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 * optimization: SUBSEQ is 30-80% faster for strings and vectors
8 whose element-type or simplicity is not fully known at
10 * bug fix: some sequence functions elided bounds checking when
12 * bug fix: too liberal weakening of union-type checks when SPEED >
14 * bug fix: more bogus fixnum declarations in ROOM implementation
17 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
18 * new feature: MAKE-HASH-TABLE now experimentally accepts a
19 :SYNCHRONIZED argument, which makes the hash-table safe for
20 concurrent accesses (but not iteration.) See also:
21 SB-EXT:WITH-LOCKED-HASH-TABLE, and
22 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
23 * optimization: CONCATENATE on strings is an order of magnitue faster
24 in code compiled with (> SPEED SPACE).
25 * optimization: SUBSEQ is ~50% faster on lists.
26 * bug fix: bug 417 fixed -- source location reporting is now more robust.
27 * bug fix: SUBSEQ on a list will now correctly signal an error if if
28 END is smaller then START.
29 * bug fix: SB-PROFILE will no longer report extra consing for nested
30 calls to profiled functions.
31 * bug fix: ROOM implementation had bogus fixnum declarations which
32 could cause type-errors when calling ROOM on large images.
33 * bug fix: if file compilation is aborted, the partial fasl is now
34 deleted, and COMPILE-FILE returns NIL as the primary value.
35 * bug fix: number of thread safety issues relating to SBCL's internal
36 hash-table usage have been fixed.
37 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
38 be returned from its body when the values were being returned
39 using unknown-values return convection and the W-P-O was wrapped
40 inside an UNWIND-PROTECT.
41 * bug fix: sb-posix should now compile again under Windows, enabling
44 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
45 * incompatible change: hash-table accessor functions are no longer
46 automatically protected by locks. Concurrent accesses on the same hash-table
47 from multiple threads can give inconsistent results or even corrupt the
48 hash-table completely. Multi-threaded applications should do their own
49 locking at the correct granularity. In the current implementation it is
50 still safe to have multiple readers access the same table, but it's not
51 guaranteed that this property will be maintained in future releases.
52 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
53 and will signal an error at runtime.
54 * enhancement: SB-THREAD package now exports a semaphore interface.
55 * enhancement: CONS can now stack-allocate on x86 and
56 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
58 * enhancement: nested lists can now be stack allocated on
59 platforms providing stack allocation support.
60 * enhancement: dynamic-extent support has been extended to support
61 cases where there are multiple possible sources for the stack
63 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
64 if the mutex is uncontested on Linux.
65 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
66 as specified by the CLHS. (thanks to Tobias Rittweiler)
67 * bug fix: NaN comparison now works on x86-64.
68 * bug fix: CLOSE :ABORT T on a stream with pending output now
70 * bug fix: instances of non-standard metaclasses using standard
71 instance structure protocol sometimes missed the slot type checks
73 * bug fix: known functions can be inlined in high-debug code.
75 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
76 * minor incompatible change: the MSI installer on Windows no longer
77 associates .lisp and .fasl files with the installed SBCL.
78 * minor incompatible change: :UNIX is no longer present in *FEATURES*
79 on Windows. (thanks to Luis Oliviera)
80 * new feature: SB-CLTL2 contrib module now implements
81 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
82 * optimization: scavenging weak pointers is now more efficient,
83 requiring O(1) instead of O(N) per weak pointer to identify
84 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
85 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
86 method bodies using constant slot-names when the first argument is
87 a specializer parameter for the method.
88 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
89 names, when the class of the instance is a direct instance
90 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
91 as slow as the constant slot-name case.)
92 * optimization: member type construction is now O(N) instead
94 * optimization: UNION and NUNION are now O(N+M) for large
95 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
97 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
98 now more readable in environments like Slime which display it.
99 (thanks to Tobias C. Rittweiler)
100 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
101 was able to infer the correct argument type for the object on which
102 the CAS operation was being performed.
103 * bug fix: copy propagation interfered with parallel assignment
104 semantics in local calls. (reported by Paul Khuong)
105 * bug fix: the signed modular fixnum shift compiled to wrong code on
106 x86 and x86-64. (spotted by a slight modification to some of
109 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
110 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
111 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
112 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
114 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
115 and generic function now print the object identity for anonymous
116 instances. (thanks to Attila Lendvai)
117 * enhancement: as an extension to MOP, SBCL now supports SETF of
118 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
119 (thanks to Attila Lendvai)
120 * enhancement: sb-cover has an alternate annotation mode which puts
121 the form annotations at the car of the form, instead of annotating
122 it completely. The benefit of this mode is that it shows better
123 which forms were instrumented by the compiler.
124 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
125 by printer control variables. (reported by Dan Corkill)
126 * bug fix: system leaked memory when delayed output was performed by
127 the OS in smaller chunks then expected. (thanks to David Smith)
128 * bug fix: system leaked memory when file streams were not closed
130 * bug fix: large objects written to slow streams that were modified
131 after the write could end up with the modified state written to
132 the underlying file descriptor.
133 * bug fix: multiple threads operating in parallel on the same stream
134 could cause buffer-overflows.
135 * bug fix: source location information is stored correctly
136 (broken since 1.0.6). This bug would generally show up as the
137 Slime debugger higlighting the wrong form when the sldb-show-source
139 * bug fix: Branch forms are again annotated as branches in the sb-cover
141 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
142 on x86-64. (reported by Gregory Vanuxem)
143 * bug fix: EQUALP could return wrong results for structures with raw
144 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
145 word). (reported by Vjacheslav Fyodorov)
146 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
149 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
150 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
151 atomic compare-and-swap operations on threaded platforms.
152 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
153 allows assining a global minimum value to optimization qualities
154 (overriding proclamations and declarations).
155 * enhancement: closed over variables can be stack-allocated on x86
157 * performance bug fix: GETHASH and (SETF GETHASH) are once again
159 * optimization: slot definition lookup is now O(1). This speeds up
160 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
161 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
162 * optimization: ASSOC and MEMBER can now be open-coded for all
163 combinations of keyword arguments when second argument is constant
164 and SPEED >= SPACE. In other cases a specialized version is
166 * bug fix: using obsoleted structure instances with TYPEP and
167 generic functions now signals a sensible error.
168 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
169 (reported by Kristoffer Kvello)
170 * bug fix: backtrace construction is now more careful when making
171 lisp-objects from pointers on the stack, to avoid creating bogus
172 objects that can be seen by the GC.
173 * bug fix: defaulting of values in contexts expecting more than 7
174 variables now works on x86-64. (reported by Christopher Laux)
175 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
177 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
178 as the property-list of a symbol.
179 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
180 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
181 situations requested, are once again file-compileable. (reported
184 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
185 * MOP improvement: support for user-defined subclasses of
186 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
187 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
188 the expansion of DEFMETHOD, is responsible for generating a form
189 which creates a list of specializers when evaluated. Additional
190 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
191 debugging and introspective support.
192 * minor incompatible change: the (unsupported) spinlock interface
193 has changed: free spinlock now has the value NIL, and a held spinlock
194 has the owning thread as its value.
195 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
196 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
198 * enhancement: name of a socket-stream is now "a socket" instead of
200 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
201 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
202 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
203 for generic functions. (Thanks to Tobias C. Rittweiler)
204 * optimization: bignum printing speed has been improved by 20-40%
205 (depending on the bignum size.)
206 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
208 * bug fix: the cache used by the CLOS to store precomputed effective
209 methods, slot offsets, and constant return values is now thread and
211 * bug fix: generic function dispatch function updating is now thread
212 and interrupt safe (in the sense that the known issues have been
214 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
215 * bug fix: interning EQL-specializers is now thread and interrupt safe.
216 * bug fix: asdf systems with dependencies to the SB-POSIX or
217 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
218 * bug fix: interrupt safety of applicable method computation has been
221 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
222 * new contrib: sb-cover, an experimental code coverage tool, is included
224 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
225 significantly faster.
226 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
227 produce more efficient code on x86-64 if the compiler can determine
228 that (+ INDEX <constant>) does not require a bounds check and FOO
229 has an element type at least 8 bits wide.
230 * enhancement: a new, experimental synchronous timeout facility is
231 provided. Refer to SB-SYS:WITH-DEADLINE for details.
232 * enhancement: when a symbol name conflict error arises, the
233 conflicting symbols are always printed with a package prefix.
234 (thanks to Kevin Reid)
235 * enhancement: stepping is now once again supported on the SPARC and
236 MIPS platforms. (It is also now more likely to work on CheneyGC
238 * enhancement: sb-sprof can now also track and report accurate call
240 * bug fixes: the treatment of non-standard subclasses of
241 SB-MOP:SPECIALIZER is more correct.
242 * incompatible change: PURIFY no longer copies the data from the
243 dynamic space into the static and read-only spaces on platforms
244 that use the generational garbage collector
245 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
247 * bug fix: GC race condition occasionally resulting in crashes with
248 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
250 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
252 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
253 a STYLE-WARNING to be signalled (regression from 1.0.4.)
254 * bug fix: an asynchronous interrupt could previously leave the
255 system running with GC inhibited.
256 * bug fix: a DECLARE form evaluated at top-level now causes an error
257 rather than silently (or verbosely) returning NIL.
258 * bug fix: trying to dissassemble functions compiled at high DEBUG could
259 result in a error being signalled due to source form lookup errors.
260 (reported by Peter Graves)
262 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
263 * incompatible change: removed writer methods for host-ent-name,
264 host-ent-addresses -- changing the values did not update the DNS
266 * minor incompatible change: changed experimental JOIN-THREAD interface
267 * documentation: the manual now lists reader and writer methods
268 in class slot documentation sections. (thanks to Richard M Kreuter)
269 * documentation: unwinding from asyncronous events has been
270 documented as unsafe.
271 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
272 in multithreaded application code.
273 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
274 platforms. (thanks to James Anderson for the optimization hint)
275 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
277 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
278 variants no longer cons.
279 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
280 their NOT- variants no longer cons.
281 * optimization: EQUAL hash tables no longer use SXHASH for objects
282 of all data types, but instead use an EQL hash for types for which
283 EQUAL is the same as EQL.
284 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
285 are significantly faster.
286 * optimization: new STRING-OUTPUT-STREAM implementation is much
287 faster for large outputs and conses 30% less on average.
288 * enhancement: XREF information is now collected for references made
289 to global variables using SYMBOL-VALUE and a constant argument.
290 * enhancement: SIGINT now causes a specific condition
291 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
292 * bug fix: FILE-POSITION used to signal an error in some cases where
293 ANSI requires it to return NIL.
294 * bug fix: ADJUST-ARRAY is now interrupt-safe.
295 * bug fix: adding and removing fd-handlers is now interrupt-safe.
296 * bug fix: inlined calls to C now ensure 16byte stack alignment on
298 * bug fix: bad type declaration in the CLOS implementation has
299 been fixed. (reported by James Anderson)
300 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
301 (thanks to Lutz Euler)
302 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
303 required by the CLHS. (reported by Samium Gromoff)
304 * bug fix: dead unbound variable references now signal an error.
305 (reported by Marco Monteiro)
306 * bug fix: / with an unused value was being deleted in safe code.
307 (thanks to Marco Monteiro and Kevin Reid)
308 * bug fix: number of characters that can be written onto a single
309 line in a file is unlimited.
310 * bug fix: some GC deadlocks caused by asynchronous interrupts have
311 been fixed by inhibiting interrupts for when GC is disbled.
312 * bug fix: some interrupt safety issues with GETHASH, (SETF
313 GETHASH), CLRHASH and REMHASH have been fixed.
314 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
315 specifier no longer causes infinite recursion.
316 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
317 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
318 * bug fix: an &environment argument with the correct variable information
319 is passed to macros that are expanded during byte compilation.
320 (reported by Samium Gromoff)
321 * bug fix: a logic error was causing the hash function for bignums to
322 have excessive amounts of collisions. (reported by Faré Rideau)
323 * bug fix: modifying the contents of an array could change the return
324 value of SXHASH on that array, which is only allowed for strings
325 and bit vectors. (bug introduced in 0.9.16)
326 * improvement: the x86-64/darwin port now passes all tests (except
327 for the debugger tests) but should still be considered
328 experimental until this is fixed.
329 * improvement: a style-warning is signaled for CASE (etc) clauses with
330 duplicate keys. (thanks to Kevin Reid)
331 * improvement: macos/x86 and macos/x86-64 now use mach exception
332 handlers for dealing with illegal instructions (for trapping and
333 error handling) and memory protection violations (for GC).
335 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
336 * new platform: experimental support for x86-64/darwin (MacOS).
337 * incompatible change: the thread-safe (on most platforms) getaddrinfo
338 and getnameinfo sockets functions are used instead of gethostbyaddr
339 and gethostbyname, on platforms where the newer functions are available.
340 As a result, the ALIASES field of HOST-ENT will always be NIL on these
342 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
343 don't print the startup banner, but behave as if --noinform was passed
344 as a command line argument. (thanks to Kevin Reid)
345 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
346 * optimization: code using alien values with undeclared types is much faster.
347 * optimization: the compiler is now able to open code SEARCH in more cases.
348 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
349 * bug fix: using standardized COMMON-LISP special variables as loop
350 variables no longer signals bogus package lock violations. (reported
352 * bug fix: declaring local loop variables to be of a range-limited type
353 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
354 (reported by Andras Simon)
355 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
356 bugs remain on x86-64.)
357 * bug fix: the #= and ## reader macros now interact reasonably with
358 funcallable instances.
359 * bug fix: type-checks for function arguments were compiled using the
360 compiler policy settings of the wrong lexical environment (in 1.0.2
362 * bug fix: SHADOW accepts characters as string designators, as required
363 by the spec (thanks to Eric Marsden)
364 * bug fix: fixed GC safety issues when foreign functions are called with
365 non-base strings as arguments
366 * bug fix: more consistent error output for fatal-compiler-warnings, like
368 * bug fix: functions with proper names were showing up as NIL in
369 backtraces (reported by Edi Weitz, regression in 1.0.2)
371 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
372 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
373 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
374 produce more efficient code on the x86 if the compiler can determine
375 that (+ INDEX <constant>) does not require a bounds check and FOO
376 has an element type at least 8 bits wide.
377 * bug fix: references to undefined variables are handled the same way
378 in toplevel forms as in the normal compiler
379 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
381 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
382 (thanks to Jon Buller)
383 * improvement: faster compilation times for complex functions
384 * improvement: added readlink support to SB-POSIX (thanks to Richard
387 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
388 * improvement: experimental support for mach exception handling on
389 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
391 * improvement: support for GBK external format.
392 (thanks to Chun Tian (binghe))
393 * improvement: the debugger now displays variables that have been closed
394 over, in code compiled with (DEBUG 2) or higher
395 * improvement: support for executable cores on NetBSD (thanks to
397 * new feature: added a RESTART-FRAME debugger command
398 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
399 be used to provide an implementation for FILE-POSITION on Gray streams
400 (thanks to Eric Marsden)
401 * new feature: add syslog, openlog and closelog support to SB-POSIX
402 (thanks to Richard Kreuter)
403 * optimization: the function call overhead in code compiled with
404 a high DEBUG optimization setting is significantly
405 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
406 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
407 (thanks to Tony Martinez)
408 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
409 evaluated once (reported by Kevin Reid)
410 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
411 works again (thanks to Kilian Sprotte)
412 * bug fix: an error is signaled for tagbodies with duplicate tags
413 (thanks to Stephen Wilson)
414 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
415 * bug fix: Win32 port can now handle foreign code unwinding Lisp
416 stack frames from alien callbacks.
417 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
418 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
419 65k pages (thanks to David Woodhouse)
420 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
422 changes in sbcl-1.0.1 relative to sbcl-1.0:
423 * new platform: FreeBSD/x86-64, including support for threading.
424 * new feature: the compiler stores cross-referencing information
425 abount function calls (who-calls), macroexpansion (who-macroexpands)
426 and special variables (who-binds, who-sets, who-references) for code
427 compiled with (< SPACE 3). This information is available through the
428 sb-introspect contrib.
429 * new feature: users may subclass SEQUENCE, and have instances of
430 these classes interoperate with standard Common Lisp functions if
431 a number of methods are defined. (This feature is experimental
432 and the interface subject to change based on feedback from SBCL
433 users and the general community)
434 * improvement: sb-sprof traces call stacks to an arbitrary depth on
435 x86/x86-64, rather than the previous fixed depth of 8
436 * improvement: another pthread back-end of mutex "pthread-futex".
437 * bug fix: non-ascii command-line arguments are processed correctly
438 (thanks to Yaroslav Kavenchuk)
439 * bug fix: non-required arguments were not passed correctly when a method
440 defined using DEFMETHOD was called from a mop-generated method using
441 CALL-NEXT-METHOD (reported by Pascal Costanza)
442 * bug fix: recursion is now permitted in accessors through
443 SLOT-UNBOUND. (reported by Pascal Costanza)
444 * bug fix: an error was signaled at startup if the HOME environment
445 variable was defined, but had an empty value (reported by Peter Van Eynde)
446 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
447 signal a PROGRAM-ERROR, not a BUG.
448 * bug fix: SB-POSIX:READDIR works when built with large file support.
449 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
451 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
452 for better automatic indentation support. (thanks to Matt Pillsbury)
453 * optimization: loading generic functions no longer takes O(n^2) time,
454 proportional to the amount of methods in the generic function
455 (reported by Todd Sabin and Jeremy Brown)
456 * optimization: the FIND and POSITION family of sequence functions
457 are significantly faster on arrays whose element types have been
459 * improvements to the Windows port:
460 ** Intermittent heap corruption problems have been fixed. (thanks
461 to Alastair Bridgewater)
462 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
464 ** Lisp is able to unwind foreign exception frames from alien
465 callbacks. (thanks to Alastair Bridgewater)
467 changes in sbcl-1.0 relative to sbcl-0.9.18:
468 * improvement: experimental support for threading on FreeBSD/x86.
469 (thanks to NIIMI Satoshi)
470 * improvement: runtime option --dynamic-space-size can be used
471 to set the size of the dynamic space reserved on startup.
472 * improvement: floating point modes in effect are now saved in
473 core, and restored on startup.
474 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
475 startup, not time since first call to GET-INTERNAL-REAL-TIME.
476 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
477 threads are not running after *SAVE-HOOKS* have run.
478 * improvement: writes to CLOS instance slots are type-checked in code
479 compiled with (SAFETY 3)
480 * improvement: floating-point exception handling on FreeBSD (thanks to
482 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
483 (thanks to Zach Beane)
484 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
486 * improvement: added support for the Shift-JIS external format.
487 (contributed by NIIMI Satoshi)
488 * improvement: callbacks are supported on Linux/PPC. (thanks to
490 * bug fix: compiler bug triggered by a (non-standard) VALUES
491 declaration in a LET* was fixed. (reported by Kaersten Poeck)
492 * bug fix: file compiler no longer confuses validated and already
493 dumped structurres. (reported by Kaersten Poeck)
494 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
495 fill-pointer signals a type-error as required. (thanks to
497 * bug fix: disassemly of funcallable instances works.
498 * bug fix: single stepping on PPC.
499 * bug fix: fix thread-safety problems in the type system (generally
500 manifesting as nonsensical errors like "STRING is a bad type specifier
501 for sequences" or "The value 1 is not of type FIXNUM")
502 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
503 (reported by Josip Gracin)
504 * bug fix: an error is signaled for attempts to displace arrays with
505 incompatible element types (thanks to Mario Mommer)
506 * bug fix: more correct handling of wide characters in debug info
507 (bug reported by Attila Lendvai and fixed by Juho Snellman)
508 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
509 and don't cause extra consing
510 * optimization: MAP and MAP-INTO are significantly faster on vectors
511 whose elements types have been declared.
512 * Improvements to SB-SPROF:
513 ** Support for allocation profiling
514 ** Reduced profiling overhead, especially for long profiling runs
515 * Improvements to the Windows port:
516 ** floating point exceptions are now reported correctly.
517 ** stack exhaustion detection works partially.
518 ** more accurate GET-INTERNAL-REAL-TIME.
519 ** better RUN-PROGRAM behaviour in case of errors in the child
521 ** PROBE-FILE now simplifies pathnames correctly.
522 ** DIRECTORY now works correctly with :WILD-INFERIORS.
523 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
524 1st 1970 (although time timezone data will be unreliable).
525 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
527 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
528 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
529 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
531 * bug fix: two potential GC deadlocks affecting threaded builds.
532 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
533 returns T (reported by Anton Kazennikov)
534 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
535 with non-variable places
536 * bug fix: remove a race condition in the setting of
537 funcallable-instance functions, this should make threaded CLOS
538 code more stable against memory faults.
539 * bug fix: corruption of specials when unbinding is interrupted by an
540 asynchronous unwind (reported by Hannu Koivisto)
541 * improvement: the debugger will now also display local variables that
542 are only used once, for code compiled with a DEBUG optimization quality
545 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
546 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
547 * incompatible change: External-format support for FFI calls. The
548 SB-ALIEN:C-STRING no longer implies an ASCII
549 external-format. Instead the string is subject to external-format
550 conversion. Additionally return values of type SB-ALIEN:C-STRING
551 are by default of type (SIMPLE-ARRAY CHARACTER), not
552 SIMPLE-BASE-STRING. For an alien type that behaves like the old
553 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
554 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
555 * incompatible change: SB-EXT package no longer contains the
556 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
557 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
559 * incompatible change: the single-stepper is no longer available
560 on Alpha, Mips and Sparc platforms.
561 * minor incompatible change: the direct superclasses of
562 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
563 not (STANDARD-OBJECT FUNCTION). This makes the
564 class-precedence-lists of GENERIC-FUNCTION and
565 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
567 * minor incompatible change: the default stream external format on
568 non-unicode SBCL is no longer detected from the locale, but is always
570 * new feature: Add a version of evaluator that uses an interpreter instead
571 of the compiler. EVAL still uses the compiler by default, to switch it
572 to use the interpreter, set the value of the variable
573 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
574 * minor incompatible change: the single-stepper REPL has been merged
575 with the normal debugger (see the "Stepping" heading of the debugger help
576 for more details). The debugger command STEP will no longer switch
577 to the single-stepper REPL.
578 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
579 for a type now works.
580 * bug fix: loading of default sysinit file works. (thanks to Leonid
582 * bug fix: better detection of circularities in the file-compiler.
583 (reported by Marco Monteiro)
584 * bug fix: the CL pathname functions now work with files that have
585 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
586 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
587 systems with Unix98 pty semantics.
588 * bug fix: ASDF-INSTALL will now work with bsd tar.
589 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
591 * bug fix: timers expiring in dead threads no longer cause a
592 type-error (reported by Paul "Nonny Mouse").
593 * bug fix: thanks to more lightweight single-stepper instrumentation,
594 code compiled with (DEBUG 3) will compile and execute significantly faster,
595 and will have more accurate type-inferencing than before
596 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
597 whose bindings are modified
598 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
599 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
600 ** replace dummy implementations of CL:MACHINE-INSTANCE and
601 CL:SOFTWARE-VERSION with proper version
603 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
604 * feature: implemented the READER-METHOD-CLASS and
605 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
606 as specified by AMOP.
607 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
609 * optimization: faster LOGCOUNT implementation on x86 and x86-64
610 (thanks to Lutz Euler)
611 * optimization: hashing of general arrays and vectors has been
612 improved. (reported by Any Fingerhut)
613 * enhancement: SB-INTROSPECT is now able to find definitions of
614 profiled functions. (thanks to Troels Henriksen)
615 * enhancement: compiler-macro expansion applies now to FUNCALL forms
616 as well, allowing compiler-macros for SETF-functions to expand.
617 * enhancement: step-instrumentation no longer wraps known
618 single-value functions with multiple-value context, allowing
619 better type inference.
620 * fixed bug #337: use of MAKE-METHOD in method combination now works
621 even in the presence of user-defined method classes. (reported by
622 Bruno Haible and Pascal Costanza)
623 * fixed bug #339(c): if there are applicable methods not part of any
624 long-form method-combination group, call INVALID-METHOD-ERROR.
625 (reported by Bruno Haible)
626 * fixed bug #361: the :FUNCTION initarg in the protocol for
627 initialization of methods can now be used to override
628 internally-produced optimized functions. (reported by Bruno
630 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
631 system-provided lambda expression no longer cause warnings about
632 unbound #:|pv-table| symbols.
633 * bug fix: improved the handling of type declarations and the
634 detection of violations for keyword arguments with non-constant
636 * bug fix: potentially erronous calls to PATHNAME and
637 MERGE-PATHNAMES were being flushed in some cases.
638 (reported by Richard Kreuter)
639 * bug fix: compiled calls to TYPEP were mishandling obsolete
640 instances. (reported by James Bielman and Attila Lendvai)
641 * bug fix: format strings with ~> without matching ~< no longer
642 trigger an AVER, but signal an understandable error instead.
643 (reported by Antonio Martinez)
644 * bug fix: specifying an output-file without a pathname-type for
645 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
646 of input-file instead of "fasl". (reported by Robert Dodier)
647 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
648 (reported by James Y Knight).
649 * bug fix: compiler-macros-function did not consider the environment
650 argument for shadowing by local functions.
651 * bug fix: compiler-macros expansion was inhibited by local INLINE
653 * bug fix: inline expansions of known functions were subject to
654 step-instrumentation in high debug policies, leading to problems
656 * bug fix: compiler failed to differentiate between different CONS
658 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
659 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
660 * bug fix: #368: incorrect use of expressed vs. upgraded array
662 * bug fix: #306a: more precise unions of array types.
663 * thread-safety improvements:
664 ** CONDITION-WAIT could return early on Linux, if the thread was
665 interrupted and subsequently continued with SIGCONT.
666 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
668 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
669 * added support for the ucs-2 external format. (contributed by Ivan
671 * minor incompatible change: pretty printing of objects of type
672 (cons symbol) is, in the default pprint-dispatch-table, now
673 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
675 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
676 called later in a class's lifetime, possibly as late as when the
677 first instance of the class is created. Previously,
678 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
679 class became finalizeable.
680 * fixed bug: FILE-POSITION sometimes returned inconsistent results
681 for multibyte external-format streams. (thanks to "vbzoli")
682 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
683 with :ALLOCATION :CLASS inherited from superclasses of the
685 * fixed bug: anonymous classes can now be created using the :NAME
686 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
687 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
688 * fixed bug: core-files saved with :EXECUTABLE T can again be
689 executed when SBCL_HOME isn't set. (reported by James Knight)
690 * fixed bug: toplevel LOCALLY forms with declarations could
691 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
692 * fixed bug: printing from several different threads using different
693 values of *print-case* could cause invalid output, due to
694 some internal special variables of the printer not being bound
695 thread-locally (reported by Max Mikhanosha)
696 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
697 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
698 (reported by Pascal Costanza)
699 * fixed bug: FINALIZE-INHERITANCE is called from
700 REINITIALIZE-INSTANCE on classes when the class has previously
701 been finalized, as required by AMOP.
702 * minor code generation optimizations:
703 ** better register allocation in CLOS dispatching functions
704 ** overflow detection when coercing signed bytes to fixnums on x86-64
705 ** is now implemented with one IMUL instruction instead of three shifts
706 ** more efficient bit-vector access on x86 and x86-64
707 ** more efficient access to raw structure slots on x86 and x86-64
708 * fixed some bugs revealed by Paul Dietz' test suite:
709 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
712 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
713 * feature: thread support on Solaris/x86, and experimental thread support
715 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
717 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
718 default method for SLOT-UNBOUND.
719 * minor incompatible change: prevent the user from specializing the
720 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
721 somewhat counter to the intent of the protocol, I (CSR) think, and
722 additionally it just doesn't work in SBCL as currently
723 implemented, thanks to optimizations (that are always valid for
724 the other three SLOT-VALUEish functions, but not for the setter).
725 * minor incompatible change: the :SB-LDB feature is now enabled by
726 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
727 the low-level debugger.
728 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
729 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
730 down to an SB-SYS:FD-STREAM, allowing direct communication in
731 more cases. Temporary files and pipes are still used as a fallback
733 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
734 on the non-reentrant C functions localtime(3) and gmtime(3).
735 * bug fix: global optimization policy was not visible in LOCALLY and
737 * bug fix: class objects can be used as specializers in methods.
738 (reported by Pascal Costanza)
739 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
740 with a type error. (reported by blitz_ on #lisp)
741 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
742 failed with a type error. (reported by Pascal Bourguignon)
743 * bug fix: merging pathnames against defaults with :DIRECTORY
744 starting with '(:RELATIVE :BACK) should preserve the :BACK.
745 (reported by James Y Knight)
746 * bug fix: saving large (>2GB) cores on x86-64 now works
747 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
748 constant index and a (simple-array (signed-byte 32)) array
749 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
750 error (patch by Robert J. Macomber)
751 * feature: TIME now displays the amount of run-time spent in GC
752 * bug fix: The debugger now does a better job of respecting
753 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
754 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
755 workaround for bug 403.)
756 * bug fix: floating point exception handling now works on FreeBSD
757 (thanks to NIIMI Satoshi)
758 * fixed some bugs revealed by Paul Dietz' test suite:
759 ** MISC.641: LET-conversion were not supposed to work in late
761 * improvements to the Win32/x86 port:
762 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
763 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
765 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
766 * new feature: source path information is generated for macro-expansion
767 errors for use in IDE's like Slime (thanks to Helmut Eller)
768 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
769 * bug fix: compiler does not loop forever on an invalid type in
771 * improvement: compilation of most CLOS applications is significantly
773 * optimization: added a limited bytecode compiler for simple toplevel
774 forms, speeding up compilation and FASL loading
775 * bug fix: the statistical profiler now properly distinguishes anonymous
778 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
779 * minor incompatible change: in sbcl-0.9.11 (but not earlier
780 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
781 into subclasses of the finalized class. Now user calls to
782 FINALIZE-INHERITANCE finalize just the one class, and calls by the
783 system return before any subclasses are finalized.
784 * minor incompatible change: The reader no longer ignores errors
785 regarding non-existent packages in #+ and #- feature tests.
786 * new feature: command line options --no-sysinit, --no-userinit to
787 inhibit loading the corresponding init files
788 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
789 for sending data through UDP sockets (thanks to François-René Rideau)
790 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
791 error is signalled instead (thanks to François-René Rideau)
792 * minor incompatible change: Error signalling behaviour of lexical
793 operations violating package locks has changed slightly. Refer to
794 documentation on package locks for details.
795 * enhancement: EVAL can process IF-expressions without resorting to the
797 * bug fix: Some operations involving SIN, COS, or TAN weren't being
798 constant-folded properly due to missing out-of-line functions.
799 (reported by Mika Pihlajamäki)
800 * bug fix: LISTEN sometimes returned T even in cases where no data was
801 immediately available from the stream
802 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
803 were reversed. (reported by Levente Mészáros)
804 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
805 when given a vector displaced to an adjustable array. (reported
807 * bug fix: derivation of float boundaries from numbers outside the
808 appropriate float range (reported by John Wiseman)
809 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
811 * bug fix: precomputing cache entries for generic functions with
812 some subclasses of specializers as yet invalid does not attempt to
813 fill a cache line with a negative offset. (reported by Levente
815 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
816 allow condition class objects as arguments to DOCUMENTATION and
817 (SETF DOCUMENTATION); only find and set documentation for
818 structure names for the STRUCTURE doc-type. (suggested by Gary
820 * improvements to the Win32/x86 port:
821 ** (user-homedir-pathname) and default initialization file
822 locations now know about the user's "Documents and Settings"
823 directory (thanks to Yaroslav Kavenchuk)
824 ** run-program is implemented (thanks to Mike Thomas)
825 ** sockets support (thanks to Timothy Ritchey)
826 ** better backtrace support (thanks to Alastair Bridgewater)
827 ** sb-grovel supported
828 ** asdf-install and sb-posix work somewhat
829 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
830 * improvements to the Solaris/x86 port:
831 ** works on Solaris 11/Solaris Express
832 ** floating-point exception handling support
833 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
834 * fixed some bugs revealed by Paul Dietz' test suite:
835 ** REMOVE-METHOD returns its generic function argument even when
836 the method is not one of the generic functions' methods.
837 ** objects declared as MEMBER types can be admissible arguments to
839 ** printing characters should simply be printed by the FORMAT ~:C
841 ** compiler failure when compiling functions with hairy constant
842 defaults for optional parameters.
843 ** compiler produces wrong code when MAYBE-INLINE-expanding a
844 function, which is already optimized.
846 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
847 * new platform: experimental support for SBCL x86/Darwin, including
848 MacOS X 10.4.5 on Intel.
849 * new feature: Unicode character names are now known to the system
850 (through CHAR-NAME and NAME-CHAR).
851 * new feature: the filesystem location of SBCL's core file is
852 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
853 * minor incompatible change: the contrib modules SB-POSIX and
854 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
855 this change is to make it easier to distribute
856 location-independent binaries.
857 * bug fix: as implied by AMOP, standardized classes no longer have
858 slots named by external symbols of public packages. (reported by
860 * optimization: calling functions via constant symbols -- as in
861 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
862 via the function object as in (FUNCALL #'FOO).
863 * optimization: on x86 and x86-64, the calling convention for the
864 case of dealing with an unknown number of values has been altered
865 to be friendlier to the prediction heuristics implemented,
866 particularly with reference to CALL and RET pairing. (thanks to
867 Alastair Bridgewater)
868 * optimization: on x86, the code for access to thread-local symbol
869 values and binding/unbinding of thread-local symbols is smaller.
870 (thanks to Alastair Bridgewater)
871 * enhancement: CONSTANTP is now able to determine constantness of
872 more complex forms, including calls to constant-foldable standardized
873 functions and some special forms beyond QUOTE.
874 * fixed bug: occasional GC crashes on Solaris/x86
875 * optimization: x86-64 supports stack allocation of results of simple
876 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
877 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
878 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
879 (thanks to James Knight)
880 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
881 by the standard (thanks to Stephen Compall)
883 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
884 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
885 be used for bundling the runtime and the core file into one
886 executable binary. This feature is not currently supported on all SBCL
887 platforms. (thanks to James Bielman and NIIMI Satoshi)
888 * new feature: a generational or ephemeral garbage collector is now
889 the default on the PowerPC platform (both Linux and Darwin). The
890 old Cheney (stop and copy) collector is a build-time option.
891 (thanks to Cyrus Harmon, after Raymond Toy)
892 * minor incompatible change: the method by which SBCL finds its
893 contributed modules has changed; it no longer relies on symbolic
894 links from an $SBCL_HOME/systems directory, but searches directly
895 in the subdirectories of $SBCL_HOME.
896 * enhancement: the dynamic heap size on the Linux/PPC platform is
897 markedly larger, even using the older Cheney garbage collector.
898 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
899 certain complicated string types. (reported by Gary King)
900 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
901 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
902 * fixed bug: DOCUMENTATION on structures no longer signals an
903 error if no structure type of the right name exists. (reported by
905 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
906 applicable non-standard primary or :AROUND method on
907 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
908 no longer cause unbound variable errors. (reported by Kevin Reid)
909 * fixed bug: metacircle resolution in cases where methods have slots
910 added before the slots from STANDARD-METHOD. (reported by Jean
912 * fixed bug: the Power PC assembly code for calling into lisp
913 sometimes computed its return address wrongly (depending on the
914 15th bit of the address, fixed at link time). This used to
915 manifest itself in a segmentation violation while building PCL.
916 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
917 many others over the years)
918 * fixed bug: ROOM no longer reports silly numbers for stack usage if
919 the stack pointer (treated as a signed integer) is negative.
920 (thanks to Peter van Eynde)
922 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
923 * new platform: experimental support for the Windows operating
924 system has been added. (thanks to Alastair Bridgewater)
925 * fixed several bugs in and robustified the PPC FFI (including
926 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
927 * bug fix: allow non-simple string symbol names (reported by Paul
929 * bug fix: interrupt handling on NetBSD (thanks to Richard M
931 * bug fix: saving a core corrupted callbacks on x86/x86-64
932 * bug fix: closed a loophole in metacircularity detection and
933 grounding in the PCL implementation of CLOS.
934 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
935 (reported by tomppa on #lisp)
936 * bug fix: clear the direction flag on Lisp -> C transitions, as
937 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
938 (reported by Andrej Grozin and Hendrik Maryns)
939 * optimization: major improvements to GC efficiency on GENCGC platforms
940 * optimization: faster implementation of EQUAL
941 * optimization: emit more efficient opcodes for some common
942 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
943 * optimization: several other minor code-generation improvements on x86-64
944 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
946 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
947 * minor incompatible change: (SETF CLASS-NAME) and (SETF
948 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
949 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
950 by ANSI as a generic function, and for consistency (SETF
951 GENERIC-FUNCTION-NAME) is treated likewise.
952 * fixed bug #233.b: make constraint propagation notice when a variable
953 value is changed after it is referenced but before it is used
954 * fixed bug #296: no more arbitrary behaviour from filenames with
955 odd characters as --load arguments. (reported by Adam Warner)
956 * fixed bug #390: :CHARACTER-SET pathname components now work as
957 expected. (reported by Tim Daly Jr)
958 * fixed bug #391: complicated :TYPE intersections in slot
959 definitions no longer cause an error in PCL internals.
960 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
961 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
962 returning the number of octets which would be written to the
963 file-stream. (thanks to Robert J. Macomber)
964 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
965 arguments to the internal stream functions. (thanks to David
967 * bug fix: stack exhaustion is now handled on the x86/SunOS
968 platform. (thanks to Daisuke Homma)
969 * bug fix: more accurate ROOM results on GENCGC platforms
970 * optimization: improved type inference for arithmetic-for
971 index variables in LOOP
972 * optimization: faster floating-point SQRT on x86-64
973 * fixed some bugs revealed by Paul Dietz' test suite:
974 ** DOCUMENTATION returns NIL instead of "" for method combinations
975 that don't have a docstring
977 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
978 * minor incompatible change: (SETF CLASS-NAME) and (SETF
979 GENERIC-FUNCTION-NAME) are no longer generic functions, and
980 therefore are not customizeable by user code (as seems to be at
981 least permitted and maybe required by AMOP). As a consolation,
982 however, the SBCL implementation of these functions now calls
983 REINITIALIZE-INSTANCE as specified by AMOP.
984 * bug fix: slot-definition documentation is propagated as per ANSI
985 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
986 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
987 argument, as do eslotd initialization methods. (from Pascal
988 Costanza's "Closer" project)
989 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
990 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
992 * bug fix: it is now possible to have more than one subclass of
993 STANDARD-GENERIC-FUNCTION without causing stack overflow.
994 (reported by Bruno Haible, Pascal Costanza and others)
995 * bug fix: the dependent update protocol now works for generic
996 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
998 * bug fix: condition-class instances corresponding to
999 DEFINE-CONDITION forms are now created eagerly. (reported by
1000 Kalle Olavi Niemitalo on comp.lang.lisp)
1001 * bug fix: floating point printing is more accurate in some
1002 circumstances. (thanks to Simon Alexander)
1003 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1004 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1005 * bug fix: callbacks on OS X now preserve stack-alignment.
1006 (thanks to Cyrus Harmon)
1007 * enhancement: the x86-64 disassembler is much better at
1008 disassembling SSE instructions. (thanks to Lutz Euler)
1009 * enhancement: CHECK-TYPE on a variable now allows type-inference
1010 (thanks to Vincent Arkesteijn)
1011 * optimization: improved performance of EUC-JP external format.
1012 (thanks to NIIMI Satoshi)
1013 * optimization: performance improvements to IO on file streams of
1014 :ELEMENT-TYPE CHARACTER
1015 * optimization: much faster memory allocation on x86-64
1016 * optimization: faster garbage collections (latency and throughput)
1018 * optimization: faster bignum subtraction and division on x86 and x86-64
1019 (thanks to Lutz Euler)
1020 * optimization: more accurate type-inference for loops that use a
1021 floating point index variable or a negative step.
1023 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1024 * bug fix: add a workaround to SBCL looping infinitely at startup on
1025 Linux kernels with apparently buggy implementations of personality().
1026 (thanks to Svein Ove Aas)
1027 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1028 (thanks to David Lichteblau)
1029 * bug fix: local bindings shadow global symbol macros for
1030 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1031 * optimization: non-open coded uses of numeric comparison operators
1032 (e.g. >) no longer cons when called with more than one parameter
1033 on platforms supporting dynamic-extent allocation.
1034 * enhancement: saving cores with foreign code loaded is now supported
1035 on MIPS/Linux in addition to the previously supported platforms.
1036 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1037 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1038 * bug fix: division by zero in sb-sprof when no samples were collected
1039 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1040 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1041 (thanks to Alexander Kjeldaas)
1042 * new feature: ignoring the return values of destructive functions
1043 like NREVERSE signals a compile-time style-warning.
1044 (thanks to Kevin Reid)
1045 * new feature: the alignment of alien structure fields can be
1046 explicitly specified. (thanks to Cyrus Harmon)
1047 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1048 and dump core on SIGQUIT
1050 ** incompatible change: threads do not inherit values of specials
1051 from their parents (see manual)
1052 ** bug fix: threads stacks belonging to dead threads are freed by the
1053 next exiting thread, no need to gc to collect thread stacks anymore
1054 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1055 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1056 ** bug fix: dynamic variable and thread start related gc lossage
1057 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1059 * fixed some bugs revealed by Paul Dietz' test suite:
1060 ** SUBTYPEP is slightly more accurate on heinously complicated
1061 CONS types where some of the members have uncertain (in the
1062 NIL, NIL sense) type relationships to each other.
1064 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1065 no longer purified unless :PURIFY T is explicitly specified.
1066 ** Non-purified cores are significantly smaller than before
1068 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1069 * new feature: timers based on Zach Beane's excellent timer package
1070 * added support for the following external formats: koi8-u,
1071 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1072 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1073 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1074 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1075 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1076 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1077 * incompatible change: a threaded SBCL will no longer revert to
1078 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1079 * bug fix: interrupts are disabled until startup is complete; no
1080 more sigsegvs when receiving a signal to soon
1081 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1083 * bug fix: add a workaround for the memory randomization features in
1084 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1085 workaround will only be in effect on systems with the proc filesystem
1087 * bug fix: printing objects of type HASH-TABLE signals a
1088 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1090 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1091 have been running for over 50 days. (reported by Gilbert Baumann)
1092 * bug fix: the logic for getting names of functions gets less
1093 confused when confronded with alternate-metaclass
1094 funcallable-instances. (reported by Cyrus Harmon)
1095 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1096 consistent, even on internal alternate-metaclass objects.
1097 * bug fix: flush closure information collected by physical
1098 environment analysis prepass before the main pass. (bug reported
1100 * bug fix: compiler pack phase does not modify a hashtable, which is
1101 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1102 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1103 may be expanded. (reported by Alexander <malishev> on c.l.l)
1104 * bug fix: no more style-warnings when compiling code using LAST
1105 under high-SPEED low-SPACE settings. (reported by David Wallin
1107 * fixed some incompatibilities between SBCL's MOP and the MOP
1109 ** the METAOBJECT class is now implemented;
1110 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1111 STANDARD-OBJECT, as required;
1112 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1113 now compatible (as required by VALIDATE-SUPERCLASS); there
1114 remains a consistency requirement over the presence or absence
1115 of the FUNCTION class in the superclass list of finalized
1116 classes; see the manual for more details;
1117 ** the :AROUND method for COMPUTE-SLOTS on
1118 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1119 requested slot ordering.
1121 ** bug fix: parent thread now can be gc'ed even with a live
1123 ** bug fix: binding a special with PROGV to no value is not
1124 the same as that symbol not having been bound (thanks to
1126 ** bug fix: binding specials is thread safe (thanks to
1128 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1129 and SB-SPROF do not die with 'no handler for signal XX in
1130 interrupt_handle_now(..)' anymore
1131 ** bug fix: WITH-TIMEOUT works with multiple running threads
1132 * fixed some bugs revealed by Paul Dietz' test suite:
1133 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1134 the :method-class keyword argument.
1136 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1137 * new port: the Solaris operating system on x86 processors is now
1138 mostly supported, though some rough edges in the environment
1139 remain. (thanks to Daisuke Homma)
1140 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1141 space for its heap. (thanks to Thiemo Seufer)
1142 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1143 closures. (thanks to Thiemo Seufer)
1144 * minor incompatible change: eof selects abort in the debugger.
1145 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1146 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1148 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1149 deprecated, and will go away in a future revision of SBCL.
1150 * minor incompatible change: GC-ON and GC-OFF are no longer
1151 implemented with a counter, it does not matter how many times gc
1152 is switched on or off
1153 * bug fix: discriminating functions for generic function classes
1154 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1155 make invalid assumptions about method precedence order. (reported
1157 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1158 platforms. (thanks to Thiemo Seufer)
1159 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1160 * bug fix: correct pathname printing with printer escaping is on.
1161 (thanks to Kevin Reid)
1162 * bug fix: complex VOP definitions in "user-space" no longer trigger
1163 package locks. (reported by Zach Beane)
1164 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1165 for funcallable-instance functions, and is no different from
1167 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1168 characters in the current readtable. (reported by Nicholas Neuss)
1169 * bug fix: SUBTYPEP on various CONS types returns more a more
1170 accurate acknowledgment of its certainty.
1171 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1172 lists in some cases. This partially fixes bug 384.
1173 * flush all standard streams before prompting in the REPL and the
1175 * bug fix: signal handling and triggering gc do not conflict
1176 directly anymore, in particular a high frequency sb-sprof does
1177 not prevent gc from running
1178 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1179 approximation for timezone and DST information between the
1180 universal time epoch and the smallest negative 32-bit time_t.
1181 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1182 year 1899 so long as the encoded time is non-negative.
1183 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1184 the correct answer for non-constant rotate amounts when performing
1185 an inline 32-bit rotation.
1187 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1188 there is only one thread in the session
1189 ** bug fix: memory leak for streams created in one thread and
1190 written to in another
1191 ** bug fix: lockup when compiled with gcc4
1192 ** bug fix: race that allows the gc to be triggered when gc is
1194 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1195 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1197 ** bug fix: finished threads are gc'ed properly
1198 * fixed some bugs revealed by Paul Dietz' test suite:
1199 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1200 has arguments with a different set of applicable methods from
1201 the orignal arguments.
1202 ** The type error thrown by MAP now has a correct expected-type
1204 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1205 name a compiled function.
1206 ** Three MISC tests where a large bignum was improperly coerced to
1207 a float that couldn't represent that bignum during type
1208 derivation were fixed.
1209 ** SUBTYPEP can now handle types involving the negation of a
1210 list-form FUNCTION type.
1211 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1212 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1213 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1215 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1216 * New feature: Experimental support for bivalent streams: streams
1217 opened with :element-type :default now allow character and binary
1218 (unsigned-byte 8) I/O
1219 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1220 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1222 * bug fix: no more highly sporadic "couldn't check whether ~S is
1223 readable" when reading a stream and an interrupt hits in the middle
1224 of a select system call
1225 * compiler better recognizes complex arrays (reported by Eduardo
1227 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1228 for variables on SBCL built with linkage-tables. (reported by Luis
1230 * various error reporting improvements.
1231 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1232 (Thanks to Thiemo Seufer)
1233 * bug fix: nested reader invokations work correctly
1234 * bug fix: it is possible to have simultaneous references to foreign
1235 code and foreign data with the same name.
1237 ** added x86-64 support
1238 ** incompatible change: the threading api now works with thread
1239 objects instead of thread ids
1240 ** bug fix: threads are protected from signals and interruption when
1241 starting up or going down
1242 ** bug fix: a race where an exiting thread could lose its stack to gc
1243 ** bug fix: don't halt on infinite error in threads if possible
1244 ** fixed numerous gc deadlocks introduced in the pthread merge
1245 ** bug fix: fixed thread safety issues in read and print
1246 ** bug fix: debugger doesn't hang on session lock if interrupted at
1247 an inappropriate moment
1248 ** bug fix: run-program is now thread safe(r)
1249 ** bug fix: inner with-recursive-lock no longer releases the mutex
1250 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1251 * fixed some bugs revealed by Paul Dietz' test suite:
1252 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1253 EXPECTED-TYPE slots filled.
1254 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1256 ** MAKE-CONDITION accepts classes as type-designators.
1257 ** COMPILE may never return NIL.
1258 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1259 range before calling Unix time functions
1261 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1262 * numerous signal handling fixes to increase stability
1263 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1264 * minor incompatible change: we now correctly canonize default
1265 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1266 as per the MOP, rather than the historical (INITARG INITFUNCTION
1267 INITFORM). (reported by Bruno Haible)
1268 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1269 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1270 for more information.
1271 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1272 pathname is a directory pathname.
1273 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1274 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1276 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1277 :METACLASS argument in addition to a class name. (reported by
1278 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1279 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1280 (reported by Bruno Haible; patch from Gerd Moellmann)
1281 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1283 * bug fix: more cleanups to the floating point exception handling on
1284 x86-64 (thanks to James Knight)
1285 * bug fix: the compiler does not try to constant fold calls of
1286 COERCE and BIT-* functions when they return freshly constructed
1287 objects. (reported by Nikodemus Siivola and Paul Dietz)
1288 * optimization: improved the MIPS versions of generic (in the
1289 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1290 * optimization: direct conversion of (unsigned-byte 32) to floats on
1291 the PowerPC platform.
1292 * optimization: structure instances with raw slots now use less
1293 memory, and probably show better memory locality. (thanks to
1295 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1296 are treated as requests for stack allocation on the x86-64,
1297 Alpha-32, PPC and SPARC platforms.
1298 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1299 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1300 (thanks to Tony Martinez)
1302 ** gcing a dead thread can no longer lead to lockups
1303 ** threads block signals until they are set up properly
1304 ** errno is no longer shared by threads
1305 ** interrupt-thread restores the eflags register on x86
1306 ** fixed some lockups due to gc/thread interaction
1307 * fixed some bugs revealed by Paul Dietz' test suite:
1308 ** invalid dotted lists no longer raise a read error when
1309 *READ-SUPPRESS* is T
1310 ** signal an error if a symbol that names a declaration is used
1311 as the name of a type, or vice versa
1312 ** allow using the (declare (typespec var*)) abbreviation for
1313 (declare (type typespec var*)) with all type specifiers
1314 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1315 ** always return NIL from PPRINT-POP when OBJECT is NIL
1316 ** don't signal errors when pretty-printing malformed LABELS,
1317 FLET or MACROLET forms
1318 ** declarations in a DOLIST body are also in scope for the
1320 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1322 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1325 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1326 * fixed cross-compiler leakages that prevented building a 32-bit
1327 target with a 64-bit host compiler.
1328 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1329 opened with :IF-EXISTS :OVERWRITE.
1330 * fixed bug 281: error for an invalid qualifier in a short-form method
1331 combination method is not signalled until the faulty method is called.
1332 * bug fix: iteration variable type inferrer failed to deal with open
1333 intervals. (reported by Alan Shields)
1334 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1335 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1336 or more runtime options were provided to the sbcl binary.
1337 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1338 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1339 * optimization: there's now a fast-path for fixnum arguments in the
1340 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1341 * optimization: the code generated on x86-64 is more compact thanks
1342 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1343 * bug fix: floating-point exception handling now partly works on
1344 x86-64. (Thanks to James Knight)
1345 * improvement to the MIPS backend from Thiemo Seufer: C-style
1346 64-bit long long arguments and return values to alien functions
1348 * fixed some bugs revealed by Paul Dietz' test suite:
1349 ** the type-error signalled from WARN has a filled-in DATUM slot.
1350 ** the type-error required when a stream is not associated with
1351 a file has the stream as its datum.
1352 ** type-errors on single-floats on x86-64 no longer have
1353 :INVALID-OBJECT as the datum
1354 ** the type-errors signalled for invalid function names now have
1355 a correct expected type
1356 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1357 for structures defined with a :TYPE. Documentation strings for
1358 typed structures are no longer immediately discarded
1359 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1360 broadcast streams and synonym streams. FILE-LENGTH now also works
1361 on broadcast streams.
1363 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1364 * incompatible change: the --noprogrammer option, deprecated since
1365 version 0.7.5, has been removed. Please use the equivalent
1366 --disable-debugger option instead.
1367 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1369 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1370 inoperational for a while now) has been completely removed.
1371 * null lexical environments are now printed as #<NULL-LEXENV>,
1372 significantly reducing the amount of clutter in typical backtraces.
1373 * documentation on weak pointers, finalization, and after GC hooks
1374 has been added to the manual.
1375 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1376 as other specialized array types, is much faster. SUBSEQ and
1377 COPY-SEQ on such arrays have also been sped up.
1378 * optimization: EQL is now more efficient when at least other argument
1379 is known to be of type (OR FIXNUM (NOT NUMBER)).
1380 * fixed bug: the runtime is now less vulnerable to changes in the
1381 size of the SBCL object on OS X, and virtual memory is reserved for
1382 all spaces that need to be at a fixed address.
1383 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1384 * fixed bug: finalizers and after GC hooks that cause consing are now
1386 * fixed bug: compiler error messages and summaries are now printed to
1387 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1388 * fixed inference of the upper bound of an iteration variable.
1389 (reported by Rajat Datta).
1390 * fixed bug 211e: calling local functions with duplicated constant
1391 keyword argument no longer causes a bogus style warning about an
1393 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1394 ftype declaration to be disregarded. (reported by Dave Roberts)
1395 * fixed bug 373: caused by erronous compilation of references to alien
1396 variables in the runtime on ppc/darwin.
1397 * fixed bug 376: CONJUGATE type deriver.
1398 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1399 (reported by Baughn on #lisp)
1400 * a cross-compiler bug on non-x86 platforms has been identified and
1401 fixed. (thanks to Bruno Haible)
1402 * improvements to the MIPS runtime code for increased stability.
1403 (thanks to Thiemo Seufer)
1404 * increased the maximimum compact environment size to allow
1405 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1406 * partial workaround for bug 135: don't name the function we're
1407 calling for hairy cases of EVAL, so as not to accumulate one environment
1408 entry per call to EVAL. (reported by Kevin Reid)
1409 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1410 * optimization: type testing for non-vector arrays should be faster.
1411 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1413 * bug fix: setting 31st element of a bit vector to zero did not work
1415 * bug fix: redefining a class definition which failed due to a
1416 previous accessor / function clash now works (but see BUGS entry
1417 #380 for more problems in this area). (thanks to Zach Beane)
1418 * the long form of DEFINE-METHOD-COMBINATION disables method group
1419 checking when given a single method group with pattern *. (thanks
1421 * on x86 compiler supports stack allocation of results of simple
1422 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1423 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1424 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1425 * fixed some bugs related to Unicode integration:
1426 ** the restarts for recovering from input and output encoding
1427 errors only appear when there is in fact such an error to
1429 * fixed some bugs revealed by Paul Dietz' test suite:
1430 ** MISC.549 and similar: late transformation of unsafe type
1431 assertions into derived types caused unexpected code
1433 ** SCALE-FLOAT type deriver is less wrong.
1434 ** type derivers for EXP, LOG and similar functions compute result
1435 types for complex arguments better.
1436 ** (MISC.563) CONJUGATE type deriver works for very restricted
1438 ** out-of-line type testers for character strings are available.
1439 ** EQUAL compiler transform understands specialness of objects
1441 ** accessing double-floats stored on the stack now works on x86-64.
1442 ** debugger internals could sometimes create invalid lispobjs,
1443 resulting in GC crashes.
1444 ** MISC.548: type check weakening can convert required type into
1446 ** initialization forms for bindings are not in scope of free special
1449 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1450 * incompatible change: thread support for non-NPTL systems has
1451 been removed - locking is buggy and unreliable. A threaded
1452 SBCL build will now warn at startup and refuse to create
1453 new threads, unless futex support is detected in the kernel
1454 * incompatible change: the top level REPL now has only an ABORT
1455 restart associated with it, not TOPLEVEL and ABORT as it used to.
1456 TOP and TOPLEVEL are now available as debugger commands for
1457 returning to the top level.
1458 * incompatible change: forms evaluated in the REPL now use the
1459 global optimization policy.
1460 * incompatible change: user- and system-initialization files are
1461 no longer processed with LOAD, but by READ and EVAL; hence the
1462 global optimization policy, startup package, readtable, etc,
1464 * The .fasl file format number has been incremented because of
1465 various incompatible changes.
1466 * internal entry point details and argument counts no longer appear
1467 in backtraces unless explicitly requested by setting
1468 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1469 * built-in and standard functions no longer have names like "top
1470 level local call to FOO".
1471 * fixed bug 32: functions defined in non-null lexical environments
1472 now have more legible printed representation
1473 * fixed bug 33: functions defined in non-null lexical environemnts
1474 are now more amenable to inspection by INSPECT.
1475 * workaround for bug 354: XEPs no longer appear in backtraces unless
1476 explicitly requested.
1477 * fixed bug: receiving the signal which results from attempting to
1478 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1479 SIGBUS on other *BSDs) on architectures where the C stack is also
1480 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1481 to debugger confusion or wild instability. (reported by Cheuksan
1483 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1484 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1485 (reported by Lutz Euler)
1486 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1487 compiled code. (reported by Paul Dietz)
1488 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1489 specialized parameter to a method to something that is not TYPEP
1490 the specializer is now possible.
1491 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1492 face of package deletion.
1493 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1494 pathname. (reported by David Tolpin on c.l.l)
1495 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1496 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1497 STRING-STREAM and FILE-STREAM, respectively.
1498 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1499 than previously; in addition, its test suite is now run on build.
1500 (thanks to Luke Gorrie)
1501 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1502 * optimization: added a immediate representation for single-floats
1504 * fixed some bugs revealed by Paul Dietz' test suite:
1505 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1506 ** attempting to create a package with a colliding nickname causes
1507 correctable errors to be signalled.
1508 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1509 ** several x86-64 backend bugs related to sign-extension of immediate
1512 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1513 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1514 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1516 * fixed loading of multiply forward-referenced layouts.
1517 (thanks to Cheuksan Wang)
1518 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1519 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1520 (see COMPILE-FILE documentation for details of :PRINT option)
1521 * fixed bugs 19 and 317: fixed-format floating point printing is
1522 more accurate. This also fixes a bug reported by Adam Warner
1523 related to the ~@F format directive.
1524 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1526 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1527 dispatch table if the from-char is a dispatch macro character.
1528 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1529 by the wordsize no longer ignore the last word. (reported by Lutz
1531 * fixed bug in type checking of dynamic-extent variables. (reported
1533 * optimization: sequence traversal functions use their freedom to
1534 coerce function designators to functions.
1535 * optimization: code with many calls to CLOS methods specialized on
1536 CLOS classes has had redundant type checks removed.
1537 * optimization: added declarations to speed up operations that access
1538 the internal character database (for example STRING-UPCASE)
1539 * optimization: comparison operations between floats and sufficiently small
1540 fixnums no longer create extra rationals
1541 * fixed some bugs related to Unicode integration:
1542 ** portions of multibyte characters at the end of buffers for
1543 character-based file input are correctly transferred to the
1544 start of the buffer at the next read.
1545 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1546 passing it through to OPEN.
1547 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1548 argument given to internal calls to OPEN.
1549 ** the built-in comment readers (introduced by character sequences
1550 ";" and "#|") are more forgiving to encoding errors; they will
1551 STYLE-WARN and then attempt to resync the stream at a character
1552 boundary. (thanks to Teemu Kalvas)
1553 * fixed some bugs in the x86-64 port:
1554 ** Negative short int return values from c-calls are sign-extended
1556 ** The stack is aligned to 16-bytes for c-calls, as required by
1557 the ABI. (thanks to Cheuksan Wang)
1558 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1559 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1560 for enums. (thanks to Vincent Arkesteijn)
1561 ** Multiple small optimizations and bugfixes for floating point
1563 * fixed some bugs revealed by Paul Dietz' test suite:
1564 ** Space, Tab, Linefeed, Return and Page have the invalid
1565 secondary constituent character trait.
1566 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1568 ** WITH-INPUT-FROM-STRING should only update the index place on
1570 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1571 works more reliably.
1572 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1573 with one-dimensional arrays with fill pointers.
1574 ** TYPE-OF failed on a complex with an integer realpart and a
1576 ** compiler failure during type inference for the code of form
1577 (IF (EQL X (THE ...)) ...) (MISC.535).
1579 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1580 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1581 platform. The port supports 61-bit fixnums, large memory spaces
1582 and reloading shared object files.
1583 * enhancement: saving cores with foreign code loaded is now
1584 supported on ppc/Darwin in addition to the previously supported
1586 * enhancement: the statistical profiler now walks deeper into the
1587 call stack for more meaningful call-graphs and accrued time
1588 reports (x86/x86-64 only). It also now reports time spent in
1590 * enhancement: it is now possible to trace most individual methods
1591 of a generic function in addition to tracing the generic function
1593 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1595 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1596 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1597 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1598 produces an error. (thanks to Vincent Arkesteijn)
1599 * bug fix: NAMESTRING on pathnames with :WILD components in their
1600 directories works correctly. (thanks to Artem V. Andreev)
1601 * fixed bug 125: compiler preserves identity of closures. (reported
1603 * bug fixed: functions with &REST arguments sometimes failed with
1604 "Undefined function" when compiled with (DEBUG 3). (reported by
1606 * bug fix: overflow during compiling of setting element of a bit
1607 vector with constant index and value. (reported by Timmy Douglas)
1608 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1609 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1610 * on x86 compiler partially supports stack allocation of dynamic-extent
1612 * GO and RETURN-FROM do not check the extent of their exit points
1613 when compiled with SAFETY 0.
1614 * fixed some bugs related to Unicode integration:
1615 ** encoding and decoding errors are now much more robustly
1616 handled; it should now be possible to recover even from invalid
1617 input or output to the terminal. (thanks to Teemu Kalvas)
1618 ** provided a first cut at implementing STRING-TO-OCTETS and
1619 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1620 ** altered the SB-MD5 contributed module slightly, changing the
1621 interface just enough for it to be supportable for builds where
1622 lisp characters are not eight bits.
1623 * fixed some bugs revealed by Paul Dietz' test suite:
1624 ** the FORMATTER-generated functions for ~V[ conditionals require
1625 the correct number of arguments.
1626 ** READ-FROM-STRING returns the mandated second value when applied
1627 to displaced strings.
1628 ** the #\Rubout and #\Backspace characters are treated as invalid
1629 constituent characters by the tokenizer.
1631 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1632 * new feature: reloading changed shared object files with
1633 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1634 * new feature: references to foreign variables and functions
1635 can now be compiled and loaded before the corresponding shared
1636 object file is loaded, as long as the foreign definitions are
1637 available at runtime.
1638 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1639 supported. (thanks to Dan Debertin)
1640 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1641 just keywords. (thanks to Vincent Arkesteijn)
1642 * fixed bug #331: structure-class instances corresponding to
1643 DEFSTRUCT forms are now created eagerly.
1644 * fixed bug #345: backtraces from calls to undefined functions work
1645 on x86 as well. Related bug #61 is now also partially fixed on x86
1646 (backtraces from throws to unknown catch tags.)
1647 * bug fix: lambda-list parsing is now stricter vrt. order and number
1648 of lambda-list keywords.
1649 * bug fix: as specified by AMOP, an error is signalled if a
1650 class-option appears multiple times in a DEFCLASS form. (reported
1652 * bug fix: RANDOM can be compiled when the compiler derives the type
1653 of its numeric argument as a disjoint set of small integers.
1654 (reported by Paul Dietz)
1655 * bug fix: starting a core saved with shared objects loaded when
1656 those objects are not available no longer causes threaded SBCL to
1657 hang. (reported by Sean Ross)
1658 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1660 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1661 argument, which must always be NIL. (reported by Kalle Niemitalo)
1662 * bug fix: printing 1.0d+23 no longer results in an error.
1663 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1664 * bug fix: structure slot setters preserve evaluation order. (thanks
1666 * fixed some bugs related to Unicode integration:
1667 ** RUN-PROGRAM can allow its child to take input from a Lisp
1668 stream. (reported by Stefan Scholl)
1669 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1670 locales. (reported by Ken Causey)
1671 * fixed some bugs revealed by Paul Dietz' test suite:
1672 ** INCF, DECF and REMF evaluate their place form as specified in
1674 ** FORMATTER expands ~{ ~} iteration directives with V or #
1675 parameters correctly.
1676 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1677 consequent uses no arguments correctly.
1678 ** the system has a partial understanding of the (COMPLEX RATIO)
1680 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1683 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1684 * new feature: a build-time option (controlled by the :SB-UNICODE
1685 keyword feature, enabled by default) for building the system with
1686 support for the entire 21-bit character space defined by the
1688 * new feature: the system now has rudimentary external-format
1689 support; the primary user-visible change at this time is that
1690 characters with the high bit set (such as lower-case-e-acute) will
1691 print correctly to a terminal in a UTF-8 environment.
1692 * minor incompatible change: BASE-CHAR no longer names a class;
1693 however, CHARACTER continues to do so, as required by ANSI.
1694 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1695 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1696 has been moved to the SB-EXT package (temporarily re-exported from
1698 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1699 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1701 * fixed bug #351: better error-handlind and reporting for malformed
1703 * fixed bug #350: bignum-printing is now more memory-efficient,
1704 allowing printing of very large bignums, eg. (expt 2 10000000).
1705 (reported by Bruno Haible)
1706 * fixed bug #302: better primitive-type selection for intersection
1708 * fixed bug #308: non-graphic characters now all have names, as
1709 required. (reported by Bruno Haible)
1710 * bug fix: redefining a class with different superclasses now correctly
1711 removes it from the direct-subclasses of its previous superclasses.
1712 (reported by David Morse)
1713 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1714 new value now works. (reported by Bruno Haible)
1715 * bug fix: correct canonicalization of multiple non-standard slot
1716 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1717 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1718 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1719 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1720 now exists, an signals an error.
1721 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1722 during macroexpansion for non-iteration variables as well. (reported
1723 by Bruno Haible for CMUCL)
1724 * bug fix: Cyclic structures and unprintable objects in compiler
1725 messages no longer cause errors. (reported by Bruno Haible)
1726 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1727 and READ-FROM-STRING all now return a primary value of NIL if
1728 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1729 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1730 (reported by Bruno Haible for CMUCL)
1731 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1732 do not designate a condition. (reported by Bruno Haible for
1734 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1735 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1736 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1737 permitting resolution in favour of any of the conflicting symbols.
1738 (reported by Bruno Haible for CMUCL)
1739 * bug fix: EQUAL compiler optimizations is less aggressive on
1740 strings which can potentially compare true despite having distinct
1741 specialized array element types.
1742 * bug fix: unit enumerations can be defined without dividing by
1743 zero. (thanks to Vincent Arkesteijn)
1744 * FORMAT compile-time argument count checking has been enhanced.
1745 (report from Bruno Haible for CMUCL)
1746 * a partial workaround for the bug 262: the compiler does not try to
1747 inline-expand a local function doing RETURN-FROM from a deleted
1748 BLOCK. (thanks to Peter Denno for the bug report and to David
1749 Wragg for the simple test case)
1750 * fixed some bugs revealed by Paul Dietz' test suite:
1751 ** RENAME-PACKAGE allows all package designators as new package
1753 ** constraint propagation and lambda variable substitution are
1754 more cautious in dealing with partially deleted code.
1755 ** compiler sometimes lost reoptimization passes.
1756 ** CERROR, given a condition as condition designator, passes its
1757 remaining arguments to the continue format control without
1759 ** Case-altering FORMAT directives work correctly on non-ASCII
1761 ** The REAL type specifier handles bounds outside the floating
1762 point ranges without signalling FLOATING-POINT-OVERFLOW.
1763 ** Functions with IR1-transformations can create intercomponent
1764 references to global functions.
1765 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1767 ** FORMAT ~R treats a nil value for its first parameter correctly.
1769 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1770 * enhancement: saving cores with foreign code loaded is now
1771 supported on x86/NetBSD and sparc/Linux in addition to the previously
1772 supported platforms.
1773 * bug fix: on some platforms repeated installations caused multiple
1774 copies of HTML documentation to be installed -- should not happen
1775 any more. (reported by Stefan Scholl)
1776 * bug fix: parsing self-recursive alien record types multiple times
1777 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1778 original patch by Helmut Eller for CMUCL)
1779 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1780 (thanks to Richard Kreuter)
1781 * bug fix: defining classes whose accessors are methods on existing
1782 generic functions in other (locked) packages no longer signals
1783 bogus package lock violations. (reported by François-René Rideau)
1784 * bug fix: special variables as DEFMETHOD parameters no longer have
1785 associated bogus type declarations. (reported by David Wragg and
1787 * bug fix: read-write consistency on streams of element-type
1788 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1789 * bug fix: redefiniton of the only method of a generic function with
1790 no DEFGENERIC no longer emits a full WARNING. In addition,
1791 redefinition of generic functions with no DEFGENERIC to an
1792 incompatible lambda list now signals an error. (thanks to Zach
1794 * bug fix: DEFGENERIC now works even when there's a function of the
1795 same name in an enclosing lexical environment. (thanks to Zach
1797 * fixed compiler failure, caused by instrumenting code during
1798 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1799 * optimization: added loop analysis and improved register allocation
1800 * fixed some bugs revealed by Paul Dietz' test suite:
1801 ** POSITION on displaced vectors with non-zero displacement
1802 returns the right answer.
1803 ** (SIMPLE-STRING) is a valid type specifier for sequence
1805 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1807 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1808 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1810 ** The justification version of the FORMAT ~< directive treats
1811 non-zero minpad parameter correctly.
1813 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1814 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1815 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1816 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1817 the supported interface.
1818 * new feature: Single-stepping of code compiled with DEBUG 2 or
1819 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1820 * new feature: saving cores with foreign code loaded is now
1821 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1822 Timothy Moore's work for CMUCL)
1823 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1824 parameters to * instead of NIL if no initform is supplied.
1825 (reported by Johan Bockgaard)
1826 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1827 correctly when FUNCALL appears as the car of the form. Note:
1828 despite this FUNCALL forms are not currently subject to
1829 compiler-macro expansion. (port of Raymond Toy's fix for the
1830 same from CMUCL, reported by Johan Bockgaard)
1831 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1832 (thanks for Teemu Kalvas)
1833 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1834 case instead of terminating the iteration (thanks for Julian
1835 Squires, Sean Champ and Raymond Toy)
1836 * bug fix: incorrect expansion of defgeneric that caused a style
1837 warning. (thanks for Zach Beane)
1838 * bug fix: slot accessor effective method computation works properly
1839 for classes with multiple non-standard applicable methods on
1840 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1841 * on x86 compiler supports stack allocation of results of LIST and
1842 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1843 CMUCL implementation by Gerd Moellmann)
1844 * fixed some bugs revealed by Paul Dietz' test suite:
1845 ** FORMAT strings with both the ~<~:;~> form of the justification
1846 directive and pretty-printing directives cause an error.
1848 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1849 * incompatible change: the internal functions
1850 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1851 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1852 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1853 instead of the old functions.
1854 * new feature: on platforms where "dladdr" is available foreign
1855 function names now appear in backtraces. (based on Helmut Eller's
1857 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1858 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1860 * documentation: ASDF manual is now distributed with SBCL.
1861 * bug fix: SBCL can now load its contributed modules with REQUIRE
1862 even if the system-provided entries have been removed from
1863 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1865 * bug fix: backtraces involving undefined functions or assembly
1866 routines are more informative. (thanks to Brian Downing)
1867 * bug fix: mutually referent alien structures now work correctly.
1868 (reported by Rick Taube)
1869 * bug fix: structures defined by WITH-ALIEN can be referred to
1870 within other definitions in the same WITH-ALIEN.
1871 * bug fix: division operators (MOD, TRUNCATE and the like) with
1872 constant zero divisors and integer dividends no longer generate
1874 * bug fix: provide default methods for INPUT-STREAM-P and
1875 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1876 * bug fix: improve the detection and resolution of MOP metacycles
1877 (where effective method computation depends on the generic function
1878 having its effective method computed). (reported by Bruno Haible)
1879 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1880 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1881 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1882 * bug fix: pathnames with relative directory components can be
1883 represented relative to default pathnames.
1884 * optimization: in taking the GCD of bignums, reduce the two bignums
1885 to approximately the same size (using Euclid's algorithm) before
1886 applying the more sophisticated binary GCD. (thanks to Juho
1888 * optimization: COUNT on bitvectors now operates word-at-a-time.
1889 * optimization: ASH with a positive, but not necessarily constant,
1890 (leftwards) shift, when in a modular context, is compiled to a
1892 * fixed some bugs revealed by Paul Dietz' test suite:
1893 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1894 if the corresponding argument is NIL.
1895 ** FORMAT directives accepting a minpad parameter treat negative
1897 ** literal commas as character parameters to FORMAT directives are
1899 ** literal spaces directly after ~<Newline> directives within a
1900 format-logical-block (~:< ~@:>) do not induce :FILL-style
1901 conditional newlines.
1902 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1903 as specified: it no longer includes conditional newlines).
1904 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1906 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1907 :PER-LINE-PREFIX argument does not evaluate to a string.
1909 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1910 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1911 the manual for details; for now, package locks can be disabled by
1912 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1913 if no major problems are found then it is likely that they will be
1914 compiled in unconditionally.
1915 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1916 unsupported operators on all platforms. To load a shared library
1917 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1918 object file, link it into a shared library outside of SBCL and
1919 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1921 * bug workaround: ROOM T and the GC were not working together
1922 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1923 reporting facility aren't preserved. That reporting has been
1924 disabled, so now until and unless someone figures out how to make it
1925 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1926 * minor incompatible change: as threatened around sbcl-0.8.0, the
1927 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1928 The symbols are also exported from SB-PCL for backwards
1929 compatibility, but more so than before SB-PCL should be treated as
1930 an implementation-internal package.
1931 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1933 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1934 It is known as of this release not to work on the Alpha, however.
1935 * fixed bug #167: errors signalled due to illegal syntax in method
1936 bodies are now more legible.
1937 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1938 designators and can hence be used with TYPEP.
1939 * fixed bug #333: CHECK-TYPE now ensures that the type error
1940 signalled, if any, has the right object to be accessed by
1941 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1942 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1943 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1945 * fixed bug #334: programmatic addition of slots using specialized
1946 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1947 and :ALLOCATION :CLASS slots.
1948 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1949 float exponents. (rereported by Peter Seibel)
1950 * fixed a bug: #\Space (and other whitespace characters) are no
1951 longer considered to be macro characters in standard syntax by
1952 GET-MACRO-CHARACTER.
1953 * fixed bug: initialization of condition class metaobjects no longer
1954 causes an instance of the condition to be created. (reported by
1956 * fixed bug: it is now possible to have slots such that
1957 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1958 description is neither :INSTANCE nor :CLASS.
1959 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1960 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1961 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1962 non-standard SLOT-DEFINITION-ALLOCATION values.
1963 * fixed bugs: various race conditions handling exiting threads.
1964 CL-PPCRE's thread torture test now passes.
1965 * fixed arguably-a-bug: GC time proportional to number of threads in
1966 system even when most of them are idle
1967 * optimization: improved performance of BIT and SBIT on bit-vectors.
1968 * .fasl file incompatibility: The fasl file version number has
1969 been incremented because of changes associated with package locks.
1971 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1972 * minor incompatible change: the system no longer provides
1973 optimization hints (conditions which are TYPEP
1974 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1975 mode; these hints are still emitted when higher SPEED optimization
1977 * new contrib module: a sampling profiler (profiling by statistical
1978 sampling, rather than by instrumenting functions) is available as
1979 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1980 * the behaviour of the standard function ED is now customizeable by
1981 third parties through a hook variable: see ED's documentation
1982 string for information on the protocol.
1983 * the compiler no longer emits efficiency notes for (FUNCALL X)
1984 when the type of X is uncertain under default optimization
1986 * fixed bug 276: mutating a binding of a specialized parameter to a
1987 method to something that is not TYPEP the specializer is now
1989 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1990 exhibit the required behaviour on the x86 platform. (thanks to
1991 Peter van Eynde, Eric Marsden and Bruno Haible)
1992 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1993 even for difficult arguments. (reported by Peter Graves)
1994 * fixed bug 141a: the backquote printer now descends quoted
1996 * fixed another bug in backquote printing: no more destructive
1997 modification of the form's list structure. (reported by Brian
1999 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2000 move between its address being taken and the call to
2001 interrupt_thread, fixing a crashing race condition.
2002 * the SB-POSIX contrib implementation has been adjusted so that it
2003 no longer exhibits ridiculously poor performance when constructing
2004 instances corresponding to C structs.
2006 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2007 * minor incompatible change: the sb-grovel contrib now treats C
2008 structures as alien (in the sense of SB-ALIEN) objects rather than
2009 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2010 has implications for memory management of client code
2011 (sb-grovel-returned objects must now be manually managed) and for
2012 type safety (alien objects now have full types).
2013 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2014 used to control emission of compiler diagnostics, rather than the
2015 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2016 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2017 quality should be considered deprecated.
2018 * install.sh now installs the user manual as well
2019 * (not quite a new documentable feature, but worth considering in
2020 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2021 beginnings of a semantically meaningful condition hierarchy is
2022 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2024 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2025 designator as the defaults argument.
2026 * fixed bug: Displaced arrays whose displaced-to array has become
2027 too small now cause ARRAY-DIMENSION to signal an error, providing
2028 for safer bounds-checking. (reported by Bruno Haible)
2029 * fixed bug: DEFCLASS slot definitions with identical :READER and
2030 :WRITER names now signal a reasonable error. (reported by Thomas
2032 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2034 * fixed bug: Invalid slot specification errors now print correctly.
2035 (thanks to Zach Beane)
2036 * fixed bug 320: Shared to local slot value transfers in class
2037 redefinitions now happen corrently from superclasses as
2038 well. (reported by Bruno Haible)
2039 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2041 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2042 improper lists correctly. (reported by Bruno Haible)
2043 * fixed bug 313: source-transform for <fun-name> was erroneously
2044 applied to a call of a value of a variable with name <fun-name>.
2045 (reported by Antonio Menezes Leitao)
2046 * fixed bug 307: The obsolete instance protocol ensures that
2047 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2048 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2049 remove dead unknown-values globs from the middle of the stack.
2050 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2051 easier to resume long computations after using *BREAK-ON-SIGNALS*
2052 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2053 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2054 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2055 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2056 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2057 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2059 * fixed a (fairly theoretical) bug in string printing: if
2060 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2061 does not have array-element-type equal to the most general string
2063 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2064 function name. (thanks to Nikodemus Siivola)
2065 * fixed bug: compile-time format string checker failed on
2066 non-closed ~{. (reported by Thomas F Burdick)
2067 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2068 don't warn on legal loop code involving more than one aggregate
2069 boolean. (thanks to Nikodemus Siivola)
2070 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2071 in keyword position to MAKE-PACKAGE was not being reported as
2073 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2074 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2076 * fixed bugs 280 and 312: the checking for multiple definitions in a
2077 file is less likely to become confused by uses of inline
2079 * fixed bug: the #S reader macro performs the keyword coercion
2080 specified for slot names. (reported by Kalle Niemitalo)
2081 * fixed bug: lambda lists may contain symbols whose names start with
2082 & but are not lambda-list-keywords; their occurrence triggers a
2084 * fixed bug 321: define-method-combination argument lambda lists do
2085 not cause invalid code to be generated when &optional or &aux
2086 variables have default values. (reported by Bruno Haible)
2087 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2088 have CLOS classes; this appears also to have fixed reported
2089 instability in stack exhaustion detection.
2090 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2091 MUFFLE-WARNING when no associated restart is present is now
2093 * optimization: rearranged the expansion of various defining macros
2094 so that each expands into only one top-level form in a
2095 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2097 * optimization: used a previously little-used slot in symbols to
2098 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2100 * fixed some bugs revealed by Paul Dietz' test suite:
2101 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2103 ** Fixed an optimization bug related to inheritance of initforms
2104 from local to shared slots.
2105 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2106 ** CAST optimizer forgot to flush argument derived type.
2107 ** print/read consistency on floats is now orders of magnitude
2108 more likely. (thanks also to Bruno Haible for a similar report
2110 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2111 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2112 for objects of type REAL. Make it so.
2113 ** SXHASH is specified (infelicitously) to respect similarity,
2114 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2115 it so. (thanks to Markus Ziegler)
2116 ** on the Alpha, the compiler succeeds in compiling functions
2117 returning a known number of arguments greater than 63.
2118 ** fixed handling of invalid NIL arguments in keyword position
2120 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2122 ** provided a readably-printable representation for RANDOM-STATE
2124 ** ensured that pathnames, where they have a namestring, always
2125 print using #P"..." syntax.
2127 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2128 * Support for the forthcoming 2.0 version of the NetBSD kernel
2129 running on x86 hardware has been added. (thanks to Perry
2130 E. Metzger most immediately, and others for their past work)
2131 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2132 current release version; SBCL's OpenBSD support had been broken
2133 since about the time of OpenBSD's switch to ELF binary format.)
2134 * [placeholder for DX summary]
2135 ** user code with &REST lists declared dynamic-extent, under high
2136 speed or space and low safety and debug optimization policy.
2137 * The manual has been converted to Texinfo format and the debugger
2138 chapter from the cmucl manual has been added.
2139 * A facility has been added to extract documentation strings from
2140 sbcl and store them as Texinfo-formatted snippets for inclusion in
2141 the manual (via Texinfo's @include directive)
2142 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2143 the test case to Dave Roberts)
2144 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2145 pointers. (reported by Sean Ross)
2146 * bug fix: PROFILE output is printed nicely even for large numerical
2147 values. (thanks to Zach Beane)
2148 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2149 greater than 32 handle EOF correctly.
2150 * bug fix: on X86 an immediate argument of the IMUL instruction is
2151 correctly printed in disassembly. (thanks to Lutz Euler)
2152 * bug fix: class slots in redefined classes preserve their old
2153 values. (thanks to Bruno Haible and Nikodemus Siivola)
2154 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2155 succeeds. (reported by Marco Baringer)
2156 * fixed some bugs revealed by Paul Dietz' test suite:
2157 ** READ-SEQUENCE now works on ECHO-STREAMs.
2158 ** RATIONALIZE works more according to its specification. (thanks
2161 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2162 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2163 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2164 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2165 should matter to you only if you rebind the printer control
2166 variables and then find you want different bindings in the
2167 debugger than in the ordinary execution of your program.)
2168 * The runtime build system has been tweaked to support building
2169 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2170 assembler and linker. (thanks to Nikodemus Siivola)
2171 * Unbound, undefined, undeclared variables now trigger full
2172 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2173 more often programmer error than deliberate exploitation of undefined
2175 * optimization: the hash algorithm for strings has changed to one
2176 that is less vulnerable to spurious collisions. (thanks to Juho
2178 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2179 needless bounds checking. (thanks to Juho Snellman)
2180 * optimization: implemented multiplication as a modular
2181 (UNSIGNED-BYTE 32) operation on the PPC backend.
2182 * fixed some bugs revealed by Paul Dietz' test suite:
2183 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2185 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2186 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2187 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2189 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2191 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2193 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2194 their output stream on EOF from read.
2195 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2196 have been read to end-of-file.
2197 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2199 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2200 description of determination of which consecutive characters
2202 ** Printing the "Space" character with escaping on now yields "#\\ ",
2203 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2204 ** Reading floating-point numbers with *READ-BASE* set to a number
2205 less than 10 works correctly.
2206 ** Reading floating-point numbers with *READ-BASE* set to a number
2207 more than 10 works correctly.
2208 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2209 the readtable currently in effect.
2211 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2212 * minor incompatible change: parsing of namestrings on a physical
2213 (Unix) host has changed; numbers after the final #\. in a
2214 namestring are no longer interpreted as a version field. This is
2215 intented to be largely invisible to the user, except that the
2216 meaning of the namestring "*.*.*" has changed: it now refers to a
2217 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2218 should usually be replaced by
2219 (make-pathname :name :wild :type :wild :version :wild)
2220 with the added benefit that this is more likely to be portable.
2221 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2222 option to OPEN now signals an error if the file being opened
2223 exists; this may have an impact on existing code.
2224 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2225 signals received, on the PowerPC platforms (both Linux and
2226 Darwin). (thanks to Pierre Mai for pointing out the location of
2228 * several fixes on OS X: The system now builds and runs cleanly on
2229 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2230 (thanks to Brian Mastenbrook)
2231 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2232 second-resolution: integer multiples of 1/3600 between -24 and 24.
2233 (thanks to Vincent Arkesteijn)
2234 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2235 type when called with 1 argument; PEEK-CHAR checked type of
2236 PEEK-TYPE only after having read first character from a
2237 stream. (reported by Peter Graves)
2238 * bug fix: the garbage collector now has much better locality
2239 behaviour, and in particular no longer treats all memory as being
2240 exhausted when many small objects point to each other in a deeply
2242 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2243 recognized as being TYPEP their class.
2244 * bug fix: the PUSHNEW documentation string has been corrected.
2245 (thanks to Vincent Arkesteijn)
2246 * bug fix: defaulting of the value for the last of an atypically
2247 large number of multiple values being bound was not being
2248 performed correctly on the Alpha or PPC platforms
2249 * optimization: implemented multiplication as a modular
2250 (UNSIGNED-BYTE 32) operation on the x86 backend.
2251 * optimization: SEARCH on simple-base-strings can now be open-coded.
2252 (see also contrib/compiler-extras.lisp for inspiration for
2253 teaching the compiler about the Boyer-Moore algorithm).
2254 * value, returned by MAX (and MIN) called with several EQUALP, but
2255 not EQL, arguments now does not depend on compiler settings.
2256 * fixed some bugs revealed by Paul Dietz' test suite:
2257 ** in stack analysis liveness information is propagated from
2258 non-local entry points.
2259 ** pathwise CAST removing failed when the CAST node did not start
2261 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2262 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2264 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2265 host is already defined.
2266 ** RENAME-FILE works on streams instead of signalling an internal
2268 ** PEEK-CHAR uses the current readtable when determining whether
2269 or not a character is whitespace.
2270 ** MERGE-PATHNAMES handles the case when the pathname does not
2271 specify a name while the default-pathname specifies a version.
2272 ** Pathnames now stand a better chance of respecting print/read
2274 ** Attempting to use standardized file system operators with a
2275 pathname with invalid :DIRECTORY components signals a
2277 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2279 ** DIRECTORY on logical pathnames is more correct.
2280 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2281 signal a TYPE-ERROR if their argument is not a stream.
2282 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2283 designator argument does not designate a stream.
2284 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2285 examining the synonym.
2286 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2288 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2289 element-type larger than ([UN]SIGNED-BYTE 32).
2291 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2292 * When built with the :SB-FUTEX feature, threaded builds now take
2293 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2294 for faster/more reliable mutex and condition variable support.
2295 * Incompatible change (but one you probably shouldn't have been using
2296 anyway): the interface and code for arbitrating between multiple
2297 threads in the same user session has been redesigned.
2298 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2299 error when called without an explicit environment argument.
2300 (thanks to Nikodemus Siivola)
2301 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2302 confused about their position. (thanks to Adam Warner and Gerd
2304 * bug fix: Pretty printing backquoted forms with unquotations in the
2305 argument list position of various code constructs such as LAMBDA
2306 now works correctly. (reported by Paul Dietz)
2307 * bug fix: Pretty printing unquotations no longer loses all
2308 stream position information.
2309 * optimization: performance of string output streams is now less
2310 poor for multiple small sequence writes.
2311 * optimization: performance of CSUBTYPEP in the presence of complex
2312 expressions involving CONS and NOT many times has been improved.
2313 (reported by Paul Dietz)
2314 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2316 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2317 streams for the REPL, for the debugger, and so on. (thanks to
2319 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2320 * fixed some bugs revealed by Paul Dietz' test suite:
2321 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2322 is ignored if it is too small, rather than propagating through
2323 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2324 ** extremely complex negations of CONS types were not being
2325 sufficiently canonicalized, leading to inconsistencies in
2327 ** VALUES tranformer lost derived type.
2329 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2330 * fixed a bootstrapping bug: the build process no longer assumes
2331 that the various BOOLE-related constants have the same value in
2332 host and target lisps. (noted by Paul Dietz' test suite on an
2333 SBCL binary built from CLISP)
2334 * The system can now be dynamically linked on the MIPS platform,
2335 which enables dynamic loading of foreign code from Lisp. (thanks
2336 to Ralf Baechle for discussions on the MIPS ABI)
2337 * The system now records debugging information for its own source
2338 files in a filesystem-position-independent manner, relative to
2339 the "SYS" logical host.
2340 * fixed a compiler bug: MV-LET convertion did not check references
2341 to the "max args" entry point. (reported by Brian Downing)
2342 * tweaked disassembly notes to be less confident about proclaiming
2343 some instruction as an LRA. (thanks to Brian Downing)
2344 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2345 now each have their own history, command character, and other
2346 characteristics. (thanks to David Lichteblau)
2347 * fixed some bugs revealed by Paul Dietz' test suite:
2348 ** compiler failure in compiling LOGAND expressions including a
2350 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2351 shift greater than 32.
2352 ** FLUSH-DEST did not mark blocks for type check regeneration.
2353 ** HANDLER-CASE failed to accept declarations in handler clauses
2354 in some circumstances.
2356 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2357 * New code in contrib/sb-introspect (still probably not entirely
2358 stable yet) provides some support for smart Lisp development
2359 environments like SLIME.
2360 * The conditions signalled for errors occurring when loading .fasl
2361 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2362 in a way which should help ASDF recover gracefully.
2363 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2364 stuff has been cleaned up. If you code contrib/ stuff, this might
2365 affect you, and you can look at contrib/README, contrib/STANDARDS,
2366 and/or the 0.8.4.27 diff to check.
2367 * In full calls the compiler now does not generate checks for declared
2368 argument types for all arguments.
2369 * various threading fixes
2370 ** and some experimental patches which didn't make it into
2371 the main tree for this release, but which are shipped in
2372 contrib/experimental-thread.patch as a possible fix for some
2373 failures (deadlock, spinning...) in GC-intensive multithreaded
2375 * fixed PPC build problem (source code incompatibility of different
2376 library versions): added offsetof() hackery which attempts to divine
2377 where glibc maintainers put uc_mcontext today
2378 * fixed bug 282: compiler does not trust type assertions while passing
2379 arguments to a full call.
2380 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2381 &OPTIONAL VALUES type parameter.
2382 * fix bug 214: algorithm for noting rejected templates is now more
2383 similar to that of template seletion. (also reported by rydis on
2385 * fixed bug 141b: printing backquoted information readably and prettily
2386 inserts a space where necessary.
2387 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2388 (CHARACTER 10) are now reported as errors, rather than propagated
2389 as unknown types. (reported by piso on #lisp)
2390 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2391 with values NIL and :ERROR. (thanks to Milan Zamazal)
2392 * fixed bug 191c: CLOS now does proper keyword argument checking as
2393 described in CLHS 7.6.5 and 7.6.5.1.
2394 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2395 longer raise an error; further, using a list as a for-as-arithmetic
2396 counter now raises a meaningful error.
2397 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2398 now understood by sequence creation functions such as MAKE-SEQUENCE
2400 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2401 a STRING-INPUT-STREAM.
2402 * compiler enhancement: SIGNUM is now better able to derive the type
2404 * type declarations inside WITH-SLOTS are checked. (reported by
2406 * fixed some bugs revealed by Paul Dietz' test suite:
2407 ** incorrect optimization of TRUNCATE for a positive first
2408 argument and negative second.
2409 ** compiler failure in let-convertion during flushing dead code.
2410 ** compiler failure while deriving type of TRUNCATE on an
2411 interval, containing 0.
2412 ** ASH of a negative bignum by a negative bignum count now returns
2414 ** intersection of CONS types now canonicalizes properly, fixing
2415 inconsistencies in SUBTYPEP.
2417 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2418 * incompatible change: The --disable-debugger command line
2419 option now clobbers the debugger at a more fundamental
2420 level, by redefining #'INVOKE-DEBUGGER instead of by
2421 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2422 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2423 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2424 enter the debugger REPL and then suffer recursive errors
2425 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2426 while under the new system, BREAK in --disable-debugger mode
2427 terminates the system just as an unhandled error would.
2428 * fixed compiler performance when processing loops with a step >1;
2429 * bug fix: DOCUMENTATION now retrieves generic function
2430 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2431 support has been systematized, and now supports the methods
2432 specified by ANSI, along with a default method and a method for
2433 slot documentation. (reported by Nathan Froyd)
2434 * bug fix: effective methods associated with a generic function are
2435 no longer cached over a change of that generic function's method
2436 combination. (reported by Andreas Fuchs)
2437 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2438 names a non-existent directory. (thanks to Andreas Fuchs)
2439 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2440 the PPC platform, lead to stack corruption; this has been fixed.
2441 (reported by Rainer Joswig)
2442 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2443 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2444 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2445 arithmetic in general, is now much more reliable on the PPC
2447 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2448 platform now returns the right answer.
2449 * optimization: restored some effective method precomputation in
2450 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2451 precomputation is now tunable.
2452 * optimization: compiler-internal data structure use has been
2453 reviewed, and changes have been made that should improve the
2454 performance of the compiler by about 20%.
2455 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2456 simple-base-strings and simple-bit-vectors is improved.
2457 * optimization: the optimization of 32-bit logical and arithmetic
2458 functions introduced in version 0.8.3 on the x86 has been
2459 implemented on the mips, ppc and sparc platforms; an
2460 implementation of the same facility, but for 64-bit arithmetic,
2461 has been added for the alpha.
2462 * microoptimization: the compiler is better able to make use of the
2463 x86 LEA instruction for multiplication by constants.
2464 * bug fix: in some situations compiler did not report usage of
2465 generic arithmetic in (SPEED 3) policy.
2466 * bug 145b fix: compiler used wrong type specifier while converting
2467 MEMBER-types to numeric.
2468 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2470 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2471 index argument now works properly on the Alpha platform.
2472 * bug fix: floating point exception treatment on the Alpha platform
2474 * bug fix: FILE-POSITION works much better on string input and
2475 output streams. (thanks to Nikodemus Siivola)
2476 * bug fix: many threading/garbage collection symptoms sorted.
2477 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2478 might be pseudo-atomic.
2479 * internal change: Stopping for GC is now done with signals not
2480 ptrace. GC is now done in whichever thread wanted it, instead of
2482 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2484 * bug fix: result form in DO is not contained in the implicit
2486 * incompatible change: ICR structure is changed; the value part of
2487 CONTINUATION is now called LVAR; corresponding functions are
2488 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2489 * added type deriver for ISQRT (thanks to Robert E. Brown).
2490 * bug fix: better support for loading from the command line when an
2491 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2493 * fixed some bugs revealed by Paul Dietz' test suite:
2494 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2495 ** ROUND and FROUND now give the right answer when given very
2496 small float arguments.
2497 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2499 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2500 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2501 ** LCM should return a non-negative integer.
2502 ** PARSE-INTEGER returned the index of a terminator instead of the
2503 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2504 ** PARSE-INTEGER returned an incorrect index being applied to a
2506 ** LCM with two arguments of 0 returns 0 rather than signalling
2508 ** unsigned addition of a 32-bit constant with the high bit set no
2509 longer causes an internal compiler error on the x86.
2510 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2511 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2512 with negative last argument.
2513 ** byte specifiers with zero size and position no longer cause
2514 an error during type derivation.
2515 ** bignum multiplication on the Alpha platform now returns the
2517 * porting: The system now builds on SuSE AMD64, although it still
2518 generates a 32-bit binary.
2519 * .fasl file incompatibility: The fasl file version number has
2520 been incremented (because of the changes to internal compiler
2521 data structures referred to above).
2523 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2524 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2525 more accurately, on the Darwin kernel running on PowerPC hardware.
2526 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2527 * Compiler code deletion notes now signal a condition of type
2528 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2529 an associated MUFFLE-WARNING restart.
2530 * The compiler now performs limited argument count validation of
2531 constant format strings in FORMAT, and where appropriate in ERROR,
2532 CERROR and WARN. (thanks to Gerd Moellmann)
2533 * New ASDF-INSTALL contrib can be used for automatic download and
2534 installation of third-party Lisp code from CCLAN or other sites
2536 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2537 which forces another thread to execute a function supplied by the
2539 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2540 now accept and act upon their :ELEMENT-TYPE keyword argument.
2541 (reported by Martin Atzmueller, Edi Weitz)
2542 * bug fix: FILE-POSITION now accepts position designators up to
2543 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2544 is the greater. (thanks to Patrik Nordebo)
2545 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2546 Kalle Olavi Niemitalo)
2547 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2548 printable. (reported by Eric Marsden)
2549 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2550 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2551 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2552 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2553 (thanks to Stig Sandoe)
2554 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2555 UNREAD-CHAR now correctly includes the unread character in the
2556 target sequence. (thanks to Gerd Moellmann)
2557 * bug fix in threaded builds: the system can now be suspended and
2558 resumed by shell job control with minimal disruption.
2559 * bug fixes in times and timezones >2038AD
2560 * better handling of "where is GNU make?" problem in build scripts
2561 (thanks to Nikodemus Siivola)
2562 * new optimization: inside a named function any reference to a
2563 function with the same name is considered to be a self-reference;
2564 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2565 optimization quality.
2566 * new optimization on x86: logical functions and + now have
2567 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2568 used when the result is truncated to 32 bits.
2569 * VALUES declaration is partially enabled.
2570 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2571 * fixed some bugs revealed by Paul Dietz' test suite:
2572 ** The system now obeys the constraint imposed by
2573 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2574 types form a lattice under type intersection.
2575 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2576 ** ASSOC now ignores NIL elements in an alist.
2577 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2578 and (1+ MOST-POSITIVE-FIXNUM) answers.
2579 ** The addition of a method with invalid qualifiers to a generic
2580 function does not cause an error to be signalled immediately;
2581 a warning is signalled, and the error is generated only on
2582 calling the generic function.
2583 * changed .fasl file version number, in part to add type codes for
2584 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2585 obscure ANSI requirements
2587 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2588 * fixed bug 148: failure to inline-expand a local function left
2589 garbage, confusing the compiler.
2590 * fixed bugs 3cd: structure slot readers perform type check if the
2591 slot can have an invalid value (i.e. it is either not initialized
2592 or can be written with a less specific slot writer).
2593 * bug fix: the compiler now traps array references to elements off
2594 the end of an array; previously, the bounds checking in some
2595 circumstances could go off-by-one.
2596 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2598 * type declarations for array element types now obey the description
2599 on the CLHS page "Declaration TYPE", as per discussions on
2600 sbcl-help around 2003-05-08. This means that a declaration
2601 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2602 declaration, all references to BAR will be asserted or assumed
2603 (with THE, so dependent on compiler policy) to involve objects of
2604 type FOO. Note that no such declaration is implied in
2605 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2606 * declared types of functions from the "Conditions"
2607 chapter. (reported by Paul Dietz)
2608 * bug fix: CERROR accepts a function as its first argument.
2609 * bug fix: NTH an NTHCDR accept a bignum as index
2610 arguments. (reported by Adam Warner)
2611 * optimization: character compare routines now optimize comparing
2612 against a constant character. (reported by Gilbert Baumann)
2613 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2614 argument now works properly on the MIPS platform.
2615 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2616 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2617 declared to have a functional type, does not cause an error
2619 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2620 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2621 * fixed bug: if last continuation of a deleted block has a
2622 destination, this destination should be deleted too. (reported by
2624 * fixed a bug in the bootstrap process: the host compiler's values
2625 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2626 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2627 test case from Patrik Nordebo)
2628 * improved the ability of the disassembler on the PPC platform to
2629 provide helpful disassembly notes.
2630 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2631 the class in more cases than previously.
2632 * bug fix: FILE-POSITION now understands :START and :END for
2633 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2634 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2635 8) data. (thanks to David Lichteblau)
2636 * bug fix: it is possible to add a method to a generic function
2637 without lambda list.
2638 * bug fix: reader failed to signal END-OF-FILE inside an
2639 object representation. (reported by Nikodemus Siivola)
2640 * fixed some bugs revealed by Paul Dietz' test suite:
2641 ** LAST and [N]BUTLAST should accept a bignum.
2642 ** condition slot accessors are methods.
2643 ** (VECTOR NIL) is a subtype of STRING.
2645 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2646 * minor incompatible change: some nonsensical specialized lambda
2647 lists (used in DEFMETHOD) which were previously ignored now signal
2649 * minor incompatible change: the system is now aware of the types of
2650 variables in the COMMON-LISP package, and will signal errors for
2651 most violations of these type constraints (where previously they
2652 were silently accepted).
2653 * minor incompatible change: COMPILE-FILE now uses the freedom
2654 afforded (ANSI 3.2.2.3) to use derived function types for
2655 functions defined in the same file. This also permits the system
2656 to warn on static type mismatches and function
2657 redefinition. (Currently it does not work with high DEBUG level.)
2658 * minor incompatible change: VALUES declaration is disabled.
2659 * When issuing notes, the compiler now signals a condition of type
2660 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2661 restart for use in user handlers. It is expected that the
2662 COMPILER-NOTE condition will eventually become a condition
2663 supertype to a hierarchy of note types, which will then be
2664 handleable in a similar fashion. However, at the moment, no such
2665 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2666 but it's an implementation detail, not a classification for the
2668 * Changes in type checking closed the following bugs:
2669 ** type checking of unused values (192b, 194d, 203);
2670 ** template selection based on unsafe type assertions (192c, 236);
2671 ** type checking in branches (194bc).
2672 * A short form of VALUES type specifier has ANSI meaning (it has
2673 increased the number of situations when SBCL cannot perform type
2675 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2676 keywords or constants is permissible.
2677 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2678 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2679 classes. (thanks to Antonio Martinez)
2680 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2681 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2682 argument to a pretty-print stream. Instead, it leaves any such
2683 operations to DESCRIBE-OBJECT methods.
2684 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2686 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2687 lambda lists. (thanks to David Lichteblau)
2688 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2689 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2690 (thanks to Gerd Moellmann)
2691 * fixed bug 246: increased compilation speed of long
2692 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2694 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2696 * DEFCONSTANT now throws a condition of type
2697 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2698 constant to a non-EQL value; CONTINUE and ABORT restarts
2699 respectively change and preserve the value.
2700 * fixed bug 63: The code walker, part of the implementation of CLOS,
2701 is now better at handling symbol macros.
2702 * bug fix: There is no longer an internal implementation type named
2703 CL:LENGTH. (reported by Raymond Toy)
2704 * bug fix: In macro-like defining macros/special operators the
2705 implicit block does not enclose the lambda list.
2706 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2707 enclosed in parentheses) are not suitable as type specifiers, and
2708 their use properly signals an error now.
2709 * bug fix: An argument count mismatch for a type specifier in code
2710 being compiled no longer causes an unhandled error at compile
2711 time, but signals a compile-time warning.
2712 * fixed simple vector readable printing
2713 * bug fix: DESCRIBE takes more care over whether the class
2714 precedence list slot of a class is bound before accessing it.
2715 (reported by Markus Krummenacker)
2716 * bug fix: FORMATTER can successfully compile pretty-printer format
2717 strings which use variants of the ~* directive inside.
2718 * bug fix: SEARCH now applies its TEST predicate to the elements of
2719 the arguments in the correct order. (thanks to Wolfhard Buss)
2720 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2721 in type inference. (thanks to Robert E. Brown)
2722 * bug fix: Reading in symbols with an explicit package name of ""
2723 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2724 * fixed some bugs revealed by Paul Dietz' test suite:
2725 ** NIL is now allowed as a structure slot name.
2726 ** Arbitrary numbers, not just REALs, are allowed in certain
2727 circumstances in LOOP for-as-arithmetic clauses.
2728 ** Multiple class redefinitions before slot access no longer
2729 causes a type error.
2730 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2731 association between the name and a class.
2732 ** Generic functions with non-standard method-combination and over
2733 five methods all of which return constants no longer return NIL
2734 after the first few invocations. (thanks to Gerd Moellmann)
2735 ** CALL-NEXT-METHOD with no arguments now passes the original
2736 values of the arguments, even in the presence of assignment.
2737 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2739 ** DISASSEMBLE works with closures and funcallable instances.
2740 ** ADD-METHOD now returns the generic function, not the new method.
2741 ** FIND-METHOD signals an error if the lengths of the specializers
2742 is incompatible with the generic function, even if the ERRORP
2744 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2745 which its argument is a member.
2746 ** DEFCLASS only redefines the class named by its class-name
2747 argument if that name is the proper name of the class;
2748 otherwise, it creates a new class.
2749 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2750 of the UNBOUND-SLOT condition to the name of the slot.
2751 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2752 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2753 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2754 SLOT-MAKUNBOUND in the specified fashion.
2756 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2757 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2758 cross-compilation host. As a consequence, we can now bootstrap our
2759 way up to SBCL starting with a bare gcc toolchain and human-readable
2760 source code (first the source to CLISP, then the source to SBCL).
2761 * A contributed module containing a partial implementation of the
2762 simple-streams interface has been included. (thanks to Rudi
2764 * A contributed module implementing the RFC1321 Message Digest
2765 Algorithm, known as MD5, has been included.
2766 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2767 no longer has any effect, as the code controlled by this feature
2768 has been deleted. (As far as we know, no-one has ever built using
2769 this feature, and its semantics were confused in any case).
2770 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2771 work on conditions (as required by the ANSI specification),
2772 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2773 expected behaviour on conditions. Users should note, however,
2774 that such behaviour is not required by the ANSI specification,
2775 and so use of this behaviour may render their code unportable.
2776 * fixed some bugs revealed by Paul Dietz' test suite:
2777 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2779 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2780 ** SXHASH on condition objects no longer returns NIL.
2781 ** :ALLOCATION :CLASS slots are better treated; their values are
2782 updated on class redefinition, and initforms inherited from
2783 superclasses are applied.
2784 ** REMOVE-METHOD returns its generic function argument even when
2785 no method was removed.
2786 ** SHARED-INITIALIZE now initializes the values of the requested
2787 slots, including those with :ALLOCATION :CLASS.
2788 ** ALLOCATE-INSTANCE now works on structure classes defined via
2789 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2791 ** SLOT-EXISTS-P now works on conditions, as well as structures
2793 ** MAKE-LOAD-FORM now has the required methods on
2794 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2795 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2796 keyword, and now implements the SLOT-NAMES argument.
2797 ** methods with &OPTIONAL arguments no longer allow too many
2798 arguments to be passed in the call without error.
2799 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2800 option is consistent with the required arguments of the generic
2801 function lambda list.
2802 * bug fix: REQUIRE accepts a string designator. (Thanks to
2804 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2805 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2806 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2807 * bug fix: The compiler now checks for duplicated variables in macro
2809 * bug fix: SETQ on globals returns the correct value.
2810 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2811 not a COMPILER-ERROR (followed by some other strange error on
2812 choosing the CONTINUE restart).
2813 * bug fix: make.sh and friends are now more consistent in the way that
2814 they look for GNU "make".
2816 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2817 * experimental native threads support (on x86 Linux >=2.4 only).
2818 This is not compiled in by default: you need to add :SB-THREAD to
2819 the target features. See the "Beyond ANSI" chapter of the manual
2821 * fix for longstanding nonANSIism: The old distinction between
2822 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2823 The return value from CL:FIND-CLASS is now a CLOS class, and
2824 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2825 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2826 * An interface to the MetaObject Protocol, as described in Kiczales,
2827 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2828 MIT Press, 1991, is available from the SB-MOP package.
2829 * incompatible change: the SB-PCL package should now be considered
2830 a private implementation detail, and no longer a semi-private MOP
2832 * minor incompatible change: due to rearrangement for threads, the
2833 control stack and binding stack are now allocated at arbitrary
2834 addresses instead of being hardcoded per-port. Users affected by
2835 this probably have to be doing advanced things with shared
2836 libraries, and will know who they are.
2837 * minor incompatible change: Previously, all --eval forms used were
2838 processed with READ before any of them were processed with EVAL.
2839 Now each --eval form is processed with both READ and EVAL before
2840 the next --eval form is processed. (Thus package operations like
2841 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2842 work as the user might reasonably expect.)
2843 * minor incompatible change: *STANDARD-INPUT* is now only an
2844 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2846 * minor incompatible change: Y-OR-N-P is now character-oriented,
2847 not line oriented. Also, YES-OR-NO-P now works without errors.
2848 (thanks to Antonio Martinez)
2849 * sb-aclrepl module improvements: an integrated inspector, added
2850 repl features, and a bug fix to :trace command.
2851 * Known functions, which cannot be open coded by the backend, are
2852 considered to be able to check types of their arguments. (fixing
2853 a bug report by Nathan J. Froyd)
2854 * fixed a bug in computing method discriminating functions: It is
2855 now possible to define methods specialized on classes which have
2856 forward-referenced superclasses. (thanks to Gerd Moellmann)
2857 * fixed evaluation order in optional entries (reported by Gilbert
2859 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2860 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2861 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2862 required-by-AMOP NAME argument, as well as CLASS and
2863 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2864 * fixed bug 20: DEFMETHOD can define methods using names that are
2865 not the proper names of classes to designate class specializers.
2866 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2867 Unix file descriptors, instead of blowing up. (thanks to Antonio
2869 * Garbage collection refactoring: user-visible change is that a
2870 call to the GC function during WITHOUT-GCING will not do garbage
2871 collection until the end of the WITHOUT-GCING. If you were doing
2872 this you were probably losing anyway.
2873 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2874 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2875 TYPEP the latter but not the former.
2876 * The compiler issues a full WARNING for calls to undefined functions
2877 with names from the CL package.
2878 * MAP-INTO for a vector destination is open coded. (reported by
2879 Brian Downing on c.l.l)
2880 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2881 documentation string.
2882 * fixed some bugs revealed by Paul Dietz' test suite:
2883 ** COPY-ALIST now signals an error if its argument is a dotted
2885 ** Condition slots are now accessed more correctly in the presence
2886 of multiple initargs for a given slot.
2887 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2888 exclude restarts of the same name associated with a different
2890 ** DEFCLASS of forward-referenced classes with another
2891 forward-referenced class in the superclasses list no longer
2893 ** Condition slots are now initialized once each, not multiple
2894 times. (thanks to Gerd Moellmann)
2895 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2897 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2898 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2899 arguments contain duplicated elements.
2900 ** RESTART-CASE understands local macros.
2901 ** RESTART-CASE associates exactly its own restarts with a condition.
2902 ** ENDP in safe mode checks its argument to be of type LIST.
2903 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2904 in question is unbound.
2905 ** Optimized MAKE-INSTANCE functions no longer cause internal
2906 assertion failures in the presence of duplicate initargs.
2907 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2908 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2909 propagated even in the ctor optimized implementation of
2911 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2913 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2914 * a better implementation of SXHASH on (simple) bit vectors,
2915 measured both in execution speed and in distribution of results
2916 over the positive fixnums, has been installed. Likewise, a better
2917 implementation of EQUAL for simple bit vectors is now available.
2918 * fixed CEILING optimization for a divisor of form 2^k.
2919 * fixed bug 240 (emitting extra style warnings "using the lexical
2920 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2921 by Antonio Martinez)
2922 * fixed SXHASH, giving different results for NIL depending on type
2923 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2924 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2925 variable. (found by Rolf Wester)
2926 * SBCL does not ignore type declarations for special
2927 variables. (reported by rif on c.l.l 2003-03-05)
2928 * some bug fixes in contrib/sb-aclrepl/
2929 * fixed some bugs revealed by Paul Dietz' test suite:
2930 ** a bug in the CONS type specifier, whereby the CAR and CDR
2931 types got intertwined, has been fixed;
2932 ** the type system is now able to reason about the interaction
2933 between INTEGER and RATIO types more completely;
2934 ** APPEND, [N]REVERSE and NRECONC check that those their
2935 arguments, which must be proper lists, are really so;
2936 ** An array specialized to be unable to hold elements has been
2937 implemented, as required -- yes, really -- by ANSI;
2938 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2939 on malformed property lists;
2941 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2942 * incompatible packaging change: in line with Unix convention,
2943 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2944 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2945 * REQUIRE and PROVIDE are now optionally capable of doing something
2946 useful. See the documentation string for REQUIRE.
2947 * infrastructure for a managed SBCL contrib system: contributed
2948 modules in this release include:
2949 ** the ASDF system definition facility;
2950 ** an interface to the BSD Sockets API;
2951 ** an ACL-like convenience interface to the repl;
2952 (thanks to Kevin Rosenberg)
2953 ** an implementation of ROTATE-BYTE, with efficient implementation
2955 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2956 gives a closer approximation to the right answer than previously.
2957 (thanks to Raymond Toy)
2958 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2959 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2960 optional environment arguments, as required by ANSI.
2961 * fixed bugs in other functions taking environment objects, allowing
2962 calls with an explicit NIL environment argument to be compiled
2964 * fixed bug 228: primary return values from
2965 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2966 COMPILE or FUNCTION.
2967 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2968 :TYPE will no longer signal errors on innocuous objects.
2969 * fixed bug 231b: SETQ is better at respecting type declarations in
2970 the lexical environment.
2971 * fixed a bug in DEFCLASS: classes named by symbols with no or
2972 unprintable packages can now be defined.
2973 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2974 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2975 * fixed compiler failure related to checking types of functions.
2976 (reported by Robert E. Brown)
2977 * the compiler is now much more consistent in its error-checking
2978 treatment of bounding index arguments to sequence functions: in
2979 (SAFETY 3) code, errors will be signalled in almost all cases if
2980 invalid sequence bounding indices are passed to functions defined
2981 by ANSI to operate on sequences.
2982 * fixed a bug in the build procedure: documentation of SBCL-specific
2983 packages is now preserved and available in the final Lisp image.
2984 * lifted FDEFINITION lookup out of loops in the implementation of
2985 many list operations. (thanks to Robert E. Brown)
2986 * fixed a bug in the reader: the #n# reader macro now works for
2987 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2988 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2989 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2990 E. Brown and Tony Martinez respectively)
2991 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2992 SLOT-BOUNDP: in particular, it is now able to optimize them much
2993 better, and is now not vulnerable to having packages renamed.
2994 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2995 always called when a slot is not present in an instance. (thanks
2997 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2998 longer signal an error on unreading a character following EOF on
2999 the previous constituent stream. (thanks to Tony Martinez)
3000 * fixed some bugs revealed by Paul Dietz' test suite:
3001 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3002 not just nonnegative fixnums;
3003 ** the logical bit-array operators such as BIT-AND now accept an
3004 explicit NIL for their "opt-arg" argument (to indicate a
3005 freshly-consed result bit-array);
3006 ** ELT now signals an error on an invalid sequence index in safe
3008 ** the type system is now cleverer about negations of numeric
3009 types, and consequently understands the BIGNUM and RATIO types
3011 ** the type system is now cleverer about the interaction between
3012 INTEGER and RATIO types: while bugs still remain, many more
3013 cases are accurately computed;
3014 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3015 if it is in the last clause;
3016 ** CONSTANTLY now correctly returns a side-effect-free function in
3018 ** DECLARE is no longer treated as a special-operator; in
3019 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3020 * incremented fasl file version number due to the change in the
3021 DEFSTRUCT-SLOT-DESCRIPTION structure.
3023 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3024 * minor incompatible change: code processed by the "interpreter" or
3025 EVAL now has a compilation optimization policy of (DEBUG 2)
3026 (changed from (DEBUG 1)) to improve debuggability of interactive
3027 development, and to allow the use of the debug RETURN command in
3029 * an experimental implementation of the RETURN command for the
3030 debugger has been included. (thanks to Frederik Kuivinen)
3031 * fixed bug 62: constraints were not propagated into a loop.
3032 * fixed bug in embedded calls of SORT (reported and investigated by
3034 * fixed some bugs revealed by Paul Dietz' test suite:
3035 ** printing and reading of arrays with some dimensions having
3036 length 0 (thanks to Gerd Moellmann);
3037 ** BOA constructor with &AUX argument without a default value does
3038 not cause a type error;
3039 ** CONSTANTP now returns true for all self-evaluating objects.
3041 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3042 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3043 accessors that are related by inheritance, as specified in the
3044 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3045 Valtteri Vuorikoski)
3046 * The compiler is now able to inline functions that were defined in
3047 a complex lexical environment (e.g. inside a MACROLET).
3048 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3050 * Support for the upcoming FreeBSD-5.0 release has been included.
3051 (thanks to Dag-Erling Smorgrav)
3052 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3053 effect when it is not in a toplevel context.
3054 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3055 stand a better chance of being correct. (thanks to Gerd
3057 * fixed bug in COERCE, which now signals an error on coercing a
3058 rational to a bounded real type which excludes the expected
3060 * The compiler is now able to derive types more accurately from the
3061 COERCE and COMPILE functions.
3062 * fixed bug 223: functional binding is considered to be constant
3063 only for symbols in the CL package.
3064 * fixed bug 231: SETQ did not check the type of a variable being set
3065 (reported by Robert E. Brown)
3066 * A new optimization for MAKE-INSTANCE has been included, fixing
3067 various bugs (including relating to :ALLOCATION :CLASS slots and
3068 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3070 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3071 ** As required by ANSI, LOOP now disallows anonymous collection
3072 clauses such as COLLECT I in conjunction with aggregate boolean
3073 clauses such as THEREIS (= I 1);
3074 ** LOOP now signals an error when any variable is reused in the
3075 same loop (including the potentially useful construct analogous
3076 to WITH A = 1 WITH A = (1+ A);
3077 ** IT is only a special loop symbol within the first clause of a
3078 conditional loop clause;
3079 ** LOOP with a typed iteration variable over a hashtable now
3080 signals a type error iff it should.
3081 * fixed some other bugs revealed by Paul Dietz' test suite:
3082 ** FILE-STREAM now names the class previously known as FD-STREAM;
3083 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3084 argument) no longer signals an error;
3085 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3086 of the slot symbol, rather than using the current package
3087 ((:CONC-NAME "") continues to intern the slot's name in the
3089 * incremented fasl file version number, because of the incompatible
3090 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3091 of the new implementation of DEFINE-COMPILER-MACRO.
3093 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3094 * Support for building SBCL for MIPS platforms running in
3095 little-endian mode has now been checked in, and basic
3096 functionality on said platforms verified.
3097 * minor incompatible change: PCL now records the pathname of a file
3098 in which methods and the like are defined, rather than its
3100 * minor incompatible change: TRUENAME now considers the truename of
3101 a file naming a directory to be the pathname with :DIRECTORY
3102 component indicating that directory.
3103 * minor incompatible change: a NAMED clause in the extended form of
3104 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3105 reason for the previous behaviour is unclear.
3106 * more systematization and improvement of CLOS and MOP conformance
3107 in PCL (thanks to Gerd Moellman and Pierre Mai):
3108 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3110 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3111 primary methods with no specializers;
3112 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3114 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3115 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3116 FORWARD-REFERENCED-CLASSes; error reporting on
3117 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3119 ** SXHASH on CLOS instances now uses a slot internal to the
3120 instance to return different numbers on distinct instances,
3121 while preserving the same return value through invocations of
3123 ** DEFMETHOD signals errors when methods with longer incongruent
3124 lambda lists are added to generic functions;
3125 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3126 CLASS, as specified in AMOP;
3127 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3128 on the order returned by the primary method for classes of
3129 class STANDARD-CLASS;
3130 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3131 * fixed some bugs shown by Paul Dietz' test suite:
3132 ** DOLIST puts its body in TAGBODY;
3133 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3135 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3136 value producing form;
3137 ** if more variables are given to PROGV than values, extra
3138 variables are bound and made to have no value;
3139 ** NSUBSTITUTE on list arguments gets the right answer with
3141 ** ELT signals an error of type TYPE-ERROR when the index argument
3142 is not a valid sequence index;
3143 ** LOOP signals (at macroexpansion time) an error of type
3144 PROGRAM-ERROR when duplicate variable names are found;
3145 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3146 ** FUNCALL of special-operators now cause an error of type
3148 ** PSETQ now works as required in the presence of side-effecting
3149 symbol-macro places;
3150 ** NCONC accepts any object as its last argument;
3151 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3153 ** loop-for-as-package does not require a package to be explicitely
3155 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3157 * fixed bug 166: compiler preserves "there is a way to go"
3158 invariant when deleting code.
3159 * fixed bug 172: macro lambda lists with required arguments after
3160 &REST arguments now cause an error to be signalled. (thanks to
3162 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3163 bug. (thanks to Matthew Danish)
3164 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3166 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3167 arguments are lexically rebound. (thanks to Gerd Moellmann and
3169 * fixed bug 194: error messages are now more informative when there
3170 is no primary method applicable in a call to a generic
3171 function. (thanks to Gerd Moellmann)
3172 * fixed bug in command line argument checking (thanks to Julian
3174 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3175 (thanks to Matthew Danish)
3176 * incremented fasl file version number, because of the
3177 SXHASH-related changes in the layout of CLOS data structures
3179 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3180 * minor incompatible change: The runtime (the Unix executable named
3181 "sbcl") is now much pickier about the .core files it will load.
3182 Essentially it now requires .core files to descend from the same
3183 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3184 as the runtime does. (The intent is to prevent the crashes which
3185 can occur, and which can even be reported as mysterious failures,
3186 when people patch the sources or change the build parameters
3187 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3188 sbcl and .core files.)
3189 * fixed bug: VALUES-LIST is no longer optimized away.
3190 * fixed bug 142: The FFI conversion of C string values to Lisp
3191 string values no longer conses excessively. (thanks to Nathan
3192 Froyd porting Raymond Toy's fix to CMU CL)
3193 * began to systematize and improve MOP conformance in PCL (thanks to
3194 Nathan Froyd, Gerd Moellman and Pierre Mai):
3195 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3197 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3198 ** FINALIZE-INHERITANCE is now called on class finalization;
3199 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3200 argument precedence order.
3201 * fixed bug 202: The compiler no longer fails on functions whose
3202 derived types contradict their declared type.
3203 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3204 so it can be non-toplevel.
3205 * The fasl file version number has changed (because of the new
3206 implementation of DEFMACRO).
3207 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3208 safe code, that any length requirement by their type-specifier
3209 argument is valid. The exceptions to this are described in bug
3211 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3212 functions now check the types of their inputs as required by ANSI.
3213 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3214 introduced symbol is DECLAREd to be SPECIAL.
3215 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3216 * fixed bug: :COUNT argument to sequence functions may be negative
3217 * fixed bug: body of DO-SYMBOLS may contain declarations
3218 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3219 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3220 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3221 symbol macro only once
3222 * fixed printing of call frame when argument list is unavailable
3223 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3224 * compiler no longer signals WARNING on unknown keyword
3227 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3228 * A beta-quality port to the mips architecture running Linux,
3229 based on the old CMUCL backend, has been made. It has been tested
3230 on a big-endian kernel, and works sufficiently well to be able to
3231 rebuild itself; it has not been tested in little-endian mode.
3232 * fixed an inconsistency between gencgc.c and purify.c which made
3233 dumping/loading .core files unreliable
3234 * fixed bug 120a: The compiler now deals correctly with IFs where
3235 the consequent is the same as the alternative, instead of
3236 misderiving the return type. (thanks to Alexey Dejneka)
3237 * fixed bug 113: Logical pathnames are now dumpable (the logical
3238 host is resolved at load-time, throwing an error if it is not
3240 * fixed bug 174: FORMAT's error message is slightly clearer when a
3241 non-printing character is used in a format directive.
3242 * fixed several bugs in compiler checking of type declarations, i.e.
3243 violations of the Python "declarations are assertions" principle
3244 (thanks to Alexey Dejneka)
3245 * fixed several bugs in PCL's error checking (thanks to Gerd
3247 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3249 * fixed bug in compilation of functions as first class values
3250 (thanks to Antonio Martinez-Shotton)
3251 * The compiler's handling TYPE-ERRORs which it can prove will
3252 inevitably happen at runtime has been cleaned up and corrected
3253 in several ways. (thanks to Alexey Dejneka)
3254 * improved argument type checking for various basic arithmetic
3255 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3256 lavished on them in the past that they can be compiled in many
3257 ways in different special cases
3258 * fixed bug 181: compiler checks validity of user supplied type
3260 * cleaned up code flushing in optimization: Function calls which
3261 should signal errors for safety purposes (e.g. which ANSI says
3262 should signal errors when their arguments are of incorrect type)
3263 are no longer optimized away.
3264 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3265 * incremented fasl file version number, because changes in the
3266 implementation of sequence functions like COERCE caused
3267 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3268 in old inline expansions) to become undefined. (Actually these
3269 changes were later undone, so we might very well be binary
3270 compatible with 0.7.7 after all, but leaving the version number
3271 incremented seemed like the simplest and most conservative
3274 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3275 * An alpha-quality port to the parisc architecture running Linux,
3276 based on the old CMUCL backend, has been made. This, even more so
3277 than the other backends, should be considered still a work in
3278 progress; known problems include that the Linux kernel in 64-bit
3279 mode does not propagate the correct sigcontext structure to
3280 userspace, and consequently SBCL on a parisc64 kernel will not
3282 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3283 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3284 suppressed.) Also now that the compiler is looking at declarations
3285 in the environment, it checks optimization declarations as well,
3286 and suppresses inlining when (> DEBUG SPEED).
3287 * More fixes have been made to treatment of floating point exception
3288 treatment and other Unix signals. In particular, floating point
3289 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3290 * The detection and handling of control stack exhaustion (infinite
3291 or very deeply nested recursion) has changed. Stack exhaustion
3292 detection is now done by write-protecting pages at the OS level
3293 and applies at all optimization settings; when found, a
3294 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3295 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3296 be caught using IGNORE-ERRORS.
3297 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3298 correctly when transforming recursive LABELS functions to LETs.
3299 (thanks to Alexey Dejneka)
3300 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3301 that are names of constants or global variables.
3302 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3303 alien routines with docstrings.
3304 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3305 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3307 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3308 (thanks to Eric Marsden)
3309 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3310 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3311 * Bug fix: The SPARC backend can now compile functions involving
3312 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3313 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3314 to a functions expecting a PATHNAME-DESIGNATOR.
3315 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3316 lambda lists. (thanks to Alexey Dejneka)
3317 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3318 functions so that EXPORTing the name of the class doesn't cause
3319 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3320 to Antonio Martinez for reporting this)
3321 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3322 constant-folded without failing an assertion. (thanks to Einar
3323 Floystad Dorum for reporting this)
3324 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3325 types are now dealt with more correctly by the compiler.
3326 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3327 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3328 behaviour analogous to RENAME-FILE. This puts its behaviour more
3329 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3331 * The fasl file version number has changed again. (because of the
3332 bug fix involving the names of PCL MAKE-INSTANCE functions)
3334 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3335 * bug fix: Floating point exceptions are treated much more
3336 consistently on the x86/Linux and PPC/Linux platforms.
3337 * Array initialization with :INITIAL-ELEMENT is now much faster for
3338 cases when the compiler cannot open code the array creation, but
3339 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3340 array accesses have also seen a speed increase.
3341 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3342 specified. (This was at the root of some bad interactions between
3343 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3345 * bug fix: Internal error arguments for undefined functions are now
3346 computed correctly on the PPC/Linux platform.
3347 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3348 Raymond Toy's patch for CMU CL)
3349 * Support for the Solaris 9 operating environment has been included
3350 (thanks to Daniel Merritt)
3351 * A very ugly but hopefully complete draft of the missing FFI chapter
3352 of the manual has been created by reformatting the corresponding
3353 CMU CL manual chapter into (currently very ugly and incoherent)
3354 DocBook and bringing it up to date for SBCL behavior. Thus, the
3355 manual is now essentially complete, at least by my extreme
3356 once-and-only-once standards, whereby it's acceptable to refer to
3357 the doc strings of SB-EXT functions as the primary documentation.
3358 * The fasl file version number has changed again, due to cleanup of
3359 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3360 is no longer a static symbol.)
3362 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3363 * SBCL now builds with OpenMCL (version 0.12) as the
3364 cross-compilation host; also, more progress has been made toward
3365 bootstrapping under CLISP.
3366 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3368 * bug 158 fixed: The compiler can now deal with integer loop
3369 increments different from 1; fixing this turned out also to fix
3371 * bug 169 fixed: no more bogus warnings about using lexical bindings
3372 despite the presence of perfectly good SPECIAL declarations (thanks
3373 to David Lichteblau)
3374 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3375 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3376 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3378 * bug 180 fixed: Method combination specifications no longer ignore
3379 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3380 * bug fix: Structure type predicate functions now check their argument
3381 count as they should.
3382 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3383 correctly. (thanks to Pierre Mai)
3384 * minor incompatible change: The --noprogrammer option is deprecated
3385 in favor of the new --disable-debugger option, which is very similar.
3386 (The major difference is that it takes effect at a slightly different
3387 time at startup, causing handling of errors in --sysinit and
3388 --userinit files will be affected differently.) The
3389 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3390 been added to allow this functionality to be controlled from ordinary
3391 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3392 maintainers, who might want to run non-interactive scripts to
3393 build SBCL cores which will later be used interactively.)
3394 * minor incompatible change: The LOAD function no longer, when given
3395 a wild pathname to load, loads all files matching that pathname.
3396 Instead, an error of type FILE-ERROR is signalled.
3398 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3399 * bug 147 fixed: The compiler preserves its block link/count
3400 invariants more correctly now so that it doesn't crash. (thanks
3402 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3404 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3405 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3406 no-value-for-this-character as NIL (as specified by ANSI).
3407 * HOST-NAMESTRING on physical pathnames now returns a string that is
3408 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3409 (thanks to Christophe Rhodes)
3410 * The Alpha port handles icache flushing more correctly. (thanks to
3412 * More progress has been made toward bootstrapping under CLISP. (thanks
3413 to Christophe Rhodes)
3414 * The fasl file format has changed again, because dynamic loading
3415 on OpenBSD (which has non-ELF object files) motivated some cleanups
3416 in the way that foreign symbols are transformed and passed around.
3417 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3418 is no longer treated as whitespace by the reader, but instead as
3419 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3420 |A
\7fB|, instead of A as it used to.
3422 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3423 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3424 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3425 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3426 this as of 0.7.1.45, but was left out of the previous news section
3427 (thanks to Dan Barlow)
3428 * SBCL now runs on the Solaris operating system on SPARC architectures
3429 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3430 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3431 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3432 * SPARC backend cleanups, allowing builds of cores optimized for V8
3433 and V9 SPARCS, and also emission of code targeted to a particular
3434 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3436 * SBCL is closer to bootstrapping under CLISP, thanks to various
3437 fixes by Christophe Rhodes.
3438 * The fasl file format has changed again, to allow the compiler's
3439 INFO database to support symbol macros.
3440 * The user manual (in doc/) is formatted into HTML more nicely.
3441 (thanks to coreythomas)
3442 * The system is smarter about SUBTYPEP relationships, especially
3443 those involving NOT types (including types such as ATOM which are
3444 represented internally using NOT types). Thus SUBTYPEP is less
3445 likely to return (VALUES NIL NIL) in general, and in particular
3446 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3448 * The fasl file format has changed again, because the internal
3449 representation of types now includes a new slot to support the new
3450 SUBTYPEP-of-NOT-types logic.
3451 * (not a change in the main branch of SBCL, but a related prototype
3452 which can hopefully be merged into the main branch of SBCL in the
3453 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3454 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3455 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3456 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3457 dispatch macro character combinations. (thanks to Alexey Dejneka)
3458 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3459 that superficially look like logical namestrings correctly.
3460 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3462 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3463 default supplied-p) for &optional and &key arguments. (thanks to
3466 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3467 * incompatible change: The compiler is now less aggressive about
3468 tail call optimization, doing it only when (> SPACE DEBUG) or
3469 (> SPEED DEBUG). (This is an incompatible change because there are
3470 programs which relied on the old CMU-CL-style behavior to optimize
3471 away their unbounded recursion which will now die of stack overflow.)
3472 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3473 for non-GENCGC systems has been increased to 20M (since that
3474 seems much closer to the likely performance optimum for modern
3475 systems than the old 4M value was)
3476 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3477 and *DEBUG-PRINT-LEVEL*
3478 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3479 of CMU CL's support for SPARC, and various endianness and other
3480 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3481 * new syntactic sugar for the Unix command line: --load foo.bar is now
3482 an alternate notation for --eval '(load "foo.bar")'.
3484 ** The system now detects stack overflow and handles it gracefully,
3485 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3486 optimization settings. (This is a good thing in general, and
3487 its introduction in this version should be particularly timely
3488 for anyone whose code fails because of suppression of tail
3490 ** The system now hunts for the C variable "environ" in a more
3491 devious way, to avoid segfaults when the C library version differs
3492 between compile time and run time. (thanks to Christophe Rhodes)
3493 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3494 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3495 ** The compiler no longer issues bogus style warnings for undefined
3496 classes in the same source file as the DEFCLASSes which defined
3497 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3499 ** fixes in CONDITION class precedence list for undefined function
3500 errors (thanks to Alexei Dejneka)
3501 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3502 correctly. (thanks to Dan Barlow)
3503 ** portability fixes aiming at bootstrapping under CLISP (thanks
3504 to Dave McDonald and Christophe Rhodes)
3505 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3506 ** fixes in type translation and and type inference (thanks to
3508 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3509 ** various fixes in the new ports (thanks to Dan Barlow)
3510 * several changes related to debugging:
3511 ** suppression of tail recursion, as noted above
3512 ** stack overflow detection, as noted above
3513 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3514 is now the default. (For some time encapsulation has been more
3515 reliable than the breakpoint-based :ENCAPSULATE NIL
3516 implementation, at least on X86 systems; and I just noticed that
3517 encapsulation also seems closer to the spirit of the ANSI
3520 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3522 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3523 up properly again. (There was a packaging bug in 0.7.0 which
3524 left their definitions in SB-SYS::LOAD-FOREIGN and
3525 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3526 vital for most things which interface to C-level interfaces,
3527 like extensions working with sockets or databases or
3528 Perl-compatible regexes or whatever, and the need to fix
3529 this bug is the main reason that 0.7.1 was released so
3531 ** DEFGENERIC is now choosier about the methods it redefines, so that
3532 reLOADing a previously-LOADed file containing DEFGENERICs does
3533 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3534 cycle now works as it should. (thanks to Alexey Dejneka)
3535 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3536 fixed by Christophe Rhodes. (He actually submitted this patch
3537 months ago, and I delayed until after 0.7.0.)
3538 ** Bug 111 (internal compiler confusion about runtime checks on
3539 FUNCTION types) was fixed by Alexey Dejneka.
3540 * Some internal cleanups (getting rid of variables which aren't
3541 needed now that the byte interpreter is gone) caused the fasl
3542 file format number to change again.
3544 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3545 * major incompatible change: The default fasl file extension, i.e. the
3546 default extension for files produced by COMPILE-FILE, has changed
3547 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3549 ** There are many changes in the implementation of the compiler.
3550 SBCL is now essentially a compiler-only implementation of ANSI
3551 Common Lisp. EVAL still "interprets" a few special cases, but
3552 almost all the interesting cases are handled by creating
3553 a LAMBDA expression, calling COMPILE on it, then calling
3554 FUNCALL on the result.
3555 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3556 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3557 Since the code is newer, there might still be some new bugs
3558 (though not as many as before Martin Atzmueller's fixes:-). But
3559 the new code is substantially simpler and clearer, and hopefully
3560 any remaining bugs will be simpler, less fundamental, and more
3561 fixable then the bugs in the old code.
3562 ** The revised compiler is still a little unsteady on its feet.
3564 *** The debugging information it produces (particularly the names
3565 of FUNCTION objects) is sometimes much less useful than what
3566 the old compiler produced.
3567 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3568 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3569 within a MACROLET) has been temporarily weakened.
3570 ** There are new compiler optimizations for various functions:
3571 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3572 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3573 *** the math functions TRUNCATE, FLOOR, and CEILING
3574 *** the function-of-all-trades COERCE
3575 Mostly these should be transparent, but there's one
3576 potentially-annoying problem (bug 117): when the compiler
3577 inline-expands a function and does type analysis on the result,
3578 it can create control paths which have type mismatches, and
3579 when it can't prove that those control paths aren't taken,
3580 it will issue WARNINGs about the type mismatches. This is
3581 a particular problem in practice for the new sequence functions.
3582 It's not clear how this should be fixed, and for now, a
3583 workaround is given in the entry for 117 in the BUGS file.
3584 ** (Because of the interaction between the two previous items --
3585 occasional inlining problems and new inline expansions -- some
3586 of the new sequence function optimizations won't really kick in
3587 completely until debugging information, and then inlining, are
3588 straightened out in some future version.)
3589 * minor incompatible changes:
3590 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3591 DIRECTORY no longer implicitly promotes NIL slots of its
3592 pathname argument to :WILD. In particular, when you ask for the
3593 contents of a directory (which you used to be able to do without
3594 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3595 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3596 ** changes in behavior that ANSI explicitly defines to be
3597 implementation dependent:
3598 *** The new compiler-only implementation still conforms with ANSI,
3599 but acts a little different than before. Besides the obvious
3600 changes in performance tradeoffs (that the cost per form passed
3601 to EVAL has gone up, and the cost per form executed by EVAL
3602 has gone down), the behavior of the system changes a little
3603 because there are no longer any interpreted function objects.
3604 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3605 e.g. doing COMPILE on the output of interactive DEFUN is
3607 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3609 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3610 has changed from (:CL) to NIL.
3611 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3612 e.g. #\Newline and #\Tab, don't have names specified in the
3613 ANSI Common Lisp standard, is now based on their ASCII symbolic
3614 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3615 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3616 are no longer used for output.
3617 ** changes in internal implementation constants:
3618 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3619 4 million. (If your application spends a lot of time GCing and
3620 you have a lot of RAM, you might want to experiment with
3621 increasing it even more.)
3622 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3623 However, almost all old code should still continue to work without
3624 immediate update, as SB-C-CALL is now a (deprecated) nickname
3626 ** Old operator names in the style DEF-FOO are now deprecated in
3627 favor of new corresponding names DEFINE-FOO, for consistency with
3628 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3629 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3630 internal symbols, but a few supported extensions like
3631 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3632 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3633 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3634 etc. as you get deeper into recursive calls to the debugger
3635 command loop, instead of the old "5]", "5]]", "5]]]"
3636 sequence. (I was motivated to do this when squabbles between
3637 ILISP and SBCL left me very deeply nested in the debugger. In the
3638 short term, this change will probably provoke more ILISP/SBCL
3639 squabbles, but hopefully it will be an improvement in the long run.)
3640 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3641 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3642 function, since declaiming return types involving aliens is
3643 (1) annoyingly messy to do by hand and (2) vital to efficient
3644 compilation of code which calls such functions.
3645 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3646 longer reexported by the SB-EXT package. They're solely useful
3647 for alien code, so it seems more logical that you should get
3648 them from the SB-ALIEN package, not in SB-EXT.
3649 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3650 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3651 features. Instead, the code that they used to control is always
3652 built into the system.
3653 * many other bug fixes
3654 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3655 advantage of the new EVAL-WHEN stuff and to clean them up in
3656 general, and they are now more ANSI-compliant in a number of
3657 ways. Martin Atzmueller is responsible for a lot of this.
3658 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3660 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3661 *** correct ERROR type for various file operations
3662 *** some fixes for Lisp streams
3663 *** DEFMETHOD syntax checking
3664 *** changing old weird representation of debug information as
3665 strings (which, among their other deficiencies, don't transform
3666 correctly when you rename packages, and don't change their
3667 print representation when you change things like *PACKAGE*
3668 and *PRINT-LENGTH*) to symbols and lists of symbols
3669 He also made several improvements and fixed several bugs in DESCRIBE.
3670 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3672 *** misbehavior of WRITE-STRING/WRITE-LINE
3673 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3674 and several other LOOP problems as well
3675 *** DIRECTORY when similar filenames are present
3676 *** DEFGENERIC with :METHOD options
3677 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3678 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3679 *** argument ordering in FIND with :TEST option
3680 *** mishandled package designator argument in APROPOS-LIST
3681 *** various problems in the backquote readmacro
3682 *** a bug in APROPOS
3683 *** probably some others that I'm not describing very well here,
3684 since the CVS log documents them by reference to sbcl-devel
3685 messages, and the SourceForge archives aren't working well.:-(
3686 ** Dan Barlow improved the Alpha port (and is making progress on the
3687 PPC port, for those of you who think different).
3688 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3689 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3690 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3691 continues to work on the SPARC port (for those of you in a position
3692 to look down upon our little PC-compatible boxes from a great height).
3693 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3694 and uses that copy, rather than the current dynamic value, when
3695 it's trying to decide whether to truncate output. Thus e.g.
3696 (let ((*print-lines* 50))
3697 (pprint-logical-block (stream nil)
3699 (let ((*print-lines* 8))
3700 (print (aref possiblybigthings i) stream)))))
3701 should now truncate the logical block only at 50 lines, instead of
3702 often truncating it at 8 lines, as it did before.
3703 * The doc/cmucl/ directory, containing old CMU CL documentation
3704 from the time of the fork, is no longer part of the base system.
3705 SourceForge has shut down its anonymous FTP service, and with it
3706 my original plan for distributing the old CMU CL documentation
3707 there. For now, if you need these files you can download an old
3708 SBCL source release and extract them from it.
3709 * The fasl file version number changed again, for dozens of reasons,
3710 some of which are apparent above.
3712 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3713 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3714 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3715 patch, so that DISASSEMBLE output is much nicer.
3716 * The code in the SB-PROFILE package now seems reasonably stable.
3717 I still haven't decided what the final interface should look like
3718 (I'd like PROFILE to interact cleanly with TRACE, since both
3719 facilities use function encapsulation) but if you have a need
3720 for profiling now, you can probably use it successfully with
3721 the current CMU-CL-style interface.
3722 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3723 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3724 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3725 link now returns the dangling link itself, and for similar
3726 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3727 link itself. (In these cases the old code signalled an error and
3728 looped endlessly, respectively.) Thus, DIRECTORY now works even
3729 in the presence of dangling and cyclic symbolic links.
3730 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3731 is now a supported extension again, since the consensus on
3732 sbcl-devel was that it can be useful for ordinary development
3733 work, not just for debugging SBCL itself.
3734 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3735 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3736 that currently-defined functions might be redefined later with
3737 different return types.
3738 * Hash tables can be printed readably, as inspired by CMU CL code
3739 of Eric Marsden and SBCL code of Martin Atzmueller.
3740 * better error handling in CLOS method combination, thanks to
3741 Martin Atzmueller porting Pierre Mai's CMU CL patches
3742 * more overflow fixes for >16Mbyte I/O buffers
3743 * A bug in READ has been fixed, so that now a single Ctrl-D
3744 character suffices to cause end-of-file on character streams.
3745 In particular, now you only need one Ctrl-D at the command
3746 line (not two) to exit SBCL.
3747 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3749 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3750 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3751 does the right thing.
3752 * The implementation of some type tests, especially for CONDITION
3753 types, is now tidier and maybe faster, due to CMU CL code
3754 originally by Douglas Crosher, ported by Martin Atzmueller.
3755 * Some math functions have been fixed, and there are new
3756 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3757 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3758 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3759 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3760 likely to be built into sbcl-0.7.0.)
3761 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3762 CL, ported to SBCL by Martin Atzmueller)
3763 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3764 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3765 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3766 * There's a new slam.sh hack to shorten the edit/compile/debug
3767 cycle for low-level changes to SBCL itself, and a new
3768 :SB-AFTER-XC-CORE target feature to control the generation of
3769 the after-xc.core file needed by slam.sh.
3770 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3771 COMPILE-FILE is no longer supported, so that now every function
3772 gets an entry point, so that block compilation looks a little
3773 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3774 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3775 returns the number of bytes consed since the system started,
3776 rather than the number consed since the first time the function
3777 was called. (The new definition parallels ANSI functions like
3778 CL:GET-INTERNAL-RUN-TIME.)
3779 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3780 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3781 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3782 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3783 * Fasl file version numbers are now independent of the target CPU,
3784 since historically most system changes which required version
3785 number changes have affected all CPUs equally. Similarly,
3786 the byte fasl file version is now equal to the ordinary
3789 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3790 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3791 is no longer recognized. I apologize for this, because it was
3792 listed in SB-EXT as a supported extension, but I found that
3793 its existing behavior was poorly specified, as well as incorrectly
3794 specified, and it looked like too much of a mess to straighten it
3795 out. I have enough on my hands trying to get ANSI stuff to work..
3796 * many patches ported from CMU CL by Martin Atzmueller, with
3797 half a dozen bug fixes in pretty-printing and the debugger, and
3798 half a dozen others elsewhere
3799 * fixed bug 13: Floating point infinities are now supported again.
3800 They might still be a little bit flaky, but thanks to bug reports
3801 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3802 as flaky as they were.
3803 * The --noprogrammer command line option is now supported. (Its
3804 behavior is slightly different in detail from what the old man
3805 page claimed it would do, but it's still appropriate under the
3806 same circumstances that the man page talks about.)
3807 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3808 are now supported, and enabled by default. Thus, the compiler can
3809 handle many floating point and complex operations much less
3810 inefficiently. (Thus e.g. you can implement a complex FFT
3812 * The compiler now detects type mismatches between DECLAIM FTYPE
3813 and DEFUN better, and implements CHECK-TYPE more correctly, and
3814 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3815 thanks to patches from Martin Atzmueller.
3816 * various fixes to make the cross-compiler more portable to
3817 ANSI-conforming-but-different cross-compilation hosts (notably
3818 Lispworks for Windows, following bug reports from Arthur Lemmens)
3819 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3820 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3821 fixed thanks to Pierre Mai's CMU CL patches.
3822 * fixes to keep the system from overflowing internal counters when
3823 it tries to use i/o buffers larger than 16M bytes
3824 * fixed bug 45a: Various internal functions required to support
3825 complex special functions have been merged from CMU CL sources.
3826 (When I was first setting up SBCL, I misunderstood a compile-time
3827 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3828 * improved support for type intersection and union, fixing bug 12
3829 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3830 more obscure bugs as well
3831 * some steps toward byte-compiling non-performance-critical
3832 parts of the system, courtesy of patches from Martin Atzmueller
3833 * Christophe Rhodes has made some debian packages of sbcl at
3834 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3835 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3836 stable, but are nonetheless usable. When he's ready, I'd be happy
3837 to add them to the SourceForge "File Releases" section. (And if
3838 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3839 * new fasl file format version number (because of changes in
3840 internal representation of (OR ..) types to accommodate the new
3841 support for (AND ..) types, among other things)
3843 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3844 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3846 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3848 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3849 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3850 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3851 * DESCRIBE now works on structure objects again.
3852 * Most function call argument type mismatches are now handled as
3853 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3854 know whether the function will be redefined before the call is
3855 executed. (The compiler could flag local calls with full WARNINGs,
3856 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3857 it doesn't keep track of enough information to know whether calls
3858 are local in this sense.)
3859 * Compiler output is now more verbose, with messages truncated
3860 later than before. (There should be some supported way for users
3861 to override the default verbosity, but I haven't decided how to
3862 provide it yet, so this behavior is still controlled by the internal
3863 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3864 src/compiler/ir1util.lisp.)
3865 * Fasl file format version numbers have increased again, because
3866 support for the Gray streams extension changes the layout of the
3867 system's STREAM objects.
3868 * The Gray subclassable streams extension now works, thanks to a
3869 patch from Martin Atzmueller.
3870 * The full LOAD-FOREIGN extension (not just the primitive
3871 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3872 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3873 but like most other programs, it defaults to copying the Unix
3874 environment from the original process instead of starting the
3875 new process in an empty environment.
3876 * Extensions which manipulate the Unix environment now support
3877 an :ENVIRONMENT keyword option which doesn't smash case or
3878 do other bad things. The CMU-CL-style :ENV option is retained
3879 for porting convenience.
3880 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3881 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3883 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3885 * A patch from Martin Atzmueller seems to have solved the SIGINT
3886 problem, and as far as we know, signal-handling now works cleanly.
3887 (If you find any new bugs, please report them!)
3888 * The system no longer defaults Lisp source file names to types
3889 ".l", ".cl", or ".lsp", but only to ".lisp".
3890 * The compiler no longer uses special default file extensions for
3891 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3892 seems to expect a single default extension for all compiled code,
3893 and there's no compelling reason to try to stretch the standard
3894 to allow two different extensions.) Instead, byte-compiled files
3895 default to the same extension as native-compiled files.
3896 * Fasl file format version numbers have increased again, because
3897 a rearrangement of internal implementation packages made some
3898 dumped symbols in old fasl files unreadable in new cores.
3899 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3900 many fewer weird special cases.
3901 * Bug #17 (differing COMPILE-FILE behavior between logical and
3902 physical pathnames) has been fixed, and some related misbehavior too,
3903 thanks to a patch from Martin Atzmueller.
3904 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3905 by Tim Moore, ported to SBCL by Martin Atzmueller.
3906 * Martin Atzmueller fixed several filesystem-related problems,
3907 including bug #36, in part by porting CMU CL patches, which were
3908 written in part by Paul Werkowski.
3909 * More compiler warnings in src/runtime/ are gone, thanks to
3910 more patches from Martin Atzmueller.
3911 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3914 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3916 * DESCRIBE now works on CONDITION objects.
3917 * The debugger now handles errors which arise when trying to print
3918 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3920 * The build system now uses an additional file, customize-target-features.lisp,
3921 to allow local modifications to the target *FEATURES* list. (The point of
3922 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3923 debugging features enabled, without having to worry about propagating it
3924 into everyone's system when I do a "cvs update".) When no
3925 customize-target-features.lisp file exists, the target *FEATURES* list
3926 should be constructed the same way as before.
3927 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3928 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3929 warn about behavior which is undefined under the ANSI spec. Note: This
3930 is specified by ANSI, but it's not very popular with programmers.
3931 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3932 macro in the SBCL sources for an example of a workaround which you
3933 might use to make portable ANSI-standard code which does what you want.
3934 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3935 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3936 when it's used as a non-toplevel form. (This is required in order
3937 to implement the DEFCONSTANT-EQX macro.)
3938 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3939 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3940 funny things, due to bugs in the implementation of EVAL-WHEN
3941 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3942 non-toplevel DEFCONSTANTs are uncommon.)
3943 * The core file version number and fasl file version number have been
3944 incremented, because the old noncompliant DEFCONSTANT behavior involved
3945 calling functions which no longer exist, and because I also took the
3946 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3947 * fixed bug 1 (error handling before read-eval-print loop starts), and
3948 redid debugger restarts and related debugger commands somewhat while
3950 ** The QUIT debugger command is gone, since it did something
3951 rather different than the SB-EXT:QUIT command, and since it never
3952 worked properly outside the main toplevel read/eval/print loop.
3953 Invoking the new TOPLEVEL restart provides the same functionality.
3954 ** The GO debugger command is also gone, since you can just invoke
3955 the CONTINUE restart directly instead.
3956 ** The TOP debugger command is also gone, since it's redundant with the
3957 FRAME 0 command, and since it interfered with abbreviations for the
3959 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3961 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3962 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3963 some porting work by Martin Atzmueller)
3964 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3965 with a boatload of patches which he ported to SBCL. Now that those
3967 ** The system tries to make sure that its low-priority messages
3968 are prefixed by semicolons, to help people who like to use
3969 syntax highlighting in their ILISP buffer. (This patch
3970 was originally due to Raymond Toy.)
3971 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3972 patches originally written by Raymond Toy.
3973 ** The compiler understands coercion between single-value and
3974 multiple-VALUES type expressions better, getting rid of some very
3975 weird behavior, thanks to patches originally by Robert MacLachlan
3976 and Douglas Crosher.
3977 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3978 lambda lists, thanks to a patch originally by Pierre Mai.
3979 ** The system no longer bogusly warns about "abbreviated type
3981 ** The compiler gets less confused by inlining and RETURN-FROM,
3982 thanks to some patches originally by Tim Moore.
3983 ** The system no longer hangs when dumping circular lists to fasl
3984 files, thanks to a patch originally from Douglas Crosher.
3985 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3986 undefined function error.
3987 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3988 in structure slot initforms) for now, documented workaround instead:-|
3989 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3990 * fixed bug 5: added stubs for various Gray stream functions called
3991 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3992 installed, at least appropriate type errors are generated
3993 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3994 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3995 consistently in DEFMETHOD forms.
3996 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3997 it doesn't seem to affect SBCL after all
3998 * The C runtime system now builds with better optimization and many
3999 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4001 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4003 * The system is now under CVS at SourceForge (instead of the
4004 CVS repository on my home machine).
4005 * The new signal handling code has been tweaked to treat register
4006 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4007 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4008 downstream, e.g. in debug-int.lisp, has implicit dependencies
4009 on the unsignedness of integer representation of machine words,
4010 and that caused the system to bomb out with infinite regress
4011 when trying to recover from type errors involving signed values,
4012 e.g. (BUTLAST '(1 2 3) -1).)
4013 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4014 (This was one of the bugs Peter Van Eynde reported back in July.)
4015 * The system now uses code inspired by Colin Walters' O(N)
4016 implementation of MAP (from the cmucl-imp@cons.org mailing
4017 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4018 inline the MAP operation, and there is more than one
4019 sequence argument to the MAP call (so that it can't just
4020 do ETYPECASE once and for all based on the type of the
4021 single sequence argument). (The old non-inline implementation
4022 of the general M-argument sequence-of-length-N case required
4023 O(M*N*N) time when any of the sequence arguments were LISTs.)
4024 * The QUIT :UNIX-CODE keyword argument has been renamed to
4025 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4026 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4028 * Raymond Wiker's patches to port dynamic loading from Linux to
4029 FreeBSD have been added.
4030 * The BUGS file is now more nearly up to date, thanks in large part
4031 to Martin Atzmueller's review of it.
4032 * The debugger now flushes standard output streams before it begins
4033 its output ("debugger invoked" and so forth).
4034 * The core version number and fasl file version number have both
4035 been incremented, because of incompatible changes in the layout
4037 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4038 used to not be called for a saved Lisp image.)
4039 * Martin Atzmueller's version of a patch to fix a compiler crash,
4040 as posted on sbcl-devel 13 September 2000, has been installed.
4041 * Instead of installing Martin Atzmueller's patch for the
4042 compiler transform for SUBSEQ, I deleted the compiler transform,
4043 and transforms for some similar consing operations.
4044 * A bug in signal handling which kept TRACE from working on OpenBSD
4046 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4047 away by constant folding
4048 * The system now defines its address space constants in one place
4049 (in the Lisp sources), and propagates them automatically elsewhere
4050 (through GENESIS and the sbcl.h file). Therefore, patching the
4051 address map is less unnecessarily tedious and error-prone. The
4052 Lisp names of address space constants have also been systematized.
4053 * CVS tags like dollar-Header-dollar have been removed from
4054 the sources, because they have never saved me trouble and
4055 they've been source of trouble working with patches and other
4056 diff-related operations.
4057 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4058 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4060 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4062 * The system has been ported to OpenBSD.
4063 * The system now compiles with a simple "sh make.sh" on the systems
4064 that it's supported on. I.e., now you no longer need to tweak
4065 text in the target-features.lisp-expr and symlinks in src/runtime/
4066 by hand, the make.sh takes care of it for you.
4067 * The system is no longer so grossly inefficient when compiling code
4068 involving vectors implemented as general (not simple) vectors (VECTOR T),
4069 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4070 which dares to use the various sequence functions on non-simple
4071 vectors, takes less of a performance hit.
4072 * There is now a primitive type predicate VECTOR-T-P
4073 to test for the (VECTOR T) type, so that e.g.
4074 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4075 can now be compiled with some semblance of efficiency. (The old code
4076 turned the type declaration into a full call to %TYPEP at runtime!)
4077 * AREF on (VECTOR T) is still not fast, since it's still compiled
4078 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4079 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4080 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4081 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4082 without making full calls to SUBTYPEP at runtime.
4083 (Some analogous efficiency issues for non-simple vectors specialized to
4084 element types other than T, or for non-simple multidimensional arrays,
4085 have not been addressed. They could almost certainly be handled the
4086 same way if anyone is motivated to do so.)
4087 * The changes in array handling break binary compatibility, so
4088 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4089 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4090 instead of (VALUES T T).
4091 * By following the instructions that Dan Barlow posted to sbcl-devel
4092 on 2 July 2000, I was able to enable primitive dynamic object
4093 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4094 functionality is not implemented (since it calls ld to resolve
4095 library references automatically, requiring RUN-PROGRAM for its
4096 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4097 try to resolve library references) is now supported.
4098 * The system now flushes the standard output streams when it terminates,
4099 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4100 them at several other probably-convenient times, e.g. in each pass of
4101 the toplevel read-eval-print loop, and after evaluating a form given
4102 as an "--eval" command-line option. (These changes were motivated by a
4103 discussion of stream flushing issues on cmucl-imp in August 2000.)
4104 * The source transform for TYPEP of array types no longer assumes
4105 that an array whose element type is a not-yet-defined type
4106 is implemented as an array of T, but instead punts, so that the
4107 type will be interpreted at runtime.
4108 * There is now some support for cross-compiling in make.sh: each of
4109 the phases of make.sh has its own script. (This should be transparent
4110 to people doing ordinary, non-cross-compile builds.)
4111 * Since my laptop doesn't have hundreds of megabytes of memory like
4112 my desktop machine, I became more motivated to do some items on
4113 my to-do list in order to reduce the size of the system a little:
4114 ** Arrange for various needed-only-at-cold-init things to be
4115 uninterned after cold init. To support this, those things have
4116 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4117 symbols with such names are now uninterned after cold init).
4118 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4119 fasl files for cold load.
4120 ** Remove the old compiler structure pooling code (which used to
4121 be conditional on the target feature :SB-ALLOC) completely.
4122 ** Redo the representation of some data in cold init to be more compact.
4123 (I also looked into supporting byte compiled code at bootstrap time,
4124 which would probably reduce the size of the system a lot, but that
4125 looked too complicated, so I punted for now.)
4126 * The maximum signal nesting depth in the src/runtime/ support code has
4127 been reduced from 4096 to 256. (I don't know any reason for the very
4128 large old value. If the new smaller value turns out to break something,
4129 I'll probably just bump it back up.)
4130 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4132 * Many, many bugs reported by Peter Van Eynde have been added to
4133 the BUGS list; some have even been fixed.
4134 * While enabling dynamic object file loading, I tried to make the
4135 code easier to understand, renaming various functions and variables
4136 with less ambiguous names, and changing some function calling
4137 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4138 * While trying to figure out how to do the OpenBSD port, I tried to
4139 clean up some of the code in src/runtime/. In particular, I dropped
4140 support for non-POSIX signal handling, added various comments,
4141 tweaked the code to reduce the number of compilation warnings, and
4142 renamed some files to increase consistency.
4143 * To support the new automatic configuration functionality in make.sh,
4144 the source file target-features.lisp-expr has been replaced with the
4145 source file base-target-features.lisp-expr and the machine-generated
4146 file local-target-features.lisp-expr.
4147 * fixed a stupid quoting error in make.sh so that using CMU CL
4148 "lisp -batch" as cross-compilation host works again
4150 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4152 * DESCRIBE no longer tries to call itself recursively to describe
4153 bound/fbound values, so that it no longer fails on symbols which are
4154 bound to themselves (like keywords, T, and NIL).
4155 * DESCRIBE now works on generic functions.
4156 * The printer now prints less-screwed-up representations of closures
4157 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4158 * A private symbol is used instead of the :EMPTY keyword previously
4159 used to mark empty slots in hash tables. Thus
4160 (DEFVAR *HT* (MAKE-HASH-TABLE))
4161 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4162 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4163 now does what ANSI says that it should. (You can still get
4164 similar noncompliant behavior if bang on the hash table
4165 implementation with all the symbols you get back from
4166 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4167 This breaks binary compatibility, since tests for equality to
4168 :EMPTY are wired into things like the macroexpansion of
4169 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4171 * There's now a minimal placeholder implementation for CL:STEP,
4172 as required by ANSI.
4173 * An obscure bug in the interaction of the normal compiler, the byte
4174 compiler, inlining, and structure predicates has been patched
4175 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4176 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4178 * Missing ordinary arguments in a macro call are now detected even
4179 when the macro lambda list contains &KEY or &REST.
4180 * The debugger no longer complains about encountering the top of the
4181 stack when you type "FRAME 0" to explicitly instruct it to go to
4182 the top of the stack. And it now prints the frame you request even
4183 if it's the current frame (instead of saying "You are here.").
4184 * As specified by ANSI, the system now always prints keywords
4185 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4187 * The default initial SIZE of HASH-TABLEs is now smaller.
4188 * Type information from CLOS class dispatch is now propagated
4189 into DEFMETHOD bodies, so that e.g.
4190 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4192 is now basically equivalent to
4193 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4194 (DECLARE (TYPE SINGLE-FLOAT X))
4196 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4197 operation, without having to do run-time type dispatch.
4198 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4199 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4200 * Since I don't seem to be making big file reorganizations very often
4201 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4202 is overflowing my ability to conveniently back them up), I've finally
4203 checked the system into CVS. (The CVS repository is on my home system,
4204 not at SourceForge -- putting it on SourceForge might come later.)
4205 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4206 high-level GC-NOTIFY-FOO functions send their output. (There's
4207 still very little control of where low-level verbose GC functions
4208 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4209 less than it used to -- the GC-NOTIFY-FOO functions are now under
4210 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4211 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4212 in only one place in the source code, and propagates it automatically
4213 everywhere that it's needed. Thus e.g. when I bump the version from
4214 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4215 * The C source files now include boilerplate legalese and documentation
4216 at the head of each file (just as the Lisp source files already did).
4217 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4218 to his page will be replaced with a link to his new CLiki service.
4220 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4222 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4223 * The build process now looks for GNU make under the default name "gmake",
4224 instead of "make" as it used to. If GNU make is not available as "gmake"
4225 on your system, you can change this default behavior by setting the
4226 GNUMAKE environment variable.
4227 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4228 can build without error under CMU CL.
4230 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4232 * There is now a partial SBCL user manual (with some new text and some
4233 text cribbed from the CMU CL manual).
4234 * The beginnings of a profiler have been added (starting with the
4235 CMU CL profiler and simplifying and cleaning up). Eventually the
4236 main interface should be through the TRACE macro, but for now,
4237 it's still accessed through vaguely CMU-CL-style functions and macros
4238 exported from the package SB-PROFILE.
4239 * Some problems left over from porting CMU CL to the new
4240 cross-compilation bootstrap process have been cleaned up:
4241 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4242 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4243 part when getting rid of the compiler *BACKEND* stuff.)
4244 ** The value of *NULL-TYPE* was screwed up, because it was
4245 being initialized before the type system knew the final
4246 definition of the 'NULL type. This screwed up several key
4247 optimizations in the compiler, causing inefficiency in all sorts
4248 of places. (I found it because I wanted to understand why
4249 GET-INTERNAL-RUN-TIME was consing.)
4250 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4251 PROCLAIM FTYPE information. Unfortunately this broke binary
4252 compatibility again, since now the forms output by DEFGENERIC
4253 to refer to functions which didn't exist in 0.6.3.
4254 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4255 can use the new (as of 0.6.3) transform for SOME into MAP into
4257 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4258 values to (INTEGER 0 1000000), so that the time code will no longer
4259 occasionally get blown up by Linux returning 1000000 microseconds
4260 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4261 its output conform to the ANSI spec. (Alas, this makes its output
4262 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4263 * A full call to MAP NIL with a single sequence argument no longer conses.
4264 * fixes to problems pointed out by Martin Atzmueller:
4265 * The manual page no longer talks about multiprocessing as though
4266 it were currently supported.
4267 * The ILISP support patches have been removed from the distribution,
4268 because as of version 5.10.1, ILISP now supports SBCL without us
4269 having to maintain patches.
4270 * added a modified version of Raymond Toy's recent CMU CL patch for
4271 EQUALP comparison of HASH-TABLE
4273 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4275 * The system still can't cross-compile itself with
4276 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4277 entails), but at least it can compile more of itself that way
4278 than it used to be able to, and various buglets which were uncovered
4279 by trying to cross-compile itself that way have now been fixed.
4280 * This release breaks binary compatibility again. This time
4281 at least I've incremented the FASL file format version to 2, so that the
4282 problem can be detected reliably instead of just causing weird errors.
4283 * various new style warnings:
4284 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4285 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4286 ** using the lexical binding of a variable named in the *FOO* style
4287 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4288 as specified by ANSI.
4289 * *RANDOM-STATE* is no longer automatically initialized from
4290 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4291 default behavior of the system is to repeat its behavior every time
4292 it's run. If you'd like to change this behavior, you can always
4293 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4294 old convention there was no comparably easy way to get the system to
4295 repeat its behavior every time it was run.
4296 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4297 FTYPE declarations has been removed, in favor of their ANSI
4298 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4299 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4300 the types of their sequence arguments can be determined at compile time.
4301 This is done through a new open code expansion for MAP which eliminates
4302 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4303 argument types can be determined at compile time.
4304 * The optimizer now transforms COERCE into an identity operation when it
4305 can prove that the coerced object is already of the correct type. (This
4306 can be a win for machine generated code, including the output of other
4307 optimization transforms, such as the MAP transform above.)
4308 * Credit information has been moved from source file headers into CREDITS.
4309 * Source file headers have been made more standard.
4310 * The CASE macro now compiles without complaining even when it has
4313 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4315 * (Note that the way that the PCL macroexpansions were rewritten
4316 to accommodate the change in DEFGENERIC below breaks binary
4317 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4318 not run under sbcl-0.6.2. Once we get out of alpha releases,
4319 i.e. hit release 1.0.0, we'll probably try to maintain binary
4320 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4321 and sbcl-1.4.4. Until then, however, it might be fairly common
4322 for maintenance releases to break binary compatibility.)
4323 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4325 * The reporting of unhandled signals has been changed to print some
4326 explanatory text as well as the report form. (Previously only
4327 the report form was printed.)
4328 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4329 it defines, so that the compiler no longer issues undefined function
4330 warnings for compiled-but-not-yet-loaded generic functions.
4331 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4332 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4333 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4334 section of the ANSI spec are supported.
4335 * The "" nickname for the "KEYWORD" package has been removed.
4336 The reader still handles symbol tokens which begin with a package marker
4337 as keywords, but it doesn't expose its mechanism for doing so in the
4338 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4339 * The system now issues STYLE-WARNINGs for contradictory TYPE
4340 proclamations. (Warnings for contradictory FTYPE proclamations would
4341 be nice too, but those can't be done usefully unless the type system
4342 is made smarter about FUNCTION types.)
4343 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4344 systematized, so that "*target-*.lisp is supposed to exist only on the
4345 target and imply that there's a related file which exists on the
4346 host, and *host-*.lisp is supposed to exist only on the host and imply
4347 that there's a related file which exists on the target. This involves a
4348 lot of renaming. Hopefully the acute confusion caused by the renaming
4349 will be justified by the reduction in chronic confusion..
4350 ** runtime-type.lisp -> early-target-type.lisp
4351 ** target-type.lisp -> late-target-type.lisp
4352 ** early-host-format.lisp -> early-format.lisp
4353 ** late-host-format.lisp -> late-format.lisp
4354 ** host-error.lisp -> misc-error.lisp
4355 ** early-error.lisp -> early-target-error.lisp
4356 ** late-error.lisp -> late-target-error.lisp
4357 ** host-defboot.lisp -> early-defboot.lisp
4358 ** code/misc.lisp -> code/target-misc.lisp
4359 ** code/host-misc.lisp -> code/misc.lisp
4360 ** code/numbers.lisp -> code/target-numbers.lisp
4361 ** code/early-numbers.lisp -> numbers.lisp
4362 ** early-host-type.lisp -> early-type.lisp
4363 ** late-host-type.lisp -> late-type.lisp
4364 ** host-typep.lisp -> typep.lisp
4365 ** load.lisp -> target-load.lisp
4366 ** host-load.lisp -> load.lisp
4367 ** host-disassem.lisp -> disassem.lisp
4368 ** host-insts.lisp -> insts.lisp
4369 ** byte-comp.lisp -> target-byte-comp.lisp
4370 ** host-byte-comp.lisp -> byte-comp.lisp
4371 ** host-signal.lisp -> signal.lisp
4372 ** host-defstruct.lisp -> defstruct.lisp
4373 ** late-target-type.lisp -> deftypes-for-target.lisp
4374 Furthermore, several other previously target-only files foo.lisp (e.g.
4375 hash-table.lisp and random.lisp) have been split into a target-and-host
4376 foo.lisp file and a target-only target-foo.lisp file, with their key type
4377 definitions in the target-and-host part, so that the cross-compiler will
4378 know more about target types.
4379 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4380 gone away. In their place are various *BACKEND-FOO* variables
4381 corresponding to the slots of the old structure.
4382 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4383 into the target SBCL has been fixed.
4384 * The chill.lisp system for loading cold code into a running SBCL
4386 * Support for the CMU CL "scavenger hook" extension has been removed.
4387 (It was undocumented and unused in the CMU CL sources that SBCL was
4388 derived from, and stale in sbcl-0.6.1.)
4389 * Various errors in the cross-compiler type system were detected
4390 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4391 (enabling various consistency checks). Many of them were fixed,
4392 but some hard problems remain, so the compiler is back to
4393 running without *TYPE-SYSTEM-INITIALIZED* for now.
4394 * As part of the cross-compiler type system cleanup, I implemented
4395 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4396 * I have started adding UNCROSS calls throughout the type system
4397 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4398 on cross-compiler input files will be able to go away, and various
4400 * CONSTANTP now returns true for quoted forms (as explicitly required
4403 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4405 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4406 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4407 (MAKE-LIST -1). In the longer term, it ought to become true
4408 that declarations are assertions even at SAFETY 1. For now, it's not
4409 quite true even at SAFETY 3, but it's at least more nearly true..
4410 (Note that this change seems to increases the size of the system by
4411 O(5%) and to decrease the speed of the compiler by 20% or more.)
4412 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4413 for the problem of printing dozens of lines of distracting information
4414 about low-level system machinery as part of the top stack frame
4415 on entry to the debugger when an undefined function was called.
4416 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4417 is not reset to COMMON-LISP-USER.
4418 * Compilation of stuff related to dyncount.lisp has been made conditional
4419 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4420 smaller. The various dyncount-related symbols have been moved into
4421 a new "SB-DYNCOUNT" package.
4422 * tty-inspect.lisp has been renamed to inspect.lisp.
4423 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4424 feature has gone away. (When we eventually port to other flavors of
4425 libc and/or Unix, we'll try to make the differences between flavors
4426 invisible at the user level.)
4427 * Various other *FEATURES* tags, and/or their associated conditionals,
4428 have been removed if obsolescent, or given better documentation, or
4429 sometimes given more-mnemonic names.
4431 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4433 * tidied up "make.sh" script
4434 * tidied up system directory structure
4435 * better "clean.sh" behavior
4436 * added doc/FOR-CMUCL-DEVELOPERS
4437 * many many small tweaks to output format, e.g. removing possibly-confusing
4438 trailing #\. character in DESCRIBE-INSTANCE
4439 * (EQUALP #\A 'A) no longer signals an error.
4440 * new hashing code, including EQUALP hashing
4441 * tidied up Lisp initialization and toplevel
4442 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4443 * command line argument processing
4444 * added POSIX-GETENV function to deal with Unix-ish environment variables
4445 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4446 terminating SBCL on EOF
4447 * non-verbose GC by default
4448 * There is no more "sbcl" shell script; the sbcl file is now the C
4449 runtime executable (just like CMU CL).
4450 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4452 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4453 symbol and package information as Lisp native symbol and package objects
4454 instead of strings naming symbols and strings naming packages. This way,
4455 whenever packages are renamed (as in warm init), debug information is
4456 transformed along with everything else.
4457 * tweaked the optimization policy declarations which control the building
4458 of SBCL itself. Now, among other things, the system no longer saves
4459 source location debugging information. (This helps two problems at once
4460 by reducing SBCL size and by keeping SBCL from trying to look for its
4461 sources -- which may not exist -- when reporting errors.)
4462 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4463 debugging and testing purposes
4464 * cleaned up printing, making the printer call PRINT-OBJECT for
4465 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4466 methods, giving nearly-ANSI behavior
4467 * converted almost all special variables to use *FOO* naming convention
4468 * deleted PARSE-TIME functionality, since it can be done portably
4469 * moved some files out of cold init into warm init
4470 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4472 * regularized formatting of source files
4473 * added an install.sh script
4474 * fixed ridiculous memory usage of cross-compiler by making
4475 compiler/alloc.lisp not try to do pooling unless it can hook
4476 itself into the GC of the cross-compilation host. Now the system
4477 builds nicely on my old laptop.
4478 * added :SB-ALLOC in target-features.lisp-expr
4479 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4480 was not implemented)
4481 * re-did condition handling and note reporting in the compiler. Notes
4482 are no longer handled by signalling conditions. Style warnings
4483 and warnings are handled more correctly and reported in such a way
4484 that it's easy to find one or the other in your output (so that you
4485 can e.g. figure out which of many problems caused COMPILE-FILE to
4487 * changed the severity of several compiler warnings from full WARNING
4488 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4489 compiler note reporting so that it doesn't use the condition system
4490 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4491 COMPILE-FILE command)
4492 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4493 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4494 no longer supported, since their implementation was deeply intertwingled
4495 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4496 * removed bogus "support" for compiler macros named (SETF FOO), and
4497 removed the compiler macro for SETF INFO (but only after making a fool
4498 of myself on the cmucl-imp mailing list by posting a bogus patch for
4499 DEFINE-COMPILER-MACRO..)
4500 * Compiled files containing forms which have side effects on the Lisp
4501 reader (such as DEFPACKAGE forms) are now handled more correctly.
4502 (Compiler queuing of top level lambdas has been suppressed by setting
4503 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4504 * deleted various currently-unused source files, e.g. gengc.lisp. They
4505 may be added back at some point e.g. when porting to other architectures,
4506 but until they are it's distracting to distribute them and to try to
4508 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4509 there were so many of them they're just distractions, and UNCROSS is
4510 known to be able to handle the current sources
4511 * moved PROFILE functionality into TRACE, so that it will be clear
4512 how the wrapping and unwrapping of functions when you profile them
4513 interacts with the wrapping and unwrapping of functions when you
4514 trace them. (Actually, the functionality isn't there yet, but at least
4515 the interface specification is there. Hopefully, the functionality will
4516 arrive with some future maintenance release.)
4517 * removed host-oops.lisp
4518 * changed signature of QUIT function to allow UNIX-CODE argument
4519 * fixed READ-SEQUENCE bug
4520 * tweaked verbose GC output so that it looks more like the progress
4521 output that ANSI specifies for functions like LOAD
4522 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4523 * added <http://sbcl.sourceforge.com> to the banner information printed by