1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
3 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
4 unparsing of directory pathnames as files. Analogously,
5 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
6 filename to parse into a directory pathname.
7 * enhancement: RUN-PROGRAM allows unicode arguments and environments
8 to be used (using the default stream external format), and allows
9 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
10 * optimizations: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
11 strings and vectors whose element-type or simplicity is not fully
12 known at compile-time.
13 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
14 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
15 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
17 * bug fix: some sequence functions elided bounds checking when
19 * bug fix: too liberal weakening of union-type checks when SPEED >
21 * bug fix: more bogus fixnum declarations in ROOM implementation
24 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
25 * new feature: MAKE-HASH-TABLE now experimentally accepts a
26 :SYNCHRONIZED argument, which makes the hash-table safe for
27 concurrent accesses (but not iteration.) See also:
28 SB-EXT:WITH-LOCKED-HASH-TABLE, and
29 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
30 * optimization: CONCATENATE on strings is an order of magnitue faster
31 in code compiled with (> SPEED SPACE).
32 * optimization: SUBSEQ is ~50% faster on lists.
33 * bug fix: bug 417 fixed -- source location reporting is now more robust.
34 * bug fix: SUBSEQ on a list will now correctly signal an error if if
35 END is smaller then START.
36 * bug fix: SB-PROFILE will no longer report extra consing for nested
37 calls to profiled functions.
38 * bug fix: ROOM implementation had bogus fixnum declarations which
39 could cause type-errors when calling ROOM on large images.
40 * bug fix: if file compilation is aborted, the partial fasl is now
41 deleted, and COMPILE-FILE returns NIL as the primary value.
42 * bug fix: number of thread safety issues relating to SBCL's internal
43 hash-table usage have been fixed.
44 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
45 be returned from its body when the values were being returned
46 using unknown-values return convection and the W-P-O was wrapped
47 inside an UNWIND-PROTECT.
48 * bug fix: sb-posix should now compile again under Windows, enabling
51 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
52 * incompatible change: hash-table accessor functions are no longer
53 automatically protected by locks. Concurrent accesses on the same hash-table
54 from multiple threads can give inconsistent results or even corrupt the
55 hash-table completely. Multi-threaded applications should do their own
56 locking at the correct granularity. In the current implementation it is
57 still safe to have multiple readers access the same table, but it's not
58 guaranteed that this property will be maintained in future releases.
59 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
60 and will signal an error at runtime.
61 * enhancement: SB-THREAD package now exports a semaphore interface.
62 * enhancement: CONS can now stack-allocate on x86 and
63 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
65 * enhancement: nested lists can now be stack allocated on
66 platforms providing stack allocation support.
67 * enhancement: dynamic-extent support has been extended to support
68 cases where there are multiple possible sources for the stack
70 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
71 if the mutex is uncontested on Linux.
72 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
73 as specified by the CLHS. (thanks to Tobias Rittweiler)
74 * bug fix: NaN comparison now works on x86-64.
75 * bug fix: CLOSE :ABORT T on a stream with pending output now
77 * bug fix: instances of non-standard metaclasses using standard
78 instance structure protocol sometimes missed the slot type checks
80 * bug fix: known functions can be inlined in high-debug code.
82 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
83 * minor incompatible change: the MSI installer on Windows no longer
84 associates .lisp and .fasl files with the installed SBCL.
85 * minor incompatible change: :UNIX is no longer present in *FEATURES*
86 on Windows. (thanks to Luis Oliviera)
87 * new feature: SB-CLTL2 contrib module now implements
88 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
89 * optimization: scavenging weak pointers is now more efficient,
90 requiring O(1) instead of O(N) per weak pointer to identify
91 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
92 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
93 method bodies using constant slot-names when the first argument is
94 a specializer parameter for the method.
95 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
96 names, when the class of the instance is a direct instance
97 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
98 as slow as the constant slot-name case.)
99 * optimization: member type construction is now O(N) instead
101 * optimization: UNION and NUNION are now O(N+M) for large
102 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
104 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
105 now more readable in environments like Slime which display it.
106 (thanks to Tobias C. Rittweiler)
107 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
108 was able to infer the correct argument type for the object on which
109 the CAS operation was being performed.
110 * bug fix: copy propagation interfered with parallel assignment
111 semantics in local calls. (reported by Paul Khuong)
112 * bug fix: the signed modular fixnum shift compiled to wrong code on
113 x86 and x86-64. (spotted by a slight modification to some of
116 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
117 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
118 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
119 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
121 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
122 and generic function now print the object identity for anonymous
123 instances. (thanks to Attila Lendvai)
124 * enhancement: as an extension to MOP, SBCL now supports SETF of
125 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
126 (thanks to Attila Lendvai)
127 * enhancement: sb-cover has an alternate annotation mode which puts
128 the form annotations at the car of the form, instead of annotating
129 it completely. The benefit of this mode is that it shows better
130 which forms were instrumented by the compiler.
131 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
132 by printer control variables. (reported by Dan Corkill)
133 * bug fix: system leaked memory when delayed output was performed by
134 the OS in smaller chunks then expected. (thanks to David Smith)
135 * bug fix: system leaked memory when file streams were not closed
137 * bug fix: large objects written to slow streams that were modified
138 after the write could end up with the modified state written to
139 the underlying file descriptor.
140 * bug fix: multiple threads operating in parallel on the same stream
141 could cause buffer-overflows.
142 * bug fix: source location information is stored correctly
143 (broken since 1.0.6). This bug would generally show up as the
144 Slime debugger higlighting the wrong form when the sldb-show-source
146 * bug fix: Branch forms are again annotated as branches in the sb-cover
148 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
149 on x86-64. (reported by Gregory Vanuxem)
150 * bug fix: EQUALP could return wrong results for structures with raw
151 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
152 word). (reported by Vjacheslav Fyodorov)
153 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
156 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
157 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
158 atomic compare-and-swap operations on threaded platforms.
159 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
160 allows assining a global minimum value to optimization qualities
161 (overriding proclamations and declarations).
162 * enhancement: closed over variables can be stack-allocated on x86
164 * performance bug fix: GETHASH and (SETF GETHASH) are once again
166 * optimization: slot definition lookup is now O(1). This speeds up
167 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
168 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
169 * optimization: ASSOC and MEMBER can now be open-coded for all
170 combinations of keyword arguments when second argument is constant
171 and SPEED >= SPACE. In other cases a specialized version is
173 * bug fix: using obsoleted structure instances with TYPEP and
174 generic functions now signals a sensible error.
175 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
176 (reported by Kristoffer Kvello)
177 * bug fix: backtrace construction is now more careful when making
178 lisp-objects from pointers on the stack, to avoid creating bogus
179 objects that can be seen by the GC.
180 * bug fix: defaulting of values in contexts expecting more than 7
181 variables now works on x86-64. (reported by Christopher Laux)
182 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
184 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
185 as the property-list of a symbol.
186 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
187 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
188 situations requested, are once again file-compileable. (reported
191 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
192 * MOP improvement: support for user-defined subclasses of
193 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
194 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
195 the expansion of DEFMETHOD, is responsible for generating a form
196 which creates a list of specializers when evaluated. Additional
197 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
198 debugging and introspective support.
199 * minor incompatible change: the (unsupported) spinlock interface
200 has changed: free spinlock now has the value NIL, and a held spinlock
201 has the owning thread as its value.
202 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
203 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
205 * enhancement: name of a socket-stream is now "a socket" instead of
207 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
208 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
209 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
210 for generic functions. (Thanks to Tobias C. Rittweiler)
211 * optimization: bignum printing speed has been improved by 20-40%
212 (depending on the bignum size.)
213 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
215 * bug fix: the cache used by the CLOS to store precomputed effective
216 methods, slot offsets, and constant return values is now thread and
218 * bug fix: generic function dispatch function updating is now thread
219 and interrupt safe (in the sense that the known issues have been
221 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
222 * bug fix: interning EQL-specializers is now thread and interrupt safe.
223 * bug fix: asdf systems with dependencies to the SB-POSIX or
224 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
225 * bug fix: interrupt safety of applicable method computation has been
228 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
229 * new contrib: sb-cover, an experimental code coverage tool, is included
231 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
232 significantly faster.
233 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
234 produce more efficient code on x86-64 if the compiler can determine
235 that (+ INDEX <constant>) does not require a bounds check and FOO
236 has an element type at least 8 bits wide.
237 * enhancement: a new, experimental synchronous timeout facility is
238 provided. Refer to SB-SYS:WITH-DEADLINE for details.
239 * enhancement: when a symbol name conflict error arises, the
240 conflicting symbols are always printed with a package prefix.
241 (thanks to Kevin Reid)
242 * enhancement: stepping is now once again supported on the SPARC and
243 MIPS platforms. (It is also now more likely to work on CheneyGC
245 * enhancement: sb-sprof can now also track and report accurate call
247 * bug fixes: the treatment of non-standard subclasses of
248 SB-MOP:SPECIALIZER is more correct.
249 * incompatible change: PURIFY no longer copies the data from the
250 dynamic space into the static and read-only spaces on platforms
251 that use the generational garbage collector
252 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
254 * bug fix: GC race condition occasionally resulting in crashes with
255 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
257 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
259 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
260 a STYLE-WARNING to be signalled (regression from 1.0.4.)
261 * bug fix: an asynchronous interrupt could previously leave the
262 system running with GC inhibited.
263 * bug fix: a DECLARE form evaluated at top-level now causes an error
264 rather than silently (or verbosely) returning NIL.
265 * bug fix: trying to dissassemble functions compiled at high DEBUG could
266 result in a error being signalled due to source form lookup errors.
267 (reported by Peter Graves)
269 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
270 * incompatible change: removed writer methods for host-ent-name,
271 host-ent-addresses -- changing the values did not update the DNS
273 * minor incompatible change: changed experimental JOIN-THREAD interface
274 * documentation: the manual now lists reader and writer methods
275 in class slot documentation sections. (thanks to Richard M Kreuter)
276 * documentation: unwinding from asyncronous events has been
277 documented as unsafe.
278 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
279 in multithreaded application code.
280 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
281 platforms. (thanks to James Anderson for the optimization hint)
282 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
284 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
285 variants no longer cons.
286 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
287 their NOT- variants no longer cons.
288 * optimization: EQUAL hash tables no longer use SXHASH for objects
289 of all data types, but instead use an EQL hash for types for which
290 EQUAL is the same as EQL.
291 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
292 are significantly faster.
293 * optimization: new STRING-OUTPUT-STREAM implementation is much
294 faster for large outputs and conses 30% less on average.
295 * enhancement: XREF information is now collected for references made
296 to global variables using SYMBOL-VALUE and a constant argument.
297 * enhancement: SIGINT now causes a specific condition
298 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
299 * bug fix: FILE-POSITION used to signal an error in some cases where
300 ANSI requires it to return NIL.
301 * bug fix: ADJUST-ARRAY is now interrupt-safe.
302 * bug fix: adding and removing fd-handlers is now interrupt-safe.
303 * bug fix: inlined calls to C now ensure 16byte stack alignment on
305 * bug fix: bad type declaration in the CLOS implementation has
306 been fixed. (reported by James Anderson)
307 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
308 (thanks to Lutz Euler)
309 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
310 required by the CLHS. (reported by Samium Gromoff)
311 * bug fix: dead unbound variable references now signal an error.
312 (reported by Marco Monteiro)
313 * bug fix: / with an unused value was being deleted in safe code.
314 (thanks to Marco Monteiro and Kevin Reid)
315 * bug fix: number of characters that can be written onto a single
316 line in a file is unlimited.
317 * bug fix: some GC deadlocks caused by asynchronous interrupts have
318 been fixed by inhibiting interrupts for when GC is disbled.
319 * bug fix: some interrupt safety issues with GETHASH, (SETF
320 GETHASH), CLRHASH and REMHASH have been fixed.
321 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
322 specifier no longer causes infinite recursion.
323 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
324 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
325 * bug fix: an &environment argument with the correct variable information
326 is passed to macros that are expanded during byte compilation.
327 (reported by Samium Gromoff)
328 * bug fix: a logic error was causing the hash function for bignums to
329 have excessive amounts of collisions. (reported by Faré Rideau)
330 * bug fix: modifying the contents of an array could change the return
331 value of SXHASH on that array, which is only allowed for strings
332 and bit vectors. (bug introduced in 0.9.16)
333 * improvement: the x86-64/darwin port now passes all tests (except
334 for the debugger tests) but should still be considered
335 experimental until this is fixed.
336 * improvement: a style-warning is signaled for CASE (etc) clauses with
337 duplicate keys. (thanks to Kevin Reid)
338 * improvement: macos/x86 and macos/x86-64 now use mach exception
339 handlers for dealing with illegal instructions (for trapping and
340 error handling) and memory protection violations (for GC).
342 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
343 * new platform: experimental support for x86-64/darwin (MacOS).
344 * incompatible change: the thread-safe (on most platforms) getaddrinfo
345 and getnameinfo sockets functions are used instead of gethostbyaddr
346 and gethostbyname, on platforms where the newer functions are available.
347 As a result, the ALIASES field of HOST-ENT will always be NIL on these
349 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
350 don't print the startup banner, but behave as if --noinform was passed
351 as a command line argument. (thanks to Kevin Reid)
352 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
353 * optimization: code using alien values with undeclared types is much faster.
354 * optimization: the compiler is now able to open code SEARCH in more cases.
355 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
356 * bug fix: using standardized COMMON-LISP special variables as loop
357 variables no longer signals bogus package lock violations. (reported
359 * bug fix: declaring local loop variables to be of a range-limited type
360 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
361 (reported by Andras Simon)
362 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
363 bugs remain on x86-64.)
364 * bug fix: the #= and ## reader macros now interact reasonably with
365 funcallable instances.
366 * bug fix: type-checks for function arguments were compiled using the
367 compiler policy settings of the wrong lexical environment (in 1.0.2
369 * bug fix: SHADOW accepts characters as string designators, as required
370 by the spec (thanks to Eric Marsden)
371 * bug fix: fixed GC safety issues when foreign functions are called with
372 non-base strings as arguments
373 * bug fix: more consistent error output for fatal-compiler-warnings, like
375 * bug fix: functions with proper names were showing up as NIL in
376 backtraces (reported by Edi Weitz, regression in 1.0.2)
378 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
379 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
380 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
381 produce more efficient code on the x86 if the compiler can determine
382 that (+ INDEX <constant>) does not require a bounds check and FOO
383 has an element type at least 8 bits wide.
384 * bug fix: references to undefined variables are handled the same way
385 in toplevel forms as in the normal compiler
386 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
388 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
389 (thanks to Jon Buller)
390 * improvement: faster compilation times for complex functions
391 * improvement: added readlink support to SB-POSIX (thanks to Richard
394 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
395 * improvement: experimental support for mach exception handling on
396 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
398 * improvement: support for GBK external format.
399 (thanks to Chun Tian (binghe))
400 * improvement: the debugger now displays variables that have been closed
401 over, in code compiled with (DEBUG 2) or higher
402 * improvement: support for executable cores on NetBSD (thanks to
404 * new feature: added a RESTART-FRAME debugger command
405 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
406 be used to provide an implementation for FILE-POSITION on Gray streams
407 (thanks to Eric Marsden)
408 * new feature: add syslog, openlog and closelog support to SB-POSIX
409 (thanks to Richard Kreuter)
410 * optimization: the function call overhead in code compiled with
411 a high DEBUG optimization setting is significantly
412 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
413 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
414 (thanks to Tony Martinez)
415 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
416 evaluated once (reported by Kevin Reid)
417 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
418 works again (thanks to Kilian Sprotte)
419 * bug fix: an error is signaled for tagbodies with duplicate tags
420 (thanks to Stephen Wilson)
421 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
422 * bug fix: Win32 port can now handle foreign code unwinding Lisp
423 stack frames from alien callbacks.
424 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
425 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
426 65k pages (thanks to David Woodhouse)
427 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
429 changes in sbcl-1.0.1 relative to sbcl-1.0:
430 * new platform: FreeBSD/x86-64, including support for threading.
431 * new feature: the compiler stores cross-referencing information
432 abount function calls (who-calls), macroexpansion (who-macroexpands)
433 and special variables (who-binds, who-sets, who-references) for code
434 compiled with (< SPACE 3). This information is available through the
435 sb-introspect contrib.
436 * new feature: users may subclass SEQUENCE, and have instances of
437 these classes interoperate with standard Common Lisp functions if
438 a number of methods are defined. (This feature is experimental
439 and the interface subject to change based on feedback from SBCL
440 users and the general community)
441 * improvement: sb-sprof traces call stacks to an arbitrary depth on
442 x86/x86-64, rather than the previous fixed depth of 8
443 * improvement: another pthread back-end of mutex "pthread-futex".
444 * bug fix: non-ascii command-line arguments are processed correctly
445 (thanks to Yaroslav Kavenchuk)
446 * bug fix: non-required arguments were not passed correctly when a method
447 defined using DEFMETHOD was called from a mop-generated method using
448 CALL-NEXT-METHOD (reported by Pascal Costanza)
449 * bug fix: recursion is now permitted in accessors through
450 SLOT-UNBOUND. (reported by Pascal Costanza)
451 * bug fix: an error was signaled at startup if the HOME environment
452 variable was defined, but had an empty value (reported by Peter Van Eynde)
453 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
454 signal a PROGRAM-ERROR, not a BUG.
455 * bug fix: SB-POSIX:READDIR works when built with large file support.
456 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
458 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
459 for better automatic indentation support. (thanks to Matt Pillsbury)
460 * optimization: loading generic functions no longer takes O(n^2) time,
461 proportional to the amount of methods in the generic function
462 (reported by Todd Sabin and Jeremy Brown)
463 * optimization: the FIND and POSITION family of sequence functions
464 are significantly faster on arrays whose element types have been
466 * improvements to the Windows port:
467 ** Intermittent heap corruption problems have been fixed. (thanks
468 to Alastair Bridgewater)
469 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
471 ** Lisp is able to unwind foreign exception frames from alien
472 callbacks. (thanks to Alastair Bridgewater)
474 changes in sbcl-1.0 relative to sbcl-0.9.18:
475 * improvement: experimental support for threading on FreeBSD/x86.
476 (thanks to NIIMI Satoshi)
477 * improvement: runtime option --dynamic-space-size can be used
478 to set the size of the dynamic space reserved on startup.
479 * improvement: floating point modes in effect are now saved in
480 core, and restored on startup.
481 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
482 startup, not time since first call to GET-INTERNAL-REAL-TIME.
483 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
484 threads are not running after *SAVE-HOOKS* have run.
485 * improvement: writes to CLOS instance slots are type-checked in code
486 compiled with (SAFETY 3)
487 * improvement: floating-point exception handling on FreeBSD (thanks to
489 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
490 (thanks to Zach Beane)
491 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
493 * improvement: added support for the Shift-JIS external format.
494 (contributed by NIIMI Satoshi)
495 * improvement: callbacks are supported on Linux/PPC. (thanks to
497 * bug fix: compiler bug triggered by a (non-standard) VALUES
498 declaration in a LET* was fixed. (reported by Kaersten Poeck)
499 * bug fix: file compiler no longer confuses validated and already
500 dumped structurres. (reported by Kaersten Poeck)
501 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
502 fill-pointer signals a type-error as required. (thanks to
504 * bug fix: disassemly of funcallable instances works.
505 * bug fix: single stepping on PPC.
506 * bug fix: fix thread-safety problems in the type system (generally
507 manifesting as nonsensical errors like "STRING is a bad type specifier
508 for sequences" or "The value 1 is not of type FIXNUM")
509 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
510 (reported by Josip Gracin)
511 * bug fix: an error is signaled for attempts to displace arrays with
512 incompatible element types (thanks to Mario Mommer)
513 * bug fix: more correct handling of wide characters in debug info
514 (bug reported by Attila Lendvai and fixed by Juho Snellman)
515 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
516 and don't cause extra consing
517 * optimization: MAP and MAP-INTO are significantly faster on vectors
518 whose elements types have been declared.
519 * Improvements to SB-SPROF:
520 ** Support for allocation profiling
521 ** Reduced profiling overhead, especially for long profiling runs
522 * Improvements to the Windows port:
523 ** floating point exceptions are now reported correctly.
524 ** stack exhaustion detection works partially.
525 ** more accurate GET-INTERNAL-REAL-TIME.
526 ** better RUN-PROGRAM behaviour in case of errors in the child
528 ** PROBE-FILE now simplifies pathnames correctly.
529 ** DIRECTORY now works correctly with :WILD-INFERIORS.
530 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
531 1st 1970 (although time timezone data will be unreliable).
532 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
534 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
535 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
536 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
538 * bug fix: two potential GC deadlocks affecting threaded builds.
539 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
540 returns T (reported by Anton Kazennikov)
541 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
542 with non-variable places
543 * bug fix: remove a race condition in the setting of
544 funcallable-instance functions, this should make threaded CLOS
545 code more stable against memory faults.
546 * bug fix: corruption of specials when unbinding is interrupted by an
547 asynchronous unwind (reported by Hannu Koivisto)
548 * improvement: the debugger will now also display local variables that
549 are only used once, for code compiled with a DEBUG optimization quality
552 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
553 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
554 * incompatible change: External-format support for FFI calls. The
555 SB-ALIEN:C-STRING no longer implies an ASCII
556 external-format. Instead the string is subject to external-format
557 conversion. Additionally return values of type SB-ALIEN:C-STRING
558 are by default of type (SIMPLE-ARRAY CHARACTER), not
559 SIMPLE-BASE-STRING. For an alien type that behaves like the old
560 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
561 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
562 * incompatible change: SB-EXT package no longer contains the
563 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
564 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
566 * incompatible change: the single-stepper is no longer available
567 on Alpha, Mips and Sparc platforms.
568 * minor incompatible change: the direct superclasses of
569 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
570 not (STANDARD-OBJECT FUNCTION). This makes the
571 class-precedence-lists of GENERIC-FUNCTION and
572 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
574 * minor incompatible change: the default stream external format on
575 non-unicode SBCL is no longer detected from the locale, but is always
577 * new feature: Add a version of evaluator that uses an interpreter instead
578 of the compiler. EVAL still uses the compiler by default, to switch it
579 to use the interpreter, set the value of the variable
580 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
581 * minor incompatible change: the single-stepper REPL has been merged
582 with the normal debugger (see the "Stepping" heading of the debugger help
583 for more details). The debugger command STEP will no longer switch
584 to the single-stepper REPL.
585 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
586 for a type now works.
587 * bug fix: loading of default sysinit file works. (thanks to Leonid
589 * bug fix: better detection of circularities in the file-compiler.
590 (reported by Marco Monteiro)
591 * bug fix: the CL pathname functions now work with files that have
592 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
593 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
594 systems with Unix98 pty semantics.
595 * bug fix: ASDF-INSTALL will now work with bsd tar.
596 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
598 * bug fix: timers expiring in dead threads no longer cause a
599 type-error (reported by Paul "Nonny Mouse").
600 * bug fix: thanks to more lightweight single-stepper instrumentation,
601 code compiled with (DEBUG 3) will compile and execute significantly faster,
602 and will have more accurate type-inferencing than before
603 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
604 whose bindings are modified
605 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
606 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
607 ** replace dummy implementations of CL:MACHINE-INSTANCE and
608 CL:SOFTWARE-VERSION with proper version
610 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
611 * feature: implemented the READER-METHOD-CLASS and
612 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
613 as specified by AMOP.
614 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
616 * optimization: faster LOGCOUNT implementation on x86 and x86-64
617 (thanks to Lutz Euler)
618 * optimization: hashing of general arrays and vectors has been
619 improved. (reported by Any Fingerhut)
620 * enhancement: SB-INTROSPECT is now able to find definitions of
621 profiled functions. (thanks to Troels Henriksen)
622 * enhancement: compiler-macro expansion applies now to FUNCALL forms
623 as well, allowing compiler-macros for SETF-functions to expand.
624 * enhancement: step-instrumentation no longer wraps known
625 single-value functions with multiple-value context, allowing
626 better type inference.
627 * fixed bug #337: use of MAKE-METHOD in method combination now works
628 even in the presence of user-defined method classes. (reported by
629 Bruno Haible and Pascal Costanza)
630 * fixed bug #339(c): if there are applicable methods not part of any
631 long-form method-combination group, call INVALID-METHOD-ERROR.
632 (reported by Bruno Haible)
633 * fixed bug #361: the :FUNCTION initarg in the protocol for
634 initialization of methods can now be used to override
635 internally-produced optimized functions. (reported by Bruno
637 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
638 system-provided lambda expression no longer cause warnings about
639 unbound #:|pv-table| symbols.
640 * bug fix: improved the handling of type declarations and the
641 detection of violations for keyword arguments with non-constant
643 * bug fix: potentially erronous calls to PATHNAME and
644 MERGE-PATHNAMES were being flushed in some cases.
645 (reported by Richard Kreuter)
646 * bug fix: compiled calls to TYPEP were mishandling obsolete
647 instances. (reported by James Bielman and Attila Lendvai)
648 * bug fix: format strings with ~> without matching ~< no longer
649 trigger an AVER, but signal an understandable error instead.
650 (reported by Antonio Martinez)
651 * bug fix: specifying an output-file without a pathname-type for
652 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
653 of input-file instead of "fasl". (reported by Robert Dodier)
654 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
655 (reported by James Y Knight).
656 * bug fix: compiler-macros-function did not consider the environment
657 argument for shadowing by local functions.
658 * bug fix: compiler-macros expansion was inhibited by local INLINE
660 * bug fix: inline expansions of known functions were subject to
661 step-instrumentation in high debug policies, leading to problems
663 * bug fix: compiler failed to differentiate between different CONS
665 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
666 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
667 * bug fix: #368: incorrect use of expressed vs. upgraded array
669 * bug fix: #306a: more precise unions of array types.
670 * thread-safety improvements:
671 ** CONDITION-WAIT could return early on Linux, if the thread was
672 interrupted and subsequently continued with SIGCONT.
673 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
675 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
676 * added support for the ucs-2 external format. (contributed by Ivan
678 * minor incompatible change: pretty printing of objects of type
679 (cons symbol) is, in the default pprint-dispatch-table, now
680 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
682 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
683 called later in a class's lifetime, possibly as late as when the
684 first instance of the class is created. Previously,
685 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
686 class became finalizeable.
687 * fixed bug: FILE-POSITION sometimes returned inconsistent results
688 for multibyte external-format streams. (thanks to "vbzoli")
689 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
690 with :ALLOCATION :CLASS inherited from superclasses of the
692 * fixed bug: anonymous classes can now be created using the :NAME
693 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
694 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
695 * fixed bug: core-files saved with :EXECUTABLE T can again be
696 executed when SBCL_HOME isn't set. (reported by James Knight)
697 * fixed bug: toplevel LOCALLY forms with declarations could
698 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
699 * fixed bug: printing from several different threads using different
700 values of *print-case* could cause invalid output, due to
701 some internal special variables of the printer not being bound
702 thread-locally (reported by Max Mikhanosha)
703 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
704 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
705 (reported by Pascal Costanza)
706 * fixed bug: FINALIZE-INHERITANCE is called from
707 REINITIALIZE-INSTANCE on classes when the class has previously
708 been finalized, as required by AMOP.
709 * minor code generation optimizations:
710 ** better register allocation in CLOS dispatching functions
711 ** overflow detection when coercing signed bytes to fixnums on x86-64
712 ** is now implemented with one IMUL instruction instead of three shifts
713 ** more efficient bit-vector access on x86 and x86-64
714 ** more efficient access to raw structure slots on x86 and x86-64
715 * fixed some bugs revealed by Paul Dietz' test suite:
716 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
719 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
720 * feature: thread support on Solaris/x86, and experimental thread support
722 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
724 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
725 default method for SLOT-UNBOUND.
726 * minor incompatible change: prevent the user from specializing the
727 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
728 somewhat counter to the intent of the protocol, I (CSR) think, and
729 additionally it just doesn't work in SBCL as currently
730 implemented, thanks to optimizations (that are always valid for
731 the other three SLOT-VALUEish functions, but not for the setter).
732 * minor incompatible change: the :SB-LDB feature is now enabled by
733 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
734 the low-level debugger.
735 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
736 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
737 down to an SB-SYS:FD-STREAM, allowing direct communication in
738 more cases. Temporary files and pipes are still used as a fallback
740 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
741 on the non-reentrant C functions localtime(3) and gmtime(3).
742 * bug fix: global optimization policy was not visible in LOCALLY and
744 * bug fix: class objects can be used as specializers in methods.
745 (reported by Pascal Costanza)
746 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
747 with a type error. (reported by blitz_ on #lisp)
748 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
749 failed with a type error. (reported by Pascal Bourguignon)
750 * bug fix: merging pathnames against defaults with :DIRECTORY
751 starting with '(:RELATIVE :BACK) should preserve the :BACK.
752 (reported by James Y Knight)
753 * bug fix: saving large (>2GB) cores on x86-64 now works
754 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
755 constant index and a (simple-array (signed-byte 32)) array
756 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
757 error (patch by Robert J. Macomber)
758 * feature: TIME now displays the amount of run-time spent in GC
759 * bug fix: The debugger now does a better job of respecting
760 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
761 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
762 workaround for bug 403.)
763 * bug fix: floating point exception handling now works on FreeBSD
764 (thanks to NIIMI Satoshi)
765 * fixed some bugs revealed by Paul Dietz' test suite:
766 ** MISC.641: LET-conversion were not supposed to work in late
768 * improvements to the Win32/x86 port:
769 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
770 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
772 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
773 * new feature: source path information is generated for macro-expansion
774 errors for use in IDE's like Slime (thanks to Helmut Eller)
775 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
776 * bug fix: compiler does not loop forever on an invalid type in
778 * improvement: compilation of most CLOS applications is significantly
780 * optimization: added a limited bytecode compiler for simple toplevel
781 forms, speeding up compilation and FASL loading
782 * bug fix: the statistical profiler now properly distinguishes anonymous
785 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
786 * minor incompatible change: in sbcl-0.9.11 (but not earlier
787 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
788 into subclasses of the finalized class. Now user calls to
789 FINALIZE-INHERITANCE finalize just the one class, and calls by the
790 system return before any subclasses are finalized.
791 * minor incompatible change: The reader no longer ignores errors
792 regarding non-existent packages in #+ and #- feature tests.
793 * new feature: command line options --no-sysinit, --no-userinit to
794 inhibit loading the corresponding init files
795 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
796 for sending data through UDP sockets (thanks to François-René Rideau)
797 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
798 error is signalled instead (thanks to François-René Rideau)
799 * minor incompatible change: Error signalling behaviour of lexical
800 operations violating package locks has changed slightly. Refer to
801 documentation on package locks for details.
802 * enhancement: EVAL can process IF-expressions without resorting to the
804 * bug fix: Some operations involving SIN, COS, or TAN weren't being
805 constant-folded properly due to missing out-of-line functions.
806 (reported by Mika Pihlajamäki)
807 * bug fix: LISTEN sometimes returned T even in cases where no data was
808 immediately available from the stream
809 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
810 were reversed. (reported by Levente Mészáros)
811 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
812 when given a vector displaced to an adjustable array. (reported
814 * bug fix: derivation of float boundaries from numbers outside the
815 appropriate float range (reported by John Wiseman)
816 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
818 * bug fix: precomputing cache entries for generic functions with
819 some subclasses of specializers as yet invalid does not attempt to
820 fill a cache line with a negative offset. (reported by Levente
822 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
823 allow condition class objects as arguments to DOCUMENTATION and
824 (SETF DOCUMENTATION); only find and set documentation for
825 structure names for the STRUCTURE doc-type. (suggested by Gary
827 * improvements to the Win32/x86 port:
828 ** (user-homedir-pathname) and default initialization file
829 locations now know about the user's "Documents and Settings"
830 directory (thanks to Yaroslav Kavenchuk)
831 ** run-program is implemented (thanks to Mike Thomas)
832 ** sockets support (thanks to Timothy Ritchey)
833 ** better backtrace support (thanks to Alastair Bridgewater)
834 ** sb-grovel supported
835 ** asdf-install and sb-posix work somewhat
836 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
837 * improvements to the Solaris/x86 port:
838 ** works on Solaris 11/Solaris Express
839 ** floating-point exception handling support
840 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
841 * fixed some bugs revealed by Paul Dietz' test suite:
842 ** REMOVE-METHOD returns its generic function argument even when
843 the method is not one of the generic functions' methods.
844 ** objects declared as MEMBER types can be admissible arguments to
846 ** printing characters should simply be printed by the FORMAT ~:C
848 ** compiler failure when compiling functions with hairy constant
849 defaults for optional parameters.
850 ** compiler produces wrong code when MAYBE-INLINE-expanding a
851 function, which is already optimized.
853 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
854 * new platform: experimental support for SBCL x86/Darwin, including
855 MacOS X 10.4.5 on Intel.
856 * new feature: Unicode character names are now known to the system
857 (through CHAR-NAME and NAME-CHAR).
858 * new feature: the filesystem location of SBCL's core file is
859 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
860 * minor incompatible change: the contrib modules SB-POSIX and
861 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
862 this change is to make it easier to distribute
863 location-independent binaries.
864 * bug fix: as implied by AMOP, standardized classes no longer have
865 slots named by external symbols of public packages. (reported by
867 * optimization: calling functions via constant symbols -- as in
868 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
869 via the function object as in (FUNCALL #'FOO).
870 * optimization: on x86 and x86-64, the calling convention for the
871 case of dealing with an unknown number of values has been altered
872 to be friendlier to the prediction heuristics implemented,
873 particularly with reference to CALL and RET pairing. (thanks to
874 Alastair Bridgewater)
875 * optimization: on x86, the code for access to thread-local symbol
876 values and binding/unbinding of thread-local symbols is smaller.
877 (thanks to Alastair Bridgewater)
878 * enhancement: CONSTANTP is now able to determine constantness of
879 more complex forms, including calls to constant-foldable standardized
880 functions and some special forms beyond QUOTE.
881 * fixed bug: occasional GC crashes on Solaris/x86
882 * optimization: x86-64 supports stack allocation of results of simple
883 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
884 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
885 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
886 (thanks to James Knight)
887 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
888 by the standard (thanks to Stephen Compall)
890 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
891 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
892 be used for bundling the runtime and the core file into one
893 executable binary. This feature is not currently supported on all SBCL
894 platforms. (thanks to James Bielman and NIIMI Satoshi)
895 * new feature: a generational or ephemeral garbage collector is now
896 the default on the PowerPC platform (both Linux and Darwin). The
897 old Cheney (stop and copy) collector is a build-time option.
898 (thanks to Cyrus Harmon, after Raymond Toy)
899 * minor incompatible change: the method by which SBCL finds its
900 contributed modules has changed; it no longer relies on symbolic
901 links from an $SBCL_HOME/systems directory, but searches directly
902 in the subdirectories of $SBCL_HOME.
903 * enhancement: the dynamic heap size on the Linux/PPC platform is
904 markedly larger, even using the older Cheney garbage collector.
905 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
906 certain complicated string types. (reported by Gary King)
907 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
908 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
909 * fixed bug: DOCUMENTATION on structures no longer signals an
910 error if no structure type of the right name exists. (reported by
912 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
913 applicable non-standard primary or :AROUND method on
914 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
915 no longer cause unbound variable errors. (reported by Kevin Reid)
916 * fixed bug: metacircle resolution in cases where methods have slots
917 added before the slots from STANDARD-METHOD. (reported by Jean
919 * fixed bug: the Power PC assembly code for calling into lisp
920 sometimes computed its return address wrongly (depending on the
921 15th bit of the address, fixed at link time). This used to
922 manifest itself in a segmentation violation while building PCL.
923 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
924 many others over the years)
925 * fixed bug: ROOM no longer reports silly numbers for stack usage if
926 the stack pointer (treated as a signed integer) is negative.
927 (thanks to Peter van Eynde)
929 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
930 * new platform: experimental support for the Windows operating
931 system has been added. (thanks to Alastair Bridgewater)
932 * fixed several bugs in and robustified the PPC FFI (including
933 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
934 * bug fix: allow non-simple string symbol names (reported by Paul
936 * bug fix: interrupt handling on NetBSD (thanks to Richard M
938 * bug fix: saving a core corrupted callbacks on x86/x86-64
939 * bug fix: closed a loophole in metacircularity detection and
940 grounding in the PCL implementation of CLOS.
941 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
942 (reported by tomppa on #lisp)
943 * bug fix: clear the direction flag on Lisp -> C transitions, as
944 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
945 (reported by Andrej Grozin and Hendrik Maryns)
946 * optimization: major improvements to GC efficiency on GENCGC platforms
947 * optimization: faster implementation of EQUAL
948 * optimization: emit more efficient opcodes for some common
949 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
950 * optimization: several other minor code-generation improvements on x86-64
951 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
953 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
954 * minor incompatible change: (SETF CLASS-NAME) and (SETF
955 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
956 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
957 by ANSI as a generic function, and for consistency (SETF
958 GENERIC-FUNCTION-NAME) is treated likewise.
959 * fixed bug #233.b: make constraint propagation notice when a variable
960 value is changed after it is referenced but before it is used
961 * fixed bug #296: no more arbitrary behaviour from filenames with
962 odd characters as --load arguments. (reported by Adam Warner)
963 * fixed bug #390: :CHARACTER-SET pathname components now work as
964 expected. (reported by Tim Daly Jr)
965 * fixed bug #391: complicated :TYPE intersections in slot
966 definitions no longer cause an error in PCL internals.
967 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
968 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
969 returning the number of octets which would be written to the
970 file-stream. (thanks to Robert J. Macomber)
971 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
972 arguments to the internal stream functions. (thanks to David
974 * bug fix: stack exhaustion is now handled on the x86/SunOS
975 platform. (thanks to Daisuke Homma)
976 * bug fix: more accurate ROOM results on GENCGC platforms
977 * optimization: improved type inference for arithmetic-for
978 index variables in LOOP
979 * optimization: faster floating-point SQRT on x86-64
980 * fixed some bugs revealed by Paul Dietz' test suite:
981 ** DOCUMENTATION returns NIL instead of "" for method combinations
982 that don't have a docstring
984 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
985 * minor incompatible change: (SETF CLASS-NAME) and (SETF
986 GENERIC-FUNCTION-NAME) are no longer generic functions, and
987 therefore are not customizeable by user code (as seems to be at
988 least permitted and maybe required by AMOP). As a consolation,
989 however, the SBCL implementation of these functions now calls
990 REINITIALIZE-INSTANCE as specified by AMOP.
991 * bug fix: slot-definition documentation is propagated as per ANSI
992 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
993 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
994 argument, as do eslotd initialization methods. (from Pascal
995 Costanza's "Closer" project)
996 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
997 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
999 * bug fix: it is now possible to have more than one subclass of
1000 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1001 (reported by Bruno Haible, Pascal Costanza and others)
1002 * bug fix: the dependent update protocol now works for generic
1003 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1004 and Pascal Costanza)
1005 * bug fix: condition-class instances corresponding to
1006 DEFINE-CONDITION forms are now created eagerly. (reported by
1007 Kalle Olavi Niemitalo on comp.lang.lisp)
1008 * bug fix: floating point printing is more accurate in some
1009 circumstances. (thanks to Simon Alexander)
1010 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1011 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1012 * bug fix: callbacks on OS X now preserve stack-alignment.
1013 (thanks to Cyrus Harmon)
1014 * enhancement: the x86-64 disassembler is much better at
1015 disassembling SSE instructions. (thanks to Lutz Euler)
1016 * enhancement: CHECK-TYPE on a variable now allows type-inference
1017 (thanks to Vincent Arkesteijn)
1018 * optimization: improved performance of EUC-JP external format.
1019 (thanks to NIIMI Satoshi)
1020 * optimization: performance improvements to IO on file streams of
1021 :ELEMENT-TYPE CHARACTER
1022 * optimization: much faster memory allocation on x86-64
1023 * optimization: faster garbage collections (latency and throughput)
1025 * optimization: faster bignum subtraction and division on x86 and x86-64
1026 (thanks to Lutz Euler)
1027 * optimization: more accurate type-inference for loops that use a
1028 floating point index variable or a negative step.
1030 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1031 * bug fix: add a workaround to SBCL looping infinitely at startup on
1032 Linux kernels with apparently buggy implementations of personality().
1033 (thanks to Svein Ove Aas)
1034 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1035 (thanks to David Lichteblau)
1036 * bug fix: local bindings shadow global symbol macros for
1037 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1038 * optimization: non-open coded uses of numeric comparison operators
1039 (e.g. >) no longer cons when called with more than one parameter
1040 on platforms supporting dynamic-extent allocation.
1041 * enhancement: saving cores with foreign code loaded is now supported
1042 on MIPS/Linux in addition to the previously supported platforms.
1043 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1044 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1045 * bug fix: division by zero in sb-sprof when no samples were collected
1046 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1047 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1048 (thanks to Alexander Kjeldaas)
1049 * new feature: ignoring the return values of destructive functions
1050 like NREVERSE signals a compile-time style-warning.
1051 (thanks to Kevin Reid)
1052 * new feature: the alignment of alien structure fields can be
1053 explicitly specified. (thanks to Cyrus Harmon)
1054 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1055 and dump core on SIGQUIT
1057 ** incompatible change: threads do not inherit values of specials
1058 from their parents (see manual)
1059 ** bug fix: threads stacks belonging to dead threads are freed by the
1060 next exiting thread, no need to gc to collect thread stacks anymore
1061 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1062 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1063 ** bug fix: dynamic variable and thread start related gc lossage
1064 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1066 * fixed some bugs revealed by Paul Dietz' test suite:
1067 ** SUBTYPEP is slightly more accurate on heinously complicated
1068 CONS types where some of the members have uncertain (in the
1069 NIL, NIL sense) type relationships to each other.
1071 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1072 no longer purified unless :PURIFY T is explicitly specified.
1073 ** Non-purified cores are significantly smaller than before
1075 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1076 * new feature: timers based on Zach Beane's excellent timer package
1077 * added support for the following external formats: koi8-u,
1078 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1079 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1080 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1081 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1082 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1083 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1084 * incompatible change: a threaded SBCL will no longer revert to
1085 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1086 * bug fix: interrupts are disabled until startup is complete; no
1087 more sigsegvs when receiving a signal to soon
1088 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1090 * bug fix: add a workaround for the memory randomization features in
1091 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1092 workaround will only be in effect on systems with the proc filesystem
1094 * bug fix: printing objects of type HASH-TABLE signals a
1095 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1097 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1098 have been running for over 50 days. (reported by Gilbert Baumann)
1099 * bug fix: the logic for getting names of functions gets less
1100 confused when confronded with alternate-metaclass
1101 funcallable-instances. (reported by Cyrus Harmon)
1102 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1103 consistent, even on internal alternate-metaclass objects.
1104 * bug fix: flush closure information collected by physical
1105 environment analysis prepass before the main pass. (bug reported
1107 * bug fix: compiler pack phase does not modify a hashtable, which is
1108 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1109 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1110 may be expanded. (reported by Alexander <malishev> on c.l.l)
1111 * bug fix: no more style-warnings when compiling code using LAST
1112 under high-SPEED low-SPACE settings. (reported by David Wallin
1114 * fixed some incompatibilities between SBCL's MOP and the MOP
1116 ** the METAOBJECT class is now implemented;
1117 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1118 STANDARD-OBJECT, as required;
1119 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1120 now compatible (as required by VALIDATE-SUPERCLASS); there
1121 remains a consistency requirement over the presence or absence
1122 of the FUNCTION class in the superclass list of finalized
1123 classes; see the manual for more details;
1124 ** the :AROUND method for COMPUTE-SLOTS on
1125 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1126 requested slot ordering.
1128 ** bug fix: parent thread now can be gc'ed even with a live
1130 ** bug fix: binding a special with PROGV to no value is not
1131 the same as that symbol not having been bound (thanks to
1133 ** bug fix: binding specials is thread safe (thanks to
1135 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1136 and SB-SPROF do not die with 'no handler for signal XX in
1137 interrupt_handle_now(..)' anymore
1138 ** bug fix: WITH-TIMEOUT works with multiple running threads
1139 * fixed some bugs revealed by Paul Dietz' test suite:
1140 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1141 the :method-class keyword argument.
1143 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1144 * new port: the Solaris operating system on x86 processors is now
1145 mostly supported, though some rough edges in the environment
1146 remain. (thanks to Daisuke Homma)
1147 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1148 space for its heap. (thanks to Thiemo Seufer)
1149 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1150 closures. (thanks to Thiemo Seufer)
1151 * minor incompatible change: eof selects abort in the debugger.
1152 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1153 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1155 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1156 deprecated, and will go away in a future revision of SBCL.
1157 * minor incompatible change: GC-ON and GC-OFF are no longer
1158 implemented with a counter, it does not matter how many times gc
1159 is switched on or off
1160 * bug fix: discriminating functions for generic function classes
1161 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1162 make invalid assumptions about method precedence order. (reported
1164 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1165 platforms. (thanks to Thiemo Seufer)
1166 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1167 * bug fix: correct pathname printing with printer escaping is on.
1168 (thanks to Kevin Reid)
1169 * bug fix: complex VOP definitions in "user-space" no longer trigger
1170 package locks. (reported by Zach Beane)
1171 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1172 for funcallable-instance functions, and is no different from
1174 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1175 characters in the current readtable. (reported by Nicholas Neuss)
1176 * bug fix: SUBTYPEP on various CONS types returns more a more
1177 accurate acknowledgment of its certainty.
1178 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1179 lists in some cases. This partially fixes bug 384.
1180 * flush all standard streams before prompting in the REPL and the
1182 * bug fix: signal handling and triggering gc do not conflict
1183 directly anymore, in particular a high frequency sb-sprof does
1184 not prevent gc from running
1185 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1186 approximation for timezone and DST information between the
1187 universal time epoch and the smallest negative 32-bit time_t.
1188 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1189 year 1899 so long as the encoded time is non-negative.
1190 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1191 the correct answer for non-constant rotate amounts when performing
1192 an inline 32-bit rotation.
1194 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1195 there is only one thread in the session
1196 ** bug fix: memory leak for streams created in one thread and
1197 written to in another
1198 ** bug fix: lockup when compiled with gcc4
1199 ** bug fix: race that allows the gc to be triggered when gc is
1201 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1202 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1204 ** bug fix: finished threads are gc'ed properly
1205 * fixed some bugs revealed by Paul Dietz' test suite:
1206 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1207 has arguments with a different set of applicable methods from
1208 the orignal arguments.
1209 ** The type error thrown by MAP now has a correct expected-type
1211 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1212 name a compiled function.
1213 ** Three MISC tests where a large bignum was improperly coerced to
1214 a float that couldn't represent that bignum during type
1215 derivation were fixed.
1216 ** SUBTYPEP can now handle types involving the negation of a
1217 list-form FUNCTION type.
1218 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1219 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1220 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1222 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1223 * New feature: Experimental support for bivalent streams: streams
1224 opened with :element-type :default now allow character and binary
1225 (unsigned-byte 8) I/O
1226 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1227 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1229 * bug fix: no more highly sporadic "couldn't check whether ~S is
1230 readable" when reading a stream and an interrupt hits in the middle
1231 of a select system call
1232 * compiler better recognizes complex arrays (reported by Eduardo
1234 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1235 for variables on SBCL built with linkage-tables. (reported by Luis
1237 * various error reporting improvements.
1238 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1239 (Thanks to Thiemo Seufer)
1240 * bug fix: nested reader invokations work correctly
1241 * bug fix: it is possible to have simultaneous references to foreign
1242 code and foreign data with the same name.
1244 ** added x86-64 support
1245 ** incompatible change: the threading api now works with thread
1246 objects instead of thread ids
1247 ** bug fix: threads are protected from signals and interruption when
1248 starting up or going down
1249 ** bug fix: a race where an exiting thread could lose its stack to gc
1250 ** bug fix: don't halt on infinite error in threads if possible
1251 ** fixed numerous gc deadlocks introduced in the pthread merge
1252 ** bug fix: fixed thread safety issues in read and print
1253 ** bug fix: debugger doesn't hang on session lock if interrupted at
1254 an inappropriate moment
1255 ** bug fix: run-program is now thread safe(r)
1256 ** bug fix: inner with-recursive-lock no longer releases the mutex
1257 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1258 * fixed some bugs revealed by Paul Dietz' test suite:
1259 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1260 EXPECTED-TYPE slots filled.
1261 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1263 ** MAKE-CONDITION accepts classes as type-designators.
1264 ** COMPILE may never return NIL.
1265 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1266 range before calling Unix time functions
1268 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1269 * numerous signal handling fixes to increase stability
1270 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1271 * minor incompatible change: we now correctly canonize default
1272 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1273 as per the MOP, rather than the historical (INITARG INITFUNCTION
1274 INITFORM). (reported by Bruno Haible)
1275 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1276 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1277 for more information.
1278 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1279 pathname is a directory pathname.
1280 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1281 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1283 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1284 :METACLASS argument in addition to a class name. (reported by
1285 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1286 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1287 (reported by Bruno Haible; patch from Gerd Moellmann)
1288 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1290 * bug fix: more cleanups to the floating point exception handling on
1291 x86-64 (thanks to James Knight)
1292 * bug fix: the compiler does not try to constant fold calls of
1293 COERCE and BIT-* functions when they return freshly constructed
1294 objects. (reported by Nikodemus Siivola and Paul Dietz)
1295 * optimization: improved the MIPS versions of generic (in the
1296 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1297 * optimization: direct conversion of (unsigned-byte 32) to floats on
1298 the PowerPC platform.
1299 * optimization: structure instances with raw slots now use less
1300 memory, and probably show better memory locality. (thanks to
1302 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1303 are treated as requests for stack allocation on the x86-64,
1304 Alpha-32, PPC and SPARC platforms.
1305 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1306 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1307 (thanks to Tony Martinez)
1309 ** gcing a dead thread can no longer lead to lockups
1310 ** threads block signals until they are set up properly
1311 ** errno is no longer shared by threads
1312 ** interrupt-thread restores the eflags register on x86
1313 ** fixed some lockups due to gc/thread interaction
1314 * fixed some bugs revealed by Paul Dietz' test suite:
1315 ** invalid dotted lists no longer raise a read error when
1316 *READ-SUPPRESS* is T
1317 ** signal an error if a symbol that names a declaration is used
1318 as the name of a type, or vice versa
1319 ** allow using the (declare (typespec var*)) abbreviation for
1320 (declare (type typespec var*)) with all type specifiers
1321 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1322 ** always return NIL from PPRINT-POP when OBJECT is NIL
1323 ** don't signal errors when pretty-printing malformed LABELS,
1324 FLET or MACROLET forms
1325 ** declarations in a DOLIST body are also in scope for the
1327 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1329 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1332 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1333 * fixed cross-compiler leakages that prevented building a 32-bit
1334 target with a 64-bit host compiler.
1335 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1336 opened with :IF-EXISTS :OVERWRITE.
1337 * fixed bug 281: error for an invalid qualifier in a short-form method
1338 combination method is not signalled until the faulty method is called.
1339 * bug fix: iteration variable type inferrer failed to deal with open
1340 intervals. (reported by Alan Shields)
1341 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1342 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1343 or more runtime options were provided to the sbcl binary.
1344 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1345 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1346 * optimization: there's now a fast-path for fixnum arguments in the
1347 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1348 * optimization: the code generated on x86-64 is more compact thanks
1349 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1350 * bug fix: floating-point exception handling now partly works on
1351 x86-64. (Thanks to James Knight)
1352 * improvement to the MIPS backend from Thiemo Seufer: C-style
1353 64-bit long long arguments and return values to alien functions
1355 * fixed some bugs revealed by Paul Dietz' test suite:
1356 ** the type-error signalled from WARN has a filled-in DATUM slot.
1357 ** the type-error required when a stream is not associated with
1358 a file has the stream as its datum.
1359 ** type-errors on single-floats on x86-64 no longer have
1360 :INVALID-OBJECT as the datum
1361 ** the type-errors signalled for invalid function names now have
1362 a correct expected type
1363 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1364 for structures defined with a :TYPE. Documentation strings for
1365 typed structures are no longer immediately discarded
1366 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1367 broadcast streams and synonym streams. FILE-LENGTH now also works
1368 on broadcast streams.
1370 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1371 * incompatible change: the --noprogrammer option, deprecated since
1372 version 0.7.5, has been removed. Please use the equivalent
1373 --disable-debugger option instead.
1374 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1376 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1377 inoperational for a while now) has been completely removed.
1378 * null lexical environments are now printed as #<NULL-LEXENV>,
1379 significantly reducing the amount of clutter in typical backtraces.
1380 * documentation on weak pointers, finalization, and after GC hooks
1381 has been added to the manual.
1382 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1383 as other specialized array types, is much faster. SUBSEQ and
1384 COPY-SEQ on such arrays have also been sped up.
1385 * optimization: EQL is now more efficient when at least other argument
1386 is known to be of type (OR FIXNUM (NOT NUMBER)).
1387 * fixed bug: the runtime is now less vulnerable to changes in the
1388 size of the SBCL object on OS X, and virtual memory is reserved for
1389 all spaces that need to be at a fixed address.
1390 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1391 * fixed bug: finalizers and after GC hooks that cause consing are now
1393 * fixed bug: compiler error messages and summaries are now printed to
1394 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1395 * fixed inference of the upper bound of an iteration variable.
1396 (reported by Rajat Datta).
1397 * fixed bug 211e: calling local functions with duplicated constant
1398 keyword argument no longer causes a bogus style warning about an
1400 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1401 ftype declaration to be disregarded. (reported by Dave Roberts)
1402 * fixed bug 373: caused by erronous compilation of references to alien
1403 variables in the runtime on ppc/darwin.
1404 * fixed bug 376: CONJUGATE type deriver.
1405 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1406 (reported by Baughn on #lisp)
1407 * a cross-compiler bug on non-x86 platforms has been identified and
1408 fixed. (thanks to Bruno Haible)
1409 * improvements to the MIPS runtime code for increased stability.
1410 (thanks to Thiemo Seufer)
1411 * increased the maximimum compact environment size to allow
1412 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1413 * partial workaround for bug 135: don't name the function we're
1414 calling for hairy cases of EVAL, so as not to accumulate one environment
1415 entry per call to EVAL. (reported by Kevin Reid)
1416 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1417 * optimization: type testing for non-vector arrays should be faster.
1418 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1420 * bug fix: setting 31st element of a bit vector to zero did not work
1422 * bug fix: redefining a class definition which failed due to a
1423 previous accessor / function clash now works (but see BUGS entry
1424 #380 for more problems in this area). (thanks to Zach Beane)
1425 * the long form of DEFINE-METHOD-COMBINATION disables method group
1426 checking when given a single method group with pattern *. (thanks
1428 * on x86 compiler supports stack allocation of results of simple
1429 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1430 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1431 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1432 * fixed some bugs related to Unicode integration:
1433 ** the restarts for recovering from input and output encoding
1434 errors only appear when there is in fact such an error to
1436 * fixed some bugs revealed by Paul Dietz' test suite:
1437 ** MISC.549 and similar: late transformation of unsafe type
1438 assertions into derived types caused unexpected code
1440 ** SCALE-FLOAT type deriver is less wrong.
1441 ** type derivers for EXP, LOG and similar functions compute result
1442 types for complex arguments better.
1443 ** (MISC.563) CONJUGATE type deriver works for very restricted
1445 ** out-of-line type testers for character strings are available.
1446 ** EQUAL compiler transform understands specialness of objects
1448 ** accessing double-floats stored on the stack now works on x86-64.
1449 ** debugger internals could sometimes create invalid lispobjs,
1450 resulting in GC crashes.
1451 ** MISC.548: type check weakening can convert required type into
1453 ** initialization forms for bindings are not in scope of free special
1456 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1457 * incompatible change: thread support for non-NPTL systems has
1458 been removed - locking is buggy and unreliable. A threaded
1459 SBCL build will now warn at startup and refuse to create
1460 new threads, unless futex support is detected in the kernel
1461 * incompatible change: the top level REPL now has only an ABORT
1462 restart associated with it, not TOPLEVEL and ABORT as it used to.
1463 TOP and TOPLEVEL are now available as debugger commands for
1464 returning to the top level.
1465 * incompatible change: forms evaluated in the REPL now use the
1466 global optimization policy.
1467 * incompatible change: user- and system-initialization files are
1468 no longer processed with LOAD, but by READ and EVAL; hence the
1469 global optimization policy, startup package, readtable, etc,
1471 * The .fasl file format number has been incremented because of
1472 various incompatible changes.
1473 * internal entry point details and argument counts no longer appear
1474 in backtraces unless explicitly requested by setting
1475 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1476 * built-in and standard functions no longer have names like "top
1477 level local call to FOO".
1478 * fixed bug 32: functions defined in non-null lexical environments
1479 now have more legible printed representation
1480 * fixed bug 33: functions defined in non-null lexical environemnts
1481 are now more amenable to inspection by INSPECT.
1482 * workaround for bug 354: XEPs no longer appear in backtraces unless
1483 explicitly requested.
1484 * fixed bug: receiving the signal which results from attempting to
1485 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1486 SIGBUS on other *BSDs) on architectures where the C stack is also
1487 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1488 to debugger confusion or wild instability. (reported by Cheuksan
1490 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1491 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1492 (reported by Lutz Euler)
1493 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1494 compiled code. (reported by Paul Dietz)
1495 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1496 specialized parameter to a method to something that is not TYPEP
1497 the specializer is now possible.
1498 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1499 face of package deletion.
1500 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1501 pathname. (reported by David Tolpin on c.l.l)
1502 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1503 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1504 STRING-STREAM and FILE-STREAM, respectively.
1505 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1506 than previously; in addition, its test suite is now run on build.
1507 (thanks to Luke Gorrie)
1508 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1509 * optimization: added a immediate representation for single-floats
1511 * fixed some bugs revealed by Paul Dietz' test suite:
1512 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1513 ** attempting to create a package with a colliding nickname causes
1514 correctable errors to be signalled.
1515 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1516 ** several x86-64 backend bugs related to sign-extension of immediate
1519 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1520 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1521 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1523 * fixed loading of multiply forward-referenced layouts.
1524 (thanks to Cheuksan Wang)
1525 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1526 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1527 (see COMPILE-FILE documentation for details of :PRINT option)
1528 * fixed bugs 19 and 317: fixed-format floating point printing is
1529 more accurate. This also fixes a bug reported by Adam Warner
1530 related to the ~@F format directive.
1531 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1533 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1534 dispatch table if the from-char is a dispatch macro character.
1535 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1536 by the wordsize no longer ignore the last word. (reported by Lutz
1538 * fixed bug in type checking of dynamic-extent variables. (reported
1540 * optimization: sequence traversal functions use their freedom to
1541 coerce function designators to functions.
1542 * optimization: code with many calls to CLOS methods specialized on
1543 CLOS classes has had redundant type checks removed.
1544 * optimization: added declarations to speed up operations that access
1545 the internal character database (for example STRING-UPCASE)
1546 * optimization: comparison operations between floats and sufficiently small
1547 fixnums no longer create extra rationals
1548 * fixed some bugs related to Unicode integration:
1549 ** portions of multibyte characters at the end of buffers for
1550 character-based file input are correctly transferred to the
1551 start of the buffer at the next read.
1552 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1553 passing it through to OPEN.
1554 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1555 argument given to internal calls to OPEN.
1556 ** the built-in comment readers (introduced by character sequences
1557 ";" and "#|") are more forgiving to encoding errors; they will
1558 STYLE-WARN and then attempt to resync the stream at a character
1559 boundary. (thanks to Teemu Kalvas)
1560 * fixed some bugs in the x86-64 port:
1561 ** Negative short int return values from c-calls are sign-extended
1563 ** The stack is aligned to 16-bytes for c-calls, as required by
1564 the ABI. (thanks to Cheuksan Wang)
1565 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1566 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1567 for enums. (thanks to Vincent Arkesteijn)
1568 ** Multiple small optimizations and bugfixes for floating point
1570 * fixed some bugs revealed by Paul Dietz' test suite:
1571 ** Space, Tab, Linefeed, Return and Page have the invalid
1572 secondary constituent character trait.
1573 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1575 ** WITH-INPUT-FROM-STRING should only update the index place on
1577 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1578 works more reliably.
1579 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1580 with one-dimensional arrays with fill pointers.
1581 ** TYPE-OF failed on a complex with an integer realpart and a
1583 ** compiler failure during type inference for the code of form
1584 (IF (EQL X (THE ...)) ...) (MISC.535).
1586 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1587 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1588 platform. The port supports 61-bit fixnums, large memory spaces
1589 and reloading shared object files.
1590 * enhancement: saving cores with foreign code loaded is now
1591 supported on ppc/Darwin in addition to the previously supported
1593 * enhancement: the statistical profiler now walks deeper into the
1594 call stack for more meaningful call-graphs and accrued time
1595 reports (x86/x86-64 only). It also now reports time spent in
1597 * enhancement: it is now possible to trace most individual methods
1598 of a generic function in addition to tracing the generic function
1600 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1602 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1603 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1604 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1605 produces an error. (thanks to Vincent Arkesteijn)
1606 * bug fix: NAMESTRING on pathnames with :WILD components in their
1607 directories works correctly. (thanks to Artem V. Andreev)
1608 * fixed bug 125: compiler preserves identity of closures. (reported
1610 * bug fixed: functions with &REST arguments sometimes failed with
1611 "Undefined function" when compiled with (DEBUG 3). (reported by
1613 * bug fix: overflow during compiling of setting element of a bit
1614 vector with constant index and value. (reported by Timmy Douglas)
1615 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1616 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1617 * on x86 compiler partially supports stack allocation of dynamic-extent
1619 * GO and RETURN-FROM do not check the extent of their exit points
1620 when compiled with SAFETY 0.
1621 * fixed some bugs related to Unicode integration:
1622 ** encoding and decoding errors are now much more robustly
1623 handled; it should now be possible to recover even from invalid
1624 input or output to the terminal. (thanks to Teemu Kalvas)
1625 ** provided a first cut at implementing STRING-TO-OCTETS and
1626 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1627 ** altered the SB-MD5 contributed module slightly, changing the
1628 interface just enough for it to be supportable for builds where
1629 lisp characters are not eight bits.
1630 * fixed some bugs revealed by Paul Dietz' test suite:
1631 ** the FORMATTER-generated functions for ~V[ conditionals require
1632 the correct number of arguments.
1633 ** READ-FROM-STRING returns the mandated second value when applied
1634 to displaced strings.
1635 ** the #\Rubout and #\Backspace characters are treated as invalid
1636 constituent characters by the tokenizer.
1638 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1639 * new feature: reloading changed shared object files with
1640 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1641 * new feature: references to foreign variables and functions
1642 can now be compiled and loaded before the corresponding shared
1643 object file is loaded, as long as the foreign definitions are
1644 available at runtime.
1645 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1646 supported. (thanks to Dan Debertin)
1647 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1648 just keywords. (thanks to Vincent Arkesteijn)
1649 * fixed bug #331: structure-class instances corresponding to
1650 DEFSTRUCT forms are now created eagerly.
1651 * fixed bug #345: backtraces from calls to undefined functions work
1652 on x86 as well. Related bug #61 is now also partially fixed on x86
1653 (backtraces from throws to unknown catch tags.)
1654 * bug fix: lambda-list parsing is now stricter vrt. order and number
1655 of lambda-list keywords.
1656 * bug fix: as specified by AMOP, an error is signalled if a
1657 class-option appears multiple times in a DEFCLASS form. (reported
1659 * bug fix: RANDOM can be compiled when the compiler derives the type
1660 of its numeric argument as a disjoint set of small integers.
1661 (reported by Paul Dietz)
1662 * bug fix: starting a core saved with shared objects loaded when
1663 those objects are not available no longer causes threaded SBCL to
1664 hang. (reported by Sean Ross)
1665 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1667 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1668 argument, which must always be NIL. (reported by Kalle Niemitalo)
1669 * bug fix: printing 1.0d+23 no longer results in an error.
1670 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1671 * bug fix: structure slot setters preserve evaluation order. (thanks
1673 * fixed some bugs related to Unicode integration:
1674 ** RUN-PROGRAM can allow its child to take input from a Lisp
1675 stream. (reported by Stefan Scholl)
1676 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1677 locales. (reported by Ken Causey)
1678 * fixed some bugs revealed by Paul Dietz' test suite:
1679 ** INCF, DECF and REMF evaluate their place form as specified in
1681 ** FORMATTER expands ~{ ~} iteration directives with V or #
1682 parameters correctly.
1683 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1684 consequent uses no arguments correctly.
1685 ** the system has a partial understanding of the (COMPLEX RATIO)
1687 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1690 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1691 * new feature: a build-time option (controlled by the :SB-UNICODE
1692 keyword feature, enabled by default) for building the system with
1693 support for the entire 21-bit character space defined by the
1695 * new feature: the system now has rudimentary external-format
1696 support; the primary user-visible change at this time is that
1697 characters with the high bit set (such as lower-case-e-acute) will
1698 print correctly to a terminal in a UTF-8 environment.
1699 * minor incompatible change: BASE-CHAR no longer names a class;
1700 however, CHARACTER continues to do so, as required by ANSI.
1701 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1702 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1703 has been moved to the SB-EXT package (temporarily re-exported from
1705 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1706 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1708 * fixed bug #351: better error-handlind and reporting for malformed
1710 * fixed bug #350: bignum-printing is now more memory-efficient,
1711 allowing printing of very large bignums, eg. (expt 2 10000000).
1712 (reported by Bruno Haible)
1713 * fixed bug #302: better primitive-type selection for intersection
1715 * fixed bug #308: non-graphic characters now all have names, as
1716 required. (reported by Bruno Haible)
1717 * bug fix: redefining a class with different superclasses now correctly
1718 removes it from the direct-subclasses of its previous superclasses.
1719 (reported by David Morse)
1720 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1721 new value now works. (reported by Bruno Haible)
1722 * bug fix: correct canonicalization of multiple non-standard slot
1723 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1724 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1725 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1726 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1727 now exists, an signals an error.
1728 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1729 during macroexpansion for non-iteration variables as well. (reported
1730 by Bruno Haible for CMUCL)
1731 * bug fix: Cyclic structures and unprintable objects in compiler
1732 messages no longer cause errors. (reported by Bruno Haible)
1733 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1734 and READ-FROM-STRING all now return a primary value of NIL if
1735 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1736 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1737 (reported by Bruno Haible for CMUCL)
1738 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1739 do not designate a condition. (reported by Bruno Haible for
1741 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1742 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1743 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1744 permitting resolution in favour of any of the conflicting symbols.
1745 (reported by Bruno Haible for CMUCL)
1746 * bug fix: EQUAL compiler optimizations is less aggressive on
1747 strings which can potentially compare true despite having distinct
1748 specialized array element types.
1749 * bug fix: unit enumerations can be defined without dividing by
1750 zero. (thanks to Vincent Arkesteijn)
1751 * FORMAT compile-time argument count checking has been enhanced.
1752 (report from Bruno Haible for CMUCL)
1753 * a partial workaround for the bug 262: the compiler does not try to
1754 inline-expand a local function doing RETURN-FROM from a deleted
1755 BLOCK. (thanks to Peter Denno for the bug report and to David
1756 Wragg for the simple test case)
1757 * fixed some bugs revealed by Paul Dietz' test suite:
1758 ** RENAME-PACKAGE allows all package designators as new package
1760 ** constraint propagation and lambda variable substitution are
1761 more cautious in dealing with partially deleted code.
1762 ** compiler sometimes lost reoptimization passes.
1763 ** CERROR, given a condition as condition designator, passes its
1764 remaining arguments to the continue format control without
1766 ** Case-altering FORMAT directives work correctly on non-ASCII
1768 ** The REAL type specifier handles bounds outside the floating
1769 point ranges without signalling FLOATING-POINT-OVERFLOW.
1770 ** Functions with IR1-transformations can create intercomponent
1771 references to global functions.
1772 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1774 ** FORMAT ~R treats a nil value for its first parameter correctly.
1776 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1777 * enhancement: saving cores with foreign code loaded is now
1778 supported on x86/NetBSD and sparc/Linux in addition to the previously
1779 supported platforms.
1780 * bug fix: on some platforms repeated installations caused multiple
1781 copies of HTML documentation to be installed -- should not happen
1782 any more. (reported by Stefan Scholl)
1783 * bug fix: parsing self-recursive alien record types multiple times
1784 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1785 original patch by Helmut Eller for CMUCL)
1786 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1787 (thanks to Richard Kreuter)
1788 * bug fix: defining classes whose accessors are methods on existing
1789 generic functions in other (locked) packages no longer signals
1790 bogus package lock violations. (reported by François-René Rideau)
1791 * bug fix: special variables as DEFMETHOD parameters no longer have
1792 associated bogus type declarations. (reported by David Wragg and
1794 * bug fix: read-write consistency on streams of element-type
1795 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1796 * bug fix: redefiniton of the only method of a generic function with
1797 no DEFGENERIC no longer emits a full WARNING. In addition,
1798 redefinition of generic functions with no DEFGENERIC to an
1799 incompatible lambda list now signals an error. (thanks to Zach
1801 * bug fix: DEFGENERIC now works even when there's a function of the
1802 same name in an enclosing lexical environment. (thanks to Zach
1804 * fixed compiler failure, caused by instrumenting code during
1805 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1806 * optimization: added loop analysis and improved register allocation
1807 * fixed some bugs revealed by Paul Dietz' test suite:
1808 ** POSITION on displaced vectors with non-zero displacement
1809 returns the right answer.
1810 ** (SIMPLE-STRING) is a valid type specifier for sequence
1812 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1814 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1815 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1817 ** The justification version of the FORMAT ~< directive treats
1818 non-zero minpad parameter correctly.
1820 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1821 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1822 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1823 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1824 the supported interface.
1825 * new feature: Single-stepping of code compiled with DEBUG 2 or
1826 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1827 * new feature: saving cores with foreign code loaded is now
1828 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1829 Timothy Moore's work for CMUCL)
1830 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1831 parameters to * instead of NIL if no initform is supplied.
1832 (reported by Johan Bockgaard)
1833 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1834 correctly when FUNCALL appears as the car of the form. Note:
1835 despite this FUNCALL forms are not currently subject to
1836 compiler-macro expansion. (port of Raymond Toy's fix for the
1837 same from CMUCL, reported by Johan Bockgaard)
1838 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1839 (thanks for Teemu Kalvas)
1840 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1841 case instead of terminating the iteration (thanks for Julian
1842 Squires, Sean Champ and Raymond Toy)
1843 * bug fix: incorrect expansion of defgeneric that caused a style
1844 warning. (thanks for Zach Beane)
1845 * bug fix: slot accessor effective method computation works properly
1846 for classes with multiple non-standard applicable methods on
1847 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1848 * on x86 compiler supports stack allocation of results of LIST and
1849 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1850 CMUCL implementation by Gerd Moellmann)
1851 * fixed some bugs revealed by Paul Dietz' test suite:
1852 ** FORMAT strings with both the ~<~:;~> form of the justification
1853 directive and pretty-printing directives cause an error.
1855 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1856 * incompatible change: the internal functions
1857 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1858 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1859 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1860 instead of the old functions.
1861 * new feature: on platforms where "dladdr" is available foreign
1862 function names now appear in backtraces. (based on Helmut Eller's
1864 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1865 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1867 * documentation: ASDF manual is now distributed with SBCL.
1868 * bug fix: SBCL can now load its contributed modules with REQUIRE
1869 even if the system-provided entries have been removed from
1870 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1872 * bug fix: backtraces involving undefined functions or assembly
1873 routines are more informative. (thanks to Brian Downing)
1874 * bug fix: mutually referent alien structures now work correctly.
1875 (reported by Rick Taube)
1876 * bug fix: structures defined by WITH-ALIEN can be referred to
1877 within other definitions in the same WITH-ALIEN.
1878 * bug fix: division operators (MOD, TRUNCATE and the like) with
1879 constant zero divisors and integer dividends no longer generate
1881 * bug fix: provide default methods for INPUT-STREAM-P and
1882 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1883 * bug fix: improve the detection and resolution of MOP metacycles
1884 (where effective method computation depends on the generic function
1885 having its effective method computed). (reported by Bruno Haible)
1886 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1887 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1888 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1889 * bug fix: pathnames with relative directory components can be
1890 represented relative to default pathnames.
1891 * optimization: in taking the GCD of bignums, reduce the two bignums
1892 to approximately the same size (using Euclid's algorithm) before
1893 applying the more sophisticated binary GCD. (thanks to Juho
1895 * optimization: COUNT on bitvectors now operates word-at-a-time.
1896 * optimization: ASH with a positive, but not necessarily constant,
1897 (leftwards) shift, when in a modular context, is compiled to a
1899 * fixed some bugs revealed by Paul Dietz' test suite:
1900 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1901 if the corresponding argument is NIL.
1902 ** FORMAT directives accepting a minpad parameter treat negative
1904 ** literal commas as character parameters to FORMAT directives are
1906 ** literal spaces directly after ~<Newline> directives within a
1907 format-logical-block (~:< ~@:>) do not induce :FILL-style
1908 conditional newlines.
1909 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1910 as specified: it no longer includes conditional newlines).
1911 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1913 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1914 :PER-LINE-PREFIX argument does not evaluate to a string.
1916 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1917 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1918 the manual for details; for now, package locks can be disabled by
1919 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1920 if no major problems are found then it is likely that they will be
1921 compiled in unconditionally.
1922 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1923 unsupported operators on all platforms. To load a shared library
1924 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1925 object file, link it into a shared library outside of SBCL and
1926 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1928 * bug workaround: ROOM T and the GC were not working together
1929 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1930 reporting facility aren't preserved. That reporting has been
1931 disabled, so now until and unless someone figures out how to make it
1932 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1933 * minor incompatible change: as threatened around sbcl-0.8.0, the
1934 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1935 The symbols are also exported from SB-PCL for backwards
1936 compatibility, but more so than before SB-PCL should be treated as
1937 an implementation-internal package.
1938 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1940 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1941 It is known as of this release not to work on the Alpha, however.
1942 * fixed bug #167: errors signalled due to illegal syntax in method
1943 bodies are now more legible.
1944 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1945 designators and can hence be used with TYPEP.
1946 * fixed bug #333: CHECK-TYPE now ensures that the type error
1947 signalled, if any, has the right object to be accessed by
1948 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1949 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1950 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1952 * fixed bug #334: programmatic addition of slots using specialized
1953 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1954 and :ALLOCATION :CLASS slots.
1955 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1956 float exponents. (rereported by Peter Seibel)
1957 * fixed a bug: #\Space (and other whitespace characters) are no
1958 longer considered to be macro characters in standard syntax by
1959 GET-MACRO-CHARACTER.
1960 * fixed bug: initialization of condition class metaobjects no longer
1961 causes an instance of the condition to be created. (reported by
1963 * fixed bug: it is now possible to have slots such that
1964 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1965 description is neither :INSTANCE nor :CLASS.
1966 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1967 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1968 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1969 non-standard SLOT-DEFINITION-ALLOCATION values.
1970 * fixed bugs: various race conditions handling exiting threads.
1971 CL-PPCRE's thread torture test now passes.
1972 * fixed arguably-a-bug: GC time proportional to number of threads in
1973 system even when most of them are idle
1974 * optimization: improved performance of BIT and SBIT on bit-vectors.
1975 * .fasl file incompatibility: The fasl file version number has
1976 been incremented because of changes associated with package locks.
1978 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1979 * minor incompatible change: the system no longer provides
1980 optimization hints (conditions which are TYPEP
1981 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1982 mode; these hints are still emitted when higher SPEED optimization
1984 * new contrib module: a sampling profiler (profiling by statistical
1985 sampling, rather than by instrumenting functions) is available as
1986 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1987 * the behaviour of the standard function ED is now customizeable by
1988 third parties through a hook variable: see ED's documentation
1989 string for information on the protocol.
1990 * the compiler no longer emits efficiency notes for (FUNCALL X)
1991 when the type of X is uncertain under default optimization
1993 * fixed bug 276: mutating a binding of a specialized parameter to a
1994 method to something that is not TYPEP the specializer is now
1996 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1997 exhibit the required behaviour on the x86 platform. (thanks to
1998 Peter van Eynde, Eric Marsden and Bruno Haible)
1999 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2000 even for difficult arguments. (reported by Peter Graves)
2001 * fixed bug 141a: the backquote printer now descends quoted
2003 * fixed another bug in backquote printing: no more destructive
2004 modification of the form's list structure. (reported by Brian
2006 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2007 move between its address being taken and the call to
2008 interrupt_thread, fixing a crashing race condition.
2009 * the SB-POSIX contrib implementation has been adjusted so that it
2010 no longer exhibits ridiculously poor performance when constructing
2011 instances corresponding to C structs.
2013 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2014 * minor incompatible change: the sb-grovel contrib now treats C
2015 structures as alien (in the sense of SB-ALIEN) objects rather than
2016 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2017 has implications for memory management of client code
2018 (sb-grovel-returned objects must now be manually managed) and for
2019 type safety (alien objects now have full types).
2020 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2021 used to control emission of compiler diagnostics, rather than the
2022 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2023 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2024 quality should be considered deprecated.
2025 * install.sh now installs the user manual as well
2026 * (not quite a new documentable feature, but worth considering in
2027 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2028 beginnings of a semantically meaningful condition hierarchy is
2029 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2031 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2032 designator as the defaults argument.
2033 * fixed bug: Displaced arrays whose displaced-to array has become
2034 too small now cause ARRAY-DIMENSION to signal an error, providing
2035 for safer bounds-checking. (reported by Bruno Haible)
2036 * fixed bug: DEFCLASS slot definitions with identical :READER and
2037 :WRITER names now signal a reasonable error. (reported by Thomas
2039 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2041 * fixed bug: Invalid slot specification errors now print correctly.
2042 (thanks to Zach Beane)
2043 * fixed bug 320: Shared to local slot value transfers in class
2044 redefinitions now happen corrently from superclasses as
2045 well. (reported by Bruno Haible)
2046 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2048 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2049 improper lists correctly. (reported by Bruno Haible)
2050 * fixed bug 313: source-transform for <fun-name> was erroneously
2051 applied to a call of a value of a variable with name <fun-name>.
2052 (reported by Antonio Menezes Leitao)
2053 * fixed bug 307: The obsolete instance protocol ensures that
2054 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2055 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2056 remove dead unknown-values globs from the middle of the stack.
2057 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2058 easier to resume long computations after using *BREAK-ON-SIGNALS*
2059 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2060 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2061 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2062 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2063 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2064 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2066 * fixed a (fairly theoretical) bug in string printing: if
2067 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2068 does not have array-element-type equal to the most general string
2070 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2071 function name. (thanks to Nikodemus Siivola)
2072 * fixed bug: compile-time format string checker failed on
2073 non-closed ~{. (reported by Thomas F Burdick)
2074 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2075 don't warn on legal loop code involving more than one aggregate
2076 boolean. (thanks to Nikodemus Siivola)
2077 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2078 in keyword position to MAKE-PACKAGE was not being reported as
2080 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2081 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2083 * fixed bugs 280 and 312: the checking for multiple definitions in a
2084 file is less likely to become confused by uses of inline
2086 * fixed bug: the #S reader macro performs the keyword coercion
2087 specified for slot names. (reported by Kalle Niemitalo)
2088 * fixed bug: lambda lists may contain symbols whose names start with
2089 & but are not lambda-list-keywords; their occurrence triggers a
2091 * fixed bug 321: define-method-combination argument lambda lists do
2092 not cause invalid code to be generated when &optional or &aux
2093 variables have default values. (reported by Bruno Haible)
2094 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2095 have CLOS classes; this appears also to have fixed reported
2096 instability in stack exhaustion detection.
2097 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2098 MUFFLE-WARNING when no associated restart is present is now
2100 * optimization: rearranged the expansion of various defining macros
2101 so that each expands into only one top-level form in a
2102 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2104 * optimization: used a previously little-used slot in symbols to
2105 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2107 * fixed some bugs revealed by Paul Dietz' test suite:
2108 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2110 ** Fixed an optimization bug related to inheritance of initforms
2111 from local to shared slots.
2112 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2113 ** CAST optimizer forgot to flush argument derived type.
2114 ** print/read consistency on floats is now orders of magnitude
2115 more likely. (thanks also to Bruno Haible for a similar report
2117 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2118 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2119 for objects of type REAL. Make it so.
2120 ** SXHASH is specified (infelicitously) to respect similarity,
2121 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2122 it so. (thanks to Markus Ziegler)
2123 ** on the Alpha, the compiler succeeds in compiling functions
2124 returning a known number of arguments greater than 63.
2125 ** fixed handling of invalid NIL arguments in keyword position
2127 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2129 ** provided a readably-printable representation for RANDOM-STATE
2131 ** ensured that pathnames, where they have a namestring, always
2132 print using #P"..." syntax.
2134 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2135 * Support for the forthcoming 2.0 version of the NetBSD kernel
2136 running on x86 hardware has been added. (thanks to Perry
2137 E. Metzger most immediately, and others for their past work)
2138 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2139 current release version; SBCL's OpenBSD support had been broken
2140 since about the time of OpenBSD's switch to ELF binary format.)
2141 * [placeholder for DX summary]
2142 ** user code with &REST lists declared dynamic-extent, under high
2143 speed or space and low safety and debug optimization policy.
2144 * The manual has been converted to Texinfo format and the debugger
2145 chapter from the cmucl manual has been added.
2146 * A facility has been added to extract documentation strings from
2147 sbcl and store them as Texinfo-formatted snippets for inclusion in
2148 the manual (via Texinfo's @include directive)
2149 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2150 the test case to Dave Roberts)
2151 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2152 pointers. (reported by Sean Ross)
2153 * bug fix: PROFILE output is printed nicely even for large numerical
2154 values. (thanks to Zach Beane)
2155 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2156 greater than 32 handle EOF correctly.
2157 * bug fix: on X86 an immediate argument of the IMUL instruction is
2158 correctly printed in disassembly. (thanks to Lutz Euler)
2159 * bug fix: class slots in redefined classes preserve their old
2160 values. (thanks to Bruno Haible and Nikodemus Siivola)
2161 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2162 succeeds. (reported by Marco Baringer)
2163 * fixed some bugs revealed by Paul Dietz' test suite:
2164 ** READ-SEQUENCE now works on ECHO-STREAMs.
2165 ** RATIONALIZE works more according to its specification. (thanks
2168 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2169 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2170 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2171 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2172 should matter to you only if you rebind the printer control
2173 variables and then find you want different bindings in the
2174 debugger than in the ordinary execution of your program.)
2175 * The runtime build system has been tweaked to support building
2176 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2177 assembler and linker. (thanks to Nikodemus Siivola)
2178 * Unbound, undefined, undeclared variables now trigger full
2179 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2180 more often programmer error than deliberate exploitation of undefined
2182 * optimization: the hash algorithm for strings has changed to one
2183 that is less vulnerable to spurious collisions. (thanks to Juho
2185 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2186 needless bounds checking. (thanks to Juho Snellman)
2187 * optimization: implemented multiplication as a modular
2188 (UNSIGNED-BYTE 32) operation on the PPC backend.
2189 * fixed some bugs revealed by Paul Dietz' test suite:
2190 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2192 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2193 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2194 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2196 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2198 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2200 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2201 their output stream on EOF from read.
2202 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2203 have been read to end-of-file.
2204 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2206 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2207 description of determination of which consecutive characters
2209 ** Printing the "Space" character with escaping on now yields "#\\ ",
2210 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2211 ** Reading floating-point numbers with *READ-BASE* set to a number
2212 less than 10 works correctly.
2213 ** Reading floating-point numbers with *READ-BASE* set to a number
2214 more than 10 works correctly.
2215 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2216 the readtable currently in effect.
2218 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2219 * minor incompatible change: parsing of namestrings on a physical
2220 (Unix) host has changed; numbers after the final #\. in a
2221 namestring are no longer interpreted as a version field. This is
2222 intented to be largely invisible to the user, except that the
2223 meaning of the namestring "*.*.*" has changed: it now refers to a
2224 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2225 should usually be replaced by
2226 (make-pathname :name :wild :type :wild :version :wild)
2227 with the added benefit that this is more likely to be portable.
2228 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2229 option to OPEN now signals an error if the file being opened
2230 exists; this may have an impact on existing code.
2231 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2232 signals received, on the PowerPC platforms (both Linux and
2233 Darwin). (thanks to Pierre Mai for pointing out the location of
2235 * several fixes on OS X: The system now builds and runs cleanly on
2236 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2237 (thanks to Brian Mastenbrook)
2238 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2239 second-resolution: integer multiples of 1/3600 between -24 and 24.
2240 (thanks to Vincent Arkesteijn)
2241 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2242 type when called with 1 argument; PEEK-CHAR checked type of
2243 PEEK-TYPE only after having read first character from a
2244 stream. (reported by Peter Graves)
2245 * bug fix: the garbage collector now has much better locality
2246 behaviour, and in particular no longer treats all memory as being
2247 exhausted when many small objects point to each other in a deeply
2249 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2250 recognized as being TYPEP their class.
2251 * bug fix: the PUSHNEW documentation string has been corrected.
2252 (thanks to Vincent Arkesteijn)
2253 * bug fix: defaulting of the value for the last of an atypically
2254 large number of multiple values being bound was not being
2255 performed correctly on the Alpha or PPC platforms
2256 * optimization: implemented multiplication as a modular
2257 (UNSIGNED-BYTE 32) operation on the x86 backend.
2258 * optimization: SEARCH on simple-base-strings can now be open-coded.
2259 (see also contrib/compiler-extras.lisp for inspiration for
2260 teaching the compiler about the Boyer-Moore algorithm).
2261 * value, returned by MAX (and MIN) called with several EQUALP, but
2262 not EQL, arguments now does not depend on compiler settings.
2263 * fixed some bugs revealed by Paul Dietz' test suite:
2264 ** in stack analysis liveness information is propagated from
2265 non-local entry points.
2266 ** pathwise CAST removing failed when the CAST node did not start
2268 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2269 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2271 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2272 host is already defined.
2273 ** RENAME-FILE works on streams instead of signalling an internal
2275 ** PEEK-CHAR uses the current readtable when determining whether
2276 or not a character is whitespace.
2277 ** MERGE-PATHNAMES handles the case when the pathname does not
2278 specify a name while the default-pathname specifies a version.
2279 ** Pathnames now stand a better chance of respecting print/read
2281 ** Attempting to use standardized file system operators with a
2282 pathname with invalid :DIRECTORY components signals a
2284 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2286 ** DIRECTORY on logical pathnames is more correct.
2287 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2288 signal a TYPE-ERROR if their argument is not a stream.
2289 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2290 designator argument does not designate a stream.
2291 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2292 examining the synonym.
2293 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2295 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2296 element-type larger than ([UN]SIGNED-BYTE 32).
2298 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2299 * When built with the :SB-FUTEX feature, threaded builds now take
2300 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2301 for faster/more reliable mutex and condition variable support.
2302 * Incompatible change (but one you probably shouldn't have been using
2303 anyway): the interface and code for arbitrating between multiple
2304 threads in the same user session has been redesigned.
2305 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2306 error when called without an explicit environment argument.
2307 (thanks to Nikodemus Siivola)
2308 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2309 confused about their position. (thanks to Adam Warner and Gerd
2311 * bug fix: Pretty printing backquoted forms with unquotations in the
2312 argument list position of various code constructs such as LAMBDA
2313 now works correctly. (reported by Paul Dietz)
2314 * bug fix: Pretty printing unquotations no longer loses all
2315 stream position information.
2316 * optimization: performance of string output streams is now less
2317 poor for multiple small sequence writes.
2318 * optimization: performance of CSUBTYPEP in the presence of complex
2319 expressions involving CONS and NOT many times has been improved.
2320 (reported by Paul Dietz)
2321 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2323 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2324 streams for the REPL, for the debugger, and so on. (thanks to
2326 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2327 * fixed some bugs revealed by Paul Dietz' test suite:
2328 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2329 is ignored if it is too small, rather than propagating through
2330 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2331 ** extremely complex negations of CONS types were not being
2332 sufficiently canonicalized, leading to inconsistencies in
2334 ** VALUES tranformer lost derived type.
2336 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2337 * fixed a bootstrapping bug: the build process no longer assumes
2338 that the various BOOLE-related constants have the same value in
2339 host and target lisps. (noted by Paul Dietz' test suite on an
2340 SBCL binary built from CLISP)
2341 * The system can now be dynamically linked on the MIPS platform,
2342 which enables dynamic loading of foreign code from Lisp. (thanks
2343 to Ralf Baechle for discussions on the MIPS ABI)
2344 * The system now records debugging information for its own source
2345 files in a filesystem-position-independent manner, relative to
2346 the "SYS" logical host.
2347 * fixed a compiler bug: MV-LET convertion did not check references
2348 to the "max args" entry point. (reported by Brian Downing)
2349 * tweaked disassembly notes to be less confident about proclaiming
2350 some instruction as an LRA. (thanks to Brian Downing)
2351 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2352 now each have their own history, command character, and other
2353 characteristics. (thanks to David Lichteblau)
2354 * fixed some bugs revealed by Paul Dietz' test suite:
2355 ** compiler failure in compiling LOGAND expressions including a
2357 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2358 shift greater than 32.
2359 ** FLUSH-DEST did not mark blocks for type check regeneration.
2360 ** HANDLER-CASE failed to accept declarations in handler clauses
2361 in some circumstances.
2363 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2364 * New code in contrib/sb-introspect (still probably not entirely
2365 stable yet) provides some support for smart Lisp development
2366 environments like SLIME.
2367 * The conditions signalled for errors occurring when loading .fasl
2368 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2369 in a way which should help ASDF recover gracefully.
2370 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2371 stuff has been cleaned up. If you code contrib/ stuff, this might
2372 affect you, and you can look at contrib/README, contrib/STANDARDS,
2373 and/or the 0.8.4.27 diff to check.
2374 * In full calls the compiler now does not generate checks for declared
2375 argument types for all arguments.
2376 * various threading fixes
2377 ** and some experimental patches which didn't make it into
2378 the main tree for this release, but which are shipped in
2379 contrib/experimental-thread.patch as a possible fix for some
2380 failures (deadlock, spinning...) in GC-intensive multithreaded
2382 * fixed PPC build problem (source code incompatibility of different
2383 library versions): added offsetof() hackery which attempts to divine
2384 where glibc maintainers put uc_mcontext today
2385 * fixed bug 282: compiler does not trust type assertions while passing
2386 arguments to a full call.
2387 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2388 &OPTIONAL VALUES type parameter.
2389 * fix bug 214: algorithm for noting rejected templates is now more
2390 similar to that of template seletion. (also reported by rydis on
2392 * fixed bug 141b: printing backquoted information readably and prettily
2393 inserts a space where necessary.
2394 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2395 (CHARACTER 10) are now reported as errors, rather than propagated
2396 as unknown types. (reported by piso on #lisp)
2397 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2398 with values NIL and :ERROR. (thanks to Milan Zamazal)
2399 * fixed bug 191c: CLOS now does proper keyword argument checking as
2400 described in CLHS 7.6.5 and 7.6.5.1.
2401 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2402 longer raise an error; further, using a list as a for-as-arithmetic
2403 counter now raises a meaningful error.
2404 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2405 now understood by sequence creation functions such as MAKE-SEQUENCE
2407 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2408 a STRING-INPUT-STREAM.
2409 * compiler enhancement: SIGNUM is now better able to derive the type
2411 * type declarations inside WITH-SLOTS are checked. (reported by
2413 * fixed some bugs revealed by Paul Dietz' test suite:
2414 ** incorrect optimization of TRUNCATE for a positive first
2415 argument and negative second.
2416 ** compiler failure in let-convertion during flushing dead code.
2417 ** compiler failure while deriving type of TRUNCATE on an
2418 interval, containing 0.
2419 ** ASH of a negative bignum by a negative bignum count now returns
2421 ** intersection of CONS types now canonicalizes properly, fixing
2422 inconsistencies in SUBTYPEP.
2424 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2425 * incompatible change: The --disable-debugger command line
2426 option now clobbers the debugger at a more fundamental
2427 level, by redefining #'INVOKE-DEBUGGER instead of by
2428 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2429 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2430 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2431 enter the debugger REPL and then suffer recursive errors
2432 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2433 while under the new system, BREAK in --disable-debugger mode
2434 terminates the system just as an unhandled error would.
2435 * fixed compiler performance when processing loops with a step >1;
2436 * bug fix: DOCUMENTATION now retrieves generic function
2437 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2438 support has been systematized, and now supports the methods
2439 specified by ANSI, along with a default method and a method for
2440 slot documentation. (reported by Nathan Froyd)
2441 * bug fix: effective methods associated with a generic function are
2442 no longer cached over a change of that generic function's method
2443 combination. (reported by Andreas Fuchs)
2444 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2445 names a non-existent directory. (thanks to Andreas Fuchs)
2446 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2447 the PPC platform, lead to stack corruption; this has been fixed.
2448 (reported by Rainer Joswig)
2449 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2450 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2451 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2452 arithmetic in general, is now much more reliable on the PPC
2454 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2455 platform now returns the right answer.
2456 * optimization: restored some effective method precomputation in
2457 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2458 precomputation is now tunable.
2459 * optimization: compiler-internal data structure use has been
2460 reviewed, and changes have been made that should improve the
2461 performance of the compiler by about 20%.
2462 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2463 simple-base-strings and simple-bit-vectors is improved.
2464 * optimization: the optimization of 32-bit logical and arithmetic
2465 functions introduced in version 0.8.3 on the x86 has been
2466 implemented on the mips, ppc and sparc platforms; an
2467 implementation of the same facility, but for 64-bit arithmetic,
2468 has been added for the alpha.
2469 * microoptimization: the compiler is better able to make use of the
2470 x86 LEA instruction for multiplication by constants.
2471 * bug fix: in some situations compiler did not report usage of
2472 generic arithmetic in (SPEED 3) policy.
2473 * bug 145b fix: compiler used wrong type specifier while converting
2474 MEMBER-types to numeric.
2475 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2477 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2478 index argument now works properly on the Alpha platform.
2479 * bug fix: floating point exception treatment on the Alpha platform
2481 * bug fix: FILE-POSITION works much better on string input and
2482 output streams. (thanks to Nikodemus Siivola)
2483 * bug fix: many threading/garbage collection symptoms sorted.
2484 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2485 might be pseudo-atomic.
2486 * internal change: Stopping for GC is now done with signals not
2487 ptrace. GC is now done in whichever thread wanted it, instead of
2489 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2491 * bug fix: result form in DO is not contained in the implicit
2493 * incompatible change: ICR structure is changed; the value part of
2494 CONTINUATION is now called LVAR; corresponding functions are
2495 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2496 * added type deriver for ISQRT (thanks to Robert E. Brown).
2497 * bug fix: better support for loading from the command line when an
2498 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2500 * fixed some bugs revealed by Paul Dietz' test suite:
2501 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2502 ** ROUND and FROUND now give the right answer when given very
2503 small float arguments.
2504 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2506 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2507 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2508 ** LCM should return a non-negative integer.
2509 ** PARSE-INTEGER returned the index of a terminator instead of the
2510 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2511 ** PARSE-INTEGER returned an incorrect index being applied to a
2513 ** LCM with two arguments of 0 returns 0 rather than signalling
2515 ** unsigned addition of a 32-bit constant with the high bit set no
2516 longer causes an internal compiler error on the x86.
2517 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2518 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2519 with negative last argument.
2520 ** byte specifiers with zero size and position no longer cause
2521 an error during type derivation.
2522 ** bignum multiplication on the Alpha platform now returns the
2524 * porting: The system now builds on SuSE AMD64, although it still
2525 generates a 32-bit binary.
2526 * .fasl file incompatibility: The fasl file version number has
2527 been incremented (because of the changes to internal compiler
2528 data structures referred to above).
2530 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2531 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2532 more accurately, on the Darwin kernel running on PowerPC hardware.
2533 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2534 * Compiler code deletion notes now signal a condition of type
2535 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2536 an associated MUFFLE-WARNING restart.
2537 * The compiler now performs limited argument count validation of
2538 constant format strings in FORMAT, and where appropriate in ERROR,
2539 CERROR and WARN. (thanks to Gerd Moellmann)
2540 * New ASDF-INSTALL contrib can be used for automatic download and
2541 installation of third-party Lisp code from CCLAN or other sites
2543 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2544 which forces another thread to execute a function supplied by the
2546 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2547 now accept and act upon their :ELEMENT-TYPE keyword argument.
2548 (reported by Martin Atzmueller, Edi Weitz)
2549 * bug fix: FILE-POSITION now accepts position designators up to
2550 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2551 is the greater. (thanks to Patrik Nordebo)
2552 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2553 Kalle Olavi Niemitalo)
2554 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2555 printable. (reported by Eric Marsden)
2556 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2557 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2558 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2559 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2560 (thanks to Stig Sandoe)
2561 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2562 UNREAD-CHAR now correctly includes the unread character in the
2563 target sequence. (thanks to Gerd Moellmann)
2564 * bug fix in threaded builds: the system can now be suspended and
2565 resumed by shell job control with minimal disruption.
2566 * bug fixes in times and timezones >2038AD
2567 * better handling of "where is GNU make?" problem in build scripts
2568 (thanks to Nikodemus Siivola)
2569 * new optimization: inside a named function any reference to a
2570 function with the same name is considered to be a self-reference;
2571 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2572 optimization quality.
2573 * new optimization on x86: logical functions and + now have
2574 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2575 used when the result is truncated to 32 bits.
2576 * VALUES declaration is partially enabled.
2577 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2578 * fixed some bugs revealed by Paul Dietz' test suite:
2579 ** The system now obeys the constraint imposed by
2580 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2581 types form a lattice under type intersection.
2582 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2583 ** ASSOC now ignores NIL elements in an alist.
2584 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2585 and (1+ MOST-POSITIVE-FIXNUM) answers.
2586 ** The addition of a method with invalid qualifiers to a generic
2587 function does not cause an error to be signalled immediately;
2588 a warning is signalled, and the error is generated only on
2589 calling the generic function.
2590 * changed .fasl file version number, in part to add type codes for
2591 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2592 obscure ANSI requirements
2594 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2595 * fixed bug 148: failure to inline-expand a local function left
2596 garbage, confusing the compiler.
2597 * fixed bugs 3cd: structure slot readers perform type check if the
2598 slot can have an invalid value (i.e. it is either not initialized
2599 or can be written with a less specific slot writer).
2600 * bug fix: the compiler now traps array references to elements off
2601 the end of an array; previously, the bounds checking in some
2602 circumstances could go off-by-one.
2603 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2605 * type declarations for array element types now obey the description
2606 on the CLHS page "Declaration TYPE", as per discussions on
2607 sbcl-help around 2003-05-08. This means that a declaration
2608 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2609 declaration, all references to BAR will be asserted or assumed
2610 (with THE, so dependent on compiler policy) to involve objects of
2611 type FOO. Note that no such declaration is implied in
2612 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2613 * declared types of functions from the "Conditions"
2614 chapter. (reported by Paul Dietz)
2615 * bug fix: CERROR accepts a function as its first argument.
2616 * bug fix: NTH an NTHCDR accept a bignum as index
2617 arguments. (reported by Adam Warner)
2618 * optimization: character compare routines now optimize comparing
2619 against a constant character. (reported by Gilbert Baumann)
2620 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2621 argument now works properly on the MIPS platform.
2622 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2623 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2624 declared to have a functional type, does not cause an error
2626 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2627 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2628 * fixed bug: if last continuation of a deleted block has a
2629 destination, this destination should be deleted too. (reported by
2631 * fixed a bug in the bootstrap process: the host compiler's values
2632 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2633 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2634 test case from Patrik Nordebo)
2635 * improved the ability of the disassembler on the PPC platform to
2636 provide helpful disassembly notes.
2637 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2638 the class in more cases than previously.
2639 * bug fix: FILE-POSITION now understands :START and :END for
2640 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2641 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2642 8) data. (thanks to David Lichteblau)
2643 * bug fix: it is possible to add a method to a generic function
2644 without lambda list.
2645 * bug fix: reader failed to signal END-OF-FILE inside an
2646 object representation. (reported by Nikodemus Siivola)
2647 * fixed some bugs revealed by Paul Dietz' test suite:
2648 ** LAST and [N]BUTLAST should accept a bignum.
2649 ** condition slot accessors are methods.
2650 ** (VECTOR NIL) is a subtype of STRING.
2652 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2653 * minor incompatible change: some nonsensical specialized lambda
2654 lists (used in DEFMETHOD) which were previously ignored now signal
2656 * minor incompatible change: the system is now aware of the types of
2657 variables in the COMMON-LISP package, and will signal errors for
2658 most violations of these type constraints (where previously they
2659 were silently accepted).
2660 * minor incompatible change: COMPILE-FILE now uses the freedom
2661 afforded (ANSI 3.2.2.3) to use derived function types for
2662 functions defined in the same file. This also permits the system
2663 to warn on static type mismatches and function
2664 redefinition. (Currently it does not work with high DEBUG level.)
2665 * minor incompatible change: VALUES declaration is disabled.
2666 * When issuing notes, the compiler now signals a condition of type
2667 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2668 restart for use in user handlers. It is expected that the
2669 COMPILER-NOTE condition will eventually become a condition
2670 supertype to a hierarchy of note types, which will then be
2671 handleable in a similar fashion. However, at the moment, no such
2672 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2673 but it's an implementation detail, not a classification for the
2675 * Changes in type checking closed the following bugs:
2676 ** type checking of unused values (192b, 194d, 203);
2677 ** template selection based on unsafe type assertions (192c, 236);
2678 ** type checking in branches (194bc).
2679 * A short form of VALUES type specifier has ANSI meaning (it has
2680 increased the number of situations when SBCL cannot perform type
2682 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2683 keywords or constants is permissible.
2684 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2685 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2686 classes. (thanks to Antonio Martinez)
2687 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2688 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2689 argument to a pretty-print stream. Instead, it leaves any such
2690 operations to DESCRIBE-OBJECT methods.
2691 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2693 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2694 lambda lists. (thanks to David Lichteblau)
2695 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2696 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2697 (thanks to Gerd Moellmann)
2698 * fixed bug 246: increased compilation speed of long
2699 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2701 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2703 * DEFCONSTANT now throws a condition of type
2704 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2705 constant to a non-EQL value; CONTINUE and ABORT restarts
2706 respectively change and preserve the value.
2707 * fixed bug 63: The code walker, part of the implementation of CLOS,
2708 is now better at handling symbol macros.
2709 * bug fix: There is no longer an internal implementation type named
2710 CL:LENGTH. (reported by Raymond Toy)
2711 * bug fix: In macro-like defining macros/special operators the
2712 implicit block does not enclose the lambda list.
2713 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2714 enclosed in parentheses) are not suitable as type specifiers, and
2715 their use properly signals an error now.
2716 * bug fix: An argument count mismatch for a type specifier in code
2717 being compiled no longer causes an unhandled error at compile
2718 time, but signals a compile-time warning.
2719 * fixed simple vector readable printing
2720 * bug fix: DESCRIBE takes more care over whether the class
2721 precedence list slot of a class is bound before accessing it.
2722 (reported by Markus Krummenacker)
2723 * bug fix: FORMATTER can successfully compile pretty-printer format
2724 strings which use variants of the ~* directive inside.
2725 * bug fix: SEARCH now applies its TEST predicate to the elements of
2726 the arguments in the correct order. (thanks to Wolfhard Buss)
2727 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2728 in type inference. (thanks to Robert E. Brown)
2729 * bug fix: Reading in symbols with an explicit package name of ""
2730 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2731 * fixed some bugs revealed by Paul Dietz' test suite:
2732 ** NIL is now allowed as a structure slot name.
2733 ** Arbitrary numbers, not just REALs, are allowed in certain
2734 circumstances in LOOP for-as-arithmetic clauses.
2735 ** Multiple class redefinitions before slot access no longer
2736 causes a type error.
2737 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2738 association between the name and a class.
2739 ** Generic functions with non-standard method-combination and over
2740 five methods all of which return constants no longer return NIL
2741 after the first few invocations. (thanks to Gerd Moellmann)
2742 ** CALL-NEXT-METHOD with no arguments now passes the original
2743 values of the arguments, even in the presence of assignment.
2744 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2746 ** DISASSEMBLE works with closures and funcallable instances.
2747 ** ADD-METHOD now returns the generic function, not the new method.
2748 ** FIND-METHOD signals an error if the lengths of the specializers
2749 is incompatible with the generic function, even if the ERRORP
2751 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2752 which its argument is a member.
2753 ** DEFCLASS only redefines the class named by its class-name
2754 argument if that name is the proper name of the class;
2755 otherwise, it creates a new class.
2756 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2757 of the UNBOUND-SLOT condition to the name of the slot.
2758 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2759 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2760 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2761 SLOT-MAKUNBOUND in the specified fashion.
2763 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2764 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2765 cross-compilation host. As a consequence, we can now bootstrap our
2766 way up to SBCL starting with a bare gcc toolchain and human-readable
2767 source code (first the source to CLISP, then the source to SBCL).
2768 * A contributed module containing a partial implementation of the
2769 simple-streams interface has been included. (thanks to Rudi
2771 * A contributed module implementing the RFC1321 Message Digest
2772 Algorithm, known as MD5, has been included.
2773 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2774 no longer has any effect, as the code controlled by this feature
2775 has been deleted. (As far as we know, no-one has ever built using
2776 this feature, and its semantics were confused in any case).
2777 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2778 work on conditions (as required by the ANSI specification),
2779 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2780 expected behaviour on conditions. Users should note, however,
2781 that such behaviour is not required by the ANSI specification,
2782 and so use of this behaviour may render their code unportable.
2783 * fixed some bugs revealed by Paul Dietz' test suite:
2784 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2786 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2787 ** SXHASH on condition objects no longer returns NIL.
2788 ** :ALLOCATION :CLASS slots are better treated; their values are
2789 updated on class redefinition, and initforms inherited from
2790 superclasses are applied.
2791 ** REMOVE-METHOD returns its generic function argument even when
2792 no method was removed.
2793 ** SHARED-INITIALIZE now initializes the values of the requested
2794 slots, including those with :ALLOCATION :CLASS.
2795 ** ALLOCATE-INSTANCE now works on structure classes defined via
2796 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2798 ** SLOT-EXISTS-P now works on conditions, as well as structures
2800 ** MAKE-LOAD-FORM now has the required methods on
2801 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2802 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2803 keyword, and now implements the SLOT-NAMES argument.
2804 ** methods with &OPTIONAL arguments no longer allow too many
2805 arguments to be passed in the call without error.
2806 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2807 option is consistent with the required arguments of the generic
2808 function lambda list.
2809 * bug fix: REQUIRE accepts a string designator. (Thanks to
2811 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2812 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2813 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2814 * bug fix: The compiler now checks for duplicated variables in macro
2816 * bug fix: SETQ on globals returns the correct value.
2817 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2818 not a COMPILER-ERROR (followed by some other strange error on
2819 choosing the CONTINUE restart).
2820 * bug fix: make.sh and friends are now more consistent in the way that
2821 they look for GNU "make".
2823 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2824 * experimental native threads support (on x86 Linux >=2.4 only).
2825 This is not compiled in by default: you need to add :SB-THREAD to
2826 the target features. See the "Beyond ANSI" chapter of the manual
2828 * fix for longstanding nonANSIism: The old distinction between
2829 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2830 The return value from CL:FIND-CLASS is now a CLOS class, and
2831 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2832 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2833 * An interface to the MetaObject Protocol, as described in Kiczales,
2834 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2835 MIT Press, 1991, is available from the SB-MOP package.
2836 * incompatible change: the SB-PCL package should now be considered
2837 a private implementation detail, and no longer a semi-private MOP
2839 * minor incompatible change: due to rearrangement for threads, the
2840 control stack and binding stack are now allocated at arbitrary
2841 addresses instead of being hardcoded per-port. Users affected by
2842 this probably have to be doing advanced things with shared
2843 libraries, and will know who they are.
2844 * minor incompatible change: Previously, all --eval forms used were
2845 processed with READ before any of them were processed with EVAL.
2846 Now each --eval form is processed with both READ and EVAL before
2847 the next --eval form is processed. (Thus package operations like
2848 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2849 work as the user might reasonably expect.)
2850 * minor incompatible change: *STANDARD-INPUT* is now only an
2851 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2853 * minor incompatible change: Y-OR-N-P is now character-oriented,
2854 not line oriented. Also, YES-OR-NO-P now works without errors.
2855 (thanks to Antonio Martinez)
2856 * sb-aclrepl module improvements: an integrated inspector, added
2857 repl features, and a bug fix to :trace command.
2858 * Known functions, which cannot be open coded by the backend, are
2859 considered to be able to check types of their arguments. (fixing
2860 a bug report by Nathan J. Froyd)
2861 * fixed a bug in computing method discriminating functions: It is
2862 now possible to define methods specialized on classes which have
2863 forward-referenced superclasses. (thanks to Gerd Moellmann)
2864 * fixed evaluation order in optional entries (reported by Gilbert
2866 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2867 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2868 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2869 required-by-AMOP NAME argument, as well as CLASS and
2870 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2871 * fixed bug 20: DEFMETHOD can define methods using names that are
2872 not the proper names of classes to designate class specializers.
2873 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2874 Unix file descriptors, instead of blowing up. (thanks to Antonio
2876 * Garbage collection refactoring: user-visible change is that a
2877 call to the GC function during WITHOUT-GCING will not do garbage
2878 collection until the end of the WITHOUT-GCING. If you were doing
2879 this you were probably losing anyway.
2880 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2881 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2882 TYPEP the latter but not the former.
2883 * The compiler issues a full WARNING for calls to undefined functions
2884 with names from the CL package.
2885 * MAP-INTO for a vector destination is open coded. (reported by
2886 Brian Downing on c.l.l)
2887 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2888 documentation string.
2889 * fixed some bugs revealed by Paul Dietz' test suite:
2890 ** COPY-ALIST now signals an error if its argument is a dotted
2892 ** Condition slots are now accessed more correctly in the presence
2893 of multiple initargs for a given slot.
2894 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2895 exclude restarts of the same name associated with a different
2897 ** DEFCLASS of forward-referenced classes with another
2898 forward-referenced class in the superclasses list no longer
2900 ** Condition slots are now initialized once each, not multiple
2901 times. (thanks to Gerd Moellmann)
2902 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2904 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2905 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2906 arguments contain duplicated elements.
2907 ** RESTART-CASE understands local macros.
2908 ** RESTART-CASE associates exactly its own restarts with a condition.
2909 ** ENDP in safe mode checks its argument to be of type LIST.
2910 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2911 in question is unbound.
2912 ** Optimized MAKE-INSTANCE functions no longer cause internal
2913 assertion failures in the presence of duplicate initargs.
2914 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2915 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2916 propagated even in the ctor optimized implementation of
2918 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2920 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2921 * a better implementation of SXHASH on (simple) bit vectors,
2922 measured both in execution speed and in distribution of results
2923 over the positive fixnums, has been installed. Likewise, a better
2924 implementation of EQUAL for simple bit vectors is now available.
2925 * fixed CEILING optimization for a divisor of form 2^k.
2926 * fixed bug 240 (emitting extra style warnings "using the lexical
2927 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2928 by Antonio Martinez)
2929 * fixed SXHASH, giving different results for NIL depending on type
2930 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2931 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2932 variable. (found by Rolf Wester)
2933 * SBCL does not ignore type declarations for special
2934 variables. (reported by rif on c.l.l 2003-03-05)
2935 * some bug fixes in contrib/sb-aclrepl/
2936 * fixed some bugs revealed by Paul Dietz' test suite:
2937 ** a bug in the CONS type specifier, whereby the CAR and CDR
2938 types got intertwined, has been fixed;
2939 ** the type system is now able to reason about the interaction
2940 between INTEGER and RATIO types more completely;
2941 ** APPEND, [N]REVERSE and NRECONC check that those their
2942 arguments, which must be proper lists, are really so;
2943 ** An array specialized to be unable to hold elements has been
2944 implemented, as required -- yes, really -- by ANSI;
2945 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2946 on malformed property lists;
2948 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2949 * incompatible packaging change: in line with Unix convention,
2950 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2951 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2952 * REQUIRE and PROVIDE are now optionally capable of doing something
2953 useful. See the documentation string for REQUIRE.
2954 * infrastructure for a managed SBCL contrib system: contributed
2955 modules in this release include:
2956 ** the ASDF system definition facility;
2957 ** an interface to the BSD Sockets API;
2958 ** an ACL-like convenience interface to the repl;
2959 (thanks to Kevin Rosenberg)
2960 ** an implementation of ROTATE-BYTE, with efficient implementation
2962 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2963 gives a closer approximation to the right answer than previously.
2964 (thanks to Raymond Toy)
2965 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2966 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2967 optional environment arguments, as required by ANSI.
2968 * fixed bugs in other functions taking environment objects, allowing
2969 calls with an explicit NIL environment argument to be compiled
2971 * fixed bug 228: primary return values from
2972 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2973 COMPILE or FUNCTION.
2974 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2975 :TYPE will no longer signal errors on innocuous objects.
2976 * fixed bug 231b: SETQ is better at respecting type declarations in
2977 the lexical environment.
2978 * fixed a bug in DEFCLASS: classes named by symbols with no or
2979 unprintable packages can now be defined.
2980 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2981 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2982 * fixed compiler failure related to checking types of functions.
2983 (reported by Robert E. Brown)
2984 * the compiler is now much more consistent in its error-checking
2985 treatment of bounding index arguments to sequence functions: in
2986 (SAFETY 3) code, errors will be signalled in almost all cases if
2987 invalid sequence bounding indices are passed to functions defined
2988 by ANSI to operate on sequences.
2989 * fixed a bug in the build procedure: documentation of SBCL-specific
2990 packages is now preserved and available in the final Lisp image.
2991 * lifted FDEFINITION lookup out of loops in the implementation of
2992 many list operations. (thanks to Robert E. Brown)
2993 * fixed a bug in the reader: the #n# reader macro now works for
2994 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2995 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2996 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2997 E. Brown and Tony Martinez respectively)
2998 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2999 SLOT-BOUNDP: in particular, it is now able to optimize them much
3000 better, and is now not vulnerable to having packages renamed.
3001 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3002 always called when a slot is not present in an instance. (thanks
3004 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3005 longer signal an error on unreading a character following EOF on
3006 the previous constituent stream. (thanks to Tony Martinez)
3007 * fixed some bugs revealed by Paul Dietz' test suite:
3008 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3009 not just nonnegative fixnums;
3010 ** the logical bit-array operators such as BIT-AND now accept an
3011 explicit NIL for their "opt-arg" argument (to indicate a
3012 freshly-consed result bit-array);
3013 ** ELT now signals an error on an invalid sequence index in safe
3015 ** the type system is now cleverer about negations of numeric
3016 types, and consequently understands the BIGNUM and RATIO types
3018 ** the type system is now cleverer about the interaction between
3019 INTEGER and RATIO types: while bugs still remain, many more
3020 cases are accurately computed;
3021 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3022 if it is in the last clause;
3023 ** CONSTANTLY now correctly returns a side-effect-free function in
3025 ** DECLARE is no longer treated as a special-operator; in
3026 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3027 * incremented fasl file version number due to the change in the
3028 DEFSTRUCT-SLOT-DESCRIPTION structure.
3030 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3031 * minor incompatible change: code processed by the "interpreter" or
3032 EVAL now has a compilation optimization policy of (DEBUG 2)
3033 (changed from (DEBUG 1)) to improve debuggability of interactive
3034 development, and to allow the use of the debug RETURN command in
3036 * an experimental implementation of the RETURN command for the
3037 debugger has been included. (thanks to Frederik Kuivinen)
3038 * fixed bug 62: constraints were not propagated into a loop.
3039 * fixed bug in embedded calls of SORT (reported and investigated by
3041 * fixed some bugs revealed by Paul Dietz' test suite:
3042 ** printing and reading of arrays with some dimensions having
3043 length 0 (thanks to Gerd Moellmann);
3044 ** BOA constructor with &AUX argument without a default value does
3045 not cause a type error;
3046 ** CONSTANTP now returns true for all self-evaluating objects.
3048 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3049 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3050 accessors that are related by inheritance, as specified in the
3051 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3052 Valtteri Vuorikoski)
3053 * The compiler is now able to inline functions that were defined in
3054 a complex lexical environment (e.g. inside a MACROLET).
3055 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3057 * Support for the upcoming FreeBSD-5.0 release has been included.
3058 (thanks to Dag-Erling Smorgrav)
3059 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3060 effect when it is not in a toplevel context.
3061 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3062 stand a better chance of being correct. (thanks to Gerd
3064 * fixed bug in COERCE, which now signals an error on coercing a
3065 rational to a bounded real type which excludes the expected
3067 * The compiler is now able to derive types more accurately from the
3068 COERCE and COMPILE functions.
3069 * fixed bug 223: functional binding is considered to be constant
3070 only for symbols in the CL package.
3071 * fixed bug 231: SETQ did not check the type of a variable being set
3072 (reported by Robert E. Brown)
3073 * A new optimization for MAKE-INSTANCE has been included, fixing
3074 various bugs (including relating to :ALLOCATION :CLASS slots and
3075 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3077 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3078 ** As required by ANSI, LOOP now disallows anonymous collection
3079 clauses such as COLLECT I in conjunction with aggregate boolean
3080 clauses such as THEREIS (= I 1);
3081 ** LOOP now signals an error when any variable is reused in the
3082 same loop (including the potentially useful construct analogous
3083 to WITH A = 1 WITH A = (1+ A);
3084 ** IT is only a special loop symbol within the first clause of a
3085 conditional loop clause;
3086 ** LOOP with a typed iteration variable over a hashtable now
3087 signals a type error iff it should.
3088 * fixed some other bugs revealed by Paul Dietz' test suite:
3089 ** FILE-STREAM now names the class previously known as FD-STREAM;
3090 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3091 argument) no longer signals an error;
3092 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3093 of the slot symbol, rather than using the current package
3094 ((:CONC-NAME "") continues to intern the slot's name in the
3096 * incremented fasl file version number, because of the incompatible
3097 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3098 of the new implementation of DEFINE-COMPILER-MACRO.
3100 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3101 * Support for building SBCL for MIPS platforms running in
3102 little-endian mode has now been checked in, and basic
3103 functionality on said platforms verified.
3104 * minor incompatible change: PCL now records the pathname of a file
3105 in which methods and the like are defined, rather than its
3107 * minor incompatible change: TRUENAME now considers the truename of
3108 a file naming a directory to be the pathname with :DIRECTORY
3109 component indicating that directory.
3110 * minor incompatible change: a NAMED clause in the extended form of
3111 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3112 reason for the previous behaviour is unclear.
3113 * more systematization and improvement of CLOS and MOP conformance
3114 in PCL (thanks to Gerd Moellman and Pierre Mai):
3115 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3117 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3118 primary methods with no specializers;
3119 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3121 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3122 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3123 FORWARD-REFERENCED-CLASSes; error reporting on
3124 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3126 ** SXHASH on CLOS instances now uses a slot internal to the
3127 instance to return different numbers on distinct instances,
3128 while preserving the same return value through invocations of
3130 ** DEFMETHOD signals errors when methods with longer incongruent
3131 lambda lists are added to generic functions;
3132 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3133 CLASS, as specified in AMOP;
3134 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3135 on the order returned by the primary method for classes of
3136 class STANDARD-CLASS;
3137 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3138 * fixed some bugs shown by Paul Dietz' test suite:
3139 ** DOLIST puts its body in TAGBODY;
3140 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3142 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3143 value producing form;
3144 ** if more variables are given to PROGV than values, extra
3145 variables are bound and made to have no value;
3146 ** NSUBSTITUTE on list arguments gets the right answer with
3148 ** ELT signals an error of type TYPE-ERROR when the index argument
3149 is not a valid sequence index;
3150 ** LOOP signals (at macroexpansion time) an error of type
3151 PROGRAM-ERROR when duplicate variable names are found;
3152 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3153 ** FUNCALL of special-operators now cause an error of type
3155 ** PSETQ now works as required in the presence of side-effecting
3156 symbol-macro places;
3157 ** NCONC accepts any object as its last argument;
3158 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3160 ** loop-for-as-package does not require a package to be explicitely
3162 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3164 * fixed bug 166: compiler preserves "there is a way to go"
3165 invariant when deleting code.
3166 * fixed bug 172: macro lambda lists with required arguments after
3167 &REST arguments now cause an error to be signalled. (thanks to
3169 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3170 bug. (thanks to Matthew Danish)
3171 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3173 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3174 arguments are lexically rebound. (thanks to Gerd Moellmann and
3176 * fixed bug 194: error messages are now more informative when there
3177 is no primary method applicable in a call to a generic
3178 function. (thanks to Gerd Moellmann)
3179 * fixed bug in command line argument checking (thanks to Julian
3181 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3182 (thanks to Matthew Danish)
3183 * incremented fasl file version number, because of the
3184 SXHASH-related changes in the layout of CLOS data structures
3186 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3187 * minor incompatible change: The runtime (the Unix executable named
3188 "sbcl") is now much pickier about the .core files it will load.
3189 Essentially it now requires .core files to descend from the same
3190 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3191 as the runtime does. (The intent is to prevent the crashes which
3192 can occur, and which can even be reported as mysterious failures,
3193 when people patch the sources or change the build parameters
3194 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3195 sbcl and .core files.)
3196 * fixed bug: VALUES-LIST is no longer optimized away.
3197 * fixed bug 142: The FFI conversion of C string values to Lisp
3198 string values no longer conses excessively. (thanks to Nathan
3199 Froyd porting Raymond Toy's fix to CMU CL)
3200 * began to systematize and improve MOP conformance in PCL (thanks to
3201 Nathan Froyd, Gerd Moellman and Pierre Mai):
3202 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3204 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3205 ** FINALIZE-INHERITANCE is now called on class finalization;
3206 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3207 argument precedence order.
3208 * fixed bug 202: The compiler no longer fails on functions whose
3209 derived types contradict their declared type.
3210 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3211 so it can be non-toplevel.
3212 * The fasl file version number has changed (because of the new
3213 implementation of DEFMACRO).
3214 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3215 safe code, that any length requirement by their type-specifier
3216 argument is valid. The exceptions to this are described in bug
3218 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3219 functions now check the types of their inputs as required by ANSI.
3220 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3221 introduced symbol is DECLAREd to be SPECIAL.
3222 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3223 * fixed bug: :COUNT argument to sequence functions may be negative
3224 * fixed bug: body of DO-SYMBOLS may contain declarations
3225 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3226 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3227 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3228 symbol macro only once
3229 * fixed printing of call frame when argument list is unavailable
3230 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3231 * compiler no longer signals WARNING on unknown keyword
3234 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3235 * A beta-quality port to the mips architecture running Linux,
3236 based on the old CMUCL backend, has been made. It has been tested
3237 on a big-endian kernel, and works sufficiently well to be able to
3238 rebuild itself; it has not been tested in little-endian mode.
3239 * fixed an inconsistency between gencgc.c and purify.c which made
3240 dumping/loading .core files unreliable
3241 * fixed bug 120a: The compiler now deals correctly with IFs where
3242 the consequent is the same as the alternative, instead of
3243 misderiving the return type. (thanks to Alexey Dejneka)
3244 * fixed bug 113: Logical pathnames are now dumpable (the logical
3245 host is resolved at load-time, throwing an error if it is not
3247 * fixed bug 174: FORMAT's error message is slightly clearer when a
3248 non-printing character is used in a format directive.
3249 * fixed several bugs in compiler checking of type declarations, i.e.
3250 violations of the Python "declarations are assertions" principle
3251 (thanks to Alexey Dejneka)
3252 * fixed several bugs in PCL's error checking (thanks to Gerd
3254 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3256 * fixed bug in compilation of functions as first class values
3257 (thanks to Antonio Martinez-Shotton)
3258 * The compiler's handling TYPE-ERRORs which it can prove will
3259 inevitably happen at runtime has been cleaned up and corrected
3260 in several ways. (thanks to Alexey Dejneka)
3261 * improved argument type checking for various basic arithmetic
3262 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3263 lavished on them in the past that they can be compiled in many
3264 ways in different special cases
3265 * fixed bug 181: compiler checks validity of user supplied type
3267 * cleaned up code flushing in optimization: Function calls which
3268 should signal errors for safety purposes (e.g. which ANSI says
3269 should signal errors when their arguments are of incorrect type)
3270 are no longer optimized away.
3271 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3272 * incremented fasl file version number, because changes in the
3273 implementation of sequence functions like COERCE caused
3274 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3275 in old inline expansions) to become undefined. (Actually these
3276 changes were later undone, so we might very well be binary
3277 compatible with 0.7.7 after all, but leaving the version number
3278 incremented seemed like the simplest and most conservative
3281 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3282 * An alpha-quality port to the parisc architecture running Linux,
3283 based on the old CMUCL backend, has been made. This, even more so
3284 than the other backends, should be considered still a work in
3285 progress; known problems include that the Linux kernel in 64-bit
3286 mode does not propagate the correct sigcontext structure to
3287 userspace, and consequently SBCL on a parisc64 kernel will not
3289 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3290 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3291 suppressed.) Also now that the compiler is looking at declarations
3292 in the environment, it checks optimization declarations as well,
3293 and suppresses inlining when (> DEBUG SPEED).
3294 * More fixes have been made to treatment of floating point exception
3295 treatment and other Unix signals. In particular, floating point
3296 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3297 * The detection and handling of control stack exhaustion (infinite
3298 or very deeply nested recursion) has changed. Stack exhaustion
3299 detection is now done by write-protecting pages at the OS level
3300 and applies at all optimization settings; when found, a
3301 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3302 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3303 be caught using IGNORE-ERRORS.
3304 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3305 correctly when transforming recursive LABELS functions to LETs.
3306 (thanks to Alexey Dejneka)
3307 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3308 that are names of constants or global variables.
3309 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3310 alien routines with docstrings.
3311 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3312 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3314 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3315 (thanks to Eric Marsden)
3316 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3317 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3318 * Bug fix: The SPARC backend can now compile functions involving
3319 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3320 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3321 to a functions expecting a PATHNAME-DESIGNATOR.
3322 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3323 lambda lists. (thanks to Alexey Dejneka)
3324 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3325 functions so that EXPORTing the name of the class doesn't cause
3326 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3327 to Antonio Martinez for reporting this)
3328 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3329 constant-folded without failing an assertion. (thanks to Einar
3330 Floystad Dorum for reporting this)
3331 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3332 types are now dealt with more correctly by the compiler.
3333 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3334 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3335 behaviour analogous to RENAME-FILE. This puts its behaviour more
3336 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3338 * The fasl file version number has changed again. (because of the
3339 bug fix involving the names of PCL MAKE-INSTANCE functions)
3341 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3342 * bug fix: Floating point exceptions are treated much more
3343 consistently on the x86/Linux and PPC/Linux platforms.
3344 * Array initialization with :INITIAL-ELEMENT is now much faster for
3345 cases when the compiler cannot open code the array creation, but
3346 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3347 array accesses have also seen a speed increase.
3348 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3349 specified. (This was at the root of some bad interactions between
3350 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3352 * bug fix: Internal error arguments for undefined functions are now
3353 computed correctly on the PPC/Linux platform.
3354 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3355 Raymond Toy's patch for CMU CL)
3356 * Support for the Solaris 9 operating environment has been included
3357 (thanks to Daniel Merritt)
3358 * A very ugly but hopefully complete draft of the missing FFI chapter
3359 of the manual has been created by reformatting the corresponding
3360 CMU CL manual chapter into (currently very ugly and incoherent)
3361 DocBook and bringing it up to date for SBCL behavior. Thus, the
3362 manual is now essentially complete, at least by my extreme
3363 once-and-only-once standards, whereby it's acceptable to refer to
3364 the doc strings of SB-EXT functions as the primary documentation.
3365 * The fasl file version number has changed again, due to cleanup of
3366 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3367 is no longer a static symbol.)
3369 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3370 * SBCL now builds with OpenMCL (version 0.12) as the
3371 cross-compilation host; also, more progress has been made toward
3372 bootstrapping under CLISP.
3373 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3375 * bug 158 fixed: The compiler can now deal with integer loop
3376 increments different from 1; fixing this turned out also to fix
3378 * bug 169 fixed: no more bogus warnings about using lexical bindings
3379 despite the presence of perfectly good SPECIAL declarations (thanks
3380 to David Lichteblau)
3381 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3382 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3383 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3385 * bug 180 fixed: Method combination specifications no longer ignore
3386 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3387 * bug fix: Structure type predicate functions now check their argument
3388 count as they should.
3389 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3390 correctly. (thanks to Pierre Mai)
3391 * minor incompatible change: The --noprogrammer option is deprecated
3392 in favor of the new --disable-debugger option, which is very similar.
3393 (The major difference is that it takes effect at a slightly different
3394 time at startup, causing handling of errors in --sysinit and
3395 --userinit files will be affected differently.) The
3396 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3397 been added to allow this functionality to be controlled from ordinary
3398 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3399 maintainers, who might want to run non-interactive scripts to
3400 build SBCL cores which will later be used interactively.)
3401 * minor incompatible change: The LOAD function no longer, when given
3402 a wild pathname to load, loads all files matching that pathname.
3403 Instead, an error of type FILE-ERROR is signalled.
3405 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3406 * bug 147 fixed: The compiler preserves its block link/count
3407 invariants more correctly now so that it doesn't crash. (thanks
3409 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3411 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3412 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3413 no-value-for-this-character as NIL (as specified by ANSI).
3414 * HOST-NAMESTRING on physical pathnames now returns a string that is
3415 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3416 (thanks to Christophe Rhodes)
3417 * The Alpha port handles icache flushing more correctly. (thanks to
3419 * More progress has been made toward bootstrapping under CLISP. (thanks
3420 to Christophe Rhodes)
3421 * The fasl file format has changed again, because dynamic loading
3422 on OpenBSD (which has non-ELF object files) motivated some cleanups
3423 in the way that foreign symbols are transformed and passed around.
3424 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3425 is no longer treated as whitespace by the reader, but instead as
3426 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3427 |A
\7fB|, instead of A as it used to.
3429 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3430 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3431 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3432 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3433 this as of 0.7.1.45, but was left out of the previous news section
3434 (thanks to Dan Barlow)
3435 * SBCL now runs on the Solaris operating system on SPARC architectures
3436 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3437 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3438 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3439 * SPARC backend cleanups, allowing builds of cores optimized for V8
3440 and V9 SPARCS, and also emission of code targeted to a particular
3441 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3443 * SBCL is closer to bootstrapping under CLISP, thanks to various
3444 fixes by Christophe Rhodes.
3445 * The fasl file format has changed again, to allow the compiler's
3446 INFO database to support symbol macros.
3447 * The user manual (in doc/) is formatted into HTML more nicely.
3448 (thanks to coreythomas)
3449 * The system is smarter about SUBTYPEP relationships, especially
3450 those involving NOT types (including types such as ATOM which are
3451 represented internally using NOT types). Thus SUBTYPEP is less
3452 likely to return (VALUES NIL NIL) in general, and in particular
3453 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3455 * The fasl file format has changed again, because the internal
3456 representation of types now includes a new slot to support the new
3457 SUBTYPEP-of-NOT-types logic.
3458 * (not a change in the main branch of SBCL, but a related prototype
3459 which can hopefully be merged into the main branch of SBCL in the
3460 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3461 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3462 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3463 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3464 dispatch macro character combinations. (thanks to Alexey Dejneka)
3465 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3466 that superficially look like logical namestrings correctly.
3467 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3469 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3470 default supplied-p) for &optional and &key arguments. (thanks to
3473 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3474 * incompatible change: The compiler is now less aggressive about
3475 tail call optimization, doing it only when (> SPACE DEBUG) or
3476 (> SPEED DEBUG). (This is an incompatible change because there are
3477 programs which relied on the old CMU-CL-style behavior to optimize
3478 away their unbounded recursion which will now die of stack overflow.)
3479 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3480 for non-GENCGC systems has been increased to 20M (since that
3481 seems much closer to the likely performance optimum for modern
3482 systems than the old 4M value was)
3483 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3484 and *DEBUG-PRINT-LEVEL*
3485 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3486 of CMU CL's support for SPARC, and various endianness and other
3487 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3488 * new syntactic sugar for the Unix command line: --load foo.bar is now
3489 an alternate notation for --eval '(load "foo.bar")'.
3491 ** The system now detects stack overflow and handles it gracefully,
3492 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3493 optimization settings. (This is a good thing in general, and
3494 its introduction in this version should be particularly timely
3495 for anyone whose code fails because of suppression of tail
3497 ** The system now hunts for the C variable "environ" in a more
3498 devious way, to avoid segfaults when the C library version differs
3499 between compile time and run time. (thanks to Christophe Rhodes)
3500 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3501 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3502 ** The compiler no longer issues bogus style warnings for undefined
3503 classes in the same source file as the DEFCLASSes which defined
3504 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3506 ** fixes in CONDITION class precedence list for undefined function
3507 errors (thanks to Alexei Dejneka)
3508 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3509 correctly. (thanks to Dan Barlow)
3510 ** portability fixes aiming at bootstrapping under CLISP (thanks
3511 to Dave McDonald and Christophe Rhodes)
3512 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3513 ** fixes in type translation and and type inference (thanks to
3515 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3516 ** various fixes in the new ports (thanks to Dan Barlow)
3517 * several changes related to debugging:
3518 ** suppression of tail recursion, as noted above
3519 ** stack overflow detection, as noted above
3520 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3521 is now the default. (For some time encapsulation has been more
3522 reliable than the breakpoint-based :ENCAPSULATE NIL
3523 implementation, at least on X86 systems; and I just noticed that
3524 encapsulation also seems closer to the spirit of the ANSI
3527 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3529 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3530 up properly again. (There was a packaging bug in 0.7.0 which
3531 left their definitions in SB-SYS::LOAD-FOREIGN and
3532 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3533 vital for most things which interface to C-level interfaces,
3534 like extensions working with sockets or databases or
3535 Perl-compatible regexes or whatever, and the need to fix
3536 this bug is the main reason that 0.7.1 was released so
3538 ** DEFGENERIC is now choosier about the methods it redefines, so that
3539 reLOADing a previously-LOADed file containing DEFGENERICs does
3540 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3541 cycle now works as it should. (thanks to Alexey Dejneka)
3542 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3543 fixed by Christophe Rhodes. (He actually submitted this patch
3544 months ago, and I delayed until after 0.7.0.)
3545 ** Bug 111 (internal compiler confusion about runtime checks on
3546 FUNCTION types) was fixed by Alexey Dejneka.
3547 * Some internal cleanups (getting rid of variables which aren't
3548 needed now that the byte interpreter is gone) caused the fasl
3549 file format number to change again.
3551 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3552 * major incompatible change: The default fasl file extension, i.e. the
3553 default extension for files produced by COMPILE-FILE, has changed
3554 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3556 ** There are many changes in the implementation of the compiler.
3557 SBCL is now essentially a compiler-only implementation of ANSI
3558 Common Lisp. EVAL still "interprets" a few special cases, but
3559 almost all the interesting cases are handled by creating
3560 a LAMBDA expression, calling COMPILE on it, then calling
3561 FUNCALL on the result.
3562 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3563 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3564 Since the code is newer, there might still be some new bugs
3565 (though not as many as before Martin Atzmueller's fixes:-). But
3566 the new code is substantially simpler and clearer, and hopefully
3567 any remaining bugs will be simpler, less fundamental, and more
3568 fixable then the bugs in the old code.
3569 ** The revised compiler is still a little unsteady on its feet.
3571 *** The debugging information it produces (particularly the names
3572 of FUNCTION objects) is sometimes much less useful than what
3573 the old compiler produced.
3574 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3575 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3576 within a MACROLET) has been temporarily weakened.
3577 ** There are new compiler optimizations for various functions:
3578 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3579 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3580 *** the math functions TRUNCATE, FLOOR, and CEILING
3581 *** the function-of-all-trades COERCE
3582 Mostly these should be transparent, but there's one
3583 potentially-annoying problem (bug 117): when the compiler
3584 inline-expands a function and does type analysis on the result,
3585 it can create control paths which have type mismatches, and
3586 when it can't prove that those control paths aren't taken,
3587 it will issue WARNINGs about the type mismatches. This is
3588 a particular problem in practice for the new sequence functions.
3589 It's not clear how this should be fixed, and for now, a
3590 workaround is given in the entry for 117 in the BUGS file.
3591 ** (Because of the interaction between the two previous items --
3592 occasional inlining problems and new inline expansions -- some
3593 of the new sequence function optimizations won't really kick in
3594 completely until debugging information, and then inlining, are
3595 straightened out in some future version.)
3596 * minor incompatible changes:
3597 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3598 DIRECTORY no longer implicitly promotes NIL slots of its
3599 pathname argument to :WILD. In particular, when you ask for the
3600 contents of a directory (which you used to be able to do without
3601 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3602 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3603 ** changes in behavior that ANSI explicitly defines to be
3604 implementation dependent:
3605 *** The new compiler-only implementation still conforms with ANSI,
3606 but acts a little different than before. Besides the obvious
3607 changes in performance tradeoffs (that the cost per form passed
3608 to EVAL has gone up, and the cost per form executed by EVAL
3609 has gone down), the behavior of the system changes a little
3610 because there are no longer any interpreted function objects.
3611 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3612 e.g. doing COMPILE on the output of interactive DEFUN is
3614 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3616 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3617 has changed from (:CL) to NIL.
3618 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3619 e.g. #\Newline and #\Tab, don't have names specified in the
3620 ANSI Common Lisp standard, is now based on their ASCII symbolic
3621 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3622 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3623 are no longer used for output.
3624 ** changes in internal implementation constants:
3625 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3626 4 million. (If your application spends a lot of time GCing and
3627 you have a lot of RAM, you might want to experiment with
3628 increasing it even more.)
3629 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3630 However, almost all old code should still continue to work without
3631 immediate update, as SB-C-CALL is now a (deprecated) nickname
3633 ** Old operator names in the style DEF-FOO are now deprecated in
3634 favor of new corresponding names DEFINE-FOO, for consistency with
3635 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3636 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3637 internal symbols, but a few supported extensions like
3638 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3639 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3640 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3641 etc. as you get deeper into recursive calls to the debugger
3642 command loop, instead of the old "5]", "5]]", "5]]]"
3643 sequence. (I was motivated to do this when squabbles between
3644 ILISP and SBCL left me very deeply nested in the debugger. In the
3645 short term, this change will probably provoke more ILISP/SBCL
3646 squabbles, but hopefully it will be an improvement in the long run.)
3647 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3648 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3649 function, since declaiming return types involving aliens is
3650 (1) annoyingly messy to do by hand and (2) vital to efficient
3651 compilation of code which calls such functions.
3652 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3653 longer reexported by the SB-EXT package. They're solely useful
3654 for alien code, so it seems more logical that you should get
3655 them from the SB-ALIEN package, not in SB-EXT.
3656 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3657 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3658 features. Instead, the code that they used to control is always
3659 built into the system.
3660 * many other bug fixes
3661 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3662 advantage of the new EVAL-WHEN stuff and to clean them up in
3663 general, and they are now more ANSI-compliant in a number of
3664 ways. Martin Atzmueller is responsible for a lot of this.
3665 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3667 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3668 *** correct ERROR type for various file operations
3669 *** some fixes for Lisp streams
3670 *** DEFMETHOD syntax checking
3671 *** changing old weird representation of debug information as
3672 strings (which, among their other deficiencies, don't transform
3673 correctly when you rename packages, and don't change their
3674 print representation when you change things like *PACKAGE*
3675 and *PRINT-LENGTH*) to symbols and lists of symbols
3676 He also made several improvements and fixed several bugs in DESCRIBE.
3677 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3679 *** misbehavior of WRITE-STRING/WRITE-LINE
3680 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3681 and several other LOOP problems as well
3682 *** DIRECTORY when similar filenames are present
3683 *** DEFGENERIC with :METHOD options
3684 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3685 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3686 *** argument ordering in FIND with :TEST option
3687 *** mishandled package designator argument in APROPOS-LIST
3688 *** various problems in the backquote readmacro
3689 *** a bug in APROPOS
3690 *** probably some others that I'm not describing very well here,
3691 since the CVS log documents them by reference to sbcl-devel
3692 messages, and the SourceForge archives aren't working well.:-(
3693 ** Dan Barlow improved the Alpha port (and is making progress on the
3694 PPC port, for those of you who think different).
3695 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3696 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3697 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3698 continues to work on the SPARC port (for those of you in a position
3699 to look down upon our little PC-compatible boxes from a great height).
3700 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3701 and uses that copy, rather than the current dynamic value, when
3702 it's trying to decide whether to truncate output. Thus e.g.
3703 (let ((*print-lines* 50))
3704 (pprint-logical-block (stream nil)
3706 (let ((*print-lines* 8))
3707 (print (aref possiblybigthings i) stream)))))
3708 should now truncate the logical block only at 50 lines, instead of
3709 often truncating it at 8 lines, as it did before.
3710 * The doc/cmucl/ directory, containing old CMU CL documentation
3711 from the time of the fork, is no longer part of the base system.
3712 SourceForge has shut down its anonymous FTP service, and with it
3713 my original plan for distributing the old CMU CL documentation
3714 there. For now, if you need these files you can download an old
3715 SBCL source release and extract them from it.
3716 * The fasl file version number changed again, for dozens of reasons,
3717 some of which are apparent above.
3719 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3720 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3721 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3722 patch, so that DISASSEMBLE output is much nicer.
3723 * The code in the SB-PROFILE package now seems reasonably stable.
3724 I still haven't decided what the final interface should look like
3725 (I'd like PROFILE to interact cleanly with TRACE, since both
3726 facilities use function encapsulation) but if you have a need
3727 for profiling now, you can probably use it successfully with
3728 the current CMU-CL-style interface.
3729 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3730 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3731 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3732 link now returns the dangling link itself, and for similar
3733 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3734 link itself. (In these cases the old code signalled an error and
3735 looped endlessly, respectively.) Thus, DIRECTORY now works even
3736 in the presence of dangling and cyclic symbolic links.
3737 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3738 is now a supported extension again, since the consensus on
3739 sbcl-devel was that it can be useful for ordinary development
3740 work, not just for debugging SBCL itself.
3741 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3742 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3743 that currently-defined functions might be redefined later with
3744 different return types.
3745 * Hash tables can be printed readably, as inspired by CMU CL code
3746 of Eric Marsden and SBCL code of Martin Atzmueller.
3747 * better error handling in CLOS method combination, thanks to
3748 Martin Atzmueller porting Pierre Mai's CMU CL patches
3749 * more overflow fixes for >16Mbyte I/O buffers
3750 * A bug in READ has been fixed, so that now a single Ctrl-D
3751 character suffices to cause end-of-file on character streams.
3752 In particular, now you only need one Ctrl-D at the command
3753 line (not two) to exit SBCL.
3754 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3756 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3757 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3758 does the right thing.
3759 * The implementation of some type tests, especially for CONDITION
3760 types, is now tidier and maybe faster, due to CMU CL code
3761 originally by Douglas Crosher, ported by Martin Atzmueller.
3762 * Some math functions have been fixed, and there are new
3763 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3764 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3765 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3766 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3767 likely to be built into sbcl-0.7.0.)
3768 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3769 CL, ported to SBCL by Martin Atzmueller)
3770 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3771 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3772 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3773 * There's a new slam.sh hack to shorten the edit/compile/debug
3774 cycle for low-level changes to SBCL itself, and a new
3775 :SB-AFTER-XC-CORE target feature to control the generation of
3776 the after-xc.core file needed by slam.sh.
3777 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3778 COMPILE-FILE is no longer supported, so that now every function
3779 gets an entry point, so that block compilation looks a little
3780 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3781 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3782 returns the number of bytes consed since the system started,
3783 rather than the number consed since the first time the function
3784 was called. (The new definition parallels ANSI functions like
3785 CL:GET-INTERNAL-RUN-TIME.)
3786 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3787 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3788 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3789 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3790 * Fasl file version numbers are now independent of the target CPU,
3791 since historically most system changes which required version
3792 number changes have affected all CPUs equally. Similarly,
3793 the byte fasl file version is now equal to the ordinary
3796 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3797 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3798 is no longer recognized. I apologize for this, because it was
3799 listed in SB-EXT as a supported extension, but I found that
3800 its existing behavior was poorly specified, as well as incorrectly
3801 specified, and it looked like too much of a mess to straighten it
3802 out. I have enough on my hands trying to get ANSI stuff to work..
3803 * many patches ported from CMU CL by Martin Atzmueller, with
3804 half a dozen bug fixes in pretty-printing and the debugger, and
3805 half a dozen others elsewhere
3806 * fixed bug 13: Floating point infinities are now supported again.
3807 They might still be a little bit flaky, but thanks to bug reports
3808 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3809 as flaky as they were.
3810 * The --noprogrammer command line option is now supported. (Its
3811 behavior is slightly different in detail from what the old man
3812 page claimed it would do, but it's still appropriate under the
3813 same circumstances that the man page talks about.)
3814 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3815 are now supported, and enabled by default. Thus, the compiler can
3816 handle many floating point and complex operations much less
3817 inefficiently. (Thus e.g. you can implement a complex FFT
3819 * The compiler now detects type mismatches between DECLAIM FTYPE
3820 and DEFUN better, and implements CHECK-TYPE more correctly, and
3821 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3822 thanks to patches from Martin Atzmueller.
3823 * various fixes to make the cross-compiler more portable to
3824 ANSI-conforming-but-different cross-compilation hosts (notably
3825 Lispworks for Windows, following bug reports from Arthur Lemmens)
3826 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3827 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3828 fixed thanks to Pierre Mai's CMU CL patches.
3829 * fixes to keep the system from overflowing internal counters when
3830 it tries to use i/o buffers larger than 16M bytes
3831 * fixed bug 45a: Various internal functions required to support
3832 complex special functions have been merged from CMU CL sources.
3833 (When I was first setting up SBCL, I misunderstood a compile-time
3834 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3835 * improved support for type intersection and union, fixing bug 12
3836 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3837 more obscure bugs as well
3838 * some steps toward byte-compiling non-performance-critical
3839 parts of the system, courtesy of patches from Martin Atzmueller
3840 * Christophe Rhodes has made some debian packages of sbcl at
3841 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3842 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3843 stable, but are nonetheless usable. When he's ready, I'd be happy
3844 to add them to the SourceForge "File Releases" section. (And if
3845 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3846 * new fasl file format version number (because of changes in
3847 internal representation of (OR ..) types to accommodate the new
3848 support for (AND ..) types, among other things)
3850 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3851 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3853 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3855 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3856 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3857 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3858 * DESCRIBE now works on structure objects again.
3859 * Most function call argument type mismatches are now handled as
3860 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3861 know whether the function will be redefined before the call is
3862 executed. (The compiler could flag local calls with full WARNINGs,
3863 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3864 it doesn't keep track of enough information to know whether calls
3865 are local in this sense.)
3866 * Compiler output is now more verbose, with messages truncated
3867 later than before. (There should be some supported way for users
3868 to override the default verbosity, but I haven't decided how to
3869 provide it yet, so this behavior is still controlled by the internal
3870 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3871 src/compiler/ir1util.lisp.)
3872 * Fasl file format version numbers have increased again, because
3873 support for the Gray streams extension changes the layout of the
3874 system's STREAM objects.
3875 * The Gray subclassable streams extension now works, thanks to a
3876 patch from Martin Atzmueller.
3877 * The full LOAD-FOREIGN extension (not just the primitive
3878 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3879 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3880 but like most other programs, it defaults to copying the Unix
3881 environment from the original process instead of starting the
3882 new process in an empty environment.
3883 * Extensions which manipulate the Unix environment now support
3884 an :ENVIRONMENT keyword option which doesn't smash case or
3885 do other bad things. The CMU-CL-style :ENV option is retained
3886 for porting convenience.
3887 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3888 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3890 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3892 * A patch from Martin Atzmueller seems to have solved the SIGINT
3893 problem, and as far as we know, signal-handling now works cleanly.
3894 (If you find any new bugs, please report them!)
3895 * The system no longer defaults Lisp source file names to types
3896 ".l", ".cl", or ".lsp", but only to ".lisp".
3897 * The compiler no longer uses special default file extensions for
3898 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3899 seems to expect a single default extension for all compiled code,
3900 and there's no compelling reason to try to stretch the standard
3901 to allow two different extensions.) Instead, byte-compiled files
3902 default to the same extension as native-compiled files.
3903 * Fasl file format version numbers have increased again, because
3904 a rearrangement of internal implementation packages made some
3905 dumped symbols in old fasl files unreadable in new cores.
3906 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3907 many fewer weird special cases.
3908 * Bug #17 (differing COMPILE-FILE behavior between logical and
3909 physical pathnames) has been fixed, and some related misbehavior too,
3910 thanks to a patch from Martin Atzmueller.
3911 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3912 by Tim Moore, ported to SBCL by Martin Atzmueller.
3913 * Martin Atzmueller fixed several filesystem-related problems,
3914 including bug #36, in part by porting CMU CL patches, which were
3915 written in part by Paul Werkowski.
3916 * More compiler warnings in src/runtime/ are gone, thanks to
3917 more patches from Martin Atzmueller.
3918 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3921 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3923 * DESCRIBE now works on CONDITION objects.
3924 * The debugger now handles errors which arise when trying to print
3925 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3927 * The build system now uses an additional file, customize-target-features.lisp,
3928 to allow local modifications to the target *FEATURES* list. (The point of
3929 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3930 debugging features enabled, without having to worry about propagating it
3931 into everyone's system when I do a "cvs update".) When no
3932 customize-target-features.lisp file exists, the target *FEATURES* list
3933 should be constructed the same way as before.
3934 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3935 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3936 warn about behavior which is undefined under the ANSI spec. Note: This
3937 is specified by ANSI, but it's not very popular with programmers.
3938 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3939 macro in the SBCL sources for an example of a workaround which you
3940 might use to make portable ANSI-standard code which does what you want.
3941 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3942 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3943 when it's used as a non-toplevel form. (This is required in order
3944 to implement the DEFCONSTANT-EQX macro.)
3945 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3946 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3947 funny things, due to bugs in the implementation of EVAL-WHEN
3948 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3949 non-toplevel DEFCONSTANTs are uncommon.)
3950 * The core file version number and fasl file version number have been
3951 incremented, because the old noncompliant DEFCONSTANT behavior involved
3952 calling functions which no longer exist, and because I also took the
3953 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3954 * fixed bug 1 (error handling before read-eval-print loop starts), and
3955 redid debugger restarts and related debugger commands somewhat while
3957 ** The QUIT debugger command is gone, since it did something
3958 rather different than the SB-EXT:QUIT command, and since it never
3959 worked properly outside the main toplevel read/eval/print loop.
3960 Invoking the new TOPLEVEL restart provides the same functionality.
3961 ** The GO debugger command is also gone, since you can just invoke
3962 the CONTINUE restart directly instead.
3963 ** The TOP debugger command is also gone, since it's redundant with the
3964 FRAME 0 command, and since it interfered with abbreviations for the
3966 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3968 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3969 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3970 some porting work by Martin Atzmueller)
3971 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3972 with a boatload of patches which he ported to SBCL. Now that those
3974 ** The system tries to make sure that its low-priority messages
3975 are prefixed by semicolons, to help people who like to use
3976 syntax highlighting in their ILISP buffer. (This patch
3977 was originally due to Raymond Toy.)
3978 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3979 patches originally written by Raymond Toy.
3980 ** The compiler understands coercion between single-value and
3981 multiple-VALUES type expressions better, getting rid of some very
3982 weird behavior, thanks to patches originally by Robert MacLachlan
3983 and Douglas Crosher.
3984 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3985 lambda lists, thanks to a patch originally by Pierre Mai.
3986 ** The system no longer bogusly warns about "abbreviated type
3988 ** The compiler gets less confused by inlining and RETURN-FROM,
3989 thanks to some patches originally by Tim Moore.
3990 ** The system no longer hangs when dumping circular lists to fasl
3991 files, thanks to a patch originally from Douglas Crosher.
3992 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3993 undefined function error.
3994 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3995 in structure slot initforms) for now, documented workaround instead:-|
3996 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3997 * fixed bug 5: added stubs for various Gray stream functions called
3998 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3999 installed, at least appropriate type errors are generated
4000 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4001 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4002 consistently in DEFMETHOD forms.
4003 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4004 it doesn't seem to affect SBCL after all
4005 * The C runtime system now builds with better optimization and many
4006 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4008 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4010 * The system is now under CVS at SourceForge (instead of the
4011 CVS repository on my home machine).
4012 * The new signal handling code has been tweaked to treat register
4013 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4014 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4015 downstream, e.g. in debug-int.lisp, has implicit dependencies
4016 on the unsignedness of integer representation of machine words,
4017 and that caused the system to bomb out with infinite regress
4018 when trying to recover from type errors involving signed values,
4019 e.g. (BUTLAST '(1 2 3) -1).)
4020 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4021 (This was one of the bugs Peter Van Eynde reported back in July.)
4022 * The system now uses code inspired by Colin Walters' O(N)
4023 implementation of MAP (from the cmucl-imp@cons.org mailing
4024 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4025 inline the MAP operation, and there is more than one
4026 sequence argument to the MAP call (so that it can't just
4027 do ETYPECASE once and for all based on the type of the
4028 single sequence argument). (The old non-inline implementation
4029 of the general M-argument sequence-of-length-N case required
4030 O(M*N*N) time when any of the sequence arguments were LISTs.)
4031 * The QUIT :UNIX-CODE keyword argument has been renamed to
4032 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4033 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4035 * Raymond Wiker's patches to port dynamic loading from Linux to
4036 FreeBSD have been added.
4037 * The BUGS file is now more nearly up to date, thanks in large part
4038 to Martin Atzmueller's review of it.
4039 * The debugger now flushes standard output streams before it begins
4040 its output ("debugger invoked" and so forth).
4041 * The core version number and fasl file version number have both
4042 been incremented, because of incompatible changes in the layout
4044 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4045 used to not be called for a saved Lisp image.)
4046 * Martin Atzmueller's version of a patch to fix a compiler crash,
4047 as posted on sbcl-devel 13 September 2000, has been installed.
4048 * Instead of installing Martin Atzmueller's patch for the
4049 compiler transform for SUBSEQ, I deleted the compiler transform,
4050 and transforms for some similar consing operations.
4051 * A bug in signal handling which kept TRACE from working on OpenBSD
4053 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4054 away by constant folding
4055 * The system now defines its address space constants in one place
4056 (in the Lisp sources), and propagates them automatically elsewhere
4057 (through GENESIS and the sbcl.h file). Therefore, patching the
4058 address map is less unnecessarily tedious and error-prone. The
4059 Lisp names of address space constants have also been systematized.
4060 * CVS tags like dollar-Header-dollar have been removed from
4061 the sources, because they have never saved me trouble and
4062 they've been source of trouble working with patches and other
4063 diff-related operations.
4064 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4065 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4067 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4069 * The system has been ported to OpenBSD.
4070 * The system now compiles with a simple "sh make.sh" on the systems
4071 that it's supported on. I.e., now you no longer need to tweak
4072 text in the target-features.lisp-expr and symlinks in src/runtime/
4073 by hand, the make.sh takes care of it for you.
4074 * The system is no longer so grossly inefficient when compiling code
4075 involving vectors implemented as general (not simple) vectors (VECTOR T),
4076 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4077 which dares to use the various sequence functions on non-simple
4078 vectors, takes less of a performance hit.
4079 * There is now a primitive type predicate VECTOR-T-P
4080 to test for the (VECTOR T) type, so that e.g.
4081 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4082 can now be compiled with some semblance of efficiency. (The old code
4083 turned the type declaration into a full call to %TYPEP at runtime!)
4084 * AREF on (VECTOR T) is still not fast, since it's still compiled
4085 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4086 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4087 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4088 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4089 without making full calls to SUBTYPEP at runtime.
4090 (Some analogous efficiency issues for non-simple vectors specialized to
4091 element types other than T, or for non-simple multidimensional arrays,
4092 have not been addressed. They could almost certainly be handled the
4093 same way if anyone is motivated to do so.)
4094 * The changes in array handling break binary compatibility, so
4095 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4096 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4097 instead of (VALUES T T).
4098 * By following the instructions that Dan Barlow posted to sbcl-devel
4099 on 2 July 2000, I was able to enable primitive dynamic object
4100 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4101 functionality is not implemented (since it calls ld to resolve
4102 library references automatically, requiring RUN-PROGRAM for its
4103 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4104 try to resolve library references) is now supported.
4105 * The system now flushes the standard output streams when it terminates,
4106 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4107 them at several other probably-convenient times, e.g. in each pass of
4108 the toplevel read-eval-print loop, and after evaluating a form given
4109 as an "--eval" command-line option. (These changes were motivated by a
4110 discussion of stream flushing issues on cmucl-imp in August 2000.)
4111 * The source transform for TYPEP of array types no longer assumes
4112 that an array whose element type is a not-yet-defined type
4113 is implemented as an array of T, but instead punts, so that the
4114 type will be interpreted at runtime.
4115 * There is now some support for cross-compiling in make.sh: each of
4116 the phases of make.sh has its own script. (This should be transparent
4117 to people doing ordinary, non-cross-compile builds.)
4118 * Since my laptop doesn't have hundreds of megabytes of memory like
4119 my desktop machine, I became more motivated to do some items on
4120 my to-do list in order to reduce the size of the system a little:
4121 ** Arrange for various needed-only-at-cold-init things to be
4122 uninterned after cold init. To support this, those things have
4123 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4124 symbols with such names are now uninterned after cold init).
4125 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4126 fasl files for cold load.
4127 ** Remove the old compiler structure pooling code (which used to
4128 be conditional on the target feature :SB-ALLOC) completely.
4129 ** Redo the representation of some data in cold init to be more compact.
4130 (I also looked into supporting byte compiled code at bootstrap time,
4131 which would probably reduce the size of the system a lot, but that
4132 looked too complicated, so I punted for now.)
4133 * The maximum signal nesting depth in the src/runtime/ support code has
4134 been reduced from 4096 to 256. (I don't know any reason for the very
4135 large old value. If the new smaller value turns out to break something,
4136 I'll probably just bump it back up.)
4137 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4139 * Many, many bugs reported by Peter Van Eynde have been added to
4140 the BUGS list; some have even been fixed.
4141 * While enabling dynamic object file loading, I tried to make the
4142 code easier to understand, renaming various functions and variables
4143 with less ambiguous names, and changing some function calling
4144 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4145 * While trying to figure out how to do the OpenBSD port, I tried to
4146 clean up some of the code in src/runtime/. In particular, I dropped
4147 support for non-POSIX signal handling, added various comments,
4148 tweaked the code to reduce the number of compilation warnings, and
4149 renamed some files to increase consistency.
4150 * To support the new automatic configuration functionality in make.sh,
4151 the source file target-features.lisp-expr has been replaced with the
4152 source file base-target-features.lisp-expr and the machine-generated
4153 file local-target-features.lisp-expr.
4154 * fixed a stupid quoting error in make.sh so that using CMU CL
4155 "lisp -batch" as cross-compilation host works again
4157 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4159 * DESCRIBE no longer tries to call itself recursively to describe
4160 bound/fbound values, so that it no longer fails on symbols which are
4161 bound to themselves (like keywords, T, and NIL).
4162 * DESCRIBE now works on generic functions.
4163 * The printer now prints less-screwed-up representations of closures
4164 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4165 * A private symbol is used instead of the :EMPTY keyword previously
4166 used to mark empty slots in hash tables. Thus
4167 (DEFVAR *HT* (MAKE-HASH-TABLE))
4168 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4169 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4170 now does what ANSI says that it should. (You can still get
4171 similar noncompliant behavior if bang on the hash table
4172 implementation with all the symbols you get back from
4173 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4174 This breaks binary compatibility, since tests for equality to
4175 :EMPTY are wired into things like the macroexpansion of
4176 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4178 * There's now a minimal placeholder implementation for CL:STEP,
4179 as required by ANSI.
4180 * An obscure bug in the interaction of the normal compiler, the byte
4181 compiler, inlining, and structure predicates has been patched
4182 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4183 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4185 * Missing ordinary arguments in a macro call are now detected even
4186 when the macro lambda list contains &KEY or &REST.
4187 * The debugger no longer complains about encountering the top of the
4188 stack when you type "FRAME 0" to explicitly instruct it to go to
4189 the top of the stack. And it now prints the frame you request even
4190 if it's the current frame (instead of saying "You are here.").
4191 * As specified by ANSI, the system now always prints keywords
4192 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4194 * The default initial SIZE of HASH-TABLEs is now smaller.
4195 * Type information from CLOS class dispatch is now propagated
4196 into DEFMETHOD bodies, so that e.g.
4197 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4199 is now basically equivalent to
4200 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4201 (DECLARE (TYPE SINGLE-FLOAT X))
4203 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4204 operation, without having to do run-time type dispatch.
4205 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4206 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4207 * Since I don't seem to be making big file reorganizations very often
4208 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4209 is overflowing my ability to conveniently back them up), I've finally
4210 checked the system into CVS. (The CVS repository is on my home system,
4211 not at SourceForge -- putting it on SourceForge might come later.)
4212 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4213 high-level GC-NOTIFY-FOO functions send their output. (There's
4214 still very little control of where low-level verbose GC functions
4215 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4216 less than it used to -- the GC-NOTIFY-FOO functions are now under
4217 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4218 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4219 in only one place in the source code, and propagates it automatically
4220 everywhere that it's needed. Thus e.g. when I bump the version from
4221 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4222 * The C source files now include boilerplate legalese and documentation
4223 at the head of each file (just as the Lisp source files already did).
4224 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4225 to his page will be replaced with a link to his new CLiki service.
4227 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4229 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4230 * The build process now looks for GNU make under the default name "gmake",
4231 instead of "make" as it used to. If GNU make is not available as "gmake"
4232 on your system, you can change this default behavior by setting the
4233 GNUMAKE environment variable.
4234 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4235 can build without error under CMU CL.
4237 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4239 * There is now a partial SBCL user manual (with some new text and some
4240 text cribbed from the CMU CL manual).
4241 * The beginnings of a profiler have been added (starting with the
4242 CMU CL profiler and simplifying and cleaning up). Eventually the
4243 main interface should be through the TRACE macro, but for now,
4244 it's still accessed through vaguely CMU-CL-style functions and macros
4245 exported from the package SB-PROFILE.
4246 * Some problems left over from porting CMU CL to the new
4247 cross-compilation bootstrap process have been cleaned up:
4248 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4249 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4250 part when getting rid of the compiler *BACKEND* stuff.)
4251 ** The value of *NULL-TYPE* was screwed up, because it was
4252 being initialized before the type system knew the final
4253 definition of the 'NULL type. This screwed up several key
4254 optimizations in the compiler, causing inefficiency in all sorts
4255 of places. (I found it because I wanted to understand why
4256 GET-INTERNAL-RUN-TIME was consing.)
4257 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4258 PROCLAIM FTYPE information. Unfortunately this broke binary
4259 compatibility again, since now the forms output by DEFGENERIC
4260 to refer to functions which didn't exist in 0.6.3.
4261 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4262 can use the new (as of 0.6.3) transform for SOME into MAP into
4264 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4265 values to (INTEGER 0 1000000), so that the time code will no longer
4266 occasionally get blown up by Linux returning 1000000 microseconds
4267 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4268 its output conform to the ANSI spec. (Alas, this makes its output
4269 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4270 * A full call to MAP NIL with a single sequence argument no longer conses.
4271 * fixes to problems pointed out by Martin Atzmueller:
4272 * The manual page no longer talks about multiprocessing as though
4273 it were currently supported.
4274 * The ILISP support patches have been removed from the distribution,
4275 because as of version 5.10.1, ILISP now supports SBCL without us
4276 having to maintain patches.
4277 * added a modified version of Raymond Toy's recent CMU CL patch for
4278 EQUALP comparison of HASH-TABLE
4280 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4282 * The system still can't cross-compile itself with
4283 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4284 entails), but at least it can compile more of itself that way
4285 than it used to be able to, and various buglets which were uncovered
4286 by trying to cross-compile itself that way have now been fixed.
4287 * This release breaks binary compatibility again. This time
4288 at least I've incremented the FASL file format version to 2, so that the
4289 problem can be detected reliably instead of just causing weird errors.
4290 * various new style warnings:
4291 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4292 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4293 ** using the lexical binding of a variable named in the *FOO* style
4294 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4295 as specified by ANSI.
4296 * *RANDOM-STATE* is no longer automatically initialized from
4297 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4298 default behavior of the system is to repeat its behavior every time
4299 it's run. If you'd like to change this behavior, you can always
4300 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4301 old convention there was no comparably easy way to get the system to
4302 repeat its behavior every time it was run.
4303 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4304 FTYPE declarations has been removed, in favor of their ANSI
4305 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4306 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4307 the types of their sequence arguments can be determined at compile time.
4308 This is done through a new open code expansion for MAP which eliminates
4309 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4310 argument types can be determined at compile time.
4311 * The optimizer now transforms COERCE into an identity operation when it
4312 can prove that the coerced object is already of the correct type. (This
4313 can be a win for machine generated code, including the output of other
4314 optimization transforms, such as the MAP transform above.)
4315 * Credit information has been moved from source file headers into CREDITS.
4316 * Source file headers have been made more standard.
4317 * The CASE macro now compiles without complaining even when it has
4320 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4322 * (Note that the way that the PCL macroexpansions were rewritten
4323 to accommodate the change in DEFGENERIC below breaks binary
4324 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4325 not run under sbcl-0.6.2. Once we get out of alpha releases,
4326 i.e. hit release 1.0.0, we'll probably try to maintain binary
4327 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4328 and sbcl-1.4.4. Until then, however, it might be fairly common
4329 for maintenance releases to break binary compatibility.)
4330 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4332 * The reporting of unhandled signals has been changed to print some
4333 explanatory text as well as the report form. (Previously only
4334 the report form was printed.)
4335 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4336 it defines, so that the compiler no longer issues undefined function
4337 warnings for compiled-but-not-yet-loaded generic functions.
4338 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4339 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4340 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4341 section of the ANSI spec are supported.
4342 * The "" nickname for the "KEYWORD" package has been removed.
4343 The reader still handles symbol tokens which begin with a package marker
4344 as keywords, but it doesn't expose its mechanism for doing so in the
4345 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4346 * The system now issues STYLE-WARNINGs for contradictory TYPE
4347 proclamations. (Warnings for contradictory FTYPE proclamations would
4348 be nice too, but those can't be done usefully unless the type system
4349 is made smarter about FUNCTION types.)
4350 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4351 systematized, so that "*target-*.lisp is supposed to exist only on the
4352 target and imply that there's a related file which exists on the
4353 host, and *host-*.lisp is supposed to exist only on the host and imply
4354 that there's a related file which exists on the target. This involves a
4355 lot of renaming. Hopefully the acute confusion caused by the renaming
4356 will be justified by the reduction in chronic confusion..
4357 ** runtime-type.lisp -> early-target-type.lisp
4358 ** target-type.lisp -> late-target-type.lisp
4359 ** early-host-format.lisp -> early-format.lisp
4360 ** late-host-format.lisp -> late-format.lisp
4361 ** host-error.lisp -> misc-error.lisp
4362 ** early-error.lisp -> early-target-error.lisp
4363 ** late-error.lisp -> late-target-error.lisp
4364 ** host-defboot.lisp -> early-defboot.lisp
4365 ** code/misc.lisp -> code/target-misc.lisp
4366 ** code/host-misc.lisp -> code/misc.lisp
4367 ** code/numbers.lisp -> code/target-numbers.lisp
4368 ** code/early-numbers.lisp -> numbers.lisp
4369 ** early-host-type.lisp -> early-type.lisp
4370 ** late-host-type.lisp -> late-type.lisp
4371 ** host-typep.lisp -> typep.lisp
4372 ** load.lisp -> target-load.lisp
4373 ** host-load.lisp -> load.lisp
4374 ** host-disassem.lisp -> disassem.lisp
4375 ** host-insts.lisp -> insts.lisp
4376 ** byte-comp.lisp -> target-byte-comp.lisp
4377 ** host-byte-comp.lisp -> byte-comp.lisp
4378 ** host-signal.lisp -> signal.lisp
4379 ** host-defstruct.lisp -> defstruct.lisp
4380 ** late-target-type.lisp -> deftypes-for-target.lisp
4381 Furthermore, several other previously target-only files foo.lisp (e.g.
4382 hash-table.lisp and random.lisp) have been split into a target-and-host
4383 foo.lisp file and a target-only target-foo.lisp file, with their key type
4384 definitions in the target-and-host part, so that the cross-compiler will
4385 know more about target types.
4386 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4387 gone away. In their place are various *BACKEND-FOO* variables
4388 corresponding to the slots of the old structure.
4389 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4390 into the target SBCL has been fixed.
4391 * The chill.lisp system for loading cold code into a running SBCL
4393 * Support for the CMU CL "scavenger hook" extension has been removed.
4394 (It was undocumented and unused in the CMU CL sources that SBCL was
4395 derived from, and stale in sbcl-0.6.1.)
4396 * Various errors in the cross-compiler type system were detected
4397 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4398 (enabling various consistency checks). Many of them were fixed,
4399 but some hard problems remain, so the compiler is back to
4400 running without *TYPE-SYSTEM-INITIALIZED* for now.
4401 * As part of the cross-compiler type system cleanup, I implemented
4402 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4403 * I have started adding UNCROSS calls throughout the type system
4404 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4405 on cross-compiler input files will be able to go away, and various
4407 * CONSTANTP now returns true for quoted forms (as explicitly required
4410 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4412 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4413 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4414 (MAKE-LIST -1). In the longer term, it ought to become true
4415 that declarations are assertions even at SAFETY 1. For now, it's not
4416 quite true even at SAFETY 3, but it's at least more nearly true..
4417 (Note that this change seems to increases the size of the system by
4418 O(5%) and to decrease the speed of the compiler by 20% or more.)
4419 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4420 for the problem of printing dozens of lines of distracting information
4421 about low-level system machinery as part of the top stack frame
4422 on entry to the debugger when an undefined function was called.
4423 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4424 is not reset to COMMON-LISP-USER.
4425 * Compilation of stuff related to dyncount.lisp has been made conditional
4426 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4427 smaller. The various dyncount-related symbols have been moved into
4428 a new "SB-DYNCOUNT" package.
4429 * tty-inspect.lisp has been renamed to inspect.lisp.
4430 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4431 feature has gone away. (When we eventually port to other flavors of
4432 libc and/or Unix, we'll try to make the differences between flavors
4433 invisible at the user level.)
4434 * Various other *FEATURES* tags, and/or their associated conditionals,
4435 have been removed if obsolescent, or given better documentation, or
4436 sometimes given more-mnemonic names.
4438 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4440 * tidied up "make.sh" script
4441 * tidied up system directory structure
4442 * better "clean.sh" behavior
4443 * added doc/FOR-CMUCL-DEVELOPERS
4444 * many many small tweaks to output format, e.g. removing possibly-confusing
4445 trailing #\. character in DESCRIBE-INSTANCE
4446 * (EQUALP #\A 'A) no longer signals an error.
4447 * new hashing code, including EQUALP hashing
4448 * tidied up Lisp initialization and toplevel
4449 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4450 * command line argument processing
4451 * added POSIX-GETENV function to deal with Unix-ish environment variables
4452 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4453 terminating SBCL on EOF
4454 * non-verbose GC by default
4455 * There is no more "sbcl" shell script; the sbcl file is now the C
4456 runtime executable (just like CMU CL).
4457 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4459 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4460 symbol and package information as Lisp native symbol and package objects
4461 instead of strings naming symbols and strings naming packages. This way,
4462 whenever packages are renamed (as in warm init), debug information is
4463 transformed along with everything else.
4464 * tweaked the optimization policy declarations which control the building
4465 of SBCL itself. Now, among other things, the system no longer saves
4466 source location debugging information. (This helps two problems at once
4467 by reducing SBCL size and by keeping SBCL from trying to look for its
4468 sources -- which may not exist -- when reporting errors.)
4469 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4470 debugging and testing purposes
4471 * cleaned up printing, making the printer call PRINT-OBJECT for
4472 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4473 methods, giving nearly-ANSI behavior
4474 * converted almost all special variables to use *FOO* naming convention
4475 * deleted PARSE-TIME functionality, since it can be done portably
4476 * moved some files out of cold init into warm init
4477 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4479 * regularized formatting of source files
4480 * added an install.sh script
4481 * fixed ridiculous memory usage of cross-compiler by making
4482 compiler/alloc.lisp not try to do pooling unless it can hook
4483 itself into the GC of the cross-compilation host. Now the system
4484 builds nicely on my old laptop.
4485 * added :SB-ALLOC in target-features.lisp-expr
4486 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4487 was not implemented)
4488 * re-did condition handling and note reporting in the compiler. Notes
4489 are no longer handled by signalling conditions. Style warnings
4490 and warnings are handled more correctly and reported in such a way
4491 that it's easy to find one or the other in your output (so that you
4492 can e.g. figure out which of many problems caused COMPILE-FILE to
4494 * changed the severity of several compiler warnings from full WARNING
4495 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4496 compiler note reporting so that it doesn't use the condition system
4497 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4498 COMPILE-FILE command)
4499 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4500 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4501 no longer supported, since their implementation was deeply intertwingled
4502 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4503 * removed bogus "support" for compiler macros named (SETF FOO), and
4504 removed the compiler macro for SETF INFO (but only after making a fool
4505 of myself on the cmucl-imp mailing list by posting a bogus patch for
4506 DEFINE-COMPILER-MACRO..)
4507 * Compiled files containing forms which have side effects on the Lisp
4508 reader (such as DEFPACKAGE forms) are now handled more correctly.
4509 (Compiler queuing of top level lambdas has been suppressed by setting
4510 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4511 * deleted various currently-unused source files, e.g. gengc.lisp. They
4512 may be added back at some point e.g. when porting to other architectures,
4513 but until they are it's distracting to distribute them and to try to
4515 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4516 there were so many of them they're just distractions, and UNCROSS is
4517 known to be able to handle the current sources
4518 * moved PROFILE functionality into TRACE, so that it will be clear
4519 how the wrapping and unwrapping of functions when you profile them
4520 interacts with the wrapping and unwrapping of functions when you
4521 trace them. (Actually, the functionality isn't there yet, but at least
4522 the interface specification is there. Hopefully, the functionality will
4523 arrive with some future maintenance release.)
4524 * removed host-oops.lisp
4525 * changed signature of QUIT function to allow UNIX-CODE argument
4526 * fixed READ-SEQUENCE bug
4527 * tweaked verbose GC output so that it looks more like the progress
4528 output that ANSI specifies for functions like LOAD
4529 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4530 * added <http://sbcl.sourceforge.com> to the banner information printed by