and special variables (who-binds, who-sets, who-references) for code
608 compiled with (< SPACE 3). This information is available through the
609 sb-introspect contrib.
610 * new feature: users may subclass SEQUENCE, and have instances of
611 these classes interoperate with standard Common Lisp functions if
612 a number of methods are defined. (This feature is experimental
613 and the interface subject to change based on feedback from SBCL
614 users and the general community)
615 * improvement: sb-sprof traces call stacks to an arbitrary depth on
616 x86/x86-64, rather than the previous fixed depth of 8
617 * improvement: another pthread back-end of mutex "pthread-futex".
618 * bug fix: non-ascii command-line arguments are processed correctly
619 (thanks to Yaroslav Kavenchuk)
620 * bug fix: non-required arguments were not passed correctly when a method
621 defined using DEFMETHOD was called from a mop-generated method using
622 CALL-NEXT-METHOD (reported by Pascal Costanza)
623 * bug fix: recursion is now permitted in accessors through
624 SLOT-UNBOUND. (reported by Pascal Costanza)
625 * bug fix: an error was signaled at startup if the HOME environment
626 variable was defined, but had an empty value (reported by Peter Van Eynde)
627 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
628 signal a PROGRAM-ERROR, not a BUG.
629 * bug fix: SB-POSIX:READDIR works when built with large file support.
630 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
632 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
633 for better automatic indentation support. (thanks to Matt Pillsbury)
634 * optimization: loading generic functions no longer takes O(n^2) time,
635 proportional to the amount of methods in the generic function
636 (reported by Todd Sabin and Jeremy Brown)
637 * optimization: the FIND and POSITION family of sequence functions
638 are significantly faster on arrays whose element types have been
640 * improvements to the Windows port:
641 ** Intermittent heap corruption problems have been fixed. (thanks
642 to Alastair Bridgewater)
643 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
645 ** Lisp is able to unwind foreign exception frames from alien
646 callbacks. (thanks to Alastair Bridgewater)
648 changes in sbcl-1.0 relative to sbcl-0.9.18:
649 * improvement: experimental support for threading on FreeBSD/x86.
650 (thanks to NIIMI Satoshi)
651 * improvement: runtime option --dynamic-space-size can be used
652 to set the size of the dynamic space reserved on startup.
653 * improvement: floating point modes in effect are now saved in
654 core, and restored on startup.
655 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
656 startup, not time since first call to GET-INTERNAL-REAL-TIME.
657 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
658 threads are not running after *SAVE-HOOKS* have run.
659 * improvement: writes to CLOS instance slots are type-checked in code
660 compiled with (SAFETY 3)
661 * improvement: floating-point exception handling on FreeBSD (thanks to
663 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
664 (thanks to Zach Beane)
665 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
667 * improvement: added support for the Shift-JIS external format.
668 (contributed by NIIMI Satoshi)
669 * improvement: callbacks are supported on Linux/PPC. (thanks to
671 * bug fix: compiler bug triggered by a (non-standard) VALUES
672 declaration in a LET* was fixed. (reported by Kaersten Poeck)
673 * bug fix: file compiler no longer confuses validated and already
674 dumped structurres. (reported by Kaersten Poeck)
675 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
676 fill-pointer signals a type-error as required. (thanks to
678 * bug fix: disassemly of funcallable instances works.
679 * bug fix: single stepping on PPC.
680 * bug fix: fix thread-safety problems in the type system (generally
681 manifesting as nonsensical errors like "STRING is a bad type specifier
682 for sequences" or "The value 1 is not of type FIXNUM")
683 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
684 (reported by Josip Gracin)
685 * bug fix: an error is signaled for attempts to displace arrays with
686 incompatible element types (thanks to Mario Mommer)
687 * bug fix: more correct handling of wide characters in debug info
688 (bug reported by Attila Lendvai and fixed by Juho Snellman)
689 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
690 and don't cause extra consing
691 * optimization: MAP and MAP-INTO are significantly faster on vectors
692 whose elements types have been declared.
693 * Improvements to SB-SPROF:
694 ** Support for allocation profiling
695 ** Reduced profiling overhead, especially for long profiling runs
696 * Improvements to the Windows port:
697 ** floating point exceptions are now reported correctly.
698 ** stack exhaustion detection works partially.
699 ** more accurate GET-INTERNAL-REAL-TIME.
700 ** better RUN-PROGRAM behaviour in case of errors in the child
702 ** PROBE-FILE now simplifies pathnames correctly.
703 ** DIRECTORY now works correctly with :WILD-INFERIORS.
704 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
705 1st 1970 (although time timezone data will be unreliable).
706 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
708 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
709 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
710 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
712 * bug fix: two potential GC deadlocks affecting threaded builds.
713 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
714 returns T (reported by Anton Kazennikov)
715 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
716 with non-variable places
717 * bug fix: remove a race condition in the setting of
718 funcallable-instance functions, this should make threaded CLOS
719 code more stable against memory faults.
720 * bug fix: corruption of specials when unbinding is interrupted by an
721 asynchronous unwind (reported by Hannu Koivisto)
722 * improvement: the debugger will now also display local variables that
723 are only used once, for code compiled with a DEBUG optimization quality
726 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
727 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
728 * incompatible change: External-format support for FFI calls. The
729 SB-ALIEN:C-STRING no longer implies an ASCII
730 external-format. Instead the string is subject to external-format
731 conversion. Additionally return values of type SB-ALIEN:C-STRING
732 are by default of type (SIMPLE-ARRAY CHARACTER), not
733 SIMPLE-BASE-STRING. For an alien type that behaves like the old
734 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
735 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
736 * incompatible change: SB-EXT package no longer contains the
737 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
738 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
740 * incompatible change: the single-stepper is no longer available
741 on Alpha, Mips and Sparc platforms.
742 * minor incompatible change: the direct superclasses of
743 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
744 not (STANDARD-OBJECT FUNCTION). This makes the
745 class-precedence-lists of GENERIC-FUNCTION and
746 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
748 * minor incompatible change: the default stream external format on
749 non-unicode SBCL is no longer detected from the locale, but is always
751 * new feature: Add a version of evaluator that uses an interpreter instead
752 of the compiler. EVAL still uses the compiler by default, to switch it
753 to use the interpreter, set the value of the variable
754 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
755 * minor incompatible change: the single-stepper REPL has been merged
756 with the normal debugger (see the "Stepping" heading of the debugger help
757 for more details). The debugger command STEP will no longer switch
758 to the single-stepper REPL.
759 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
760 for a type now works.
761 * bug fix: loading of default sysinit file works. (thanks to Leonid
763 * bug fix: better detection of circularities in the file-compiler.
764 (reported by Marco Monteiro)
765 * bug fix: the CL pathname functions now work with files that have
766 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
767 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
768 systems with Unix98 pty semantics.
769 * bug fix: ASDF-INSTALL will now work with bsd tar.
770 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
772 * bug fix: timers expiring in dead threads no longer cause a
773 type-error (reported by Paul "Nonny Mouse").
774 * bug fix: thanks to more lightweight single-stepper instrumentation,
775 code compiled with (DEBUG 3) will compile and execute significantly faster,
776 and will have more accurate type-inferencing than before
777 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
778 whose bindings are modified
779 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
780 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
781 ** replace dummy implementations of CL:MACHINE-INSTANCE and
782 CL:SOFTWARE-VERSION with proper version
784 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
785 * feature: implemented the READER-METHOD-CLASS and
786 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
787 as specified by AMOP.
788 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
790 * optimization: faster LOGCOUNT implementation on x86 and x86-64
791 (thanks to Lutz Euler)
792 * optimization: hashing of general arrays and vectors has been
793 improved. (reported by Any Fingerhut)
794 * enhancement: SB-INTROSPECT is now able to find definitions of
795 profiled functions. (thanks to Troels Henriksen)
796 * enhancement: compiler-macro expansion applies now to FUNCALL forms
797 as well, allowing compiler-macros for SETF-functions to expand.
798 * enhancement: step-instrumentation no longer wraps known
799 single-value functions with multiple-value context, allowing
800 better type inference.
801 * fixed bug #337: use of MAKE-METHOD in method combination now works
802 even in the presence of user-defined method classes. (reported by
803 Bruno Haible and Pascal Costanza)
804 * fixed bug #339(c): if there are applicable methods not part of any
805 long-form method-combination group, call INVALID-METHOD-ERROR.
806 (reported by Bruno Haible)
807 * fixed bug #361: the :FUNCTION initarg in the protocol for
808 initialization of methods can now be used to override
809 internally-produced optimized functions. (reported by Bruno
811 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
812 system-provided lambda expression no longer cause warnings about
813 unbound #:|pv-table| symbols.
814 * bug fix: improved the handling of type declarations and the
815 detection of violations for keyword arguments with non-constant
817 * bug fix: potentially erronous calls to PATHNAME and
818 MERGE-PATHNAMES were being flushed in some cases.
819 (reported by Richard Kreuter)
820 * bug fix: compiled calls to TYPEP were mishandling obsolete
821 instances. (reported by James Bielman and Attila Lendvai)
822 * bug fix: format strings with ~> without matching ~< no longer
823 trigger an AVER, but signal an understandable error instead.
824 (reported by Antonio Martinez)
825 * bug fix: specifying an output-file without a pathname-type for
826 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
827 of input-file instead of "fasl". (reported by Robert Dodier)
828 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
829 (reported by James Y Knight).
830 * bug fix: compiler-macros-function did not consider the environment
831 argument for shadowing by local functions.
832 * bug fix: compiler-macros expansion was inhibited by local INLINE
834 * bug fix: inline expansions of known functions were subject to
835 step-instrumentation in high debug policies, leading to problems
837 * bug fix: compiler failed to differentiate between different CONS
839 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
840 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
841 * bug fix: #368: incorrect use of expressed vs. upgraded array
843 * bug fix: #306a: more precise unions of array types.
844 * thread-safety improvements:
845 ** CONDITION-WAIT could return early on Linux, if the thread was
846 interrupted and subsequently continued with SIGCONT.
847 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
849 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
850 * added support for the ucs-2 external format. (contributed by Ivan
852 * minor incompatible change: pretty printing of objects of type
853 (cons symbol) is, in the default pprint-dispatch-table, now
854 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
856 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
857 called later in a class's lifetime, possibly as late as when the
858 first instance of the class is created. Previously,
859 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
860 class became finalizeable.
861 * fixed bug: FILE-POSITION sometimes returned inconsistent results
862 for multibyte external-format streams. (thanks to "vbzoli")
863 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
864 with :ALLOCATION :CLASS inherited from superclasses of the
866 * fixed bug: anonymous classes can now be created using the :NAME
867 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
868 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
869 * fixed bug: core-files saved with :EXECUTABLE T can again be
870 executed when SBCL_HOME isn't set. (reported by James Knight)
871 * fixed bug: toplevel LOCALLY forms with declarations could
872 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
873 * fixed bug: printing from several different threads using different
874 values of *print-case* could cause invalid output, due to
875 some internal special variables of the printer not being bound
876 thread-locally (reported by Max Mikhanosha)
877 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
878 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
879 (reported by Pascal Costanza)
880 * fixed bug: FINALIZE-INHERITANCE is called from
881 REINITIALIZE-INSTANCE on classes when the class has previously
882 been finalized, as required by AMOP.
883 * minor code generation optimizations:
884 ** better register allocation in CLOS dispatching functions
885 ** overflow detection when coercing signed bytes to fixnums on x86-64
886 ** is now implemented with one IMUL instruction instead of three shifts
887 ** more efficient bit-vector access on x86 and x86-64
888 ** more efficient access to raw structure slots on x86 and x86-64
889 * fixed some bugs revealed by Paul Dietz' test suite:
890 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
893 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
894 * feature: thread support on Solaris/x86, and experimental thread support
896 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
898 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
899 default method for SLOT-UNBOUND.
900 * minor incompatible change: prevent the user from specializing the
901 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
902 somewhat counter to the intent of the protocol, I (CSR) think, and
903 additionally it just doesn't work in SBCL as currently
904 implemented, thanks to optimizations (that are always valid for
905 the other three SLOT-VALUEish functions, but not for the setter).
906 * minor incompatible change: the :SB-LDB feature is now enabled by
907 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
908 the low-level debugger.
909 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
910 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
911 down to an SB-SYS:FD-STREAM, allowing direct communication in
912 more cases. Temporary files and pipes are still used as a fallback
914 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
915 on the non-reentrant C functions localtime(3) and gmtime(3).
916 * bug fix: global optimization policy was not visible in LOCALLY and
918 * bug fix: class objects can be used as specializers in methods.
919 (reported by Pascal Costanza)
920 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
921 with a type error. (reported by blitz_ on #lisp)
922 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
923 failed with a type error. (reported by Pascal Bourguignon)
924 * bug fix: merging pathnames against defaults with :DIRECTORY
925 starting with '(:RELATIVE :BACK) should preserve the :BACK.
926 (reported by James Y Knight)
927 * bug fix: saving large (>2GB) cores on x86-64 now works
928 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
929 constant index and a (simple-array (signed-byte 32)) array
930 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
931 error (patch by Robert J. Macomber)
932 * feature: TIME now displays the amount of run-time spent in GC
933 * bug fix: The debugger now does a better job of respecting
934 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
935 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
936 workaround for bug 403.)
937 * bug fix: floating point exception handling now works on FreeBSD
938 (thanks to NIIMI Satoshi)
939 * fixed some bugs revealed by Paul Dietz' test suite:
940 ** MISC.641: LET-conversion were not supposed to work in late
942 * improvements to the Win32/x86 port:
943 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
944 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
946 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
947 * new feature: source path information is generated for macro-expansion
948 errors for use in IDE's like Slime (thanks to Helmut Eller)
949 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
950 * bug fix: compiler does not loop forever on an invalid type in
952 * improvement: compilation of most CLOS applications is significantly
954 * optimization: added a limited bytecode compiler for simple toplevel
955 forms, speeding up compilation and FASL loading
956 * bug fix: the statistical profiler now properly distinguishes anonymous
959 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
960 * minor incompatible change: in sbcl-0.9.11 (but not earlier
961 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
962 into subclasses of the finalized class. Now user calls to
963 FINALIZE-INHERITANCE finalize just the one class, and calls by the
964 system return before any subclasses are finalized.
965 * minor incompatible change: The reader no longer ignores errors
966 regarding non-existent packages in #+ and #- feature tests.
967 * new feature: command line options --no-sysinit, --no-userinit to
968 inhibit loading the corresponding init files
969 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
970 for sending data through UDP sockets (thanks to François-René Rideau)
971 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
972 error is signalled instead (thanks to François-René Rideau)
973 * minor incompatible change: Error signalling behaviour of lexical
974 operations violating package locks has changed slightly. Refer to
975 documentation on package locks for details.
976 * enhancement: EVAL can process IF-expressions without resorting to the
978 * bug fix: Some operations involving SIN, COS, or TAN weren't being
979 constant-folded properly due to missing out-of-line functions.
980 (reported by Mika Pihlajamäki)
981 * bug fix: LISTEN sometimes returned T even in cases where no data was
982 immediately available from the stream
983 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
984 were reversed. (reported by Levente Mészáros)
985 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
986 when given a vector displaced to an adjustable array. (reported
988 * bug fix: derivation of float boundaries from numbers outside the
989 appropriate float range (reported by John Wiseman)
990 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
992 * bug fix: precomputing cache entries for generic functions with
993 some subclasses of specializers as yet invalid does not attempt to
994 fill a cache line with a negative offset. (reported by Levente
996 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
997 allow condition class objects as arguments to DOCUMENTATION and
998 (SETF DOCUMENTATION); only find and set documentation for
999 structure names for the STRUCTURE doc-type. (suggested by Gary
1001 * improvements to the Win32/x86 port:
1002 ** (user-homedir-pathname) and default initialization file
1003 locations now know about the user's "Documents and Settings"
1004 directory (thanks to Yaroslav Kavenchuk)
1005 ** run-program is implemented (thanks to Mike Thomas)
1006 ** sockets support (thanks to Timothy Ritchey)
1007 ** better backtrace support (thanks to Alastair Bridgewater)
1008 ** sb-grovel supported
1009 ** asdf-install and sb-posix work somewhat
1010 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1011 * improvements to the Solaris/x86 port:
1012 ** works on Solaris 11/Solaris Express
1013 ** floating-point exception handling support
1014 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1015 * fixed some bugs revealed by Paul Dietz' test suite:
1016 ** REMOVE-METHOD returns its generic function argument even when
1017 the method is not one of the generic functions' methods.
1018 ** objects declared as MEMBER types can be admissible arguments to
1019 structure accessors.
1020 ** printing characters should simply be printed by the FORMAT ~:C
1022 ** compiler failure when compiling functions with hairy constant
1023 defaults for optional parameters.
1024 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1025 function, which is already optimized.
1027 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1028 * new platform: experimental support for SBCL x86/Darwin, including
1029 MacOS X 10.4.5 on Intel.
1030 * new feature: Unicode character names are now known to the system
1031 (through CHAR-NAME and NAME-CHAR).
1032 * new feature: the filesystem location of SBCL's core file is
1033 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1034 * minor incompatible change: the contrib modules SB-POSIX and
1035 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1036 this change is to make it easier to distribute
1037 location-independent binaries.
1038 * bug fix: as implied by AMOP, standardized classes no longer have
1039 slots named by external symbols of public packages. (reported by
1041 * optimization: calling functions via constant symbols -- as in
1042 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1043 via the function object as in (FUNCALL #'FOO).
1044 * optimization: on x86 and x86-64, the calling convention for the
1045 case of dealing with an unknown number of values has been altered
1046 to be friendlier to the prediction heuristics implemented,
1047 particularly with reference to CALL and RET pairing. (thanks to
1048 Alastair Bridgewater)
1049 * optimization: on x86, the code for access to thread-local symbol
1050 values and binding/unbinding of thread-local symbols is smaller.
1051 (thanks to Alastair Bridgewater)
1052 * enhancement: CONSTANTP is now able to determine constantness of
1053 more complex forms, including calls to constant-foldable standardized
1054 functions and some special forms beyond QUOTE.
1055 * fixed bug: occasional GC crashes on Solaris/x86
1056 * optimization: x86-64 supports stack allocation of results of simple
1057 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1058 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1059 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1060 (thanks to James Knight)
1061 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1062 by the standard (thanks to Stephen Compall)
1064 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1065 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1066 be used for bundling the runtime and the core file into one
1067 executable binary. This feature is not currently supported on all SBCL
1068 platforms. (thanks to James Bielman and NIIMI Satoshi)
1069 * new feature: a generational or ephemeral garbage collector is now
1070 the default on the PowerPC platform (both Linux and Darwin). The
1071 old Cheney (stop and copy) collector is a build-time option.
1072 (thanks to Cyrus Harmon, after Raymond Toy)
1073 * minor incompatible change: the method by which SBCL finds its
1074 contributed modules has changed; it no longer relies on symbolic
1075 links from an $SBCL_HOME/systems directory, but searches directly
1076 in the subdirectories of $SBCL_HOME.
1077 * enhancement: the dynamic heap size on the Linux/PPC platform is
1078 markedly larger, even using the older Cheney garbage collector.
1079 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1080 certain complicated string types. (reported by Gary King)
1081 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1082 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1083 * fixed bug: DOCUMENTATION on structures no longer signals an
1084 error if no structure type of the right name exists. (reported by
1086 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1087 applicable non-standard primary or :AROUND method on
1088 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1089 no longer cause unbound variable errors. (reported by Kevin Reid)
1090 * fixed bug: metacircle resolution in cases where methods have slots
1091 added before the slots from STANDARD-METHOD. (reported by Jean
1093 * fixed bug: the Power PC assembly code for calling into lisp
1094 sometimes computed its return address wrongly (depending on the
1095 15th bit of the address, fixed at link time). This used to
1096 manifest itself in a segmentation violation while building PCL.
1097 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1098 many others over the years)
1099 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1100 the stack pointer (treated as a signed integer) is negative.
1101 (thanks to Peter van Eynde)
1103 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1104 * new platform: experimental support for the Windows operating
1105 system has been added. (thanks to Alastair Bridgewater)
1106 * fixed several bugs in and robustified the PPC FFI (including
1107 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1108 * bug fix: allow non-simple string symbol names (reported by Paul
1110 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1112 * bug fix: saving a core corrupted callbacks on x86/x86-64
1113 * bug fix: closed a loophole in metacircularity detection and
1114 grounding in the PCL implementation of CLOS.
1115 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1116 (reported by tomppa on #lisp)
1117 * bug fix: clear the direction flag on Lisp -> C transitions, as
1118 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1119 (reported by Andrej Grozin and Hendrik Maryns)
1120 * optimization: major improvements to GC efficiency on GENCGC platforms
1121 * optimization: faster implementation of EQUAL
1122 * optimization: emit more efficient opcodes for some common
1123 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1124 * optimization: several other minor code-generation improvements on x86-64
1125 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1127 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1128 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1129 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1130 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1131 by ANSI as a generic function, and for consistency (SETF
1132 GENERIC-FUNCTION-NAME) is treated likewise.
1133 * fixed bug #233.b: make constraint propagation notice when a variable
1134 value is changed after it is referenced but before it is used
1135 * fixed bug #296: no more arbitrary behaviour from filenames with
1136 odd characters as --load arguments. (reported by Adam Warner)
1137 * fixed bug #390: :CHARACTER-SET pathname components now work as
1138 expected. (reported by Tim Daly Jr)
1139 * fixed bug #391: complicated :TYPE intersections in slot
1140 definitions no longer cause an error in PCL internals.
1141 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1142 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1143 returning the number of octets which would be written to the
1144 file-stream. (thanks to Robert J. Macomber)
1145 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1146 arguments to the internal stream functions. (thanks to David
1148 * bug fix: stack exhaustion is now handled on the x86/SunOS
1149 platform. (thanks to Daisuke Homma)
1150 * bug fix: more accurate ROOM results on GENCGC platforms
1151 * optimization: improved type inference for arithmetic-for
1152 index variables in LOOP
1153 * optimization: faster floating-point SQRT on x86-64
1154 * fixed some bugs revealed by Paul Dietz' test suite:
1155 ** DOCUMENTATION returns NIL instead of "" for method combinations
1156 that don't have a docstring
1158 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1159 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1160 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1161 therefore are not customizeable by user code (as seems to be at
1162 least permitted and maybe required by AMOP). As a consolation,
1163 however, the SBCL implementation of these functions now calls
1164 REINITIALIZE-INSTANCE as specified by AMOP.
1165 * bug fix: slot-definition documentation is propagated as per ANSI
1166 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1167 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1168 argument, as do eslotd initialization methods. (from Pascal
1169 Costanza's "Closer" project)
1170 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1171 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1173 * bug fix: it is now possible to have more than one subclass of
1174 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1175 (reported by Bruno Haible, Pascal Costanza and others)
1176 * bug fix: the dependent update protocol now works for generic
1177 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1178 and Pascal Costanza)
1179 * bug fix: condition-class instances corresponding to
1180 DEFINE-CONDITION forms are now created eagerly. (reported by
1181 Kalle Olavi Niemitalo on comp.lang.lisp)
1182 * bug fix: floating point printing is more accurate in some
1183 circumstances. (thanks to Simon Alexander)
1184 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1185 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1186 * bug fix: callbacks on OS X now preserve stack-alignment.
1187 (thanks to Cyrus Harmon)
1188 * enhancement: the x86-64 disassembler is much better at
1189 disassembling SSE instructions. (thanks to Lutz Euler)
1190 * enhancement: CHECK-TYPE on a variable now allows type-inference
1191 (thanks to Vincent Arkesteijn)
1192 * optimization: improved performance of EUC-JP external format.
1193 (thanks to NIIMI Satoshi)
1194 * optimization: performance improvements to IO on file streams of
1195 :ELEMENT-TYPE CHARACTER
1196 * optimization: much faster memory allocation on x86-64
1197 * optimization: faster garbage collections (latency and throughput)
1199 * optimization: faster bignum subtraction and division on x86 and x86-64
1200 (thanks to Lutz Euler)
1201 * optimization: more accurate type-inference for loops that use a
1202 floating point index variable or a negative step.
1204 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1205 * bug fix: add a workaround to SBCL looping infinitely at startup on
1206 Linux kernels with apparently buggy implementations of personality().
1207 (thanks to Svein Ove Aas)
1208 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1209 (thanks to David Lichteblau)
1210 * bug fix: local bindings shadow global symbol macros for
1211 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1212 * optimization: non-open coded uses of numeric comparison operators
1213 (e.g. >) no longer cons when called with more than one parameter
1214 on platforms supporting dynamic-extent allocation.
1215 * enhancement: saving cores with foreign code loaded is now supported
1216 on MIPS/Linux in addition to the previously supported platforms.
1217 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1218 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1219 * bug fix: division by zero in sb-sprof when no samples were collected
1220 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1221 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1222 (thanks to Alexander Kjeldaas)
1223 * new feature: ignoring the return values of destructive functions
1224 like NREVERSE signals a compile-time style-warning.
1225 (thanks to Kevin Reid)
1226 * new feature: the alignment of alien structure fields can be
1227 explicitly specified. (thanks to Cyrus Harmon)
1228 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1229 and dump core on SIGQUIT
1231 ** incompatible change: threads do not inherit values of specials
1232 from their parents (see manual)
1233 ** bug fix: threads stacks belonging to dead threads are freed by the
1234 next exiting thread, no need to gc to collect thread stacks anymore
1235 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1236 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1237 ** bug fix: dynamic variable and thread start related gc lossage
1238 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1240 * fixed some bugs revealed by Paul Dietz' test suite:
1241 ** SUBTYPEP is slightly more accurate on heinously complicated
1242 CONS types where some of the members have uncertain (in the
1243 NIL, NIL sense) type relationships to each other.
1245 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1246 no longer purified unless :PURIFY T is explicitly specified.
1247 ** Non-purified cores are significantly smaller than before
1249 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1250 * new feature: timers based on Zach Beane's excellent timer package
1251 * added support for the following external formats: koi8-u,
1252 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1253 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1254 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1255 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1256 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1257 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1258 * incompatible change: a threaded SBCL will no longer revert to
1259 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1260 * bug fix: interrupts are disabled until startup is complete; no
1261 more sigsegvs when receiving a signal to soon
1262 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1264 * bug fix: add a workaround for the memory randomization features in
1265 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1266 workaround will only be in effect on systems with the proc filesystem
1268 * bug fix: printing objects of type HASH-TABLE signals a
1269 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1271 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1272 have been running for over 50 days. (reported by Gilbert Baumann)
1273 * bug fix: the logic for getting names of functions gets less
1274 confused when confronded with alternate-metaclass
1275 funcallable-instances. (reported by Cyrus Harmon)
1276 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1277 consistent, even on internal alternate-metaclass objects.
1278 * bug fix: flush closure information collected by physical
1279 environment analysis prepass before the main pass. (bug reported
1281 * bug fix: compiler pack phase does not modify a hashtable, which is
1282 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1283 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1284 may be expanded. (reported by Alexander <malishev> on c.l.l)
1285 * bug fix: no more style-warnings when compiling code using LAST
1286 under high-SPEED low-SPACE settings. (reported by David Wallin
1288 * fixed some incompatibilities between SBCL's MOP and the MOP
1290 ** the METAOBJECT class is now implemented;
1291 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1292 STANDARD-OBJECT, as required;
1293 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1294 now compatible (as required by VALIDATE-SUPERCLASS); there
1295 remains a consistency requirement over the presence or absence
1296 of the FUNCTION class in the superclass list of finalized
1297 classes; see the manual for more details;
1298 ** the :AROUND method for COMPUTE-SLOTS on
1299 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1300 requested slot ordering.
1302 ** bug fix: parent thread now can be gc'ed even with a live
1304 ** bug fix: binding a special with PROGV to no value is not
1305 the same as that symbol not having been bound (thanks to
1307 ** bug fix: binding specials is thread safe (thanks to
1309 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1310 and SB-SPROF do not die with 'no handler for signal XX in
1311 interrupt_handle_now(..)' anymore
1312 ** bug fix: WITH-TIMEOUT works with multiple running threads
1313 * fixed some bugs revealed by Paul Dietz' test suite:
1314 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1315 the :method-class keyword argument.
1317 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1318 * new port: the Solaris operating system on x86 processors is now
1319 mostly supported, though some rough edges in the environment
1320 remain. (thanks to Daisuke Homma)
1321 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1322 space for its heap. (thanks to Thiemo Seufer)
1323 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1324 closures. (thanks to Thiemo Seufer)
1325 * minor incompatible change: eof selects abort in the debugger.
1326 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1327 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1329 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1330 deprecated, and will go away in a future revision of SBCL.
1331 * minor incompatible change: GC-ON and GC-OFF are no longer
1332 implemented with a counter, it does not matter how many times gc
1333 is switched on or off
1334 * bug fix: discriminating functions for generic function classes
1335 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1336 make invalid assumptions about method precedence order. (reported
1338 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1339 platforms. (thanks to Thiemo Seufer)
1340 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1341 * bug fix: correct pathname printing with printer escaping is on.
1342 (thanks to Kevin Reid)
1343 * bug fix: complex VOP definitions in "user-space" no longer trigger
1344 package locks. (reported by Zach Beane)
1345 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1346 for funcallable-instance functions, and is no different from
1348 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1349 characters in the current readtable. (reported by Nicholas Neuss)
1350 * bug fix: SUBTYPEP on various CONS types returns more a more
1351 accurate acknowledgment of its certainty.
1352 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1353 lists in some cases. This partially fixes bug 384.
1354 * flush all standard streams before prompting in the REPL and the
1356 * bug fix: signal handling and triggering gc do not conflict
1357 directly anymore, in particular a high frequency sb-sprof does
1358 not prevent gc from running
1359 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1360 approximation for timezone and DST information between the
1361 universal time epoch and the smallest negative 32-bit time_t.
1362 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1363 year 1899 so long as the encoded time is non-negative.
1364 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1365 the correct answer for non-constant rotate amounts when performing
1366 an inline 32-bit rotation.
1368 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1369 there is only one thread in the session
1370 ** bug fix: memory leak for streams created in one thread and
1371 written to in another
1372 ** bug fix: lockup when compiled with gcc4
1373 ** bug fix: race that allows the gc to be triggered when gc is
1375 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1376 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1378 ** bug fix: finished threads are gc'ed properly
1379 * fixed some bugs revealed by Paul Dietz' test suite:
1380 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1381 has arguments with a different set of applicable methods from
1382 the orignal arguments.
1383 ** The type error thrown by MAP now has a correct expected-type
1385 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1386 name a compiled function.
1387 ** Three MISC tests where a large bignum was improperly coerced to
1388 a float that couldn't represent that bignum during type
1389 derivation were fixed.
1390 ** SUBTYPEP can now handle types involving the negation of a
1391 list-form FUNCTION type.
1392 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1393 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1394 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1396 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1397 * New feature: Experimental support for bivalent streams: streams
1398 opened with :element-type :default now allow character and binary
1399 (unsigned-byte 8) I/O
1400 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1401 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1403 * bug fix: no more highly sporadic "couldn't check whether ~S is
1404 readable" when reading a stream and an interrupt hits in the middle
1405 of a select system call
1406 * compiler better recognizes complex arrays (reported by Eduardo
1408 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1409 for variables on SBCL built with linkage-tables. (reported by Luis
1411 * various error reporting improvements.
1412 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1413 (Thanks to Thiemo Seufer)
1414 * bug fix: nested reader invokations work correctly
1415 * bug fix: it is possible to have simultaneous references to foreign
1416 code and foreign data with the same name.
1418 ** added x86-64 support
1419 ** incompatible change: the threading api now works with thread
1420 objects instead of thread ids
1421 ** bug fix: threads are protected from signals and interruption when
1422 starting up or going down
1423 ** bug fix: a race where an exiting thread could lose its stack to gc
1424 ** bug fix: don't halt on infinite error in threads if possible
1425 ** fixed numerous gc deadlocks introduced in the pthread merge
1426 ** bug fix: fixed thread safety issues in read and print
1427 ** bug fix: debugger doesn't hang on session lock if interrupted at
1428 an inappropriate moment
1429 ** bug fix: run-program is now thread safe(r)
1430 ** bug fix: inner with-recursive-lock no longer releases the mutex
1431 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1432 * fixed some bugs revealed by Paul Dietz' test suite:
1433 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1434 EXPECTED-TYPE slots filled.
1435 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1437 ** MAKE-CONDITION accepts classes as type-designators.
1438 ** COMPILE may never return NIL.
1439 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1440 range before calling Unix time functions
1442 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1443 * numerous signal handling fixes to increase stability
1444 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1445 * minor incompatible change: we now correctly canonize default
1446 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1447 as per the MOP, rather than the historical (INITARG INITFUNCTION
1448 INITFORM). (reported by Bruno Haible)
1449 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1450 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1451 for more information.
1452 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1453 pathname is a directory pathname.
1454 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1455 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1457 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1458 :METACLASS argument in addition to a class name. (reported by
1459 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1460 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1461 (reported by Bruno Haible; patch from Gerd Moellmann)
1462 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1464 * bug fix: more cleanups to the floating point exception handling on
1465 x86-64 (thanks to James Knight)
1466 * bug fix: the compiler does not try to constant fold calls of
1467 COERCE and BIT-* functions when they return freshly constructed
1468 objects. (reported by Nikodemus Siivola and Paul Dietz)
1469 * optimization: improved the MIPS versions of generic (in the
1470 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1471 * optimization: direct conversion of (unsigned-byte 32) to floats on
1472 the PowerPC platform.
1473 * optimization: structure instances with raw slots now use less
1474 memory, and probably show better memory locality. (thanks to
1476 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1477 are treated as requests for stack allocation on the x86-64,
1478 Alpha-32, PPC and SPARC platforms.
1479 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1480 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1481 (thanks to Tony Martinez)
1483 ** gcing a dead thread can no longer lead to lockups
1484 ** threads block signals until they are set up properly
1485 ** errno is no longer shared by threads
1486 ** interrupt-thread restores the eflags register on x86
1487 ** fixed some lockups due to gc/thread interaction
1488 * fixed some bugs revealed by Paul Dietz' test suite:
1489 ** invalid dotted lists no longer raise a read error when
1490 *READ-SUPPRESS* is T
1491 ** signal an error if a symbol that names a declaration is used
1492 as the name of a type, or vice versa
1493 ** allow using the (declare (typespec var*)) abbreviation for
1494 (declare (type typespec var*)) with all type specifiers
1495 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1496 ** always return NIL from PPRINT-POP when OBJECT is NIL
1497 ** don't signal errors when pretty-printing malformed LABELS,
1498 FLET or MACROLET forms
1499 ** declarations in a DOLIST body are also in scope for the
1501 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1503 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1506 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1507 * fixed cross-compiler leakages that prevented building a 32-bit
1508 target with a 64-bit host compiler.
1509 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1510 opened with :IF-EXISTS :OVERWRITE.
1511 * fixed bug 281: error for an invalid qualifier in a short-form method
1512 combination method is not signalled until the faulty method is called.
1513 * bug fix: iteration variable type inferrer failed to deal with open
1514 intervals. (reported by Alan Shields)
1515 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1516 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1517 or more runtime options were provided to the sbcl binary.
1518 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1519 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1520 * optimization: there's now a fast-path for fixnum arguments in the
1521 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1522 * optimization: the code generated on x86-64 is more compact thanks
1523 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1524 * bug fix: floating-point exception handling now partly works on
1525 x86-64. (Thanks to James Knight)
1526 * improvement to the MIPS backend from Thiemo Seufer: C-style
1527 64-bit long long arguments and return values to alien functions
1529 * fixed some bugs revealed by Paul Dietz' test suite:
1530 ** the type-error signalled from WARN has a filled-in DATUM slot.
1531 ** the type-error required when a stream is not associated with
1532 a file has the stream as its datum.
1533 ** type-errors on single-floats on x86-64 no longer have
1534 :INVALID-OBJECT as the datum
1535 ** the type-errors signalled for invalid function names now have
1536 a correct expected type
1537 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1538 for structures defined with a :TYPE. Documentation strings for
1539 typed structures are no longer immediately discarded
1540 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1541 broadcast streams and synonym streams. FILE-LENGTH now also works
1542 on broadcast streams.
1544 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1545 * incompatible change: the --noprogrammer option, deprecated since
1546 version 0.7.5, has been removed. Please use the equivalent
1547 --disable-debugger option instead.
1548 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1550 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1551 inoperational for a while now) has been completely removed.
1552 * null lexical environments are now printed as #<NULL-LEXENV>,
1553 significantly reducing the amount of clutter in typical backtraces.
1554 * documentation on weak pointers, finalization, and after GC hooks
1555 has been added to the manual.
1556 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1557 as other specialized array types, is much faster. SUBSEQ and
1558 COPY-SEQ on such arrays have also been sped up.
1559 * optimization: EQL is now more efficient when at least other argument
1560 is known to be of type (OR FIXNUM (NOT NUMBER)).
1561 * fixed bug: the runtime is now less vulnerable to changes in the
1562 size of the SBCL object on OS X, and virtual memory is reserved for
1563 all spaces that need to be at a fixed address.
1564 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1565 * fixed bug: finalizers and after GC hooks that cause consing are now
1567 * fixed bug: compiler error messages and summaries are now printed to
1568 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1569 * fixed inference of the upper bound of an iteration variable.
1570 (reported by Rajat Datta).
1571 * fixed bug 211e: calling local functions with duplicated constant
1572 keyword argument no longer causes a bogus style warning about an
1574 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1575 ftype declaration to be disregarded. (reported by Dave Roberts)
1576 * fixed bug 373: caused by erronous compilation of references to alien
1577 variables in the runtime on ppc/darwin.
1578 * fixed bug 376: CONJUGATE type deriver.
1579 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1580 (reported by Baughn on #lisp)
1581 * a cross-compiler bug on non-x86 platforms has been identified and
1582 fixed. (thanks to Bruno Haible)
1583 * improvements to the MIPS runtime code for increased stability.
1584 (thanks to Thiemo Seufer)
1585 * increased the maximimum compact environment size to allow
1586 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1587 * partial workaround for bug 135: don't name the function we're
1588 calling for hairy cases of EVAL, so as not to accumulate one environment
1589 entry per call to EVAL. (reported by Kevin Reid)
1590 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1591 * optimization: type testing for non-vector arrays should be faster.
1592 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1594 * bug fix: setting 31st element of a bit vector to zero did not work
1596 * bug fix: redefining a class definition which failed due to a
1597 previous accessor / function clash now works (but see BUGS entry
1598 #380 for more problems in this area). (thanks to Zach Beane)
1599 * the long form of DEFINE-METHOD-COMBINATION disables method group
1600 checking when given a single method group with pattern *. (thanks
1602 * on x86 compiler supports stack allocation of results of simple
1603 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1604 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1605 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1606 * fixed some bugs related to Unicode integration:
1607 ** the restarts for recovering from input and output encoding
1608 errors only appear when there is in fact such an error to
1610 * fixed some bugs revealed by Paul Dietz' test suite:
1611 ** MISC.549 and similar: late transformation of unsafe type
1612 assertions into derived types caused unexpected code
1614 ** SCALE-FLOAT type deriver is less wrong.
1615 ** type derivers for EXP, LOG and similar functions compute result
1616 types for complex arguments better.
1617 ** (MISC.563) CONJUGATE type deriver works for very restricted
1619 ** out-of-line type testers for character strings are available.
1620 ** EQUAL compiler transform understands specialness of objects
1622 ** accessing double-floats stored on the stack now works on x86-64.
1623 ** debugger internals could sometimes create invalid lispobjs,
1624 resulting in GC crashes.
1625 ** MISC.548: type check weakening can convert required type into
1627 ** initialization forms for bindings are not in scope of free special
1630 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1631 * incompatible change: thread support for non-NPTL systems has
1632 been removed - locking is buggy and unreliable. A threaded
1633 SBCL build will now warn at startup and refuse to create
1634 new threads, unless futex support is detected in the kernel
1635 * incompatible change: the top level REPL now has only an ABORT
1636 restart associated with it, not TOPLEVEL and ABORT as it used to.
1637 TOP and TOPLEVEL are now available as debugger commands for
1638 returning to the top level.
1639 * incompatible change: forms evaluated in the REPL now use the
1640 global optimization policy.
1641 * incompatible change: user- and system-initialization files are
1642 no longer processed with LOAD, but by READ and EVAL; hence the
1643 global optimization policy, startup package, readtable, etc,
1645 * The .fasl file format number has been incremented because of
1646 various incompatible changes.
1647 * internal entry point details and argument counts no longer appear
1648 in backtraces unless explicitly requested by setting
1649 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1650 * built-in and standard functions no longer have names like "top
1651 level local call to FOO".
1652 * fixed bug 32: functions defined in non-null lexical environments
1653 now have more legible printed representation
1654 * fixed bug 33: functions defined in non-null lexical environemnts
1655 are now more amenable to inspection by INSPECT.
1656 * workaround for bug 354: XEPs no longer appear in backtraces unless
1657 explicitly requested.
1658 * fixed bug: receiving the signal which results from attempting to
1659 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1660 SIGBUS on other *BSDs) on architectures where the C stack is also
1661 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1662 to debugger confusion or wild instability. (reported by Cheuksan
1664 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1665 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1666 (reported by Lutz Euler)
1667 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1668 compiled code. (reported by Paul Dietz)
1669 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1670 specialized parameter to a method to something that is not TYPEP
1671 the specializer is now possible.
1672 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1673 face of package deletion.
1674 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1675 pathname. (reported by David Tolpin on c.l.l)
1676 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1677 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1678 STRING-STREAM and FILE-STREAM, respectively.
1679 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1680 than previously; in addition, its test suite is now run on build.
1681 (thanks to Luke Gorrie)
1682 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1683 * optimization: added a immediate representation for single-floats
1685 * fixed some bugs revealed by Paul Dietz' test suite:
1686 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1687 ** attempting to create a package with a colliding nickname causes
1688 correctable errors to be signalled.
1689 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1690 ** several x86-64 backend bugs related to sign-extension of immediate
1693 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1694 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1695 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1697 * fixed loading of multiply forward-referenced layouts.
1698 (thanks to Cheuksan Wang)
1699 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1700 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1701 (see COMPILE-FILE documentation for details of :PRINT option)
1702 * fixed bugs 19 and 317: fixed-format floating point printing is
1703 more accurate. This also fixes a bug reported by Adam Warner
1704 related to the ~@F format directive.
1705 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1707 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1708 dispatch table if the from-char is a dispatch macro character.
1709 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1710 by the wordsize no longer ignore the last word. (reported by Lutz
1712 * fixed bug in type checking of dynamic-extent variables. (reported
1714 * optimization: sequence traversal functions use their freedom to
1715 coerce function designators to functions.
1716 * optimization: code with many calls to CLOS methods specialized on
1717 CLOS classes has had redundant type checks removed.
1718 * optimization: added declarations to speed up operations that access
1719 the internal character database (for example STRING-UPCASE)
1720 * optimization: comparison operations between floats and sufficiently small
1721 fixnums no longer create extra rationals
1722 * fixed some bugs related to Unicode integration:
1723 ** portions of multibyte characters at the end of buffers for
1724 character-based file input are correctly transferred to the
1725 start of the buffer at the next read.
1726 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1727 passing it through to OPEN.
1728 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1729 argument given to internal calls to OPEN.
1730 ** the built-in comment readers (introduced by character sequences
1731 ";" and "#|") are more forgiving to encoding errors; they will
1732 STYLE-WARN and then attempt to resync the stream at a character
1733 boundary. (thanks to Teemu Kalvas)
1734 * fixed some bugs in the x86-64 port:
1735 ** Negative short int return values from c-calls are sign-extended
1737 ** The stack is aligned to 16-bytes for c-calls, as required by
1738 the ABI. (thanks to Cheuksan Wang)
1739 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1740 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1741 for enums. (thanks to Vincent Arkesteijn)
1742 ** Multiple small optimizations and bugfixes for floating point
1744 * fixed some bugs revealed by Paul Dietz' test suite:
1745 ** Space, Tab, Linefeed, Return and Page have the invalid
1746 secondary constituent character trait.
1747 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1749 ** WITH-INPUT-FROM-STRING should only update the index place on
1751 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1752 works more reliably.
1753 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1754 with one-dimensional arrays with fill pointers.
1755 ** TYPE-OF failed on a complex with an integer realpart and a
1757 ** compiler failure during type inference for the code of form
1758 (IF (EQL X (THE ...)) ...) (MISC.535).
1760 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1761 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1762 platform. The port supports 61-bit fixnums, large memory spaces
1763 and reloading shared object files.
1764 * enhancement: saving cores with foreign code loaded is now
1765 supported on ppc/Darwin in addition to the previously supported
1767 * enhancement: the statistical profiler now walks deeper into the
1768 call stack for more meaningful call-graphs and accrued time
1769 reports (x86/x86-64 only). It also now reports time spent in
1771 * enhancement: it is now possible to trace most individual methods
1772 of a generic function in addition to tracing the generic function
1774 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1776 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1777 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1778 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1779 produces an error. (thanks to Vincent Arkesteijn)
1780 * bug fix: NAMESTRING on pathnames with :WILD components in their
1781 directories works correctly. (thanks to Artem V. Andreev)
1782 * fixed bug 125: compiler preserves identity of closures. (reported
1784 * bug fixed: functions with &REST arguments sometimes failed with
1785 "Undefined function" when compiled with (DEBUG 3). (reported by
1787 * bug fix: overflow during compiling of setting element of a bit
1788 vector with constant index and value. (reported by Timmy Douglas)
1789 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1790 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1791 * on x86 compiler partially supports stack allocation of dynamic-extent
1793 * GO and RETURN-FROM do not check the extent of their exit points
1794 when compiled with SAFETY 0.
1795 * fixed some bugs related to Unicode integration:
1796 ** encoding and decoding errors are now much more robustly
1797 handled; it should now be possible to recover even from invalid
1798 input or output to the terminal. (thanks to Teemu Kalvas)
1799 ** provided a first cut at implementing STRING-TO-OCTETS and
1800 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1801 ** altered the SB-MD5 contributed module slightly, changing the
1802 interface just enough for it to be supportable for builds where
1803 lisp characters are not eight bits.
1804 * fixed some bugs revealed by Paul Dietz' test suite:
1805 ** the FORMATTER-generated functions for ~V[ conditionals require
1806 the correct number of arguments.
1807 ** READ-FROM-STRING returns the mandated second value when applied
1808 to displaced strings.
1809 ** the #\Rubout and #\Backspace characters are treated as invalid
1810 constituent characters by the tokenizer.
1812 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1813 * new feature: reloading changed shared object files with
1814 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1815 * new feature: references to foreign variables and functions
1816 can now be compiled and loaded before the corresponding shared
1817 object file is loaded, as long as the foreign definitions are
1818 available at runtime.
1819 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1820 supported. (thanks to Dan Debertin)
1821 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1822 just keywords. (thanks to Vincent Arkesteijn)
1823 * fixed bug #331: structure-class instances corresponding to
1824 DEFSTRUCT forms are now created eagerly.
1825 * fixed bug #345: backtraces from calls to undefined functions work
1826 on x86 as well. Related bug #61 is now also partially fixed on x86
1827 (backtraces from throws to unknown catch tags.)
1828 * bug fix: lambda-list parsing is now stricter vrt. order and number
1829 of lambda-list keywords.
1830 * bug fix: as specified by AMOP, an error is signalled if a
1831 class-option appears multiple times in a DEFCLASS form. (reported
1833 * bug fix: RANDOM can be compiled when the compiler derives the type
1834 of its numeric argument as a disjoint set of small integers.
1835 (reported by Paul Dietz)
1836 * bug fix: starting a core saved with shared objects loaded when
1837 those objects are not available no longer causes threaded SBCL to
1838 hang. (reported by Sean Ross)
1839 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1841 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1842 argument, which must always be NIL. (reported by Kalle Niemitalo)
1843 * bug fix: printing 1.0d+23 no longer results in an error.
1844 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1845 * bug fix: structure slot setters preserve evaluation order. (thanks
1847 * fixed some bugs related to Unicode integration:
1848 ** RUN-PROGRAM can allow its child to take input from a Lisp
1849 stream. (reported by Stefan Scholl)
1850 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1851 locales. (reported by Ken Causey)
1852 * fixed some bugs revealed by Paul Dietz' test suite:
1853 ** INCF, DECF and REMF evaluate their place form as specified in
1855 ** FORMATTER expands ~{ ~} iteration directives with V or #
1856 parameters correctly.
1857 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1858 consequent uses no arguments correctly.
1859 ** the system has a partial understanding of the (COMPLEX RATIO)
1861 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1864 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1865 * new feature: a build-time option (controlled by the :SB-UNICODE
1866 keyword feature, enabled by default) for building the system with
1867 support for the entire 21-bit character space defined by the
1869 * new feature: the system now has rudimentary external-format
1870 support; the primary user-visible change at this time is that
1871 characters with the high bit set (such as lower-case-e-acute) will
1872 print correctly to a terminal in a UTF-8 environment.
1873 * minor incompatible change: BASE-CHAR no longer names a class;
1874 however, CHARACTER continues to do so, as required by ANSI.
1875 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1876 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1877 has been moved to the SB-EXT package (temporarily re-exported from
1879 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1880 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1882 * fixed bug #351: better error-handlind and reporting for malformed
1884 * fixed bug #350: bignum-printing is now more memory-efficient,
1885 allowing printing of very large bignums, eg. (expt 2 10000000).
1886 (reported by Bruno Haible)
1887 * fixed bug #302: better primitive-type selection for intersection
1889 * fixed bug #308: non-graphic characters now all have names, as
1890 required. (reported by Bruno Haible)
1891 * bug fix: redefining a class with different superclasses now correctly
1892 removes it from the direct-subclasses of its previous superclasses.
1893 (reported by David Morse)
1894 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1895 new value now works. (reported by Bruno Haible)
1896 * bug fix: correct canonicalization of multiple non-standard slot
1897 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1898 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1899 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1900 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1901 now exists, an signals an error.
1902 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1903 during macroexpansion for non-iteration variables as well. (reported
1904 by Bruno Haible for CMUCL)
1905 * bug fix: Cyclic structures and unprintable objects in compiler
1906 messages no longer cause errors. (reported by Bruno Haible)
1907 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1908 and READ-FROM-STRING all now return a primary value of NIL if
1909 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1910 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1911 (reported by Bruno Haible for CMUCL)
1912 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1913 do not designate a condition. (reported by Bruno Haible for
1915 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1916 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1917 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1918 permitting resolution in favour of any of the conflicting symbols.
1919 (reported by Bruno Haible for CMUCL)
1920 * bug fix: EQUAL compiler optimizations is less aggressive on
1921 strings which can potentially compare true despite having distinct
1922 specialized array element types.
1923 * bug fix: unit enumerations can be defined without dividing by
1924 zero. (thanks to Vincent Arkesteijn)
1925 * FORMAT compile-time argument count checking has been enhanced.
1926 (report from Bruno Haible for CMUCL)
1927 * a partial workaround for the bug 262: the compiler does not try to
1928 inline-expand a local function doing RETURN-FROM from a deleted
1929 BLOCK. (thanks to Peter Denno for the bug report and to David
1930 Wragg for the simple test case)
1931 * fixed some bugs revealed by Paul Dietz' test suite:
1932 ** RENAME-PACKAGE allows all package designators as new package
1934 ** constraint propagation and lambda variable substitution are
1935 more cautious in dealing with partially deleted code.
1936 ** compiler sometimes lost reoptimization passes.
1937 ** CERROR, given a condition as condition designator, passes its
1938 remaining arguments to the continue format control without
1940 ** Case-altering FORMAT directives work correctly on non-ASCII
1942 ** The REAL type specifier handles bounds outside the floating
1943 point ranges without signalling FLOATING-POINT-OVERFLOW.
1944 ** Functions with IR1-transformations can create intercomponent
1945 references to global functions.
1946 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1948 ** FORMAT ~R treats a nil value for its first parameter correctly.
1950 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1951 * enhancement: saving cores with foreign code loaded is now
1952 supported on x86/NetBSD and sparc/Linux in addition to the previously
1953 supported platforms.
1954 * bug fix: on some platforms repeated installations caused multiple
1955 copies of HTML documentation to be installed -- should not happen
1956 any more. (reported by Stefan Scholl)
1957 * bug fix: parsing self-recursive alien record types multiple times
1958 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1959 original patch by Helmut Eller for CMUCL)
1960 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1961 (thanks to Richard Kreuter)
1962 * bug fix: defining classes whose accessors are methods on existing
1963 generic functions in other (locked) packages no longer signals
1964 bogus package lock violations. (reported by François-René Rideau)
1965 * bug fix: special variables as DEFMETHOD parameters no longer have
1966 associated bogus type declarations. (reported by David Wragg and
1968 * bug fix: read-write consistency on streams of element-type
1969 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1970 * bug fix: redefiniton of the only method of a generic function with
1971 no DEFGENERIC no longer emits a full WARNING. In addition,
1972 redefinition of generic functions with no DEFGENERIC to an
1973 incompatible lambda list now signals an error. (thanks to Zach
1975 * bug fix: DEFGENERIC now works even when there's a function of the
1976 same name in an enclosing lexical environment. (thanks to Zach
1978 * fixed compiler failure, caused by instrumenting code during
1979 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1980 * optimization: added loop analysis and improved register allocation
1981 * fixed some bugs revealed by Paul Dietz' test suite:
1982 ** POSITION on displaced vectors with non-zero displacement
1983 returns the right answer.
1984 ** (SIMPLE-STRING) is a valid type specifier for sequence
1986 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1988 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1989 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1991 ** The justification version of the FORMAT ~< directive treats
1992 non-zero minpad parameter correctly.
1994 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1995 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1996 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1997 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1998 the supported interface.
1999 * new feature: Single-stepping of code compiled with DEBUG 2 or
2000 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2001 * new feature: saving cores with foreign code loaded is now
2002 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2003 Timothy Moore's work for CMUCL)
2004 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2005 parameters to * instead of NIL if no initform is supplied.
2006 (reported by Johan Bockgaard)
2007 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2008 correctly when FUNCALL appears as the car of the form. Note:
2009 despite this FUNCALL forms are not currently subject to
2010 compiler-macro expansion. (port of Raymond Toy's fix for the
2011 same from CMUCL, reported by Johan Bockgaard)
2012 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2013 (thanks for Teemu Kalvas)
2014 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2015 case instead of terminating the iteration (thanks for Julian
2016 Squires, Sean Champ and Raymond Toy)
2017 * bug fix: incorrect expansion of defgeneric that caused a style
2018 warning. (thanks for Zach Beane)
2019 * bug fix: slot accessor effective method computation works properly
2020 for classes with multiple non-standard applicable methods on
2021 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2022 * on x86 compiler supports stack allocation of results of LIST and
2023 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2024 CMUCL implementation by Gerd Moellmann)
2025 * fixed some bugs revealed by Paul Dietz' test suite:
2026 ** FORMAT strings with both the ~<~:;~> form of the justification
2027 directive and pretty-printing directives cause an error.
2029 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2030 * incompatible change: the internal functions
2031 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2032 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2033 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2034 instead of the old functions.
2035 * new feature: on platforms where "dladdr" is available foreign
2036 function names now appear in backtraces. (based on Helmut Eller's
2038 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2039 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2041 * documentation: ASDF manual is now distributed with SBCL.
2042 * bug fix: SBCL can now load its contributed modules with REQUIRE
2043 even if the system-provided entries have been removed from
2044 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2046 * bug fix: backtraces involving undefined functions or assembly
2047 routines are more informative. (thanks to Brian Downing)
2048 * bug fix: mutually referent alien structures now work correctly.
2049 (reported by Rick Taube)
2050 * bug fix: structures defined by WITH-ALIEN can be referred to
2051 within other definitions in the same WITH-ALIEN.
2052 * bug fix: division operators (MOD, TRUNCATE and the like) with
2053 constant zero divisors and integer dividends no longer generate
2055 * bug fix: provide default methods for INPUT-STREAM-P and
2056 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2057 * bug fix: improve the detection and resolution of MOP metacycles
2058 (where effective method computation depends on the generic function
2059 having its effective method computed). (reported by Bruno Haible)
2060 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2061 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2062 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2063 * bug fix: pathnames with relative directory components can be
2064 represented relative to default pathnames.
2065 * optimization: in taking the GCD of bignums, reduce the two bignums
2066 to approximately the same size (using Euclid's algorithm) before
2067 applying the more sophisticated binary GCD. (thanks to Juho
2069 * optimization: COUNT on bitvectors now operates word-at-a-time.
2070 * optimization: ASH with a positive, but not necessarily constant,
2071 (leftwards) shift, when in a modular context, is compiled to a
2073 * fixed some bugs revealed by Paul Dietz' test suite:
2074 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2075 if the corresponding argument is NIL.
2076 ** FORMAT directives accepting a minpad parameter treat negative
2078 ** literal commas as character parameters to FORMAT directives are
2080 ** literal spaces directly after ~<Newline> directives within a
2081 format-logical-block (~:< ~@:>) do not induce :FILL-style
2082 conditional newlines.
2083 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2084 as specified: it no longer includes conditional newlines).
2085 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2087 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2088 :PER-LINE-PREFIX argument does not evaluate to a string.
2090 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2091 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2092 the manual for details; for now, package locks can be disabled by
2093 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2094 if no major problems are found then it is likely that they will be
2095 compiled in unconditionally.
2096 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2097 unsupported operators on all platforms. To load a shared library
2098 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2099 object file, link it into a shared library outside of SBCL and
2100 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2102 * bug workaround: ROOM T and the GC were not working together
2103 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2104 reporting facility aren't preserved. That reporting has been
2105 disabled, so now until and unless someone figures out how to make it
2106 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2107 * minor incompatible change: as threatened around sbcl-0.8.0, the
2108 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2109 The symbols are also exported from SB-PCL for backwards
2110 compatibility, but more so than before SB-PCL should be treated as
2111 an implementation-internal package.
2112 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2114 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2115 It is known as of this release not to work on the Alpha, however.
2116 * fixed bug #167: errors signalled due to illegal syntax in method
2117 bodies are now more legible.
2118 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2119 designators and can hence be used with TYPEP.
2120 * fixed bug #333: CHECK-TYPE now ensures that the type error
2121 signalled, if any, has the right object to be accessed by
2122 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2123 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2124 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2126 * fixed bug #334: programmatic addition of slots using specialized
2127 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2128 and :ALLOCATION :CLASS slots.
2129 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2130 float exponents. (rereported by Peter Seibel)
2131 * fixed a bug: #\Space (and other whitespace characters) are no
2132 longer considered to be macro characters in standard syntax by
2133 GET-MACRO-CHARACTER.
2134 * fixed bug: initialization of condition class metaobjects no longer
2135 causes an instance of the condition to be created. (reported by
2137 * fixed bug: it is now possible to have slots such that
2138 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2139 description is neither :INSTANCE nor :CLASS.
2140 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2141 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2142 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2143 non-standard SLOT-DEFINITION-ALLOCATION values.
2144 * fixed bugs: various race conditions handling exiting threads.
2145 CL-PPCRE's thread torture test now passes.
2146 * fixed arguably-a-bug: GC time proportional to number of threads in
2147 system even when most of them are idle
2148 * optimization: improved performance of BIT and SBIT on bit-vectors.
2149 * .fasl file incompatibility: The fasl file version number has
2150 been incremented because of changes associated with package locks.
2152 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2153 * minor incompatible change: the system no longer provides
2154 optimization hints (conditions which are TYPEP
2155 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2156 mode; these hints are still emitted when higher SPEED optimization
2158 * new contrib module: a sampling profiler (profiling by statistical
2159 sampling, rather than by instrumenting functions) is available as
2160 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2161 * the behaviour of the standard function ED is now customizeable by
2162 third parties through a hook variable: see ED's documentation
2163 string for information on the protocol.
2164 * the compiler no longer emits efficiency notes for (FUNCALL X)
2165 when the type of X is uncertain under default optimization
2167 * fixed bug 276: mutating a binding of a specialized parameter to a
2168 method to something that is not TYPEP the specializer is now
2170 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2171 exhibit the required behaviour on the x86 platform. (thanks to
2172 Peter van Eynde, Eric Marsden and Bruno Haible)
2173 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2174 even for difficult arguments. (reported by Peter Graves)
2175 * fixed bug 141a: the backquote printer now descends quoted
2177 * fixed another bug in backquote printing: no more destructive
2178 modification of the form's list structure. (reported by Brian
2180 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2181 move between its address being taken and the call to
2182 interrupt_thread, fixing a crashing race condition.
2183 * the SB-POSIX contrib implementation has been adjusted so that it
2184 no longer exhibits ridiculously poor performance when constructing
2185 instances corresponding to C structs.
2187 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2188 * minor incompatible change: the sb-grovel contrib now treats C
2189 structures as alien (in the sense of SB-ALIEN) objects rather than
2190 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2191 has implications for memory management of client code
2192 (sb-grovel-returned objects must now be manually managed) and for
2193 type safety (alien objects now have full types).
2194 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2195 used to control emission of compiler diagnostics, rather than the
2196 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2197 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2198 quality should be considered deprecated.
2199 * install.sh now installs the user manual as well
2200 * (not quite a new documentable feature, but worth considering in
2201 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2202 beginnings of a semantically meaningful condition hierarchy is
2203 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2205 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2206 designator as the defaults argument.
2207 * fixed bug: Displaced arrays whose displaced-to array has become
2208 too small now cause ARRAY-DIMENSION to signal an error, providing
2209 for safer bounds-checking. (reported by Bruno Haible)
2210 * fixed bug: DEFCLASS slot definitions with identical :READER and
2211 :WRITER names now signal a reasonable error. (reported by Thomas
2213 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2215 * fixed bug: Invalid slot specification errors now print correctly.
2216 (thanks to Zach Beane)
2217 * fixed bug 320: Shared to local slot value transfers in class
2218 redefinitions now happen corrently from superclasses as
2219 well. (reported by Bruno Haible)
2220 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2222 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2223 improper lists correctly. (reported by Bruno Haible)
2224 * fixed bug 313: source-transform for <fun-name> was erroneously
2225 applied to a call of a value of a variable with name <fun-name>.
2226 (reported by Antonio Menezes Leitao)
2227 * fixed bug 307: The obsolete instance protocol ensures that
2228 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2229 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2230 remove dead unknown-values globs from the middle of the stack.
2231 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2232 easier to resume long computations after using *BREAK-ON-SIGNALS*
2233 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2234 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2235 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2236 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2237 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2238 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2240 * fixed a (fairly theoretical) bug in string printing: if
2241 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2242 does not have array-element-type equal to the most general string
2244 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2245 function name. (thanks to Nikodemus Siivola)
2246 * fixed bug: compile-time format string checker failed on
2247 non-closed ~{. (reported by Thomas F Burdick)
2248 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2249 don't warn on legal loop code involving more than one aggregate
2250 boolean. (thanks to Nikodemus Siivola)
2251 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2252 in keyword position to MAKE-PACKAGE was not being reported as
2254 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2255 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2257 * fixed bugs 280 and 312: the checking for multiple definitions in a
2258 file is less likely to become confused by uses of inline
2260 * fixed bug: the #S reader macro performs the keyword coercion
2261 specified for slot names. (reported by Kalle Niemitalo)
2262 * fixed bug: lambda lists may contain symbols whose names start with
2263 & but are not lambda-list-keywords; their occurrence triggers a
2265 * fixed bug 321: define-method-combination argument lambda lists do
2266 not cause invalid code to be generated when &optional or &aux
2267 variables have default values. (reported by Bruno Haible)
2268 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2269 have CLOS classes; this appears also to have fixed reported
2270 instability in stack exhaustion detection.
2271 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2272 MUFFLE-WARNING when no associated restart is present is now
2274 * optimization: rearranged the expansion of various defining macros
2275 so that each expands into only one top-level form in a
2276 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2278 * optimization: used a previously little-used slot in symbols to
2279 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2281 * fixed some bugs revealed by Paul Dietz' test suite:
2282 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2284 ** Fixed an optimization bug related to inheritance of initforms
2285 from local to shared slots.
2286 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2287 ** CAST optimizer forgot to flush argument derived type.
2288 ** print/read consistency on floats is now orders of magnitude
2289 more likely. (thanks also to Bruno Haible for a similar report
2291 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2292 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2293 for objects of type REAL. Make it so.
2294 ** SXHASH is specified (infelicitously) to respect similarity,
2295 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2296 it so. (thanks to Markus Ziegler)
2297 ** on the Alpha, the compiler succeeds in compiling functions
2298 returning a known number of arguments greater than 63.
2299 ** fixed handling of invalid NIL arguments in keyword position
2301 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2303 ** provided a readably-printable representation for RANDOM-STATE
2305 ** ensured that pathnames, where they have a namestring, always
2306 print using #P"..." syntax.
2308 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2309 * Support for the forthcoming 2.0 version of the NetBSD kernel
2310 running on x86 hardware has been added. (thanks to Perry
2311 E. Metzger most immediately, and others for their past work)
2312 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2313 current release version; SBCL's OpenBSD support had been broken
2314 since about the time of OpenBSD's switch to ELF binary format.)
2315 * [placeholder for DX summary]
2316 ** user code with &REST lists declared dynamic-extent, under high
2317 speed or space and low safety and debug optimization policy.
2318 * The manual has been converted to Texinfo format and the debugger
2319 chapter from the cmucl manual has been added.
2320 * A facility has been added to extract documentation strings from
2321 sbcl and store them as Texinfo-formatted snippets for inclusion in
2322 the manual (via Texinfo's @include directive)
2323 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2324 the test case to Dave Roberts)
2325 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2326 pointers. (reported by Sean Ross)
2327 * bug fix: PROFILE output is printed nicely even for large numerical
2328 values. (thanks to Zach Beane)
2329 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2330 greater than 32 handle EOF correctly.
2331 * bug fix: on X86 an immediate argument of the IMUL instruction is
2332 correctly printed in disassembly. (thanks to Lutz Euler)
2333 * bug fix: class slots in redefined classes preserve their old
2334 values. (thanks to Bruno Haible and Nikodemus Siivola)
2335 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2336 succeeds. (reported by Marco Baringer)
2337 * fixed some bugs revealed by Paul Dietz' test suite:
2338 ** READ-SEQUENCE now works on ECHO-STREAMs.
2339 ** RATIONALIZE works more according to its specification. (thanks
2342 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2343 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2344 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2345 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2346 should matter to you only if you rebind the printer control
2347 variables and then find you want different bindings in the
2348 debugger than in the ordinary execution of your program.)
2349 * The runtime build system has been tweaked to support building
2350 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2351 assembler and linker. (thanks to Nikodemus Siivola)
2352 * Unbound, undefined, undeclared variables now trigger full
2353 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2354 more often programmer error than deliberate exploitation of undefined
2356 * optimization: the hash algorithm for strings has changed to one
2357 that is less vulnerable to spurious collisions. (thanks to Juho
2359 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2360 needless bounds checking. (thanks to Juho Snellman)
2361 * optimization: implemented multiplication as a modular
2362 (UNSIGNED-BYTE 32) operation on the PPC backend.
2363 * fixed some bugs revealed by Paul Dietz' test suite:
2364 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2366 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2367 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2368 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2370 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2372 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2374 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2375 their output stream on EOF from read.
2376 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2377 have been read to end-of-file.
2378 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2380 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2381 description of determination of which consecutive characters
2383 ** Printing the "Space" character with escaping on now yields "#\\ ",
2384 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2385 ** Reading floating-point numbers with *READ-BASE* set to a number
2386 less than 10 works correctly.
2387 ** Reading floating-point numbers with *READ-BASE* set to a number
2388 more than 10 works correctly.
2389 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2390 the readtable currently in effect.
2392 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2393 * minor incompatible change: parsing of namestrings on a physical
2394 (Unix) host has changed; numbers after the final #\. in a
2395 namestring are no longer interpreted as a version field. This is
2396 intented to be largely invisible to the user, except that the
2397 meaning of the namestring "*.*.*" has changed: it now refers to a
2398 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2399 should usually be replaced by
2400 (make-pathname :name :wild :type :wild :version :wild)
2401 with the added benefit that this is more likely to be portable.
2402 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2403 option to OPEN now signals an error if the file being opened
2404 exists; this may have an impact on existing code.
2405 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2406 signals received, on the PowerPC platforms (both Linux and
2407 Darwin). (thanks to Pierre Mai for pointing out the location of
2409 * several fixes on OS X: The system now builds and runs cleanly on
2410 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2411 (thanks to Brian Mastenbrook)
2412 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2413 second-resolution: integer multiples of 1/3600 between -24 and 24.
2414 (thanks to Vincent Arkesteijn)
2415 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2416 type when called with 1 argument; PEEK-CHAR checked type of
2417 PEEK-TYPE only after having read first character from a
2418 stream. (reported by Peter Graves)
2419 * bug fix: the garbage collector now has much better locality
2420 behaviour, and in particular no longer treats all memory as being
2421 exhausted when many small objects point to each other in a deeply
2423 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2424 recognized as being TYPEP their class.
2425 * bug fix: the PUSHNEW documentation string has been corrected.
2426 (thanks to Vincent Arkesteijn)
2427 * bug fix: defaulting of the value for the last of an atypically
2428 large number of multiple values being bound was not being
2429 performed correctly on the Alpha or PPC platforms
2430 * optimization: implemented multiplication as a modular
2431 (UNSIGNED-BYTE 32) operation on the x86 backend.
2432 * optimization: SEARCH on simple-base-strings can now be open-coded.
2433 (see also contrib/compiler-extras.lisp for inspiration for
2434 teaching the compiler about the Boyer-Moore algorithm).
2435 * value, returned by MAX (and MIN) called with several EQUALP, but
2436 not EQL, arguments now does not depend on compiler settings.
2437 * fixed some bugs revealed by Paul Dietz' test suite:
2438 ** in stack analysis liveness information is propagated from
2439 non-local entry points.
2440 ** pathwise CAST removing failed when the CAST node did not start
2442 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2443 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2445 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2446 host is already defined.
2447 ** RENAME-FILE works on streams instead of signalling an internal
2449 ** PEEK-CHAR uses the current readtable when determining whether
2450 or not a character is whitespace.
2451 ** MERGE-PATHNAMES handles the case when the pathname does not
2452 specify a name while the default-pathname specifies a version.
2453 ** Pathnames now stand a better chance of respecting print/read
2455 ** Attempting to use standardized file system operators with a
2456 pathname with invalid :DIRECTORY components signals a
2458 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2460 ** DIRECTORY on logical pathnames is more correct.
2461 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2462 signal a TYPE-ERROR if their argument is not a stream.
2463 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2464 designator argument does not designate a stream.
2465 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2466 examining the synonym.
2467 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2469 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2470 element-type larger than ([UN]SIGNED-BYTE 32).
2472 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2473 * When built with the :SB-FUTEX feature, threaded builds now take
2474 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2475 for faster/more reliable mutex and condition variable support.
2476 * Incompatible change (but one you probably shouldn't have been using
2477 anyway): the interface and code for arbitrating between multiple
2478 threads in the same user session has been redesigned.
2479 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2480 error when called without an explicit environment argument.
2481 (thanks to Nikodemus Siivola)
2482 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2483 confused about their position. (thanks to Adam Warner and Gerd
2485 * bug fix: Pretty printing backquoted forms with unquotations in the
2486 argument list position of various code constructs such as LAMBDA
2487 now works correctly. (reported by Paul Dietz)
2488 * bug fix: Pretty printing unquotations no longer loses all
2489 stream position information.
2490 * optimization: performance of string output streams is now less
2491 poor for multiple small sequence writes.
2492 * optimization: performance of CSUBTYPEP in the presence of complex
2493 expressions involving CONS and NOT many times has been improved.
2494 (reported by Paul Dietz)
2495 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2497 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2498 streams for the REPL, for the debugger, and so on. (thanks to
2500 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2501 * fixed some bugs revealed by Paul Dietz' test suite:
2502 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2503 is ignored if it is too small, rather than propagating through
2504 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2505 ** extremely complex negations of CONS types were not being
2506 sufficiently canonicalized, leading to inconsistencies in
2508 ** VALUES tranformer lost derived type.
2510 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2511 * fixed a bootstrapping bug: the build process no longer assumes
2512 that the various BOOLE-related constants have the same value in
2513 host and target lisps. (noted by Paul Dietz' test suite on an
2514 SBCL binary built from CLISP)
2515 * The system can now be dynamically linked on the MIPS platform,
2516 which enables dynamic loading of foreign code from Lisp. (thanks
2517 to Ralf Baechle for discussions on the MIPS ABI)
2518 * The system now records debugging information for its own source
2519 files in a filesystem-position-independent manner, relative to
2520 the "SYS" logical host.
2521 * fixed a compiler bug: MV-LET convertion did not check references
2522 to the "max args" entry point. (reported by Brian Downing)
2523 * tweaked disassembly notes to be less confident about proclaiming
2524 some instruction as an LRA. (thanks to Brian Downing)
2525 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2526 now each have their own history, command character, and other
2527 characteristics. (thanks to David Lichteblau)
2528 * fixed some bugs revealed by Paul Dietz' test suite:
2529 ** compiler failure in compiling LOGAND expressions including a
2531 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2532 shift greater than 32.
2533 ** FLUSH-DEST did not mark blocks for type check regeneration.
2534 ** HANDLER-CASE failed to accept declarations in handler clauses
2535 in some circumstances.
2537 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2538 * New code in contrib/sb-introspect (still probably not entirely
2539 stable yet) provides some support for smart Lisp development
2540 environments like SLIME.
2541 * The conditions signalled for errors occurring when loading .fasl
2542 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2543 in a way which should help ASDF recover gracefully.
2544 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2545 stuff has been cleaned up. If you code contrib/ stuff, this might
2546 affect you, and you can look at contrib/README, contrib/STANDARDS,
2547 and/or the 0.8.4.27 diff to check.
2548 * In full calls the compiler now does not generate checks for declared
2549 argument types for all arguments.
2550 * various threading fixes
2551 ** and some experimental patches which didn't make it into
2552 the main tree for this release, but which are shipped in
2553 contrib/experimental-thread.patch as a possible fix for some
2554 failures (deadlock, spinning...) in GC-intensive multithreaded
2556 * fixed PPC build problem (source code incompatibility of different
2557 library versions): added offsetof() hackery which attempts to divine
2558 where glibc maintainers put uc_mcontext today
2559 * fixed bug 282: compiler does not trust type assertions while passing
2560 arguments to a full call.
2561 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2562 &OPTIONAL VALUES type parameter.
2563 * fix bug 214: algorithm for noting rejected templates is now more
2564 similar to that of template seletion. (also reported by rydis on
2566 * fixed bug 141b: printing backquoted information readably and prettily
2567 inserts a space where necessary.
2568 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2569 (CHARACTER 10) are now reported as errors, rather than propagated
2570 as unknown types. (reported by piso on #lisp)
2571 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2572 with values NIL and :ERROR. (thanks to Milan Zamazal)
2573 * fixed bug 191c: CLOS now does proper keyword argument checking as
2574 described in CLHS 7.6.5 and 7.6.5.1.
2575 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2576 longer raise an error; further, using a list as a for-as-arithmetic
2577 counter now raises a meaningful error.
2578 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2579 now understood by sequence creation functions such as MAKE-SEQUENCE
2581 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2582 a STRING-INPUT-STREAM.
2583 * compiler enhancement: SIGNUM is now better able to derive the type
2585 * type declarations inside WITH-SLOTS are checked. (reported by
2587 * fixed some bugs revealed by Paul Dietz' test suite:
2588 ** incorrect optimization of TRUNCATE for a positive first
2589 argument and negative second.
2590 ** compiler failure in let-convertion during flushing dead code.
2591 ** compiler failure while deriving type of TRUNCATE on an
2592 interval, containing 0.
2593 ** ASH of a negative bignum by a negative bignum count now returns
2595 ** intersection of CONS types now canonicalizes properly, fixing
2596 inconsistencies in SUBTYPEP.
2598 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2599 * incompatible change: The --disable-debugger command line
2600 option now clobbers the debugger at a more fundamental
2601 level, by redefining #'INVOKE-DEBUGGER instead of by
2602 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2603 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2604 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2605 enter the debugger REPL and then suffer recursive errors
2606 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2607 while under the new system, BREAK in --disable-debugger mode
2608 terminates the system just as an unhandled error would.
2609 * fixed compiler performance when processing loops with a step >1;
2610 * bug fix: DOCUMENTATION now retrieves generic function
2611 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2612 support has been systematized, and now supports the methods
2613 specified by ANSI, along with a default method and a method for
2614 slot documentation. (reported by Nathan Froyd)
2615 * bug fix: effective methods associated with a generic function are
2616 no longer cached over a change of that generic function's method
2617 combination. (reported by Andreas Fuchs)
2618 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2619 names a non-existent directory. (thanks to Andreas Fuchs)
2620 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2621 the PPC platform, lead to stack corruption; this has been fixed.
2622 (reported by Rainer Joswig)
2623 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2624 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2625 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2626 arithmetic in general, is now much more reliable on the PPC
2628 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2629 platform now returns the right answer.
2630 * optimization: restored some effective method precomputation in
2631 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2632 precomputation is now tunable.
2633 * optimization: compiler-internal data structure use has been
2634 reviewed, and changes have been made that should improve the
2635 performance of the compiler by about 20%.
2636 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2637 simple-base-strings and simple-bit-vectors is improved.
2638 * optimization: the optimization of 32-bit logical and arithmetic
2639 functions introduced in version 0.8.3 on the x86 has been
2640 implemented on the mips, ppc and sparc platforms; an
2641 implementation of the same facility, but for 64-bit arithmetic,
2642 has been added for the alpha.
2643 * microoptimization: the compiler is better able to make use of the
2644 x86 LEA instruction for multiplication by constants.
2645 * bug fix: in some situations compiler did not report usage of
2646 generic arithmetic in (SPEED 3) policy.
2647 * bug 145b fix: compiler used wrong type specifier while converting
2648 MEMBER-types to numeric.
2649 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2651 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2652 index argument now works properly on the Alpha platform.
2653 * bug fix: floating point exception treatment on the Alpha platform
2655 * bug fix: FILE-POSITION works much better on string input and
2656 output streams. (thanks to Nikodemus Siivola)
2657 * bug fix: many threading/garbage collection symptoms sorted.
2658 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2659 might be pseudo-atomic.
2660 * internal change: Stopping for GC is now done with signals not
2661 ptrace. GC is now done in whichever thread wanted it, instead of
2663 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2665 * bug fix: result form in DO is not contained in the implicit
2667 * incompatible change: ICR structure is changed; the value part of
2668 CONTINUATION is now called LVAR; corresponding functions are
2669 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2670 * added type deriver for ISQRT (thanks to Robert E. Brown).
2671 * bug fix: better support for loading from the command line when an
2672 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2674 * fixed some bugs revealed by Paul Dietz' test suite:
2675 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2676 ** ROUND and FROUND now give the right answer when given very
2677 small float arguments.
2678 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2680 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2681 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2682 ** LCM should return a non-negative integer.
2683 ** PARSE-INTEGER returned the index of a terminator instead of the
2684 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2685 ** PARSE-INTEGER returned an incorrect index being applied to a
2687 ** LCM with two arguments of 0 returns 0 rather than signalling
2689 ** unsigned addition of a 32-bit constant with the high bit set no
2690 longer causes an internal compiler error on the x86.
2691 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2692 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2693 with negative last argument.
2694 ** byte specifiers with zero size and position no longer cause
2695 an error during type derivation.
2696 ** bignum multiplication on the Alpha platform now returns the
2698 * porting: The system now builds on SuSE AMD64, although it still
2699 generates a 32-bit binary.
2700 * .fasl file incompatibility: The fasl file version number has
2701 been incremented (because of the changes to internal compiler
2702 data structures referred to above).
2704 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2705 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2706 more accurately, on the Darwin kernel running on PowerPC hardware.
2707 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2708 * Compiler code deletion notes now signal a condition of type
2709 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2710 an associated MUFFLE-WARNING restart.
2711 * The compiler now performs limited argument count validation of
2712 constant format strings in FORMAT, and where appropriate in ERROR,
2713 CERROR and WARN. (thanks to Gerd Moellmann)
2714 * New ASDF-INSTALL contrib can be used for automatic download and
2715 installation of third-party Lisp code from CCLAN or other sites
2717 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2718 which forces another thread to execute a function supplied by the
2720 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2721 now accept and act upon their :ELEMENT-TYPE keyword argument.
2722 (reported by Martin Atzmueller, Edi Weitz)
2723 * bug fix: FILE-POSITION now accepts position designators up to
2724 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2725 is the greater. (thanks to Patrik Nordebo)
2726 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2727 Kalle Olavi Niemitalo)
2728 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2729 printable. (reported by Eric Marsden)
2730 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2731 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2732 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2733 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2734 (thanks to Stig Sandoe)
2735 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2736 UNREAD-CHAR now correctly includes the unread character in the
2737 target sequence. (thanks to Gerd Moellmann)
2738 * bug fix in threaded builds: the system can now be suspended and
2739 resumed by shell job control with minimal disruption.
2740 * bug fixes in times and timezones >2038AD
2741 * better handling of "where is GNU make?" problem in build scripts
2742 (thanks to Nikodemus Siivola)
2743 * new optimization: inside a named function any reference to a
2744 function with the same name is considered to be a self-reference;
2745 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2746 optimization quality.
2747 * new optimization on x86: logical functions and + now have
2748 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2749 used when the result is truncated to 32 bits.
2750 * VALUES declaration is partially enabled.
2751 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2752 * fixed some bugs revealed by Paul Dietz' test suite:
2753 ** The system now obeys the constraint imposed by
2754 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2755 types form a lattice under type intersection.
2756 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2757 ** ASSOC now ignores NIL elements in an alist.
2758 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2759 and (1+ MOST-POSITIVE-FIXNUM) answers.
2760 ** The addition of a method with invalid qualifiers to a generic
2761 function does not cause an error to be signalled immediately;
2762 a warning is signalled, and the error is generated only on
2763 calling the generic function.
2764 * changed .fasl file version number, in part to add type codes for
2765 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2766 obscure ANSI requirements
2768 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2769 * fixed bug 148: failure to inline-expand a local function left
2770 garbage, confusing the compiler.
2771 * fixed bugs 3cd: structure slot readers perform type check if the
2772 slot can have an invalid value (i.e. it is either not initialized
2773 or can be written with a less specific slot writer).
2774 * bug fix: the compiler now traps array references to elements off
2775 the end of an array; previously, the bounds checking in some
2776 circumstances could go off-by-one.
2777 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2779 * type declarations for array element types now obey the description
2780 on the CLHS page "Declaration TYPE", as per discussions on
2781 sbcl-help around 2003-05-08. This means that a declaration
2782 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2783 declaration, all references to BAR will be asserted or assumed
2784 (with THE, so dependent on compiler policy) to involve objects of
2785 type FOO. Note that no such declaration is implied in
2786 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2787 * declared types of functions from the "Conditions"
2788 chapter. (reported by Paul Dietz)
2789 * bug fix: CERROR accepts a function as its first argument.
2790 * bug fix: NTH an NTHCDR accept a bignum as index
2791 arguments. (reported by Adam Warner)
2792 * optimization: character compare routines now optimize comparing
2793 against a constant character. (reported by Gilbert Baumann)
2794 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2795 argument now works properly on the MIPS platform.
2796 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2797 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2798 declared to have a functional type, does not cause an error
2800 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2801 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2802 * fixed bug: if last continuation of a deleted block has a
2803 destination, this destination should be deleted too. (reported by
2805 * fixed a bug in the bootstrap process: the host compiler's values
2806 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2807 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2808 test case from Patrik Nordebo)
2809 * improved the ability of the disassembler on the PPC platform to
2810 provide helpful disassembly notes.
2811 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2812 the class in more cases than previously.
2813 * bug fix: FILE-POSITION now understands :START and :END for
2814 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2815 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2816 8) data. (thanks to David Lichteblau)
2817 * bug fix: it is possible to add a method to a generic function
2818 without lambda list.
2819 * bug fix: reader failed to signal END-OF-FILE inside an
2820 object representation. (reported by Nikodemus Siivola)
2821 * fixed some bugs revealed by Paul Dietz' test suite:
2822 ** LAST and [N]BUTLAST should accept a bignum.
2823 ** condition slot accessors are methods.
2824 ** (VECTOR NIL) is a subtype of STRING.
2826 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2827 * minor incompatible change: some nonsensical specialized lambda
2828 lists (used in DEFMETHOD) which were previously ignored now signal
2830 * minor incompatible change: the system is now aware of the types of
2831 variables in the COMMON-LISP package, and will signal errors for
2832 most violations of these type constraints (where previously they
2833 were silently accepted).
2834 * minor incompatible change: COMPILE-FILE now uses the freedom
2835 afforded (ANSI 3.2.2.3) to use derived function types for
2836 functions defined in the same file. This also permits the system
2837 to warn on static type mismatches and function
2838 redefinition. (Currently it does not work with high DEBUG level.)
2839 * minor incompatible change: VALUES declaration is disabled.
2840 * When issuing notes, the compiler now signals a condition of type
2841 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2842 restart for use in user handlers. It is expected that the
2843 COMPILER-NOTE condition will eventually become a condition
2844 supertype to a hierarchy of note types, which will then be
2845 handleable in a similar fashion. However, at the moment, no such
2846 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2847 but it's an implementation detail, not a classification for the
2849 * Changes in type checking closed the following bugs:
2850 ** type checking of unused values (192b, 194d, 203);
2851 ** template selection based on unsafe type assertions (192c, 236);
2852 ** type checking in branches (194bc).
2853 * A short form of VALUES type specifier has ANSI meaning (it has
2854 increased the number of situations when SBCL cannot perform type
2856 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2857 keywords or constants is permissible.
2858 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2859 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2860 classes. (thanks to Antonio Martinez)
2861 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2862 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2863 argument to a pretty-print stream. Instead, it leaves any such
2864 operations to DESCRIBE-OBJECT methods.
2865 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2867 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2868 lambda lists. (thanks to David Lichteblau)
2869 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2870 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2871 (thanks to Gerd Moellmann)
2872 * fixed bug 246: increased compilation speed of long
2873 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2875 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2877 * DEFCONSTANT now throws a condition of type
2878 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2879 constant to a non-EQL value; CONTINUE and ABORT restarts
2880 respectively change and preserve the value.
2881 * fixed bug 63: The code walker, part of the implementation of CLOS,
2882 is now better at handling symbol macros.
2883 * bug fix: There is no longer an internal implementation type named
2884 CL:LENGTH. (reported by Raymond Toy)
2885 * bug fix: In macro-like defining macros/special operators the
2886 implicit block does not enclose the lambda list.
2887 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2888 enclosed in parentheses) are not suitable as type specifiers, and
2889 their use properly signals an error now.
2890 * bug fix: An argument count mismatch for a type specifier in code
2891 being compiled no longer causes an unhandled error at compile
2892 time, but signals a compile-time warning.
2893 * fixed simple vector readable printing
2894 * bug fix: DESCRIBE takes more care over whether the class
2895 precedence list slot of a class is bound before accessing it.
2896 (reported by Markus Krummenacker)
2897 * bug fix: FORMATTER can successfully compile pretty-printer format
2898 strings which use variants of the ~* directive inside.
2899 * bug fix: SEARCH now applies its TEST predicate to the elements of
2900 the arguments in the correct order. (thanks to Wolfhard Buss)
2901 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2902 in type inference. (thanks to Robert E. Brown)
2903 * bug fix: Reading in symbols with an explicit package name of ""
2904 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2905 * fixed some bugs revealed by Paul Dietz' test suite:
2906 ** NIL is now allowed as a structure slot name.
2907 ** Arbitrary numbers, not just REALs, are allowed in certain
2908 circumstances in LOOP for-as-arithmetic clauses.
2909 ** Multiple class redefinitions before slot access no longer
2910 causes a type error.
2911 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2912 association between the name and a class.
2913 ** Generic functions with non-standard method-combination and over
2914 five methods all of which return constants no longer return NIL
2915 after the first few invocations. (thanks to Gerd Moellmann)
2916 ** CALL-NEXT-METHOD with no arguments now passes the original
2917 values of the arguments, even in the presence of assignment.
2918 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2920 ** DISASSEMBLE works with closures and funcallable instances.
2921 ** ADD-METHOD now returns the generic function, not the new method.
2922 ** FIND-METHOD signals an error if the lengths of the specializers
2923 is incompatible with the generic function, even if the ERRORP
2925 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2926 which its argument is a member.
2927 ** DEFCLASS only redefines the class named by its class-name
2928 argument if that name is the proper name of the class;
2929 otherwise, it creates a new class.
2930 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2931 of the UNBOUND-SLOT condition to the name of the slot.
2932 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2933 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2934 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2935 SLOT-MAKUNBOUND in the specified fashion.
2937 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2938 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2939 cross-compilation host. As a consequence, we can now bootstrap our
2940 way up to SBCL starting with a bare gcc toolchain and human-readable
2941 source code (first the source to CLISP, then the source to SBCL).
2942 * A contributed module containing a partial implementation of the
2943 simple-streams interface has been included. (thanks to Rudi
2945 * A contributed module implementing the RFC1321 Message Digest
2946 Algorithm, known as MD5, has been included.
2947 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2948 no longer has any effect, as the code controlled by this feature
2949 has been deleted. (As far as we know, no-one has ever built using
2950 this feature, and its semantics were confused in any case).
2951 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2952 work on conditions (as required by the ANSI specification),
2953 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2954 expected behaviour on conditions. Users should note, however,
2955 that such behaviour is not required by the ANSI specification,
2956 and so use of this behaviour may render their code unportable.
2957 * fixed some bugs revealed by Paul Dietz' test suite:
2958 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2960 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2961 ** SXHASH on condition objects no longer returns NIL.
2962 ** :ALLOCATION :CLASS slots are better treated; their values are
2963 updated on class redefinition, and initforms inherited from
2964 superclasses are applied.
2965 ** REMOVE-METHOD returns its generic function argument even when
2966 no method was removed.
2967 ** SHARED-INITIALIZE now initializes the values of the requested
2968 slots, including those with :ALLOCATION :CLASS.
2969 ** ALLOCATE-INSTANCE now works on structure classes defined via
2970 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2972 ** SLOT-EXISTS-P now works on conditions, as well as structures
2974 ** MAKE-LOAD-FORM now has the required methods on
2975 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2976 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2977 keyword, and now implements the SLOT-NAMES argument.
2978 ** methods with &OPTIONAL arguments no longer allow too many
2979 arguments to be passed in the call without error.
2980 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2981 option is consistent with the required arguments of the generic
2982 function lambda list.
2983 * bug fix: REQUIRE accepts a string designator. (Thanks to
2985 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2986 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2987 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2988 * bug fix: The compiler now checks for duplicated variables in macro
2990 * bug fix: SETQ on globals returns the correct value.
2991 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2992 not a COMPILER-ERROR (followed by some other strange error on
2993 choosing the CONTINUE restart).
2994 * bug fix: make.sh and friends are now more consistent in the way that
2995 they look for GNU "make".
2997 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2998 * experimental native threads support (on x86 Linux >=2.4 only).
2999 This is not compiled in by default: you need to add :SB-THREAD to
3000 the target features. See the "Beyond ANSI" chapter of the manual
3002 * fix for longstanding nonANSIism: The old distinction between
3003 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3004 The return value from CL:FIND-CLASS is now a CLOS class, and
3005 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3006 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3007 * An interface to the MetaObject Protocol, as described in Kiczales,
3008 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3009 MIT Press, 1991, is available from the SB-MOP package.
3010 * incompatible change: the SB-PCL package should now be considered
3011 a private implementation detail, and no longer a semi-private MOP
3013 * minor incompatible change: due to rearrangement for threads, the
3014 control stack and binding stack are now allocated at arbitrary
3015 addresses instead of being hardcoded per-port. Users affected by
3016 this probably have to be doing advanced things with shared
3017 libraries, and will know who they are.
3018 * minor incompatible change: Previously, all --eval forms used were
3019 processed with READ before any of them were processed with EVAL.
3020 Now each --eval form is processed with both READ and EVAL before
3021 the next --eval form is processed. (Thus package operations like
3022 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3023 work as the user might reasonably expect.)
3024 * minor incompatible change: *STANDARD-INPUT* is now only an
3025 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3027 * minor incompatible change: Y-OR-N-P is now character-oriented,
3028 not line oriented. Also, YES-OR-NO-P now works without errors.
3029 (thanks to Antonio Martinez)
3030 * sb-aclrepl module improvements: an integrated inspector, added
3031 repl features, and a bug fix to :trace command.
3032 * Known functions, which cannot be open coded by the backend, are
3033 considered to be able to check types of their arguments. (fixing
3034 a bug report by Nathan J. Froyd)
3035 * fixed a bug in computing method discriminating functions: It is
3036 now possible to define methods specialized on classes which have
3037 forward-referenced superclasses. (thanks to Gerd Moellmann)
3038 * fixed evaluation order in optional entries (reported by Gilbert
3040 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3041 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3042 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3043 required-by-AMOP NAME argument, as well as CLASS and
3044 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3045 * fixed bug 20: DEFMETHOD can define methods using names that are
3046 not the proper names of classes to designate class specializers.
3047 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3048 Unix file descriptors, instead of blowing up. (thanks to Antonio
3050 * Garbage collection refactoring: user-visible change is that a
3051 call to the GC function during WITHOUT-GCING will not do garbage
3052 collection until the end of the WITHOUT-GCING. If you were doing
3053 this you were probably losing anyway.
3054 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3055 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3056 TYPEP the latter but not the former.
3057 * The compiler issues a full WARNING for calls to undefined functions
3058 with names from the CL package.
3059 * MAP-INTO for a vector destination is open coded. (reported by
3060 Brian Downing on c.l.l)
3061 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3062 documentation string.
3063 * fixed some bugs revealed by Paul Dietz' test suite:
3064 ** COPY-ALIST now signals an error if its argument is a dotted
3066 ** Condition slots are now accessed more correctly in the presence
3067 of multiple initargs for a given slot.
3068 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3069 exclude restarts of the same name associated with a different
3071 ** DEFCLASS of forward-referenced classes with another
3072 forward-referenced class in the superclasses list no longer
3074 ** Condition slots are now initialized once each, not multiple
3075 times. (thanks to Gerd Moellmann)
3076 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3078 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3079 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3080 arguments contain duplicated elements.
3081 ** RESTART-CASE understands local macros.
3082 ** RESTART-CASE associates exactly its own restarts with a condition.
3083 ** ENDP in safe mode checks its argument to be of type LIST.
3084 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3085 in question is unbound.
3086 ** Optimized MAKE-INSTANCE functions no longer cause internal
3087 assertion failures in the presence of duplicate initargs.
3088 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3089 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3090 propagated even in the ctor optimized implementation of
3092 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3094 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3095 * a better implementation of SXHASH on (simple) bit vectors,
3096 measured both in execution speed and in distribution of results
3097 over the positive fixnums, has been installed. Likewise, a better
3098 implementation of EQUAL for simple bit vectors is now available.
3099 * fixed CEILING optimization for a divisor of form 2^k.
3100 * fixed bug 240 (emitting extra style warnings "using the lexical
3101 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3102 by Antonio Martinez)
3103 * fixed SXHASH, giving different results for NIL depending on type
3104 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3105 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3106 variable. (found by Rolf Wester)
3107 * SBCL does not ignore type declarations for special
3108 variables. (reported by rif on c.l.l 2003-03-05)
3109 * some bug fixes in contrib/sb-aclrepl/
3110 * fixed some bugs revealed by Paul Dietz' test suite:
3111 ** a bug in the CONS type specifier, whereby the CAR and CDR
3112 types got intertwined, has been fixed;
3113 ** the type system is now able to reason about the interaction
3114 between INTEGER and RATIO types more completely;
3115 ** APPEND, [N]REVERSE and NRECONC check that those their
3116 arguments, which must be proper lists, are really so;
3117 ** An array specialized to be unable to hold elements has been
3118 implemented, as required -- yes, really -- by ANSI;
3119 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3120 on malformed property lists;
3122 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3123 * incompatible packaging change: in line with Unix convention,
3124 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3125 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3126 * REQUIRE and PROVIDE are now optionally capable of doing something
3127 useful. See the documentation string for REQUIRE.
3128 * infrastructure for a managed SBCL contrib system: contributed
3129 modules in this release include:
3130 ** the ASDF system definition facility;
3131 ** an interface to the BSD Sockets API;
3132 ** an ACL-like convenience interface to the repl;
3133 (thanks to Kevin Rosenberg)
3134 ** an implementation of ROTATE-BYTE, with efficient implementation
3136 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3137 gives a closer approximation to the right answer than previously.
3138 (thanks to Raymond Toy)
3139 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3140 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3141 optional environment arguments, as required by ANSI.
3142 * fixed bugs in other functions taking environment objects, allowing
3143 calls with an explicit NIL environment argument to be compiled
3145 * fixed bug 228: primary return values from
3146 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3147 COMPILE or FUNCTION.
3148 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3149 :TYPE will no longer signal errors on innocuous objects.
3150 * fixed bug 231b: SETQ is better at respecting type declarations in
3151 the lexical environment.
3152 * fixed a bug in DEFCLASS: classes named by symbols with no or
3153 unprintable packages can now be defined.
3154 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3155 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3156 * fixed compiler failure related to checking types of functions.
3157 (reported by Robert E. Brown)
3158 * the compiler is now much more consistent in its error-checking
3159 treatment of bounding index arguments to sequence functions: in
3160 (SAFETY 3) code, errors will be signalled in almost all cases if
3161 invalid sequence bounding indices are passed to functions defined
3162 by ANSI to operate on sequences.
3163 * fixed a bug in the build procedure: documentation of SBCL-specific
3164 packages is now preserved and available in the final Lisp image.
3165 * lifted FDEFINITION lookup out of loops in the implementation of
3166 many list operations. (thanks to Robert E. Brown)
3167 * fixed a bug in the reader: the #n# reader macro now works for
3168 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3169 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3170 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3171 E. Brown and Tony Martinez respectively)
3172 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3173 SLOT-BOUNDP: in particular, it is now able to optimize them much
3174 better, and is now not vulnerable to having packages renamed.
3175 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3176 always called when a slot is not present in an instance. (thanks
3178 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3179 longer signal an error on unreading a character following EOF on
3180 the previous constituent stream. (thanks to Tony Martinez)
3181 * fixed some bugs revealed by Paul Dietz' test suite:
3182 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3183 not just nonnegative fixnums;
3184 ** the logical bit-array operators such as BIT-AND now accept an
3185 explicit NIL for their "opt-arg" argument (to indicate a
3186 freshly-consed result bit-array);
3187 ** ELT now signals an error on an invalid sequence index in safe
3189 ** the type system is now cleverer about negations of numeric
3190 types, and consequently understands the BIGNUM and RATIO types
3192 ** the type system is now cleverer about the interaction between
3193 INTEGER and RATIO types: while bugs still remain, many more
3194 cases are accurately computed;
3195 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3196 if it is in the last clause;
3197 ** CONSTANTLY now correctly returns a side-effect-free function in
3199 ** DECLARE is no longer treated as a special-operator; in
3200 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3201 * incremented fasl file version number due to the change in the
3202 DEFSTRUCT-SLOT-DESCRIPTION structure.
3204 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3205 * minor incompatible change: code processed by the "interpreter" or
3206 EVAL now has a compilation optimization policy of (DEBUG 2)
3207 (changed from (DEBUG 1)) to improve debuggability of interactive
3208 development, and to allow the use of the debug RETURN command in
3210 * an experimental implementation of the RETURN command for the
3211 debugger has been included. (thanks to Frederik Kuivinen)
3212 * fixed bug 62: constraints were not propagated into a loop.
3213 * fixed bug in embedded calls of SORT (reported and investigated by
3215 * fixed some bugs revealed by Paul Dietz' test suite:
3216 ** printing and reading of arrays with some dimensions having
3217 length 0 (thanks to Gerd Moellmann);
3218 ** BOA constructor with &AUX argument without a default value does
3219 not cause a type error;
3220 ** CONSTANTP now returns true for all self-evaluating objects.
3222 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3223 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3224 accessors that are related by inheritance, as specified in the
3225 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3226 Valtteri Vuorikoski)
3227 * The compiler is now able to inline functions that were defined in
3228 a complex lexical environment (e.g. inside a MACROLET).
3229 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3231 * Support for the upcoming FreeBSD-5.0 release has been included.
3232 (thanks to Dag-Erling Smorgrav)
3233 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3234 effect when it is not in a toplevel context.
3235 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3236 stand a better chance of being correct. (thanks to Gerd
3238 * fixed bug in COERCE, which now signals an error on coercing a
3239 rational to a bounded real type which excludes the expected
3241 * The compiler is now able to derive types more accurately from the
3242 COERCE and COMPILE functions.
3243 * fixed bug 223: functional binding is considered to be constant
3244 only for symbols in the CL package.
3245 * fixed bug 231: SETQ did not check the type of a variable being set
3246 (reported by Robert E. Brown)
3247 * A new optimization for MAKE-INSTANCE has been included, fixing
3248 various bugs (including relating to :ALLOCATION :CLASS slots and
3249 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3251 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3252 ** As required by ANSI, LOOP now disallows anonymous collection
3253 clauses such as COLLECT I in conjunction with aggregate boolean
3254 clauses such as THEREIS (= I 1);
3255 ** LOOP now signals an error when any variable is reused in the
3256 same loop (including the potentially useful construct analogous
3257 to WITH A = 1 WITH A = (1+ A);
3258 ** IT is only a special loop symbol within the first clause of a
3259 conditional loop clause;
3260 ** LOOP with a typed iteration variable over a hashtable now
3261 signals a type error iff it should.
3262 * fixed some other bugs revealed by Paul Dietz' test suite:
3263 ** FILE-STREAM now names the class previously known as FD-STREAM;
3264 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3265 argument) no longer signals an error;
3266 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3267 of the slot symbol, rather than using the current package
3268 ((:CONC-NAME "") continues to intern the slot's name in the
3270 * incremented fasl file version number, because of the incompatible
3271 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3272 of the new implementation of DEFINE-COMPILER-MACRO.
3274 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3275 * Support for building SBCL for MIPS platforms running in
3276 little-endian mode has now been checked in, and basic
3277 functionality on said platforms verified.
3278 * minor incompatible change: PCL now records the pathname of a file
3279 in which methods and the like are defined, rather than its
3281 * minor incompatible change: TRUENAME now considers the truename of
3282 a file naming a directory to be the pathname with :DIRECTORY
3283 component indicating that directory.
3284 * minor incompatible change: a NAMED clause in the extended form of
3285 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3286 reason for the previous behaviour is unclear.
3287 * more systematization and improvement of CLOS and MOP conformance
3288 in PCL (thanks to Gerd Moellman and Pierre Mai):
3289 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3291 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3292 primary methods with no specializers;
3293 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3295 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3296 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3297 FORWARD-REFERENCED-CLASSes; error reporting on
3298 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3300 ** SXHASH on CLOS instances now uses a slot internal to the
3301 instance to return different numbers on distinct instances,
3302 while preserving the same return value through invocations of
3304 ** DEFMETHOD signals errors when methods with longer incongruent
3305 lambda lists are added to generic functions;
3306 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3307 CLASS, as specified in AMOP;
3308 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3309 on the order returned by the primary method for classes of
3310 class STANDARD-CLASS;
3311 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3312 * fixed some bugs shown by Paul Dietz' test suite:
3313 ** DOLIST puts its body in TAGBODY;
3314 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3316 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3317 value producing form;
3318 ** if more variables are given to PROGV than values, extra
3319 variables are bound and made to have no value;
3320 ** NSUBSTITUTE on list arguments gets the right answer with
3322 ** ELT signals an error of type TYPE-ERROR when the index argument
3323 is not a valid sequence index;
3324 ** LOOP signals (at macroexpansion time) an error of type
3325 PROGRAM-ERROR when duplicate variable names are found;
3326 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3327 ** FUNCALL of special-operators now cause an error of type
3329 ** PSETQ now works as required in the presence of side-effecting
3330 symbol-macro places;
3331 ** NCONC accepts any object as its last argument;
3332 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3334 ** loop-for-as-package does not require a package to be explicitely
3336 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3338 * fixed bug 166: compiler preserves "there is a way to go"
3339 invariant when deleting code.
3340 * fixed bug 172: macro lambda lists with required arguments after
3341 &REST arguments now cause an error to be signalled. (thanks to
3343 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3344 bug. (thanks to Matthew Danish)
3345 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3347 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3348 arguments are lexically rebound. (thanks to Gerd Moellmann and
3350 * fixed bug 194: error messages are now more informative when there
3351 is no primary method applicable in a call to a generic
3352 function. (thanks to Gerd Moellmann)
3353 * fixed bug in command line argument checking (thanks to Julian
3355 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3356 (thanks to Matthew Danish)
3357 * incremented fasl file version number, because of the
3358 SXHASH-related changes in the layout of CLOS data structures
3360 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3361 * minor incompatible change: The runtime (the Unix executable named
3362 "sbcl") is now much pickier about the .core files it will load.
3363 Essentially it now requires .core files to descend from the same
3364 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3365 as the runtime does. (The intent is to prevent the crashes which
3366 can occur, and which can even be reported as mysterious failures,
3367 when people patch the sources or change the build parameters
3368 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3369 sbcl and .core files.)
3370 * fixed bug: VALUES-LIST is no longer optimized away.
3371 * fixed bug 142: The FFI conversion of C string values to Lisp
3372 string values no longer conses excessively. (thanks to Nathan
3373 Froyd porting Raymond Toy's fix to CMU CL)
3374 * began to systematize and improve MOP conformance in PCL (thanks to
3375 Nathan Froyd, Gerd Moellman and Pierre Mai):
3376 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3378 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3379 ** FINALIZE-INHERITANCE is now called on class finalization;
3380 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3381 argument precedence order.
3382 * fixed bug 202: The compiler no longer fails on functions whose
3383 derived types contradict their declared type.
3384 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3385 so it can be non-toplevel.
3386 * The fasl file version number has changed (because of the new
3387 implementation of DEFMACRO).
3388 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3389 safe code, that any length requirement by their type-specifier
3390 argument is valid. The exceptions to this are described in bug
3392 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3393 functions now check the types of their inputs as required by ANSI.
3394 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3395 introduced symbol is DECLAREd to be SPECIAL.
3396 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3397 * fixed bug: :COUNT argument to sequence functions may be negative
3398 * fixed bug: body of DO-SYMBOLS may contain declarations
3399 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3400 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3401 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3402 symbol macro only once
3403 * fixed printing of call frame when argument list is unavailable
3404 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3405 * compiler no longer signals WARNING on unknown keyword
3408 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3409 * A beta-quality port to the mips architecture running Linux,
3410 based on the old CMUCL backend, has been made. It has been tested
3411 on a big-endian kernel, and works sufficiently well to be able to
3412 rebuild itself; it has not been tested in little-endian mode.
3413 * fixed an inconsistency between gencgc.c and purify.c which made
3414 dumping/loading .core files unreliable
3415 * fixed bug 120a: The compiler now deals correctly with IFs where
3416 the consequent is the same as the alternative, instead of
3417 misderiving the return type. (thanks to Alexey Dejneka)
3418 * fixed bug 113: Logical pathnames are now dumpable (the logical
3419 host is resolved at load-time, throwing an error if it is not
3421 * fixed bug 174: FORMAT's error message is slightly clearer when a
3422 non-printing character is used in a format directive.
3423 * fixed several bugs in compiler checking of type declarations, i.e.
3424 violations of the Python "declarations are assertions" principle
3425 (thanks to Alexey Dejneka)
3426 * fixed several bugs in PCL's error checking (thanks to Gerd
3428 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3430 * fixed bug in compilation of functions as first class values
3431 (thanks to Antonio Martinez-Shotton)
3432 * The compiler's handling TYPE-ERRORs which it can prove will
3433 inevitably happen at runtime has been cleaned up and corrected
3434 in several ways. (thanks to Alexey Dejneka)
3435 * improved argument type checking for various basic arithmetic
3436 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3437 lavished on them in the past that they can be compiled in many
3438 ways in different special cases
3439 * fixed bug 181: compiler checks validity of user supplied type
3441 * cleaned up code flushing in optimization: Function calls which
3442 should signal errors for safety purposes (e.g. which ANSI says
3443 should signal errors when their arguments are of incorrect type)
3444 are no longer optimized away.
3445 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3446 * incremented fasl file version number, because changes in the
3447 implementation of sequence functions like COERCE caused
3448 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3449 in old inline expansions) to become undefined. (Actually these
3450 changes were later undone, so we might very well be binary
3451 compatible with 0.7.7 after all, but leaving the version number
3452 incremented seemed like the simplest and most conservative
3455 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3456 * An alpha-quality port to the parisc architecture running Linux,
3457 based on the old CMUCL backend, has been made. This, even more so
3458 than the other backends, should be considered still a work in
3459 progress; known problems include that the Linux kernel in 64-bit
3460 mode does not propagate the correct sigcontext structure to
3461 userspace, and consequently SBCL on a parisc64 kernel will not
3463 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3464 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3465 suppressed.) Also now that the compiler is looking at declarations
3466 in the environment, it checks optimization declarations as well,
3467 and suppresses inlining when (> DEBUG SPEED).
3468 * More fixes have been made to treatment of floating point exception
3469 treatment and other Unix signals. In particular, floating point
3470 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3471 * The detection and handling of control stack exhaustion (infinite
3472 or very deeply nested recursion) has changed. Stack exhaustion
3473 detection is now done by write-protecting pages at the OS level
3474 and applies at all optimization settings; when found, a
3475 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3476 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3477 be caught using IGNORE-ERRORS.
3478 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3479 correctly when transforming recursive LABELS functions to LETs.
3480 (thanks to Alexey Dejneka)
3481 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3482 that are names of constants or global variables.
3483 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3484 alien routines with docstrings.
3485 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3486 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3488 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3489 (thanks to Eric Marsden)
3490 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3491 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3492 * Bug fix: The SPARC backend can now compile functions involving
3493 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3494 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3495 to a functions expecting a PATHNAME-DESIGNATOR.
3496 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3497 lambda lists. (thanks to Alexey Dejneka)
3498 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3499 functions so that EXPORTing the name of the class doesn't cause
3500 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3501 to Antonio Martinez for reporting this)
3502 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3503 constant-folded without failing an assertion. (thanks to Einar
3504 Floystad Dorum for reporting this)
3505 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3506 types are now dealt with more correctly by the compiler.
3507 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3508 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3509 behaviour analogous to RENAME-FILE. This puts its behaviour more
3510 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3512 * The fasl file version number has changed again. (because of the
3513 bug fix involving the names of PCL MAKE-INSTANCE functions)
3515 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3516 * bug fix: Floating point exceptions are treated much more
3517 consistently on the x86/Linux and PPC/Linux platforms.
3518 * Array initialization with :INITIAL-ELEMENT is now much faster for
3519 cases when the compiler cannot open code the array creation, but
3520 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3521 array accesses have also seen a speed increase.
3522 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3523 specified. (This was at the root of some bad interactions between
3524 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3526 * bug fix: Internal error arguments for undefined functions are now
3527 computed correctly on the PPC/Linux platform.
3528 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3529 Raymond Toy's patch for CMU CL)
3530 * Support for the Solaris 9 operating environment has been included
3531 (thanks to Daniel Merritt)
3532 * A very ugly but hopefully complete draft of the missing FFI chapter
3533 of the manual has been created by reformatting the corresponding
3534 CMU CL manual chapter into (currently very ugly and incoherent)
3535 DocBook and bringing it up to date for SBCL behavior. Thus, the
3536 manual is now essentially complete, at least by my extreme
3537 once-and-only-once standards, whereby it's acceptable to refer to
3538 the doc strings of SB-EXT functions as the primary documentation.
3539 * The fasl file version number has changed again, due to cleanup of
3540 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3541 is no longer a static symbol.)
3543 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3544 * SBCL now builds with OpenMCL (version 0.12) as the
3545 cross-compilation host; also, more progress has been made toward
3546 bootstrapping under CLISP.
3547 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3549 * bug 158 fixed: The compiler can now deal with integer loop
3550 increments different from 1; fixing this turned out also to fix
3552 * bug 169 fixed: no more bogus warnings about using lexical bindings
3553 despite the presence of perfectly good SPECIAL declarations (thanks
3554 to David Lichteblau)
3555 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3556 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3557 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3559 * bug 180 fixed: Method combination specifications no longer ignore
3560 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3561 * bug fix: Structure type predicate functions now check their argument
3562 count as they should.
3563 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3564 correctly. (thanks to Pierre Mai)
3565 * minor incompatible change: The --noprogrammer option is deprecated
3566 in favor of the new --disable-debugger option, which is very similar.
3567 (The major difference is that it takes effect at a slightly different
3568 time at startup, causing handling of errors in --sysinit and
3569 --userinit files will be affected differently.) The
3570 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3571 been added to allow this functionality to be controlled from ordinary
3572 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3573 maintainers, who might want to run non-interactive scripts to
3574 build SBCL cores which will later be used interactively.)
3575 * minor incompatible change: The LOAD function no longer, when given
3576 a wild pathname to load, loads all files matching that pathname.
3577 Instead, an error of type FILE-ERROR is signalled.
3579 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3580 * bug 147 fixed: The compiler preserves its block link/count
3581 invariants more correctly now so that it doesn't crash. (thanks
3583 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3585 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3586 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3587 no-value-for-this-character as NIL (as specified by ANSI).
3588 * HOST-NAMESTRING on physical pathnames now returns a string that is
3589 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3590 (thanks to Christophe Rhodes)
3591 * The Alpha port handles icache flushing more correctly. (thanks to
3593 * More progress has been made toward bootstrapping under CLISP. (thanks
3594 to Christophe Rhodes)
3595 * The fasl file format has changed again, because dynamic loading
3596 on OpenBSD (which has non-ELF object files) motivated some cleanups
3597 in the way that foreign symbols are transformed and passed around.
3598 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3599 is no longer treated as whitespace by the reader, but instead as
3600 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3601 |A
\7fB|, instead of A as it used to.
3603 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3604 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3605 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3606 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3607 this as of 0.7.1.45, but was left out of the previous news section
3608 (thanks to Dan Barlow)
3609 * SBCL now runs on the Solaris operating system on SPARC architectures
3610 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3611 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3612 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3613 * SPARC backend cleanups, allowing builds of cores optimized for V8
3614 and V9 SPARCS, and also emission of code targeted to a particular
3615 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3617 * SBCL is closer to bootstrapping under CLISP, thanks to various
3618 fixes by Christophe Rhodes.
3619 * The fasl file format has changed again, to allow the compiler's
3620 INFO database to support symbol macros.
3621 * The user manual (in doc/) is formatted into HTML more nicely.
3622 (thanks to coreythomas)
3623 * The system is smarter about SUBTYPEP relationships, especially
3624 those involving NOT types (including types such as ATOM which are
3625 represented internally using NOT types). Thus SUBTYPEP is less
3626 likely to return (VALUES NIL NIL) in general, and in particular
3627 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3629 * The fasl file format has changed again, because the internal
3630 representation of types now includes a new slot to support the new
3631 SUBTYPEP-of-NOT-types logic.
3632 * (not a change in the main branch of SBCL, but a related prototype
3633 which can hopefully be merged into the main branch of SBCL in the
3634 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3635 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3636 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3637 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3638 dispatch macro character combinations. (thanks to Alexey Dejneka)
3639 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3640 that superficially look like logical namestrings correctly.
3641 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3643 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3644 default supplied-p) for &optional and &key arguments. (thanks to
3647 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3648 * incompatible change: The compiler is now less aggressive about
3649 tail call optimization, doing it only when (> SPACE DEBUG) or
3650 (> SPEED DEBUG). (This is an incompatible change because there are
3651 programs which relied on the old CMU-CL-style behavior to optimize
3652 away their unbounded recursion which will now die of stack overflow.)
3653 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3654 for non-GENCGC systems has been increased to 20M (since that
3655 seems much closer to the likely performance optimum for modern
3656 systems than the old 4M value was)
3657 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3658 and *DEBUG-PRINT-LEVEL*
3659 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3660 of CMU CL's support for SPARC, and various endianness and other
3661 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3662 * new syntactic sugar for the Unix command line: --load foo.bar is now
3663 an alternate notation for --eval '(load "foo.bar")'.
3665 ** The system now detects stack overflow and handles it gracefully,
3666 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3667 optimization settings. (This is a good thing in general, and
3668 its introduction in this version should be particularly timely
3669 for anyone whose code fails because of suppression of tail
3671 ** The system now hunts for the C variable "environ" in a more
3672 devious way, to avoid segfaults when the C library version differs
3673 between compile time and run time. (thanks to Christophe Rhodes)
3674 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3675 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3676 ** The compiler no longer issues bogus style warnings for undefined
3677 classes in the same source file as the DEFCLASSes which defined
3678 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3680 ** fixes in CONDITION class precedence list for undefined function
3681 errors (thanks to Alexei Dejneka)
3682 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3683 correctly. (thanks to Dan Barlow)
3684 ** portability fixes aiming at bootstrapping under CLISP (thanks
3685 to Dave McDonald and Christophe Rhodes)
3686 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3687 ** fixes in type translation and and type inference (thanks to
3689 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3690 ** various fixes in the new ports (thanks to Dan Barlow)
3691 * several changes related to debugging:
3692 ** suppression of tail recursion, as noted above
3693 ** stack overflow detection, as noted above
3694 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3695 is now the default. (For some time encapsulation has been more
3696 reliable than the breakpoint-based :ENCAPSULATE NIL
3697 implementation, at least on X86 systems; and I just noticed that
3698 encapsulation also seems closer to the spirit of the ANSI
3701 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3703 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3704 up properly again. (There was a packaging bug in 0.7.0 which
3705 left their definitions in SB-SYS::LOAD-FOREIGN and
3706 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3707 vital for most things which interface to C-level interfaces,
3708 like extensions working with sockets or databases or
3709 Perl-compatible regexes or whatever, and the need to fix
3710 this bug is the main reason that 0.7.1 was released so
3712 ** DEFGENERIC is now choosier about the methods it redefines, so that
3713 reLOADing a previously-LOADed file containing DEFGENERICs does
3714 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3715 cycle now works as it should. (thanks to Alexey Dejneka)
3716 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3717 fixed by Christophe Rhodes. (He actually submitted this patch
3718 months ago, and I delayed until after 0.7.0.)
3719 ** Bug 111 (internal compiler confusion about runtime checks on
3720 FUNCTION types) was fixed by Alexey Dejneka.
3721 * Some internal cleanups (getting rid of variables which aren't
3722 needed now that the byte interpreter is gone) caused the fasl
3723 file format number to change again.
3725 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3726 * major incompatible change: The default fasl file extension, i.e. the
3727 default extension for files produced by COMPILE-FILE, has changed
3728 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3730 ** There are many changes in the implementation of the compiler.
3731 SBCL is now essentially a compiler-only implementation of ANSI
3732 Common Lisp. EVAL still "interprets" a few special cases, but
3733 almost all the interesting cases are handled by creating
3734 a LAMBDA expression, calling COMPILE on it, then calling
3735 FUNCALL on the result.
3736 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3737 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3738 Since the code is newer, there might still be some new bugs
3739 (though not as many as before Martin Atzmueller's fixes:-). But
3740 the new code is substantially simpler and clearer, and hopefully
3741 any remaining bugs will be simpler, less fundamental, and more
3742 fixable then the bugs in the old code.
3743 ** The revised compiler is still a little unsteady on its feet.
3745 *** The debugging information it produces (particularly the names
3746 of FUNCTION objects) is sometimes much less useful than what
3747 the old compiler produced.
3748 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3749 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3750 within a MACROLET) has been temporarily weakened.
3751 ** There are new compiler optimizations for various functions:
3752 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3753 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3754 *** the math functions TRUNCATE, FLOOR, and CEILING
3755 *** the function-of-all-trades COERCE
3756 Mostly these should be transparent, but there's one
3757 potentially-annoying problem (bug 117): when the compiler
3758 inline-expands a function and does type analysis on the result,
3759 it can create control paths which have type mismatches, and
3760 when it can't prove that those control paths aren't taken,
3761 it will issue WARNINGs about the type mismatches. This is
3762 a particular problem in practice for the new sequence functions.
3763 It's not clear how this should be fixed, and for now, a
3764 workaround is given in the entry for 117 in the BUGS file.
3765 ** (Because of the interaction between the two previous items --
3766 occasional inlining problems and new inline expansions -- some
3767 of the new sequence function optimizations won't really kick in
3768 completely until debugging information, and then inlining, are
3769 straightened out in some future version.)
3770 * minor incompatible changes:
3771 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3772 DIRECTORY no longer implicitly promotes NIL slots of its
3773 pathname argument to :WILD. In particular, when you ask for the
3774 contents of a directory (which you used to be able to do without
3775 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3776 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3777 ** changes in behavior that ANSI explicitly defines to be
3778 implementation dependent:
3779 *** The new compiler-only implementation still conforms with ANSI,
3780 but acts a little different than before. Besides the obvious
3781 changes in performance tradeoffs (that the cost per form passed
3782 to EVAL has gone up, and the cost per form executed by EVAL
3783 has gone down), the behavior of the system changes a little
3784 because there are no longer any interpreted function objects.
3785 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3786 e.g. doing COMPILE on the output of interactive DEFUN is
3788 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3790 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3791 has changed from (:CL) to NIL.
3792 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3793 e.g. #\Newline and #\Tab, don't have names specified in the
3794 ANSI Common Lisp standard, is now based on their ASCII symbolic
3795 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3796 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3797 are no longer used for output.
3798 ** changes in internal implementation constants:
3799 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3800 4 million. (If your application spends a lot of time GCing and
3801 you have a lot of RAM, you might want to experiment with
3802 increasing it even more.)
3803 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3804 However, almost all old code should still continue to work without
3805 immediate update, as SB-C-CALL is now a (deprecated) nickname
3807 ** Old operator names in the style DEF-FOO are now deprecated in
3808 favor of new corresponding names DEFINE-FOO, for consistency with
3809 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3810 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3811 internal symbols, but a few supported extensions like
3812 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3813 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3814 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3815 etc. as you get deeper into recursive calls to the debugger
3816 command loop, instead of the old "5]", "5]]", "5]]]"
3817 sequence. (I was motivated to do this when squabbles between
3818 ILISP and SBCL left me very deeply nested in the debugger. In the
3819 short term, this change will probably provoke more ILISP/SBCL
3820 squabbles, but hopefully it will be an improvement in the long run.)
3821 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3822 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3823 function, since declaiming return types involving aliens is
3824 (1) annoyingly messy to do by hand and (2) vital to efficient
3825 compilation of code which calls such functions.
3826 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3827 longer reexported by the SB-EXT package. They're solely useful
3828 for alien code, so it seems more logical that you should get
3829 them from the SB-ALIEN package, not in SB-EXT.
3830 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3831 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3832 features. Instead, the code that they used to control is always
3833 built into the system.
3834 * many other bug fixes
3835 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3836 advantage of the new EVAL-WHEN stuff and to clean them up in
3837 general, and they are now more ANSI-compliant in a number of
3838 ways. Martin Atzmueller is responsible for a lot of this.
3839 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3841 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3842 *** correct ERROR type for various file operations
3843 *** some fixes for Lisp streams
3844 *** DEFMETHOD syntax checking
3845 *** changing old weird representation of debug information as
3846 strings (which, among their other deficiencies, don't transform
3847 correctly when you rename packages, and don't change their
3848 print representation when you change things like *PACKAGE*
3849 and *PRINT-LENGTH*) to symbols and lists of symbols
3850 He also made several improvements and fixed several bugs in DESCRIBE.
3851 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3853 *** misbehavior of WRITE-STRING/WRITE-LINE
3854 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3855 and several other LOOP problems as well
3856 *** DIRECTORY when similar filenames are present
3857 *** DEFGENERIC with :METHOD options
3858 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3859 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3860 *** argument ordering in FIND with :TEST option
3861 *** mishandled package designator argument in APROPOS-LIST
3862 *** various problems in the backquote readmacro
3863 *** a bug in APROPOS
3864 *** probably some others that I'm not describing very well here,
3865 since the CVS log documents them by reference to sbcl-devel
3866 messages, and the SourceForge archives aren't working well.:-(
3867 ** Dan Barlow improved the Alpha port (and is making progress on the
3868 PPC port, for those of you who think different).
3869 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3870 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3871 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3872 continues to work on the SPARC port (for those of you in a position
3873 to look down upon our little PC-compatible boxes from a great height).
3874 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3875 and uses that copy, rather than the current dynamic value, when
3876 it's trying to decide whether to truncate output. Thus e.g.
3877 (let ((*print-lines* 50))
3878 (pprint-logical-block (stream nil)
3880 (let ((*print-lines* 8))
3881 (print (aref possiblybigthings i) stream)))))
3882 should now truncate the logical block only at 50 lines, instead of
3883 often truncating it at 8 lines, as it did before.
3884 * The doc/cmucl/ directory, containing old CMU CL documentation
3885 from the time of the fork, is no longer part of the base system.
3886 SourceForge has shut down its anonymous FTP service, and with it
3887 my original plan for distributing the old CMU CL documentation
3888 there. For now, if you need these files you can download an old
3889 SBCL source release and extract them from it.
3890 * The fasl file version number changed again, for dozens of reasons,
3891 some of which are apparent above.
3893 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3894 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3895 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3896 patch, so that DISASSEMBLE output is much nicer.
3897 * The code in the SB-PROFILE package now seems reasonably stable.
3898 I still haven't decided what the final interface should look like
3899 (I'd like PROFILE to interact cleanly with TRACE, since both
3900 facilities use function encapsulation) but if you have a need
3901 for profiling now, you can probably use it successfully with
3902 the current CMU-CL-style interface.
3903 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3904 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3905 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3906 link now returns the dangling link itself, and for similar
3907 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3908 link itself. (In these cases the old code signalled an error and
3909 looped endlessly, respectively.) Thus, DIRECTORY now works even
3910 in the presence of dangling and cyclic symbolic links.
3911 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3912 is now a supported extension again, since the consensus on
3913 sbcl-devel was that it can be useful for ordinary development
3914 work, not just for debugging SBCL itself.
3915 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3916 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3917 that currently-defined functions might be redefined later with
3918 different return types.
3919 * Hash tables can be printed readably, as inspired by CMU CL code
3920 of Eric Marsden and SBCL code of Martin Atzmueller.
3921 * better error handling in CLOS method combination, thanks to
3922 Martin Atzmueller porting Pierre Mai's CMU CL patches
3923 * more overflow fixes for >16Mbyte I/O buffers
3924 * A bug in READ has been fixed, so that now a single Ctrl-D
3925 character suffices to cause end-of-file on character streams.
3926 In particular, now you only need one Ctrl-D at the command
3927 line (not two) to exit SBCL.
3928 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3930 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3931 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3932 does the right thing.
3933 * The implementation of some type tests, especially for CONDITION
3934 types, is now tidier and maybe faster, due to CMU CL code
3935 originally by Douglas Crosher, ported by Martin Atzmueller.
3936 * Some math functions have been fixed, and there are new
3937 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3938 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3939 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3940 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3941 likely to be built into sbcl-0.7.0.)
3942 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3943 CL, ported to SBCL by Martin Atzmueller)
3944 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3945 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3946 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3947 * There's a new slam.sh hack to shorten the edit/compile/debug
3948 cycle for low-level changes to SBCL itself, and a new
3949 :SB-AFTER-XC-CORE target feature to control the generation of
3950 the after-xc.core file needed by slam.sh.
3951 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3952 COMPILE-FILE is no longer supported, so that now every function
3953 gets an entry point, so that block compilation looks a little
3954 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3955 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3956 returns the number of bytes consed since the system started,
3957 rather than the number consed since the first time the function
3958 was called. (The new definition parallels ANSI functions like
3959 CL:GET-INTERNAL-RUN-TIME.)
3960 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3961 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3962 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3963 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3964 * Fasl file version numbers are now independent of the target CPU,
3965 since historically most system changes which required version
3966 number changes have affected all CPUs equally. Similarly,
3967 the byte fasl file version is now equal to the ordinary
3970 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3971 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3972 is no longer recognized. I apologize for this, because it was
3973 listed in SB-EXT as a supported extension, but I found that
3974 its existing behavior was poorly specified, as well as incorrectly
3975 specified, and it looked like too much of a mess to straighten it
3976 out. I have enough on my hands trying to get ANSI stuff to work..
3977 * many patches ported from CMU CL by Martin Atzmueller, with
3978 half a dozen bug fixes in pretty-printing and the debugger, and
3979 half a dozen others elsewhere
3980 * fixed bug 13: Floating point infinities are now supported again.
3981 They might still be a little bit flaky, but thanks to bug reports
3982 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3983 as flaky as they were.
3984 * The --noprogrammer command line option is now supported. (Its
3985 behavior is slightly different in detail from what the old man
3986 page claimed it would do, but it's still appropriate under the
3987 same circumstances that the man page talks about.)
3988 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3989 are now supported, and enabled by default. Thus, the compiler can
3990 handle many floating point and complex operations much less
3991 inefficiently. (Thus e.g. you can implement a complex FFT
3993 * The compiler now detects type mismatches between DECLAIM FTYPE
3994 and DEFUN better, and implements CHECK-TYPE more correctly, and
3995 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3996 thanks to patches from Martin Atzmueller.
3997 * various fixes to make the cross-compiler more portable to
3998 ANSI-conforming-but-different cross-compilation hosts (notably
3999 Lispworks for Windows, following bug reports from Arthur Lemmens)
4000 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4001 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4002 fixed thanks to Pierre Mai's CMU CL patches.
4003 * fixes to keep the system from overflowing internal counters when
4004 it tries to use i/o buffers larger than 16M bytes
4005 * fixed bug 45a: Various internal functions required to support
4006 complex special functions have been merged from CMU CL sources.
4007 (When I was first setting up SBCL, I misunderstood a compile-time
4008 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4009 * improved support for type intersection and union, fixing bug 12
4010 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4011 more obscure bugs as well
4012 * some steps toward byte-compiling non-performance-critical
4013 parts of the system, courtesy of patches from Martin Atzmueller
4014 * Christophe Rhodes has made some debian packages of sbcl at
4015 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4016 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4017 stable, but are nonetheless usable. When he's ready, I'd be happy
4018 to add them to the SourceForge "File Releases" section. (And if
4019 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4020 * new fasl file format version number (because of changes in
4021 internal representation of (OR ..) types to accommodate the new
4022 support for (AND ..) types, among other things)
4024 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4025 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4027 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4029 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4030 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4031 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4032 * DESCRIBE now works on structure objects again.
4033 * Most function call argument type mismatches are now handled as
4034 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4035 know whether the function will be redefined before the call is
4036 executed. (The compiler could flag local calls with full WARNINGs,
4037 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4038 it doesn't keep track of enough information to know whether calls
4039 are local in this sense.)
4040 * Compiler output is now more verbose, with messages truncated
4041 later than before. (There should be some supported way for users
4042 to override the default verbosity, but I haven't decided how to
4043 provide it yet, so this behavior is still controlled by the internal
4044 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4045 src/compiler/ir1util.lisp.)
4046 * Fasl file format version numbers have increased again, because
4047 support for the Gray streams extension changes the layout of the
4048 system's STREAM objects.
4049 * The Gray subclassable streams extension now works, thanks to a
4050 patch from Martin Atzmueller.
4051 * The full LOAD-FOREIGN extension (not just the primitive
4052 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4053 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4054 but like most other programs, it defaults to copying the Unix
4055 environment from the original process instead of starting the
4056 new process in an empty environment.
4057 * Extensions which manipulate the Unix environment now support
4058 an :ENVIRONMENT keyword option which doesn't smash case or
4059 do other bad things. The CMU-CL-style :ENV option is retained
4060 for porting convenience.
4061 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4062 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4064 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4066 * A patch from Martin Atzmueller seems to have solved the SIGINT
4067 problem, and as far as we know, signal-handling now works cleanly.
4068 (If you find any new bugs, please report them!)
4069 * The system no longer defaults Lisp source file names to types
4070 ".l", ".cl", or ".lsp", but only to ".lisp".
4071 * The compiler no longer uses special default file extensions for
4072 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4073 seems to expect a single default extension for all compiled code,
4074 and there's no compelling reason to try to stretch the standard
4075 to allow two different extensions.) Instead, byte-compiled files
4076 default to the same extension as native-compiled files.
4077 * Fasl file format version numbers have increased again, because
4078 a rearrangement of internal implementation packages made some
4079 dumped symbols in old fasl files unreadable in new cores.
4080 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4081 many fewer weird special cases.
4082 * Bug #17 (differing COMPILE-FILE behavior between logical and
4083 physical pathnames) has been fixed, and some related misbehavior too,
4084 thanks to a patch from Martin Atzmueller.
4085 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4086 by Tim Moore, ported to SBCL by Martin Atzmueller.
4087 * Martin Atzmueller fixed several filesystem-related problems,
4088 including bug #36, in part by porting CMU CL patches, which were
4089 written in part by Paul Werkowski.
4090 * More compiler warnings in src/runtime/ are gone, thanks to
4091 more patches from Martin Atzmueller.
4092 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4095 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4097 * DESCRIBE now works on CONDITION objects.
4098 * The debugger now handles errors which arise when trying to print
4099 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4101 * The build system now uses an additional file, customize-target-features.lisp,
4102 to allow local modifications to the target *FEATURES* list. (The point of
4103 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4104 debugging features enabled, without having to worry about propagating it
4105 into everyone's system when I do a "cvs update".) When no
4106 customize-target-features.lisp file exists, the target *FEATURES* list
4107 should be constructed the same way as before.
4108 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4109 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4110 warn about behavior which is undefined under the ANSI spec. Note: This
4111 is specified by ANSI, but it's not very popular with programmers.
4112 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4113 macro in the SBCL sources for an example of a workaround which you
4114 might use to make portable ANSI-standard code which does what you want.
4115 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4116 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4117 when it's used as a non-toplevel form. (This is required in order
4118 to implement the DEFCONSTANT-EQX macro.)
4119 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4120 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4121 funny things, due to bugs in the implementation of EVAL-WHEN
4122 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4123 non-toplevel DEFCONSTANTs are uncommon.)
4124 * The core file version number and fasl file version number have been
4125 incremented, because the old noncompliant DEFCONSTANT behavior involved
4126 calling functions which no longer exist, and because I also took the
4127 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4128 * fixed bug 1 (error handling before read-eval-print loop starts), and
4129 redid debugger restarts and related debugger commands somewhat while
4131 ** The QUIT debugger command is gone, since it did something
4132 rather different than the SB-EXT:QUIT command, and since it never
4133 worked properly outside the main toplevel read/eval/print loop.
4134 Invoking the new TOPLEVEL restart provides the same functionality.
4135 ** The GO debugger command is also gone, since you can just invoke
4136 the CONTINUE restart directly instead.
4137 ** The TOP debugger command is also gone, since it's redundant with the
4138 FRAME 0 command, and since it interfered with abbreviations for the
4140 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4142 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4143 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4144 some porting work by Martin Atzmueller)
4145 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4146 with a boatload of patches which he ported to SBCL. Now that those
4148 ** The system tries to make sure that its low-priority messages
4149 are prefixed by semicolons, to help people who like to use
4150 syntax highlighting in their ILISP buffer. (This patch
4151 was originally due to Raymond Toy.)
4152 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4153 patches originally written by Raymond Toy.
4154 ** The compiler understands coercion between single-value and
4155 multiple-VALUES type expressions better, getting rid of some very
4156 weird behavior, thanks to patches originally by Robert MacLachlan
4157 and Douglas Crosher.
4158 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4159 lambda lists, thanks to a patch originally by Pierre Mai.
4160 ** The system no longer bogusly warns about "abbreviated type
4162 ** The compiler gets less confused by inlining and RETURN-FROM,
4163 thanks to some patches originally by Tim Moore.
4164 ** The system no longer hangs when dumping circular lists to fasl
4165 files, thanks to a patch originally from Douglas Crosher.
4166 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4167 undefined function error.
4168 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4169 in structure slot initforms) for now, documented workaround instead:-|
4170 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4171 * fixed bug 5: added stubs for various Gray stream functions called
4172 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4173 installed, at least appropriate type errors are generated
4174 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4175 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4176 consistently in DEFMETHOD forms.
4177 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4178 it doesn't seem to affect SBCL after all
4179 * The C runtime system now builds with better optimization and many
4180 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4182 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4184 * The system is now under CVS at SourceForge (instead of the
4185 CVS repository on my home machine).
4186 * The new signal handling code has been tweaked to treat register
4187 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4188 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4189 downstream, e.g. in debug-int.lisp, has implicit dependencies
4190 on the unsignedness of integer representation of machine words,
4191 and that caused the system to bomb out with infinite regress
4192 when trying to recover from type errors involving signed values,
4193 e.g. (BUTLAST '(1 2 3) -1).)
4194 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4195 (This was one of the bugs Peter Van Eynde reported back in July.)
4196 * The system now uses code inspired by Colin Walters' O(N)
4197 implementation of MAP (from the cmucl-imp@cons.org mailing
4198 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4199 inline the MAP operation, and there is more than one
4200 sequence argument to the MAP call (so that it can't just
4201 do ETYPECASE once and for all based on the type of the
4202 single sequence argument). (The old non-inline implementation
4203 of the general M-argument sequence-of-length-N case required
4204 O(M*N*N) time when any of the sequence arguments were LISTs.)
4205 * The QUIT :UNIX-CODE keyword argument has been renamed to
4206 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4207 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4209 * Raymond Wiker's patches to port dynamic loading from Linux to
4210 FreeBSD have been added.
4211 * The BUGS file is now more nearly up to date, thanks in large part
4212 to Martin Atzmueller's review of it.
4213 * The debugger now flushes standard output streams before it begins
4214 its output ("debugger invoked" and so forth).
4215 * The core version number and fasl file version number have both
4216 been incremented, because of incompatible changes in the layout
4218 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4219 used to not be called for a saved Lisp image.)
4220 * Martin Atzmueller's version of a patch to fix a compiler crash,
4221 as posted on sbcl-devel 13 September 2000, has been installed.
4222 * Instead of installing Martin Atzmueller's patch for the
4223 compiler transform for SUBSEQ, I deleted the compiler transform,
4224 and transforms for some similar consing operations.
4225 * A bug in signal handling which kept TRACE from working on OpenBSD
4227 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4228 away by constant folding
4229 * The system now defines its address space constants in one place
4230 (in the Lisp sources), and propagates them automatically elsewhere
4231 (through GENESIS and the sbcl.h file). Therefore, patching the
4232 address map is less unnecessarily tedious and error-prone. The
4233 Lisp names of address space constants have also been systematized.
4234 * CVS tags like dollar-Header-dollar have been removed from
4235 the sources, because they have never saved me trouble and
4236 they've been source of trouble working with patches and other
4237 diff-related operations.
4238 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4239 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4241 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4243 * The system has been ported to OpenBSD.
4244 * The system now compiles with a simple "sh make.sh" on the systems
4245 that it's supported on. I.e., now you no longer need to tweak
4246 text in the target-features.lisp-expr and symlinks in src/runtime/
4247 by hand, the make.sh takes care of it for you.
4248 * The system is no longer so grossly inefficient when compiling code
4249 involving vectors implemented as general (not simple) vectors (VECTOR T),
4250 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4251 which dares to use the various sequence functions on non-simple
4252 vectors, takes less of a performance hit.
4253 * There is now a primitive type predicate VECTOR-T-P
4254 to test for the (VECTOR T) type, so that e.g.
4255 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4256 can now be compiled with some semblance of efficiency. (The old code
4257 turned the type declaration into a full call to %TYPEP at runtime!)
4258 * AREF on (VECTOR T) is still not fast, since it's still compiled
4259 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4260 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4261 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4262 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4263 without making full calls to SUBTYPEP at runtime.
4264 (Some analogous efficiency issues for non-simple vectors specialized to
4265 element types other than T, or for non-simple multidimensional arrays,
4266 have not been addressed. They could almost certainly be handled the
4267 same way if anyone is motivated to do so.)
4268 * The changes in array handling break binary compatibility, so
4269 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4270 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4271 instead of (VALUES T T).
4272 * By following the instructions that Dan Barlow posted to sbcl-devel
4273 on 2 July 2000, I was able to enable primitive dynamic object
4274 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4275 functionality is not implemented (since it calls ld to resolve
4276 library references automatically, requiring RUN-PROGRAM for its
4277 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4278 try to resolve library references) is now supported.
4279 * The system now flushes the standard output streams when it terminates,
4280 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4281 them at several other probably-convenient times, e.g. in each pass of
4282 the toplevel read-eval-print loop, and after evaluating a form given
4283 as an "--eval" command-line option. (These changes were motivated by a
4284 discussion of stream flushing issues on cmucl-imp in August 2000.)
4285 * The source transform for TYPEP of array types no longer assumes
4286 that an array whose element type is a not-yet-defined type
4287 is implemented as an array of T, but instead punts, so that the
4288 type will be interpreted at runtime.
4289 * There is now some support for cross-compiling in make.sh: each of
4290 the phases of make.sh has its own script. (This should be transparent
4291 to people doing ordinary, non-cross-compile builds.)
4292 * Since my laptop doesn't have hundreds of megabytes of memory like
4293 my desktop machine, I became more motivated to do some items on
4294 my to-do list in order to reduce the size of the system a little:
4295 ** Arrange for various needed-only-at-cold-init things to be
4296 uninterned after cold init. To support this, those things have
4297 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4298 symbols with such names are now uninterned after cold init).
4299 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4300 fasl files for cold load.
4301 ** Remove the old compiler structure pooling code (which used to
4302 be conditional on the target feature :SB-ALLOC) completely.
4303 ** Redo the representation of some data in cold init to be more compact.
4304 (I also looked into supporting byte compiled code at bootstrap time,
4305 which would probably reduce the size of the system a lot, but that
4306 looked too complicated, so I punted for now.)
4307 * The maximum signal nesting depth in the src/runtime/ support code has
4308 been reduced from 4096 to 256. (I don't know any reason for the very
4309 large old value. If the new smaller value turns out to break something,
4310 I'll probably just bump it back up.)
4311 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4313 * Many, many bugs reported by Peter Van Eynde have been added to
4314 the BUGS list; some have even been fixed.
4315 * While enabling dynamic object file loading, I tried to make the
4316 code easier to understand, renaming various functions and variables
4317 with less ambiguous names, and changing some function calling
4318 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4319 * While trying to figure out how to do the OpenBSD port, I tried to
4320 clean up some of the code in src/runtime/. In particular, I dropped
4321 support for non-POSIX signal handling, added various comments,
4322 tweaked the code to reduce the number of compilation warnings, and
4323 renamed some files to increase consistency.
4324 * To support the new automatic configuration functionality in make.sh,
4325 the source file target-features.lisp-expr has been replaced with the
4326 source file base-target-features.lisp-expr and the machine-generated
4327 file local-target-features.lisp-expr.
4328 * fixed a stupid quoting error in make.sh so that using CMU CL
4329 "lisp -batch" as cross-compilation host works again
4331 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4333 * DESCRIBE no longer tries to call itself recursively to describe
4334 bound/fbound values, so that it no longer fails on symbols which are
4335 bound to themselves (like keywords, T, and NIL).
4336 * DESCRIBE now works on generic functions.
4337 * The printer now prints less-screwed-up representations of closures
4338 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4339 * A private symbol is used instead of the :EMPTY keyword previously
4340 used to mark empty slots in hash tables. Thus
4341 (DEFVAR *HT* (MAKE-HASH-TABLE))
4342 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4343 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4344 now does what ANSI says that it should. (You can still get
4345 similar noncompliant behavior if bang on the hash table
4346 implementation with all the symbols you get back from
4347 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4348 This breaks binary compatibility, since tests for equality to
4349 :EMPTY are wired into things like the macroexpansion of
4350 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4352 * There's now a minimal placeholder implementation for CL:STEP,
4353 as required by ANSI.
4354 * An obscure bug in the interaction of the normal compiler, the byte
4355 compiler, inlining, and structure predicates has been patched
4356 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4357 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4359 * Missing ordinary arguments in a macro call are now detected even
4360 when the macro lambda list contains &KEY or &REST.
4361 * The debugger no longer complains about encountering the top of the
4362 stack when you type "FRAME 0" to explicitly instruct it to go to
4363 the top of the stack. And it now prints the frame you request even
4364 if it's the current frame (instead of saying "You are here.").
4365 * As specified by ANSI, the system now always prints keywords
4366 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4368 * The default initial SIZE of HASH-TABLEs is now smaller.
4369 * Type information from CLOS class dispatch is now propagated
4370 into DEFMETHOD bodies, so that e.g.
4371 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4373 is now basically equivalent to
4374 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4375 (DECLARE (TYPE SINGLE-FLOAT X))
4377 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4378 operation, without having to do run-time type dispatch.
4379 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4380 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4381 * Since I don't seem to be making big file reorganizations very often
4382 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4383 is overflowing my ability to conveniently back them up), I've finally
4384 checked the system into CVS. (The CVS repository is on my home system,
4385 not at SourceForge -- putting it on SourceForge might come later.)
4386 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4387 high-level GC-NOTIFY-FOO functions send their output. (There's
4388 still very little control of where low-level verbose GC functions
4389 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4390 less than it used to -- the GC-NOTIFY-FOO functions are now under
4391 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4392 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4393 in only one place in the source code, and propagates it automatically
4394 everywhere that it's needed. Thus e.g. when I bump the version from
4395 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4396 * The C source files now include boilerplate legalese and documentation
4397 at the head of each file (just as the Lisp source files already did).
4398 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4399 to his page will be replaced with a link to his new CLiki service.
4401 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4403 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4404 * The build process now looks for GNU make under the default name "gmake",
4405 instead of "make" as it used to. If GNU make is not available as "gmake"
4406 on your system, you can change this default behavior by setting the
4407 GNUMAKE environment variable.
4408 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4409 can build without error under CMU CL.
4411 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4413 * There is now a partial SBCL user manual (with some new text and some
4414 text cribbed from the CMU CL manual).
4415 * The beginnings of a profiler have been added (starting with the
4416 CMU CL profiler and simplifying and cleaning up). Eventually the
4417 main interface should be through the TRACE macro, but for now,
4418 it's still accessed through vaguely CMU-CL-style functions and macros
4419 exported from the package SB-PROFILE.
4420 * Some problems left over from porting CMU CL to the new
4421 cross-compilation bootstrap process have been cleaned up:
4422 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4423 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4424 part when getting rid of the compiler *BACKEND* stuff.)
4425 ** The value of *NULL-TYPE* was screwed up, because it was
4426 being initialized before the type system knew the final
4427 definition of the 'NULL type. This screwed up several key
4428 optimizations in the compiler, causing inefficiency in all sorts
4429 of places. (I found it because I wanted to understand why
4430 GET-INTERNAL-RUN-TIME was consing.)
4431 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4432 PROCLAIM FTYPE information. Unfortunately this broke binary
4433 compatibility again, since now the forms output by DEFGENERIC
4434 to refer to functions which didn't exist in 0.6.3.
4435 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4436 can use the new (as of 0.6.3) transform for SOME into MAP into
4438 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4439 values to (INTEGER 0 1000000), so that the time code will no longer
4440 occasionally get blown up by Linux returning 1000000 microseconds
4441 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4442 its output conform to the ANSI spec. (Alas, this makes its output
4443 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4444 * A full call to MAP NIL with a single sequence argument no longer conses.
4445 * fixes to problems pointed out by Martin Atzmueller:
4446 * The manual page no longer talks about multiprocessing as though
4447 it were currently supported.
4448 * The ILISP support patches have been removed from the distribution,
4449 because as of version 5.10.1, ILISP now supports SBCL without us
4450 having to maintain patches.
4451 * added a modified version of Raymond Toy's recent CMU CL patch for
4452 EQUALP comparison of HASH-TABLE
4454 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4456 * The system still can't cross-compile itself with
4457 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4458 entails), but at least it can compile more of itself that way
4459 than it used to be able to, and various buglets which were uncovered
4460 by trying to cross-compile itself that way have now been fixed.
4461 * This release breaks binary compatibility again. This time
4462 at least I've incremented the FASL file format version to 2, so that the
4463 problem can be detected reliably instead of just causing weird errors.
4464 * various new style warnings:
4465 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4466 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4467 ** using the lexical binding of a variable named in the *FOO* style
4468 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4469 as specified by ANSI.
4470 * *RANDOM-STATE* is no longer automatically initialized from
4471 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4472 default behavior of the system is to repeat its behavior every time
4473 it's run. If you'd like to change this behavior, you can always
4474 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4475 old convention there was no comparably easy way to get the system to
4476 repeat its behavior every time it was run.
4477 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4478 FTYPE declarations has been removed, in favor of their ANSI
4479 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4480 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4481 the types of their sequence arguments can be determined at compile time.
4482 This is done through a new open code expansion for MAP which eliminates
4483 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4484 argument types can be determined at compile time.
4485 * The optimizer now transforms COERCE into an identity operation when it
4486 can prove that the coerced object is already of the correct type. (This
4487 can be a win for machine generated code, including the output of other
4488 optimization transforms, such as the MAP transform above.)
4489 * Credit information has been moved from source file headers into CREDITS.
4490 * Source file headers have been made more standard.
4491 * The CASE macro now compiles without complaining even when it has
4494 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4496 * (Note that the way that the PCL macroexpansions were rewritten
4497 to accommodate the change in DEFGENERIC below breaks binary
4498 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4499 not run under sbcl-0.6.2. Once we get out of alpha releases,
4500 i.e. hit release 1.0.0, we'll probably try to maintain binary
4501 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4502 and sbcl-1.4.4. Until then, however, it might be fairly common
4503 for maintenance releases to break binary compatibility.)
4504 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4506 * The reporting of unhandled signals has been changed to print some
4507 explanatory text as well as the report form. (Previously only
4508 the report form was printed.)
4509 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4510 it defines, so that the compiler no longer issues undefined function
4511 warnings for compiled-but-not-yet-loaded generic functions.
4512 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4513 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4514 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4515 section of the ANSI spec are supported.
4516 * The "" nickname for the "KEYWORD" package has been removed.
4517 The reader still handles symbol tokens which begin with a package marker
4518 as keywords, but it doesn't expose its mechanism for doing so in the
4519 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4520 * The system now issues STYLE-WARNINGs for contradictory TYPE
4521 proclamations. (Warnings for contradictory FTYPE proclamations would
4522 be nice too, but those can't be done usefully unless the type system
4523 is made smarter about FUNCTION types.)
4524 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4525 systematized, so that "*target-*.lisp is supposed to exist only on the
4526 target and imply that there's a related file which exists on the
4527 host, and *host-*.lisp is supposed to exist only on the host and imply
4528 that there's a related file which exists on the target. This involves a
4529 lot of renaming. Hopefully the acute confusion caused by the renaming
4530 will be justified by the reduction in chronic confusion..
4531 ** runtime-type.lisp -> early-target-type.lisp
4532 ** target-type.lisp -> late-target-type.lisp
4533 ** early-host-format.lisp -> early-format.lisp
4534 ** late-host-format.lisp -> late-format.lisp
4535 ** host-error.lisp -> misc-error.lisp
4536 ** early-error.lisp -> early-target-error.lisp
4537 ** late-error.lisp -> late-target-error.lisp
4538 ** host-defboot.lisp -> early-defboot.lisp
4539 ** code/misc.lisp -> code/target-misc.lisp
4540 ** code/host-misc.lisp -> code/misc.lisp
4541 ** code/numbers.lisp -> code/target-numbers.lisp
4542 ** code/early-numbers.lisp -> numbers.lisp
4543 ** early-host-type.lisp -> early-type.lisp
4544 ** late-host-type.lisp -> late-type.lisp
4545 ** host-typep.lisp -> typep.lisp
4546 ** load.lisp -> target-load.lisp
4547 ** host-load.lisp -> load.lisp
4548 ** host-disassem.lisp -> disassem.lisp
4549 ** host-insts.lisp -> insts.lisp
4550 ** byte-comp.lisp -> target-byte-comp.lisp
4551 ** host-byte-comp.lisp -> byte-comp.lisp
4552 ** host-signal.lisp -> signal.lisp
4553 ** host-defstruct.lisp -> defstruct.lisp
4554 ** late-target-type.lisp -> deftypes-for-target.lisp
4555 Furthermore, several other previously target-only files foo.lisp (e.g.
4556 hash-table.lisp and random.lisp) have been split into a target-and-host
4557 foo.lisp file and a target-only target-foo.lisp file, with their key type
4558 definitions in the target-and-host part, so that the cross-compiler will
4559 know more about target types.
4560 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4561 gone away. In their place are various *BACKEND-FOO* variables
4562 corresponding to the slots of the old structure.
4563 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4564 into the target SBCL has been fixed.
4565 * The chill.lisp system for loading cold code into a running SBCL
4567 * Support for the CMU CL "scavenger hook" extension has been removed.
4568 (It was undocumented and unused in the CMU CL sources that SBCL was
4569 derived from, and stale in sbcl-0.6.1.)
4570 * Various errors in the cross-compiler type system were detected
4571 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4572 (enabling various consistency checks). Many of them were fixed,
4573 but some hard problems remain, so the compiler is back to
4574 running without *TYPE-SYSTEM-INITIALIZED* for now.
4575 * As part of the cross-compiler type system cleanup, I implemented
4576 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4577 * I have started adding UNCROSS calls throughout the type system
4578 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4579 on cross-compiler input files will be able to go away, and various
4581 * CONSTANTP now returns true for quoted forms (as explicitly required
4584 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4586 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4587 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4588 (MAKE-LIST -1). In the longer term, it ought to become true
4589 that declarations are assertions even at SAFETY 1. For now, it's not
4590 quite true even at SAFETY 3, but it's at least more nearly true..
4591 (Note that this change seems to increases the size of the system by
4592 O(5%) and to decrease the speed of the compiler by 20% or more.)
4593 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4594 for the problem of printing dozens of lines of distracting information
4595 about low-level system machinery as part of the top stack frame
4596 on entry to the debugger when an undefined function was called.
4597 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4598 is not reset to COMMON-LISP-USER.
4599 * Compilation of stuff related to dyncount.lisp has been made conditional
4600 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4601 smaller. The various dyncount-related symbols have been moved into
4602 a new "SB-DYNCOUNT" package.
4603 * tty-inspect.lisp has been renamed to inspect.lisp.
4604 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4605 feature has gone away. (When we eventually port to other flavors of
4606 libc and/or Unix, we'll try to make the differences between flavors
4607 invisible at the user level.)
4608 * Various other *FEATURES* tags, and/or their associated conditionals,
4609 have been removed if obsolescent, or given better documentation, or
4610 sometimes given more-mnemonic names.
4612 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4614 * tidied up "make.sh" script
4615 * tidied up system directory structure
4616 * better "clean.sh" behavior
4617 * added doc/FOR-CMUCL-DEVELOPERS
4618 * many many small tweaks to output format, e.g. removing possibly-confusing
4619 trailing #\. character in DESCRIBE-INSTANCE
4620 * (EQUALP #\A 'A) no longer signals an error.
4621 * new hashing code, including EQUALP hashing
4622 * tidied up Lisp initialization and toplevel
4623 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4624 * command line argument processing
4625 * added POSIX-GETENV function to deal with Unix-ish environment variables
4626 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4627 terminating SBCL on EOF
4628 * non-verbose GC by default
4629 * There is no more "sbcl" shell script; the sbcl file is now the C
4630 runtime executable (just like CMU CL).
4631 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4633 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4634 symbol and package information as Lisp native symbol and package objects
4635 instead of strings naming symbols and strings naming packages. This way,
4636 whenever packages are renamed (as in warm init), debug information is
4637 transformed along with everything else.
4638 * tweaked the optimization policy declarations which control the building
4639 of SBCL itself. Now, among other things, the system no longer saves
4640 source location debugging information. (This helps two problems at once
4641 by reducing SBCL size and by keeping SBCL from trying to look for its
4642 sources -- which may not exist -- when reporting errors.)
4643 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4644 debugging and testing purposes
4645 * cleaned up printing, making the printer call PRINT-OBJECT for
4646 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4647 methods, giving nearly-ANSI behavior
4648 * converted almost all special variables to use *FOO* naming convention
4649 * deleted PARSE-TIME functionality, since it can be done portably
4650 * moved some files out of cold init into warm init
4651 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4653 * regularized formatting of source files
4654 * added an install.sh script
4655 * fixed ridiculous memory usage of cross-compiler by making
4656 compiler/alloc.lisp not try to do pooling unless it can hook
4657 itself into the GC of the cross-compilation host. Now the system
4658 builds nicely on my old laptop.
4659 * added :SB-ALLOC in target-features.lisp-expr
4660 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4661 was not implemented)
4662 * re-did condition handling and note reporting in the compiler. Notes
4663 are no longer handled by signalling conditions. Style warnings
4664 and warnings are handled more correctly and reported in such a way
4665 that it's easy to find one or the other in your output (so that you
4666 can e.g. figure out which of many problems caused COMPILE-FILE to
4668 * changed the severity of several compiler warnings from full WARNING
4669 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4670 compiler note reporting so that it doesn't use the condition system
4671 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4672 COMPILE-FILE command)
4673 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4674 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4675 no longer supported, since their implementation was deeply intertwingled
4676 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4677 * removed bogus "support" for compiler macros named (SETF FOO), and
4678 removed the compiler macro for SETF INFO (but only after making a fool
4679 of myself on the cmucl-imp mailing list by posting a bogus patch for
4680 DEFINE-COMPILER-MACRO..)
4681 * Compiled files containing forms which have side effects on the Lisp
4682 reader (such as DEFPACKAGE forms) are now handled more correctly.
4683 (Compiler queuing of top level lambdas has been suppressed by setting
4684 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4685 * deleted various currently-unused source files, e.g. gengc.lisp. They
4686 may be added back at some point e.g. when porting to other architectures,
4687 but until they are it's distracting to distribute them and to try to
4689 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4690 there were so many of them they're just distractions, and UNCROSS is
4691 known to be able to handle the current sources
4692 * moved PROFILE functionality into TRACE, so that it will be clear
4693 how the wrapping and unwrapping of functions when you profile them
4694 interacts with the wrapping and unwrapping of functions when you
4695 trace them. (Actually, the functionality isn't there yet, but at least
4696 the interface specification is there. Hopefully, the functionality will
4697 arrive with some future maintenance release.)
4698 * removed host-oops.lisp
4699 * changed signature of QUIT function to allow UNIX-CODE argument
4700 * fixed READ-SEQUENCE bug
4701 * tweaked verbose GC output so that it looks more like the progress
4702 output that ANSI specifies for functions like LOAD
4703 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4704 * added <http://sbcl.sourceforge.com> to the banner information printed by